diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml index d8549142f7..9403a1e044 100644 --- a/.github/workflows/release.yml +++ b/.github/workflows/release.yml @@ -42,3 +42,4 @@ jobs: AZURE_SIGNING_ENDPOINT: ${{ secrets.AZURE_SIGNING_ENDPOINT }} AZURE_SIGNING_ACCOUNT: ${{ secrets.AZURE_SIGNING_ACCOUNT }} AZURE_CERT_PROFILE: ${{ secrets.AZURE_CERT_PROFILE }} + LINUX_SIGNING_KEY: ${{ secrets.LINUX_SIGNING_KEY }} diff --git a/ATTRIBUTION.md b/ATTRIBUTION.md index a8541c1942..38c28a80ec 100644 --- a/ATTRIBUTION.md +++ b/ATTRIBUTION.md @@ -4,10 +4,11 @@ Every asset PdfCraft includes, bundles or uses to build its published material, with its author, source and licence. The policy is in [AGENTS.md](AGENTS.md) §1. The machine-readable list, with the SHA-256 of every asset, is [ATTRIBUTION.toml](ATTRIBUTION.toml). Licence texts are kept beside the assets and summarised in [NOTICE](NOTICE). -## In this repository (255) +## In this repository (258) | Asset | Title | Author | Licence | Source | Used for | |---|---|---|---|---|---| +| `docs/images/pdfcraft-ocr-verify.png` | PdfCraft screenshot: OCR Verify screen | PdfCraft contributors (generated by `cargo xtask screenshots`) | MIT OR Apache-2.0 | this repository | README: OCR review. Shows only PdfCraft UI, its Lucide icons and a synthetic scanned page built from the embedded fonts, with a real recognition (the OCR models) loaded into the Verify screen | | `assets/icons/accessibility.svg` | Lucide icon "accessibility" | Lucide Contributors (portions Cole Bemis, Feather) | ISC AND MIT | https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/accessibility.svg) | UI icon, embedded via crates/ui-egui/src/icon_data.rs | | `assets/icons/align-left.svg` | Lucide icon "align-left" | Lucide Contributors (portions Cole Bemis, Feather) | ISC AND MIT | https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/align-left.svg) | UI icon, embedded via crates/ui-egui/src/icon_data.rs | | `assets/icons/align-center.svg` | Lucide icon "align-center" | Lucide Contributors (portions Cole Bemis, Feather) | ISC AND MIT | https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/align-center.svg) | UI icon, embedded via crates/ui-egui/src/icon_data.rs | @@ -203,6 +204,7 @@ Every asset PdfCraft includes, bundles or uses to build its published material, | `vendor/hayro-interpret/assets/FoxitSymbol.pfb` | FoxitSymbol (standard-14 substitute font) | Foxit Software Inc. / PDFium Authors | BSD-3-Clause | https://pdfium.googlesource.com/pdfium (via hayro-interpret 0.7.0 assets/) | Renders non-embedded standard-14 fonts (vendored hayro-interpret) | | `vendor/hayro-interpret/assets/CGATS001Compat-v2-micro.icc` | CGATS001Compat v2 micro ICC profile | Clinton Ingram (saucecontrol/Compact-ICC-Profiles) | CC0-1.0 | https://github.com/saucecontrol/Compact-ICC-Profiles | Default CMYK colour conversion (vendored hayro-interpret) | | `crates/fonts/src/std14_metrics.rs` | Standard-14 font advance widths and encodings (as Rust tables) | Adobe (Core14 AFM metrics, PDF specification encodings); tables generated by the PdfCraft contributors | MIT OR Apache-2.0 | ISO 32000-2 Annex D and the Adobe Core14 AFM metrics. Generated by joining the published WinAnsi, Symbol and ZapfDingbats encodings (code to glyph name) with the per-glyph advance widths, then spot-checked against the specification by hand in the tests of crates/fonts/src/std14.rs. Data only: the code that reads these tables lives in std14.rs, so that ordinary edits do not churn this file's hash. | Advance widths for the fourteen standard fonts, which a PDF may name without embedding. Non-visual data, not a typeface: the item AGENTS.md 1.1 permits, held first-party so generated appearance streams and the renderer agree. | +| `crates/prc/src/acof/table.rs` | PRC Double prefix-code table (as a Rust table) | Kristof Ralovich (prc-rs); the code table itself is defined by the PRC format (ISO 14739-1, working draft clause 11.17.2) | MIT | Rows taken from prc-rs src/double.rs at commit 287b671bb5b31b05efb5ec1b98f7c359b9e4dd7d (https://github.com/ralovich/prc-rs), mechanically reformatted into (code length, code, entry) tuples; checked to be a complete prefix code by the tests in crates/prc/src/acof.rs. Data only: the code that reads the table lives in acof.rs. The MIT notice is kept in the file header. | Decodes every PRC Double (coordinates, transforms, colours) in 3D annotations' PRC streams. Non-visual data. | | `vendor/hayro-cmap/assets/cmaps.brotli` | Adobe CMap resources (predefined CJK CMaps) | Adobe | BSD-3-Clause | https://github.com/adobe-type-tools/cmap-resources (via hayro-cmap 0.1.0 assets/) | Character-code mapping tables required to read CJK PDFs. Non-visual data, allowed by AGENTS.md §1.1 | | `docs/brand/artcraft-logo-white.png` | ArtCraft brand: artcraft-logo-white | ArtCraft Team | LicenseRef-Storyteller-Trademark | ArtCraft Team brand assets (provided by the project owner, 2026-10-01; identical copies in storytold/photocraft) | ArtCraft branding next to the community links (About dialog, Help menu, home screen) and in docs. ArtCraft Team trademark, not open source; may be used only unmodified within PdfCraft; forks and modified versions must remove it. | | `docs/brand/artcraft-logo-white.svg` | ArtCraft brand: artcraft-logo-white | ArtCraft Team | LicenseRef-Storyteller-Trademark | ArtCraft Team brand assets (provided by the project owner, 2026-10-01; identical copies in storytold/photocraft) | ArtCraft branding next to the community links (About dialog, Help menu, home screen) and in docs. ArtCraft Team trademark, not open source; may be used only unmodified within PdfCraft; forks and modified versions must remove it. | @@ -262,6 +264,7 @@ Every asset PdfCraft includes, bundles or uses to build its published material, | `assets/icons/superscript.svg` | Lucide icon "superscript" | Lucide Contributors (portions Cole Bemis, Feather) | ISC AND MIT | https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/superscript.svg) | UI icon, embedded via crates/ui-egui/src/icon_data.rs | | `assets/icons/subscript.svg` | Lucide icon "subscript" | Lucide Contributors (portions Cole Bemis, Feather) | ISC AND MIT | https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/subscript.svg) | UI icon, embedded via crates/ui-egui/src/icon_data.rs | | `assets/icons/pilcrow.svg` | Lucide icon "pilcrow" | Lucide Contributors (portions Cole Bemis, Feather) | ISC AND MIT | https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/pilcrow.svg) | UI icon, embedded via crates/ui-egui/src/icon_data.rs | +| `assets/icons/a-large-small.svg` | Lucide icon "a-large-small" | Lucide Contributors (portions Cole Bemis, Feather) | ISC AND MIT | https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/a-large-small.svg) | UI icon, embedded via crates/ui-egui/src/widgets/icon_data.rs | | `assets/icons/move-horizontal.svg` | Lucide icon "move-horizontal" | Lucide Contributors (portions Cole Bemis, Feather) | ISC AND MIT | https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/move-horizontal.svg) | UI icon, embedded via crates/ui-egui/src/icon_data.rs | ## Compiled in through dependencies (6) diff --git a/ATTRIBUTION.toml b/ATTRIBUTION.toml index 783ecc6c9f..99073babb1 100644 --- a/ATTRIBUTION.toml +++ b/ATTRIBUTION.toml @@ -10,6 +10,17 @@ # licence: SPDX expression from the allowlist in AGENTS.md §1.2. adobe_data = true marks the closed list of # Adobe-authored non-visual technical data allowed by AGENTS.md §1.1. +[[asset]] +path = "docs/images/pdfcraft-ocr-verify.png" +title = "PdfCraft screenshot: OCR Verify screen" +author = "PdfCraft contributors (generated by `cargo xtask screenshots`)" +source = "this repository" +licence = "MIT OR Apache-2.0" +licence_file = "LICENSE-MIT" +kind = "image" +usage = "README: OCR review. Shows only PdfCraft UI, its Lucide icons and a synthetic scanned page built from the embedded fonts, with a real recognition (the OCR models) loaded into the Verify screen" +sha256 = "ea7485b151e27ad3fd2a2dac079436ab11745d42032201a029bb20f819c6af8f" + [[asset]] path = "assets/icons/accessibility.svg" title = "Lucide icon \"accessibility\"" @@ -2156,6 +2167,17 @@ adobe_data = true usage = "Advance widths for the fourteen standard fonts, which a PDF may name without embedding. Non-visual data, not a typeface: the item AGENTS.md 1.1 permits, held first-party so generated appearance streams and the renderer agree." sha256 = "f225e0b044875b8fa6375b6b5ad935f0d99fcc1ad03ecf9f0baae2cedbfca4d2" +[[asset]] +path = "crates/prc/src/acof/table.rs" +title = "PRC Double prefix-code table (as a Rust table)" +author = "Kristof Ralovich (prc-rs); the code table itself is defined by the PRC format (ISO 14739-1, working draft clause 11.17.2)" +source = "Rows taken from prc-rs src/double.rs at commit 287b671bb5b31b05efb5ec1b98f7c359b9e4dd7d (https://github.com/ralovich/prc-rs), mechanically reformatted into (code length, code, entry) tuples; checked to be a complete prefix code by the tests in crates/prc/src/acof.rs. Data only: the code that reads the table lives in acof.rs. The MIT notice is kept in the file header." +licence = "MIT" +licence_file = "licenses/prc-rs-LICENSE.txt" +kind = "data" +usage = "Decodes every PRC Double (coordinates, transforms, colours) in 3D annotations' PRC streams. Non-visual data." +sha256 = "8e705c5cde6a3ef778e8aca69759e38845b228b3f7f12949e587b19855d2114c" + [[bundled]] crate = "epaint_default_fonts" version = "0.37.0" @@ -3189,6 +3211,17 @@ kind = "icon" usage = "UI icon, embedded via crates/ui-egui/src/icon_data.rs" sha256 = "d344d9f9222f6c1b698c0c82a753a2212d9bd7785f06c00cd6e4910fe2fee8f8" +[[asset]] +path = "assets/icons/a-large-small.svg" +title = "Lucide icon \"a-large-small\"" +author = "Lucide Contributors (portions Cole Bemis, Feather)" +source = "https://github.com/lucide-icons/lucide (npm lucide-static 1.49.0, icons/a-large-small.svg)" +licence = "ISC AND MIT" +licence_file = "assets/icons/LICENSE-lucide.txt" +kind = "icon" +usage = "UI icon, embedded via crates/ui-egui/src/widgets/icon_data.rs" +sha256 = "d4985fc71174d108e322c3b25cb6abe2837a115fe3f239a1918645db436743ec" + [[asset]] path = "assets/icons/move-horizontal.svg" title = "Lucide icon \"move-horizontal\"" diff --git a/Cargo.lock b/Cargo.lock index 4df93ed3fe..e1f271ed12 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -143,6 +143,15 @@ dependencies = [ "zerocopy", ] +[[package]] +name = "aho-corasick" +version = "1.1.5" +source = "registry+https://github.com/rust-lang/crates.io-index" +checksum = "c982642fa9e8606056828ee9a8505737230110bb1099153c79efe865c59d12ba" +dependencies = [ + "memchr", +] + [[package]] name = "aligned-vec" version = "0.6.4" @@ -1303,6 +1312,15 @@ dependencies = [ "zip", ] +[[package]] +name = "craft-web-host" +version = "0.1.0" +source = "git+https://github.com/storytold/craft-libs?rev=2e2312ddaf21bc3d85cd3395d0955c7d239306f6#2e2312ddaf21bc3d85cd3395d0955c7d239306f6" +dependencies = [ + "wasm-bindgen", + "web-sys", +] + [[package]] name = "craft-winprint-gdi" version = "0.1.0" @@ -4277,6 +4295,8 @@ dependencies = [ "pdfcraft-model", "pdfcraft-organize", "pdfcraft-platform", + "pdfcraft-prc", + "pdfcraft-prc-render", "pdfcraft-render", "pdfcraft-xfa", "png 0.18.1", @@ -4322,6 +4342,7 @@ version = "0.18.1" dependencies = [ "chrono", "hayro-syntax", + "js-sys", "pdfcraft-crypt", "pdfcraft-filters", "pdfcraft-oracle", @@ -4329,6 +4350,7 @@ dependencies = [ "proptest", "serde_json", "thiserror 2.0.21", + "wasm-bindgen", ] [[package]] @@ -4505,6 +4527,7 @@ version = "0.18.1" dependencies = [ "ocrs", "pdfcraft-fonts", + "png 0.18.1", "rten", "tempfile", "thiserror 2.0.21", @@ -4551,6 +4574,24 @@ dependencies = [ "zbus", ] +[[package]] +name = "pdfcraft-prc" +version = "0.18.1" +dependencies = [ + "flate2", + "proptest", + "thiserror 2.0.21", +] + +[[package]] +name = "pdfcraft-prc-render" +version = "0.18.1" +dependencies = [ + "flate2", + "pdfcraft-prc", + "thiserror 2.0.21", +] + [[package]] name = "pdfcraft-preflight" version = "0.18.1" @@ -4578,6 +4619,7 @@ dependencies = [ name = "pdfcraft-redact" version = "0.18.1" dependencies = [ + "aho-corasick", "pdfcraft-annot", "pdfcraft-content", "pdfcraft-cos", @@ -4588,6 +4630,7 @@ dependencies = [ "pdfcraft-model", "pdfcraft-testkit", "pdfcraft-xfa", + "sha2 0.11.1", "thiserror 2.0.21", ] @@ -4682,6 +4725,7 @@ dependencies = [ "pdfcraft-engine", "pdfcraft-fonts", "pdfcraft-model", + "pdfcraft-ocr", "pdfcraft-platform", "pdfcraft-render", "pdfcraft-xfa", @@ -4709,6 +4753,7 @@ dependencies = [ name = "pdfcraft-web" version = "0.18.1" dependencies = [ + "craft-web-host", "eframe", "egui", "js-sys", diff --git a/Cargo.toml b/Cargo.toml index 4b98f47c28..bf6a4e3146 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -63,6 +63,8 @@ pdfcraft-engine = { path = "crates/engine", version = "0.18.1" } pdfcraft-platform = { path = "crates/platform", version = "0.18.1" } pdfcraft-ui-egui = { path = "crates/ui-egui", version = "0.18.1" } pdfcraft-automation = { path = "crates/automation", version = "0.18.1", default-features = false } +pdfcraft-prc = { path = "crates/prc", version = "0.18.1" } +pdfcraft-prc-render = { path = "crates/prc-render", version = "0.18.1" } # pdf hayro = "0.7.1" diff --git a/LICENSES.toml b/LICENSES.toml index d9ceb3180e..0617ca5d57 100644 --- a/LICENSES.toml +++ b/LICENSES.toml @@ -88,6 +88,13 @@ destination = "licenses/egui-emoji-icon-font-mit-license.txt" sha256 = "b9d2c1d909aa149996fd4c91dcb92b2362a04431640c1d200959da94caf8cde1" origin = "https://raw.githubusercontent.com/emilk/egui/0.36.2/crates/epaint_default_fonts/fonts/emoji-icon-font-mit-license.txt" +[[document]] +source = "licenses/prc-rs-LICENSE.txt" +licences = ["MIT"] +destination = "licenses/prc-rs-LICENSE.txt" +sha256 = "ddb846513d92153d8fa16b18ed3e7c86953a582f0d2f49f02b29f1bd569dac58" +origin = "https://raw.githubusercontent.com/ralovich/prc-rs/287b671bb5b31b05efb5ec1b98f7c359b9e4dd7d/LICENSE" + [[document]] source = "vendor/hayro-interpret/LICENSE-APACHE" licences = ["Apache-2.0"] diff --git a/README.md b/README.md index 3c1b67936a..f488d160cc 100644 --- a/README.md +++ b/README.md @@ -94,7 +94,7 @@ Real-world typography: world scripts, vertical Japanese, colour emoji, gradients ### Fearless -Every save appends your changes and leaves the original bytes untouched. Writes are atomic, undo runs deep, and nothing is lost if you close by mistake. +Ordinary saves append your changes and leave the original bytes untouched. Edits to damaged objects whose xref and header generations disagree use a full rewrite, keeping both accepted references connected to the same repaired object. Writes are atomic, undo runs deep, and nothing is lost if you close by mistake. @@ -253,6 +253,18 @@ Documents restricted by their author show a clear notice, and PdfCraft honours t +## Scan pages, then read the words back before trusting them + +**Scan & OCR ▸ Recognize text** turns a scan into searchable text — the current page, a range, or a folder of files written as searchable copies — with optional cleanup before reading (auto-rotate, deskew, denoise, binarize; word boxes map back to the original scan) and a second engine: an installed Tesseract joins the built-in one, merged per your strategy. A searchable result applies as one undoable step; the **Editable text** output opens the **OCR Verify screen** instead, where each word carries the engine's confidence as a colour, suspects are one keypress away, and corrections, a user dictionary and Re-OCR of a region or a page happen before Accept writes the new document. **Scan & OCR ▸ Correct recognized text** reviews any recognition there before it is applied. + +

+ The OCR Verify screen: a scanned note with each recognized word boxed in its confidence colour, the info strip reporting a 95% average confidence and one low-confidence word, and the confidence legend with its thresholds below +
+ The OCR Verify screen: words coloured by confidence — the legend spells the thresholds out — reviewed and corrected before anything is applied. +

+ +--- + ## Every tool, one keystroke away Press ⌘K to search every tool and command, or browse the **All tools** catalogue. Tools that are still in development are marked with the milestone that will ship them. diff --git a/ROADMAP.md b/ROADMAP.md index a489a8bbc9..db0e495b38 100644 --- a/ROADMAP.md +++ b/ROADMAP.md @@ -22,10 +22,17 @@ > **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (typed small Edit font sizes and Fill & Sign text/date colour, #2444/#2496; estimates unchanged) > **Last reviewed:** 2026-10-25 · **Last updated:** 2026-10-25 · **Change:** minor (M7, #2750: Single-style Keep source fonts edits retain source glyphs and batch only missing glyphs in fallback fonts. Per-rewrite character eligibility and widths are cached; source and fallback edge spaces stay extractable. Estimates unchanged.) +> **Last reviewed:** 2026-10-20 · **Last updated:** 2026-10-20 · **Change:** patch (M1, #1143: repaired aliases survive preflight, snapshots and normalized saves; retained materialized revisions report generation disagreements; estimates unchanged) + +> **Last reviewed:** 2026-10-18 · **Last updated:** 2026-10-18 · **Change:** minor (M1, #1143: explicit live xref/header generation repair policy and bounded diagnostics; estimates unchanged) + +> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M1: refresh object-generation reference identity against current upstream; estimates unchanged) **Stage: alpha** · next: beta, ~25% (ready for real work 50% → 75%) and ~400–700 h away -> **Last reviewed:** 2026-10-29 | **Last updated:** 2026-10-29 | **Change:** minor (M7 #2517: a narrowed then widened paragraph keeps the typed spaces the wrap consumed, after save and reopen; estimates unchanged) +> **Last reviewed:** 2026-10-29 | **Last updated:** 2026-10-29 | **Change:** minor (M2: scans keep their strokes at screen size and are smoothed when zoomed; M7 #2517: a narrowed then widened paragraph keeps the typed spaces the wrap consumed, after save and reopen; estimates unchanged) +> **Last reviewed:** 2026-10-28 | **Last updated:** 2026-10-28 | **Change:** minor (M10, #4119 partial: Enhance scanned file cleans scanned pages in place: background removal, edge shadow removal, despeckle; estimates unchanged) > **Last reviewed:** 2026-10-27 | **Last updated:** 2026-10-27 | **Change:** minor (M10: OCR model downloads use verified ONNX pairs with legacy discovery fallback; estimates unchanged) +> **Last reviewed:** 2026-10-25 | **Last updated:** 2026-10-25 | **Change:** minor (web: same-document host mode on the canvas, `data-host="page"`; estimates unchanged) > **Last reviewed:** 2026-10-25 | **Last updated:** 2026-10-25 | **Change:** minor (M14 localization current state: Arabic logical catalog, explicit RTL label direction and mirrored stock chrome; native/browser RTL workflows unqualified. Native release Hans face via LXGW WenKai Lite; estimates unchanged) > **Last reviewed:** 2026-10-23 · **Last updated:** 2026-10-23 · **Change:** minor (M7, #2636 partial: pasting added text boxes and images in the engine and through `content_copy`; estimates unchanged) > **Last reviewed:** 2026-10-23 · **Last updated:** 2026-10-23 · **Change:** minor (M11: image optimization accounts for page-local UserUnit; estimates unchanged) @@ -224,6 +231,7 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md ## Progress log +- **2026-10-28 (M10, #4119 partial):** Scan & OCR ▸ Enhance scanned file cleans scanned pages in place: background removal (level 0–100, as the camera clean-up), edge shadow removal (book gutters, copier borders) and despeckle. Each scan image is decoded, cleaned and written back into its own object at the same size and colour space (JPEG stays JPEG, others lossless, 1-bit stays 1-bit), so the OCR text layer, comments and links stay; one undo step, shared images once, unsupported images (CMYK, indexed, JPEG 2000, JBIG2, CCITT, broken data) reported. `page_enhance_scan` drives it headlessly. Not yet an option of Recognize text. Estimates unchanged. - **2026-11-03 (M11, shared JBIG2 dictionary):** With Monochrome Images set to JBIG2, Optimize has up to 16 JBIG2 images share one symbol dictionary (`/JBIG2Globals`), the letters found on more than one page stored once, when that is smaller (`pdfcraft_filters::encode_jbig2_shared`). Eight pages of 10 pt text rendered at 300 dpi take 41% less in Helvetica and 42% in Times than coded alone, 10% in Courier; groups of 16 save 45%, of 32 only 47%. Only images redaction can store again without the dictionary share one (image masks and DeviceGray named directly, at most 64 Mi pixels), and redacting one stores the rest of its group so, which the group bounds to 15. - **2026-11-03 (M5, #5850 review):** Resolve indirect IT when recognizing typewriter contents and Fill & Sign marks. Comment listing and property lookup retain their existing direct-Subtype behavior. A real double-click Edit composer preserves the typewriter’s text as written; ordinary comment edits keep trimming. Regressions cover direct Subtype with indirect intent, save/reopen, raw posted contents and same-index comments on different pages. The existing live-movement test now commits its setup with an outside click rather than Enter. Estimates unchanged. @@ -232,8 +240,12 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md - **2026-11-03 (M7, #5835):** In image crop mode, a right-click on the image, its crop box or its handles keeps the crop and opens no menu, so the menu no longer switches as the pointer moves. A right-click elsewhere clears the selection and opens the page menu. UI tests cover all four page rotations, overlapping images, a 45° image, and a check that the Done/Cancel/Remove Crop pills at a page edge still keep the crop. +- **2026-11-02 (M12, #1297):** `model_add` embeds a U3D or PRC file, recognised by its header, as a 3D annotation with a chosen activation and a poster (an image or PDF page fitted in the box, or a labelled frame); an older file's catalog `/Version` is raised to what the model needs. Views, a rendered poster and a UI command are later steps. Estimates unchanged. + - **2026-11-02 (M10, #5743):** macOS print jobs name their title and owner (`com.apple.print.JobInfo.PMJobName`, `PMJobOwner`), as the system print dialog does, because a Kyocera filter crashes on a job without them. Checked by a test and through `doc_print` against a stand-in `lp`; not yet on a physical printer. +- **2026-11-02 (M14, #5737, #5460):** Preferences can turn off the update check at start. It stays on by default, and Help ▸ Check for updates still asks. + - **2026-11-01 (M7, #5657):** Header/footer, watermark and background additions and replacements stage the complete call before committing. Synthetic later-page failures preserve exact incremental bytes, object count, dirty state and allocation cursor; success controls retain other marks and shared/inherited resources. Estimates unchanged. - **2026-11-01 (M14, #5572):** Preferences ▸ Documents and view sets how many recent files to keep, from 0 to 50 with 12 by default. At 0 Home has no Recent section and File ▸ Open Recent is disabled. The list is cut when Preferences closes, so the number can pass through 0 on its way to another. @@ -260,9 +272,11 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md - **2026-10-26 (M2, PR #4172):** Reconciled current main, preserving its Japanese serif/weight selection and both branches' progress records. Installed CJK viewer and preview fallback remains display-only; estimates unchanged. - **2026-10-29 (M7, #2517):** Widening a paragraph rejoins a wrap with the number of spaces that wrap consumed, including a typed run and none for a Chinese or Japanese word split. The count survives an incremental save and a full rewrite, and a width-only `text_edit` uses it. Estimates unchanged. +- **2026-10-29 (M2):** Scanned pages look like the paper again. An image that doesn't ask for /Interpolate was shrunk by nearest-neighbour (since #635), which kept one sample in N per axis, so the text of a 300 dpi scan at 100% lost most of its strokes; it is now area-averaged to whole device pixels (on a 13-page 300 dpi scan every sample at 96–192 dpi is within 1 of the exact area average; it was up to 255 off). Enlarged, an image of page-scan size is smoothed bilinearly, so a zoomed scan has no staircases; icons, QR codes and pixel art keep hard pixels. Estimates unchanged. - **2026-10-26 (M11, JBIG2 symbols):** Lossless JBIG2 now stores each distinct mark of a black-and-white image once, as a symbol, and places its exact copies, when that is smaller than the generic region: on six pages of rendered 300 dpi text 28% of the generic region's size and 17% of Group 4's, and the same bits back through two independent decoders (hayro-jbig2 and Poppler). Scans, whose letters are seldom identical, gain little until refinement coding; a dictionary shared between pages (`/JBIG2Globals`) is next. Estimates unchanged. - **2026-10-25 (M5, comment replies):** One reply can be deleted from the Comments panel without its thread. On the selected card each reply has its own "…" menu (Edit, Copy text, Delete); Delete takes that reply with its pop-up and the replies to it, as `comment_delete` already did headlessly, and the comment and its other replies stay. The card stays selected, and the panel's editor stays open, on what they were on though the delete moved it in `/Annots`. Undo and Redo close the editor, since either can move what it was open on. A locked reply can't be deleted. Estimates unchanged. - **2026-10-25 (M7, #2750):** Single-style Keep source fonts edits retain source glyphs and batch only missing glyphs in fallback fonts. Per-rewrite character eligibility and widths are cached; source and fallback edge spaces stay extractable. Estimates unchanged. +- **2026-10-25 (web, same-document host mode):** A page that loads the PdfCraft web module into its own document (no iframe) can opt into the host bridge with `data-host="page"` on the app's canvas — the exact value `page`; any other value keeps the normal app with downloads, unless a same-origin framed `?host=parent` selects the parent bridge (unframed or cross-origin, `?host=parent` also keeps the normal app). It is the `?host=parent` bridge with the transport swapped: the same five messages travel as non-bubbling `CustomEvent`s on the canvas (`detail` holds the parent messages' fields without `type`; the host answers with a `pdfcraft:open` event), so saves reach the host instead of downloading and dirty and start-up-failure reporting are unchanged. Unit tests cover the opt-in rule; no parity or estimate changes. - **2026-10-24 (M10, #4158 partial):** Word, HTML, RTF and PowerPoint export keep a block that continues another on its line (an italic case name, a link, a superscript) as a run in its own style inside that paragraph, with the source's word space, instead of a paragraph of its own. Style changes that wrap onto the next line still split paragraphs (the reading blocks keep style boundaries). Estimates unchanged. - **2026-10-24 (M3, window move on Wayland, #3970 follow-up):** After a move or resize by PdfCraft's own title bar or edges, the button release given back to egui goes before the pointer leaving the window, so hover no longer sticks at the old place until the pointer moves. Found by CodeRabbit in review of #3970; a unit test covers it. Estimates unchanged. - **2026-10-24 (M14, #4253):** The annotation quick bar remembers Keep tool selected across restarts, including unpinning. Old and malformed settings keep safe defaults; the opt-in UI control channel reads and changes the same preference. Estimates unchanged. @@ -291,6 +305,11 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md Estimates unchanged. - **2026-10-21 (M3, window resize and move on Wayland):** With PdfCraft's own title bar on GNOME/Wayland, a second resize by a window edge, or a second move by the empty tab strip, works straight after the first; it waited for a click inside the window, because the compositor keeps the button's release from egui. A unit test covers the release given back; checked by hand in a test build. Estimates unchanged. - **2026-10-24 (M5, move preview):** A comment being moved is drawn where it goes while it is dragged (a rectangle or oval showed only its blue frame): the comment is rendered alone once as the drag starts and laid over the page, which shows it where it was until the move is applied. Engine `Document::comment_layer`; synthetic engine and UI regressions. Estimates unchanged. +- **2026-10-24 (M5, border shape):** Text boxes and Fill & Sign's typed text take an oval or a circle for their border (Comment Properties ▸ Border shape, Fill & Sign's toolbar): the text box keeps its place and the shape goes around it, the text centred, a circle snug around the text (its box no wider than the text needs); stored as /RD and /PCBorderShape and drawn in the appearance. comment_add / comment_edit / fill_sign_add `border_shape`. One new label in every catalog. Synthetic annot, UI and automation regressions. Estimates unchanged. +- **2026-10-24 (M5, Escape in Fill & Sign):** Escape leaves a Fill & Sign tool as it leaves a comment tool: the first press closes the open type box, the next goes back to Select, so the Text tool no longer opens a box at every click with no way out by the keyboard. A synthetic UI regression covers the Text and Checkmark tools. Estimates unchanged. +- **2026-10-23 (M5, ortho):** Shift while drawing holds a line or arrow level or plumb, whichever is nearer (a CAD drawing's ortho mode), and each new side of connected lines, a polygon or a cloud from the corner before; dragging an end or corner with Shift is held the same way (it turned to 45° steps). `measure::snap::ortho`; `comment_add ortho`. Estimates unchanged. +- **2026-10-24 (M3, arrow keys on a wide page):** → / ← scroll a page wider than the window (a plan zoomed in) sideways a line at a time and turn the page at its edge, as ↓ / ↑ do in single-page view; ⌘→ / ⌘← turn it at once, and a page that fits across turns as before (#185). A synthetic viewer regression covers both. Estimates unchanged. +- **2026-10-24 (M5, part opacity):** Comment Properties sets a text box's text, border and fill, and a shape's line and fill, each at an opacity of its own (a see-through fill with an opaque line on a plan); stored as /CA, /ca (PDF 2.0) and /PCTextOpacity and drawn with separate graphics states, so other viewers show it. Make Properties Default and the settings carry it; comment_add / comment_edit `fill_opacity`, `text_opacity`. Four new labels in every catalog. Synthetic annot, UI and automation regressions. Estimates unchanged. - **2026-10-22 (M4, #3916 partial):** Document edits and Undo/Redo cancel an Organize page drag. Pointer-release regressions check exact page order and save/reopen. - **2026-10-22 (M10, #3752 paper):** CUPS jobs ask for the chosen paper in every size. The 13 sizes the Print dialog sent no paper for (A6, A0–A2, ARCH, ANSI C–E) use PWG 5101.1 names, checked against a synthetic PPD with `cupsfilter`, and `doc_print` to a printer sends the paper too. The macOS tray half (Auto Select instead of the PPD's default tray) is a craft-print change that needs a pin bump here. Estimates unchanged. @@ -312,6 +331,7 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md - **2026-10-20 (M11, #3507):** Image merging keeps trailing comments, split operands and graphics state across content streams. Synthetic tool tests check saved text, rendered regions and exact image rows. - **2026-10-17 (M8, gap 6, #2662):** A filled cover over a long line counts when it hides half of any one word, not only 40% of the line's words or box, so a black box over one number in a long line — or an unapplied redaction mark over it — no longer audits clean while the number is still extractable; a sliver under half of every word it touches and under 40% of the line is still not flagged +- **2026-10-13 (M10, OCR review, batch and two engines):** Scan & OCR gains a review step, a second engine and Unicode output. Recognition can clean the raster first (opt-in auto-rotate, deskew, denoise, binarize; word boxes map back to the scan) and read with the built-in ocrs engine, an installed tesseract program run as a fenced external process (a validated location that fails closed instead of falling back to the PATH, allow-listed languages, a private temp directory, timeout and output caps), or both: Automatic asks tesseract when the ocrs reading is weak or of unknown confidence, and a ROVER merge keeps the more confident word (an unscored ocrs word yields only to a tesseract word of 70 or more). The OCR Verify screen (Scan & OCR ▸ Correct recognized text) reviews a recognition before it is applied: confidence-coloured word boxes, corrections, a per-language user dictionary (a plain-text file in the preferences folder, so a word added from any document stays on disk after it closes), Re-OCR of a page or a region, Accept/Reject per page or whole. Pages accepted one by one each apply once; a region re-read moves the review marks of the words it keeps, and a whole-page re-read clears that page's marks (undoable in the panel). Output is searchable (an invisible layer, one undo step) or a NEW editable document of visible text. Words beyond WinAnsi are drawn in Type3 faces from craft-fonts with a ToUnicode map in the editable output (characters no bundled face has appear as '?'); the searchable layer writes them as Type3 fonts with empty character programs and a ToUnicode map, so extraction reads them back with no fonts and no substitutions; redaction removes that layer like any other text (tests in pdfcraft-redact; the end-to-end engine test needs the craft-fonts build input and is skipped without it). Batch runs skip pages that already have text unless forced, and skip whole files that have text only with the opt-in skip-files flag (off by default; Force OCR overrides both); they refuse two inputs of the same file name and existing outputs before reading anything, and report low-confidence pages at the settings' threshold. Headless: ocr_recognize (engine, strategy, preprocessing, region; the editable output goes to an `out` that must not exist yet, and `out` is refused for a searchable run), ocr_words, ocr_review_apply, ocr_dictionary, ocr_status and ocr_recognize_files. All 21 interface catalogs carry the new strings; the Lithuanian ones were added without native review. Reading with every cleaning step off passes the render through untouched, as on main (the 300 dpi normalization runs only when a cleaning step is on: enlarging a 72 dpi render 4x on its own made the models read "due: €250" as "ue: ?250"). Still missing: the deep scan clean-up options (descreen, background and edge-shadow removal), scan images in the editable output, matching the scanned typeface, and recognition quality beyond the tested English set. - **2026-10-13 (M10, #1757):** Office export keeps each dotted-leader contents entry in one paragraph, with its page number, instead of inferring table cells from its separate text objects. - **2026-10-22 (M6, calculation scripts):** The Calculate tab and a rename read simplified field notation only from a comment, where Acrobat writes it: a string that spells `BVCALC … EVCALC` is no longer taken for the notation, or rewritten by a rename. A `/` after the condition of `if`, `while`, `for` or `with` starts a regular expression, so a rename no longer rewrites a name inside one, and a quote inside one no longer hides the `getField` calls after it. Both found in review of #3550. A `//` comment ends at `\r`, U+2028 or U+2029 as well as `\n`, so in a script with `\r` line endings the code after one is no longer hidden from a rename, or taken for the notation. Estimates unchanged. - **2026-10-20 (M6, calculation order page):** Prepare a form ▸ Calculation order, a page of its own, lists the calculated fields in the order they calculate with their formulas and values, moved with Up/Down arrows (as Acrobat's Set Field Calculation Order), Alt+↑/↓ or by dragging. A card says what is wrong with the order and holds Fix order; a loop has a card of its own that opens one of its formulas. On the page, each calculated field shows its step, and arrows show what the field in focus reads and what reads it. Estimates unchanged. @@ -322,6 +342,7 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md - **2026-10-20 (M12, #3673):** Character encoding accepts Identity-H/V fonts with Adobe CJK collections and recognizes predefined non-Identity CMaps independently. Unknown CMaps and foreign registries remain refused; synthetic checker and read-only automation regressions passed after failing before the fix. Per-glyph mapping validation and embedded-font extraction remain separate. Estimates unchanged. - **2026-10-19 (M6, renames keep formulas):** Renaming a field renames it in every calculation that reads it, in the same undo step: `AFSimple_Calculate` field lists, simplified field notation (the comment and the JavaScript after it) and `getField` calls naming it in one string literal; the rest of each script is kept as written. A sum used to drop the renamed field, notation read it as 0 and a script kept a stale value. Field Properties says how many formulas changed and `form_set_props` lists them as `formulas_updated`. Names built at run time and other events' scripts are not rewritten. Estimates unchanged. - **2026-10-19 (M6, text field table):** Prepare a form ▸ Text field table drags out an area and adds rows and columns of named text fields as one undoable edit, laid out on the page as displayed, with a live preview, per-column name patterns and widths, and refusal before anything is added when a cell would be too small or a name taken; `form_add_field_table` does the same headlessly. `form.multiple-copies` is partial: copying an existing field into a grid is still missing. Estimates unchanged. +- **2026-10-17 (research):** Added a [PDF workflow research reference](docs/research/pdf-workflows/README.md): dated public-source locators, 20 workflow research domains, a pinned feature mapping and interaction evidence. Reference documentation; no feature statuses, milestone priorities or estimates changed. - **2026-10-16 (M7, #2516):** A paragraph number from `text_paragraphs` keeps naming the paragraph it listed after edits regroup the page: `text_edit`, `text_edit_layout` and `page_render`'s `hide_text_block` find it again by its text and box, and refuse one that was merged into a neighbour (the edited paragraph included) instead of rewriting the paragraph that took its number (`stale_paragraphs` lists the numbers gone). The inline editor's switch-paragraph click holds the same key. No stable identity across save and reopen yet (#2508). - **2026-10-19 (M7, #3591):** A background added to a page that already has one is painted over the old one instead of going under it and staying hidden. Behind the content, backgrounds come first, in the order added, and then behind-content watermarks, so a watermark added after a background no longer disappears. Synthetic regressions in `crates/edit` check the order. Estimates unchanged. @@ -333,9 +354,13 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md - **2026-10-20 (M3, #3541/#3620):** Comment hover follows the individual marked lines instead of the enclosing multiline rectangle, and overlapping comments follow paint order. Synthetic UI regressions cover all four text-markup kinds, page/view rotations, neighboring highlights, legacy rectangle fallback and unchanged PDF bytes. Estimates unchanged. - **2026-10-20 (M10, #3532 partial):** Vector printing retains existing widget artwork when its appearance lacks Subtype. A print-owned copy adds Form dispatch metadata. The source stream and all explicit subtypes stay unchanged. Synthetic source/output raster checks cover resources, selected/shared appearances, placement, flags and content modes. Appearance regeneration and physical printer validation remain outside this fix. Estimates unchanged. +- **2026-10-20 (M5, several comments):** Comment Properties opened on one of several selected comments, the context menu's Colour and the quick bar's colour change all of them as one undo step, each where the property applies (a font size to text boxes, a fill to shapes, a width to lines and shapes); one that stays locked keeps its look. Copied Fill & Sign ticks, crosses, dots and lines hang on the pointer as themselves rather than as empty boxes, and copied text boxes show their text (`FillMark::strokes` is shared by the appearance and the outline). Estimates unchanged. +- **2026-10-19 (M5, comments between documents):** Comments copy and paste, within and between documents: ⌘C with comments selected copies them as XFDF (each with its replies) to the app's and the system clipboard, ⌘V / Edit ▸ Paste comments / the page menu pastes them onto the current page as new comments under names of their own (`Edit::PasteComments`, `pdfcraft_xfdf::import_comments` with a target page and fresh names, so nothing is replaced and pasting twice gives two sets; `export_xfdf_comments` writes a selection with its replies). Several comments select at once (⌘-click, ⇧-click, on the page and in the panel); Delete removes them as one step and a drag moves them together. Connected lines, polygons and clouds finish from the right-click menu's Complete (Undo last point, Cancel drawing beside it), as other editors do. Tools `comments_copy` / `comments_paste`. Estimates unchanged. - **2026-10-20 (M5, arrow keys move comments):** With a comment selected, the arrow keys move it a screen pixel and ⌘/Ctrl or Shift and an arrow ten, instead of scrolling the view or turning the page; up is up on screen at any zoom and rotation, and each press is one undo step. Text markup and locked comments stay put. Estimates unchanged. +- **2026-10-20 (M1, PR #1143 review):** Repair-aware dictionary/preflight edits and snapshot restoration preserve both accepted generations of a damaged live slot. Such edits use the existing full-save fallback; classic and packed rewrites coalesce aliases so page-tree parent links retain one object identity. Explicit generation replacement remains single-generation. Materialized xref diagnostics inspect retained primary and hybrid rows before folding, retaining the 256-object cap plus summary and discarding abandoned-index notes. Synthetic red/green controls cover preflight, undo restoration, equal values, explicit narrowing, encryption packing and four save variants. Estimates unchanged. + - **2026-10-19 (M14, #3373):** The main menu and its long submenus scroll in short windows, keeping their commands reachable. Estimates unchanged. - **2026-10-19 (M3, #3387):** Save-prompt shortcuts yield to later consent and utility dialogs; a synthetic UI regression checks that an edited document stays open. Estimates unchanged. @@ -399,6 +424,8 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md - **2026-10-15 (community, M6, #2608):** In Prepare a form the selected field moves with the arrow keys: a point per press, ten with Shift, as the object nudge and dragging it do. Each press is one undo step ("Change field properties"), and the keys no longer turn the page while a field is selected. Estimates unchanged. - **2026-10-15 (M3, closable community card):** The ArtCraft community card on Home has a close button that hides it for the rest of the run; it returns on the next launch, and its links stay in the top bar's Discord button and the Help menu. A UI regression covers closing and the fresh-launch default. Estimates unchanged. - **2026-10-13 (community, M3, #1569 review):** Pixel-only Wayland wheel, wheel-tilt and continuous input use standalone moved events, so Ctrl zoom does not remain latched after the modifier is released (#991); vendored winit patch (13), with a policy test. Estimates unchanged. +- **2026-10-18 (M1, PR #1143 repair policy):** A live in-file reference resolves when its generation matches the xref row or a same-number header at that indexed offset. Neither-match and free slots remain null, including when old object bytes remain. Opening records bounded xref/header disagreements without parsing object bodies or expanding compact xref rows. Synthetic controls cover read order, clones, overlays, dictionary edits, full classic/packed saves and an independent reader of incremental output. **Correction to the 2026-10-15 note below:** its async-opening-test sentence was not evidenced by this COS change; no fresh async-opening or GUI validation is claimed. The full-reference stream-length recursion key is defensive hardening, not an independently isolated regression fix. Estimates unchanged. +- **2026-10-15 (M1, PR #1143 refresh):** COS reference lookups and full saves distinguish complete object references, so stale generations cannot expose replacement content. Snapshot restoration also keeps edits to objects with nonzero generations before incremental saves. Synthetic regressions cover caches, overlays, snapshots, stream lengths and classic/packed saves. Current upstream async-opening test corrections are included. Estimates unchanged. - **2026-10-15 (M14, #2383 candidate):** Device-limit requests now clamp unsupported maxima and minimum alignments before enabling the adapter's full 2D texture extent. This addresses the reported V3D request of 16 shader variables on an adapter supporting 15. Synthetic regressions fail before the clamp and pass afterward, along with the existing HiDPI case and Windows GPU-required application tests. Raspberry Pi startup remains unverified; milestone and overall estimates are unchanged. - **2026-10-12 (M6):** XFDF field import and data-file extraction join all text nodes in a value, preserving text after XML comments and processing instructions. Synthetic regressions cover Unicode, CDATA, empty values and PDF save/reopen. Estimates unchanged. - **2026-10-12 (M7, picture right-click menu):** The Edit text & images menu of a picture gains Cut Image, Redact Image (a redaction mark over the picture, applied from Redact a PDF), Create Link… and Recognize Text… (for that page), in all 21 catalogs. The menu was drawn only while the pointer was over the picture, so its items outside the picture (most of a logo's menu, and Delete below a page) could not be clicked; an open menu now stays drawn. Still missing from Acrobat's menu: Arrange (bring to front, send backward), Remove crop, Draw freehand and Edit image using… (opening an external editor). @@ -777,6 +804,7 @@ Newest first: the date, what moved, and the new overall estimate. | 2026-10-20 | minor | M6, #3502: form detection preserves comments across content streams; estimates unchanged. | | 2026-10-20 | minor | M10, #3532 partial: vector printing retains existing appearances with absent Subtype; estimates unchanged. | | 2026-10-20 | minor | M11 #3507: image merging preserves comments and state across content streams. Estimates unchanged. | +| 2026-10-18 | minor | M1, #1143: a live reference may match the xref generation or a same-number header at its indexed offset; neither-match and free slots remain null. Bounded repair diagnostics and synthetic cache, free-slot, edit and classic/packed save regressions; estimates unchanged. | | 2026-10-18 | minor | M5, PR #1971: nested source `/Page` and `/Pages` dictionaries export as null without traversing page content or the page tree. Synthetic saved/reopened three-page regression retains unrelated metadata and aliases. FDF import semantics, copied-graph limits and estimates unchanged. | | 2026-10-15 | minor | M5: budget retained FDF entries before cloning; discovery keeps lazy array owners and annotation positions, with alias/cache-eviction stress fixtures. Limits cover the copied graph; source parsing/page discovery retain existing COS/parser limits, with no total source-memory bound. FDF import semantics and estimates unchanged. | | 2026-10-15 | minor | M7: the Edit text caret is as tall as the row it is on in a region with several sizes; estimates unchanged. | @@ -787,6 +815,7 @@ Newest first: the date, what moved, and the new overall estimate. | 2026-10-14 | minor | M1: raw-deflate fallback preserves decoding-limit errors in strict and tolerant decoding; synthetic filter and COS stream regressions. Estimates unchanged. | | 2026-10-16 | minor | M14: Czech catalog completed (2,674 entries) and "tag" terminology unified. Estimates unchanged. | | 2026-10-13 | minor | M9: separate page validation overlays preserve signature artwork, and snapshot tabs reuse instead of multiplying; baked captions remain unchanged (#79 open); estimates unchanged | +| 2026-10-15 | minor | M1: refresh object-generation reference identity, retaining current upstream progress and regression tests. Estimates unchanged. | | 2026-10-16 | minor | M10 #1201 partial: lossless image export applies component Decode pairs and Indexed mapping. Estimates unchanged. | | 2026-10-15 | minor | M6: CSV export and spreadsheet merging quote tabs, and import distinguishes quoted tabs from TSV delimiters without changing field identities. Estimates unchanged. | | 2026-10-13 | minor | M4: rotation queries and edits resolve indirect page-tree values; inherited, negative and large rotations verified through both saves. Estimates unchanged. | diff --git a/apps/pdfcraft-cli/tests/it/batch_exit.rs b/apps/pdfcraft-cli/tests/it/batch_exit.rs new file mode 100644 index 0000000000..6dfb9f61e9 --- /dev/null +++ b/apps/pdfcraft-cli/tests/it/batch_exit.rs @@ -0,0 +1,111 @@ +//! The exit codes the CLI has today: `ExitCode::SUCCESS` (0) or `ExitCode::FAILURE` (1). +//! +//! Pinned here so that the exit-code contract is a visible, reviewed change when it comes: a +//! failed file, an outside-root path and a usage error all exit 1 today. + +use serde_json::Value; +use std::path::{Path, PathBuf}; +use std::process::{Command, Output}; + +const BIN: &str = env!("CARGO_BIN_EXE_pdfcraft-cli"); + +/// A fresh folder for one test, so the tests can run in parallel and leave nothing behind. +fn workdir(name: &str) -> PathBuf { + let dir = std::env::temp_dir().join(format!("pdfcraft-cli-exit-{}-{name}", std::process::id())); + let _ = std::fs::remove_dir_all(&dir); + std::fs::create_dir_all(&dir).unwrap(); + dir +} + +/// `batch` with `args`, run from the working directory `dir`. +fn batch(dir: &Path, args: &[&str]) -> Output { + Command::new(BIN).current_dir(dir).arg("batch").args(args).output().unwrap() +} + +/// `run --script --report` from the working directory `dir`: one JSON document for the run. +fn run_report(dir: &Path, script: &Path) -> Output { + Command::new(BIN).current_dir(dir).args(["run", "--script"]).arg(script).arg("--report").output().unwrap() +} + +/// The one JSON document a command prints on stdout. +fn reported(output: &Output) -> Value { + serde_json::from_slice(&output.stdout).unwrap() +} + +#[test] +fn batch_exits_0_when_every_file_succeeds() { + let dir = workdir("batch-ok"); + std::fs::write(dir.join("page.txt"), "Exit code page").unwrap(); + let output = batch(&dir, &["page.txt", "--steps", "convert", "--out-dir", "out"]); + assert_eq!(output.status.code(), Some(0), "{}", String::from_utf8_lossy(&output.stderr)); + let report = reported(&output); + assert_eq!(report["completed"], 1, "{report}"); + assert_eq!(report["failed"], 0, "{report}"); + assert!(dir.join("out/page.pdf").exists(), "the converted page is written"); + std::fs::remove_dir_all(&dir).unwrap(); +} + +#[test] +fn batch_exits_1_when_one_file_fails() { + let dir = workdir("batch-fail"); + std::fs::write(dir.join("bad.pdf"), "broken").unwrap(); + let output = batch(&dir, &["bad.pdf", "--steps", "convert", "--out-dir", "out"]); + assert_eq!(output.status.code(), Some(1), "{}", String::from_utf8_lossy(&output.stderr)); + let report = reported(&output); + assert_eq!(report["failed"], 1, "{report}"); + std::fs::remove_dir_all(&dir).unwrap(); +} + +#[test] +fn batch_usage_error_exits_1_and_creates_nothing() { + let dir = workdir("batch-usage"); + std::fs::write(dir.join("page.txt"), "Exit code page").unwrap(); + // Baseline: a usage error exits 1 today. A later change to the exit-code contract will + // update this assertion on purpose. + let output = batch(&dir, &["page.txt", "--out-dir", "out"]); + assert_eq!(output.status.code(), Some(1), "{}", String::from_utf8_lossy(&output.stderr)); + assert!(output.stdout.is_empty(), "a usage error prints to stderr, not a report on stdout"); + assert!(!output.stderr.is_empty(), "a usage error says what is wrong"); + assert!(!dir.join("out").exists(), "a refused batch creates no output"); + std::fs::remove_dir_all(&dir).unwrap(); +} + +#[test] +fn batch_outside_root_exits_1_and_creates_nothing() { + let dir = workdir("batch-outside-root"); + let output = Command::new(BIN) + .current_dir(&dir) + .args(["batch", "../outside.pdf", "--steps", "convert", "--out-dir", "out", "--root"]) + .arg(&dir) + .output() + .unwrap(); + assert_eq!(output.status.code(), Some(1), "{}", String::from_utf8_lossy(&output.stderr)); + let stderr = String::from_utf8_lossy(&output.stderr); + assert!(stderr.contains("is outside the allowed directory"), "the input is refused by the root: {stderr}"); + assert!(output.stdout.is_empty(), "a refused batch prints no report"); + assert!(!dir.join("out").exists(), "a refused batch creates no output"); + std::fs::remove_dir_all(&dir).unwrap(); +} + +#[test] +fn run_script_report_exits_0_when_ok_and_1_when_a_step_fails() { + let dir = workdir("run-report"); + let script = dir.join("steps.json"); + let blank = serde_json::json!({ "tool": "doc_create", "args": { "from": "blank", "width": 100, "height": 100 } }); + let steps = serde_json::json!([blank.clone(), { "tool": "no_such_tool", "args": {} }]); + std::fs::write(&script, steps.to_string()).unwrap(); + + let failing = run_report(&dir, &script); + assert_eq!(failing.status.code(), Some(1), "{}", String::from_utf8_lossy(&failing.stderr)); + let report = reported(&failing); + assert_eq!(report["ok"], false, "{report}"); + assert_eq!(report["steps"][1]["status"], "error", "{report}"); + + std::fs::write(&script, serde_json::json!([blank]).to_string()).unwrap(); + let good = run_report(&dir, &script); + assert_eq!(good.status.code(), Some(0), "{}", String::from_utf8_lossy(&good.stderr)); + let report = reported(&good); + assert_eq!(report["ok"], true, "{report}"); + assert_eq!(report["steps"][0]["status"], "ok", "{report}"); + std::fs::remove_dir_all(&dir).unwrap(); +} diff --git a/apps/pdfcraft-cli/tests/it/main.rs b/apps/pdfcraft-cli/tests/it/main.rs index f0b87c97a7..30a6e25882 100644 --- a/apps/pdfcraft-cli/tests/it/main.rs +++ b/apps/pdfcraft-cli/tests/it/main.rs @@ -3,6 +3,7 @@ mod audit; mod automation_roots; mod batch; +mod batch_exit; mod check; mod edit; mod extract; diff --git a/apps/pdfcraft-web/Cargo.toml b/apps/pdfcraft-web/Cargo.toml index ad8243d594..56e10c0c29 100644 --- a/apps/pdfcraft-web/Cargo.toml +++ b/apps/pdfcraft-web/Cargo.toml @@ -11,10 +11,12 @@ workspace = true [dependencies] pdfcraft-ui-egui = { workspace = true } +# The host messages (storytold/craft-libs): one contract and one bridge for the web builds. +craft-web-host = { git = "https://github.com/storytold/craft-libs", rev = "2e2312ddaf21bc3d85cd3395d0955c7d239306f6", version = "0.1.0" } eframe = { workspace = true, features = ["default_fonts", "glow", "persistence", "accesskit"] } egui = { workspace = true, features = ["default_fonts", "monochrome_emoji_fonts"] } [target.'cfg(target_arch = "wasm32")'.dependencies] wasm-bindgen-futures = "0.4" -web-sys = { version = "0.3", features = ["Window", "Document", "Element", "HtmlCanvasElement", "HtmlElement", "Location", "Url", "UrlSearchParams", "Response", "Event", "MessageEvent", "EventTarget"] } +web-sys = { version = "0.3", features = ["Window", "Document", "Element", "HtmlCanvasElement", "HtmlElement", "Location", "Url", "UrlSearchParams", "Response", "Event", "MessageEvent", "EventTarget", "CustomEvent", "CustomEventInit"] } js-sys = "0.3" diff --git a/apps/pdfcraft-web/src/main.rs b/apps/pdfcraft-web/src/main.rs index 73d14f4a6c..ffa5a06068 100644 --- a/apps/pdfcraft-web/src/main.rs +++ b/apps/pdfcraft-web/src/main.rs @@ -5,18 +5,11 @@ #[cfg(target_arch = "wasm32")] fn main() { - use eframe::wasm_bindgen::JsCast; use pdfcraft_ui_egui::PdfCraftApp; let options = eframe::WebOptions { renderer: eframe::Renderer::Glow, ..Default::default() }; wasm_bindgen_futures::spawn_local(async move { - let Some(canvas) = web_sys::window() - .and_then(|w| w.document()) - .and_then(|d| d.get_element_by_id("pdfcraft")) - .and_then(|e| e.dyn_into::().ok()) - else { - return; - }; + let Some(canvas) = app_canvas() else { return }; let started = eframe::WebRunner::new() .start( canvas, @@ -54,11 +47,11 @@ fn main() { // eframe never raises a close request on a browser reload or tab close, so `guard_quit` never // prompts; ask the browser to confirm instead while a document has unsaved work (#812). warn_before_unload(app.unsaved_flag.clone()); - // `?host=parent`: a same-origin page embedding PdfCraft in an iframe (such - // as the Nextcloud app) passes files in and takes saves back (`host` below). - if query_param("host").as_deref() == Some("parent") - && let Some(bridge) = host::Bridge::new() - { + // Host modes (`host` below, over `craft-web-host`): `?host=parent`, a + // same-origin page embedding PdfCraft in an iframe (such as the Nextcloud + // app), or `data-host="page"` on the canvas, a host that runs the module in + // its own document. + if let Some(bridge) = app_canvas().as_ref().and_then(host::Bridge::connect) { // `?author=`: the host's name for the user signs their comments, // unless they chose a name in PdfCraft (the web default is "Guest"). if let Some(author) = query_param("author").map(|a| a.trim().to_string()).filter(|a| !a.is_empty()) @@ -81,9 +74,7 @@ fn main() { .await; if let Err(e) = started { eframe::web_sys::console::error_1(&e); - if query_param("host").as_deref() == Some("parent") - && let Some(bridge) = host::Bridge::new() - { + if let Some(bridge) = app_canvas().as_ref().and_then(host::Bridge::connect) { bridge.post_failed(&e.as_string().unwrap_or_else(|| format!("{e:?}"))); } } @@ -123,136 +114,116 @@ async fn fetch_bytes(url: &str) -> Result, String> { Ok(js_sys::Uint8Array::new(&buf).to_vec()) } -/// The `postMessage` bridge to a same-origin parent page. Messages are plain objects with a -/// `type`: +/// The app's canvas: the one element `index.html` and every embedding page must provide, with +/// the id the app looks up. +#[cfg(target_arch = "wasm32")] +fn app_canvas() -> Option { + use eframe::wasm_bindgen::JsCast; + web_sys::window()?.document()?.get_element_by_id("pdfcraft").and_then(|e| e.dyn_into::().ok()) +} + +/// Host modes: how the page embedding PdfCraft passes files in and takes saves back. The messages +/// and the two transports are `craft-web-host`'s (storytold/craft-libs), the same contract as the +/// other Craft apps' bridges, with the `pdfcraft:` prefix. This module is PdfCraft's part of it. /// -/// - `pdfcraft:ready` (to the parent): the app is listening; send files now. -/// - `pdfcraft:open` (from the parent): `{ name, bytes }`, where `bytes` is an `ArrayBuffer` or -/// `Uint8Array` holding a PDF. It opens in a tab, like a `?file=` URL. -/// - `pdfcraft:save` (to the parent): `{ name, saveAs, bytes }` when the user saves; `bytes` is -/// a transferred `ArrayBuffer`. The parent stores the file (and reports its own errors). -/// - `pdfcraft:dirty` (to the parent): `{ dirty }` whenever unsaved work appears or goes away. -/// - `pdfcraft:failed` (to the parent): `{ error }` when the app couldn't start. -/// -/// Only the parent window on the app's own origin is heard or answered. With a parent on -/// another origin there is no bridge: the app runs as if unframed and saves download. +/// - `?host=parent` is the `postMessage` bridge for a same-origin page that embeds PdfCraft in an +/// iframe (such as the Nextcloud app). With a parent on another origin there is no bridge: the +/// app runs as if unframed and saves download (#2167). `data-host="page"` on the app's canvas +/// is for a host that loads the module into its own document, with no iframe: the same messages +/// travel as DOM `CustomEvent`s on that canvas. `?host=parent` wins when both are present. +/// - `pdfcraft:ready` once the app is listening; send files now. PdfCraft reads bytes, so it +/// says `file: false` and a host sends `bytes`. +/// - `pdfcraft:open` (from the host): `{ name, bytes }` holding a PDF. It opens in a tab, like a +/// `?file=` URL; a file over [`host::MAX_OPEN_BYTES`] is refused before it is copied. +/// - `pdfcraft:save` (to the host): `{ name, saveAs, bytes }` when the user saves. The host stores +/// the file (and reports its own errors). +/// - `pdfcraft:dirty` whenever unsaved work appears or goes away, `pdfcraft:failed` when the app +/// couldn't start. #[cfg(target_arch = "wasm32")] mod host { - use eframe::wasm_bindgen::{JsCast, JsValue, closure::Closure}; + use std::rc::Rc; + + use craft_web_host::web::{FileData, Host, Message}; + + /// The app id in the host messages (`pdfcraft:ready`). + const APP: &str = "pdfcraft"; + + /// The name a file gets when the host sent none (the crate says `document` for a blank name). + const DEFAULT_NAME: &str = "document.pdf"; + + /// The largest file the host may hand over to open, checked before it is copied (the app's own readers have their own limits). + const MAX_OPEN_BYTES: u64 = 512 << 20; #[derive(Clone)] pub struct Bridge { - window: web_sys::Window, - parent: web_sys::Window, - origin: String, + host: Rc, } impl Bridge { - /// `None` when the page isn't framed (there is no parent to talk to), or when the parent - /// is on another origin: messages only go to this origin, so the browser would drop - /// every save without an error while the document showed as saved (#2167). - pub fn new() -> Option { - let window = web_sys::window()?; - let parent = window.parent().ok().flatten()?; - if js_sys::Object::is(&parent, &window) { - return None; - } - let origin = window.location().origin().ok()?; - // Reading another origin's location throws, which is `None` too. - if parent.location().origin().ok()? != origin { - return None; - } - Some(Self { window, parent, origin }) - } - - fn post(&self, message: &js_sys::Object, transfer: Option<&JsValue>) -> Result<(), String> { - let sent = match transfer { - Some(t) => self.parent.post_message_with_transfer(message, &self.origin, &js_sys::Array::of1(t)), - None => self.parent.post_message(message, &self.origin), - }; - sent.map_err(|e| e.as_string().unwrap_or_else(|| format!("{e:?}"))) + /// The bridge for this page, or `None` when PdfCraft isn't hosted: no `?host=parent` and + /// no `data-host="page"` on the canvas, an unframed `?host=parent` page, or a parent on + /// another origin (messages only go to this origin, so the browser would drop every save + /// without an error while the document showed as saved, #2167). + pub fn connect(canvas: &web_sys::HtmlCanvasElement) -> Option { + Some(Self { host: Rc::new(Host::connect(APP, canvas)?) }) } pub fn post_ready(&self) { - if let Some(m) = message(&[("type", "pdfcraft:ready".into()), ("version", env!("CARGO_PKG_VERSION").into())]) { - let _ = self.post(&m, None); - } + let _ = self.host.post_ready(env!("CARGO_PKG_VERSION"), false); } - /// The app couldn't start (no WebGL, for example): let the parent say so. + /// The app couldn't start (no WebGL, for example): let the host say so. pub fn post_failed(&self, error: &str) { - if let Some(m) = message(&[("type", "pdfcraft:failed".into()), ("error", error.into())]) { - let _ = self.post(&m, None); - } + let _ = self.host.post_failed(error, None); } - /// Hear `pdfcraft:open` from the parent and queue the file like a `?file=` download. + /// Hear `pdfcraft:open` and queue the file like a `?file=` download. pub fn listen(&self, inbox: pdfcraft_ui_egui::Inbox, ctx: egui::Context) { - let parent = self.parent.clone(); - let origin = self.origin.clone(); - let on_message = Closure::::new(move |event: web_sys::MessageEvent| { - if event.origin() != origin || !event.source().is_some_and(|s| js_sys::Object::is(&s, &parent)) { - return; + self.host.listen(move |message| { + if let Message::Open { name, data } = message { + queue_open(&name, &data, &inbox, &ctx); } - let data = event.data(); - if string(&data, "type").as_deref() != Some("pdfcraft:open") { - return; - } - let name = string(&data, "name").filter(|n| !n.trim().is_empty()).unwrap_or_else(|| "document.pdf".to_string()); - let Ok(bytes) = js_sys::Reflect::get(&data, &JsValue::from_str("bytes")) else { return }; - let bytes = if let Some(array) = bytes.dyn_ref::() { - array.to_vec() - } else if bytes.is_instance_of::() { - js_sys::Uint8Array::new(&bytes).to_vec() - } else { - return; - }; - if let Ok(mut q) = inbox.lock() { - q.push((name, bytes)); - } - ctx.request_repaint(); }); - if self.window.add_event_listener_with_callback("message", on_message.as_ref().unchecked_ref()).is_ok() { - // The listener lives as long as the page. - on_message.forget(); - } } pub fn save_fn(&self) -> pdfcraft_ui_egui::HostSaveFn { - let bridge = self.clone(); - Box::new(move |name: &str, bytes: &[u8], save_as: bool| { - let buffer = js_sys::Uint8Array::from(bytes).buffer(); - let m = - message(&[("type", "pdfcraft:save".into()), ("name", name.into()), ("saveAs", save_as.into()), ("bytes", buffer.clone().into())]) - .ok_or_else(|| "couldn't build the message".to_string())?; - bridge.post(&m, Some(&buffer)) - }) + let host = self.host.clone(); + Box::new(move |name: &str, bytes: &[u8], save_as: bool| host.post_save(name, bytes, None, save_as)) } pub fn dirty_fn(&self) -> pdfcraft_ui_egui::HostDirtyFn { - let bridge = self.clone(); + let host = self.host.clone(); + // A host starts out clean: the first report is a change only when it is dirty. let mut last = false; Box::new(move |dirty: bool| { if dirty != last { last = dirty; - if let Some(m) = message(&[("type", "pdfcraft:dirty".into()), ("dirty", dirty.into())]) { - let _ = bridge.post(&m, None); - } + host.post_dirty(dirty); } }) } } - fn message(fields: &[(&str, JsValue)]) -> Option { - let m = js_sys::Object::new(); - for (key, value) in fields { - js_sys::Reflect::set(&m, &JsValue::from_str(key), value).ok()?; + /// The `pdfcraft:open` payload, `{ name, bytes }` where `bytes` holds a PDF: queue it like a + /// `?file=` download and wake the app. The crate has cleaned the name (and cut it to 255 + /// characters); a blank one comes as `document` and becomes `document.pdf`. + fn queue_open(name: &str, data: &FileData, inbox: &pdfcraft_ui_egui::Inbox, ctx: &egui::Context) { + let name = if name == "document" { DEFAULT_NAME } else { name }.to_string(); + let FileData::Bytes(view) = data else { + eframe::web_sys::console::warn_1(&format!("PdfCraft: host file {name} came as a File; PdfCraft reads bytes only; not opened").into()); + return; + }; + // A file from the host is untrusted input: refuse an oversized one before copying it into the app's memory. + if data.size() > MAX_OPEN_BYTES as f64 { + eframe::web_sys::console::warn_1( + &format!("PdfCraft: host file is {} bytes, over the {} MiB host open limit; not opened", data.size(), MAX_OPEN_BYTES >> 20).into(), + ); + return; } - Some(m) - } - - fn string(data: &JsValue, key: &str) -> Option { - js_sys::Reflect::get(data, &JsValue::from_str(key)).ok()?.as_string() + if let Ok(mut q) = inbox.lock() { + q.push((name, view.to_vec())); + } + ctx.request_repaint(); } } diff --git a/apps/pdfcraft/src/main.rs b/apps/pdfcraft/src/main.rs index 5bc5c5c372..be2290d2f9 100644 --- a/apps/pdfcraft/src/main.rs +++ b/apps/pdfcraft/src/main.rs @@ -439,6 +439,8 @@ fn app_creator<'a>( if let Some(dir) = pdfcraft_ui_egui::RecoveryStore::default_dir() { app.enable_recovery(pdfcraft_ui_egui::RecoveryStore::new(dir)); } + // The OCR Verify screen stores its per-language dictionary beside recovery data. + app.ocr_verify.dictionary_dir = pdfcraft_ui_egui::default_dictionary_dir(); if let Some(state) = &cc.wgpu_render_state { let adapter = state.adapter.get_info(); let max_texture_dimension_2d = state.device.limits().max_texture_dimension_2d; diff --git a/assets/icons/a-large-small.svg b/assets/icons/a-large-small.svg new file mode 100644 index 0000000000..7716db3cf2 --- /dev/null +++ b/assets/icons/a-large-small.svg @@ -0,0 +1,18 @@ + + + + + + + diff --git a/contributors/people.toml b/contributors/people.toml index fecfb68c90..8ab196abc9 100644 --- a/contributors/people.toml +++ b/contributors/people.toml @@ -183,3 +183,6 @@ sync_github_name = true [people.Combiler01] sync_github_name = true + +[people.helenkwok] +real_name = "Helen Kwok" diff --git a/crates/annot/src/appearance.rs b/crates/annot/src/appearance.rs index 909250ef26..7148e9fc2e 100644 --- a/crates/annot/src/appearance.rs +++ b/crates/annot/src/appearance.rs @@ -232,6 +232,31 @@ fn rg_stroke(c: Rgb) -> String { format!("{} {} {} RG\n", n(c[0]), n(c[1]), n(c[2])) } +/// A FreeText's border colour where it has its own: the stroking colour in `/DA` (`RG`, or `G` +/// for grey). `None`: the border takes the text colour. +pub fn da_border(d: &Dict) -> Option { + let da = d.get(b"DA").and_then(|o| o.as_string()).map(|s| String::from_utf8_lossy(&s.bytes).into_owned())?; + let toks: Vec<&str> = da.split_whitespace().collect(); + let num = |i: usize| toks.get(i).and_then(|t| t.parse::().ok()).filter(|x| x.is_finite()); + let mut col = None; + for (i, t) in toks.iter().enumerate() { + match *t { + "RG" if i >= 3 => { + if let (Some(r), Some(g), Some(b)) = (num(i - 3), num(i - 2), num(i - 1)) { + col = Some([r, g, b].map(|x| x.clamp(0.0, 1.0))); + } + } + "G" if i >= 1 => { + if let Some(g) = num(i - 1) { + col = Some([g.clamp(0.0, 1.0); 3]); + } + } + _ => {} + } + } + col +} + /// `/DA` of a text box: (text colour, font size). Defaults to black 12 pt. pub fn parse_da(d: &Dict) -> (Rgb, f64) { let da = d.get(b"DA").and_then(|o| o.as_string()).map(|s| String::from_utf8_lossy(&s.bytes).into_owned()).unwrap_or_default(); @@ -262,23 +287,51 @@ pub fn parse_da(d: &Dict) -> (Rgb, f64) { (col.map(|x| x.clamp(0.0, 1.0)), size.min(400.0)) } -fn ext_gstate(opacity: f64, multiply: bool) -> Option { - if opacity >= 1.0 && !multiply { +/// How opaque a comment's parts are drawn, 0–1: its line or border (`/CA`); its fill or +/// background (`/ca`, which PDF 2.0 gives the non-stroking operations; without it, as the line); +/// and a text box's text (`/PCTextOpacity`, PdfCraft's own; without it, as the fill, since text +/// is filled too). A comment with only `/CA` is drawn at that opacity throughout, as before. +#[derive(Clone, Copy, Debug, PartialEq)] +pub struct PartOpacity { + pub line: f64, + pub fill: f64, + pub text: f64, +} + +/// The parts' opacities of the comment `d` (see [`PartOpacity`]). +pub fn part_opacity(d: &Dict) -> PartOpacity { + let get = |key: &[u8]| d.get(key).and_then(|o| o.as_f64()).filter(|x| x.is_finite()).map(|x| x.clamp(0.0, 1.0)); + let line = get(b"CA").unwrap_or(1.0); + let fill = get(b"ca").unwrap_or(line); + let text = get(b"PCTextOpacity").unwrap_or(fill); + PartOpacity { line, fill, text } +} + +/// `/GS0`: the line's opacity for strokes and the fill's for everything filled; `/GS1`, where the +/// text's differs from the fill's: the text's for its glyphs. +fn ext_gstate(parts: PartOpacity, multiply: bool) -> Option { + if parts.line >= 1.0 && parts.fill >= 1.0 && parts.text >= 1.0 && !multiply { return None; } - let mut gs = Dict::new(); - gs.set(b"Type".to_vec(), Object::name("ExtGState")); - if multiply { - gs.set(b"BM".to_vec(), Object::name("Multiply")); - } - gs.set(b"CA".to_vec(), Object::Real(opacity)); - gs.set(b"ca".to_vec(), Object::Real(opacity)); + let gs = |stroke: f64, fill: f64| { + let mut gs = Dict::new(); + gs.set(b"Type".to_vec(), Object::name("ExtGState")); + if multiply { + gs.set(b"BM".to_vec(), Object::name("Multiply")); + } + gs.set(b"CA".to_vec(), Object::Real(stroke)); + gs.set(b"ca".to_vec(), Object::Real(fill)); + Object::Dict(gs) + }; let mut res = Dict::new(); - res.set(b"GS0".to_vec(), Object::Dict(gs)); + res.set(b"GS0".to_vec(), gs(parts.line, parts.fill)); + if parts.text != parts.fill { + res.set(b"GS1".to_vec(), gs(parts.line, parts.text)); + } Some(res) } -fn form(bbox: [f64; 4], content: &[u8], resources: Dict) -> Stream { +pub(crate) fn form(bbox: [f64; 4], content: &[u8], resources: Dict) -> Stream { let mut d = Dict::new(); d.set(b"Type".to_vec(), Object::name("XObject")); d.set(b"Subtype".to_vec(), Object::name("Form")); @@ -288,6 +341,18 @@ fn form(bbox: [f64; 4], content: &[u8], resources: Dict) -> Stream { Stream::flate(d, content) } +/// The `cm` that draws a picture into `at`: an image (its unit square) or a form XObject whose +/// `/Matrix` maps it to `size` points (`None`, or a side that is not positive, draws nothing). +pub(crate) fn picture_place(at: [f64; 4], image: bool, size: Option<(f64, f64)>) -> Option { + let [x0, y0, x1, y1] = at; + let (w, h) = (x1 - x0, y1 - y0); + if image { + return Some(format!("{} 0 0 {} {} {} cm", n(w), n(h), n(x0), n(y0))); + } + let (sw, sh) = size.filter(|(sw, sh)| *sw > 0.0 && *sh > 0.0)?; + Some(format!("{} 0 0 {} {} {} cm", n(w / sw), n(h / sh), n(x0), n(y0))) +} + /// The page `/Rotate` an image signature's picture is turned back by (`/PCPictureRotate`, written /// when the signature is added to a turned page): 0, 90, 180 or 270. pub(crate) fn picture_turn(d: &Dict) -> i64 { @@ -301,10 +366,18 @@ fn stamp_turn(d: &Dict) -> i64 { /// The page `/Rotate` a text box's or Fill & Sign text's lines are turned back by /// (`/PCTextRotate`, written when it is added to a turned page). -fn text_turn(d: &Dict) -> i64 { +pub(crate) fn text_turn(d: &Dict) -> i64 { d.get(b"PCTextRotate").and_then(Object::as_int).unwrap_or(0).rem_euclid(360) / 90 * 90 } +/// A FreeText's `/RD`: how far its text box lies inside `/Rect` from the left, bottom, right and +/// top. A distance a hair below 0 (rounding) reads as 0; `None` when one is negative or they can't +/// be read. +pub(crate) fn inset(d: &Dict) -> Option<[f64; 4]> { + let rd: [f64; 4] = nums(d, b"RD")?.try_into().ok()?; + rd.iter().all(|x| x.is_finite() && *x > -0.001).then(|| rd.map(|x| x.max(0.0))) +} + /// The first text [`build`] would draw in the standard fonts (Helvetica, WinAnsiEncoding) that /// they can't, which would come out as `?`: what it is (for the message) and the character. The /// text of a text box, callout or Fill & Sign text, a line's caption, a measurement's label, a @@ -325,7 +398,7 @@ pub fn undrawable_text(d: &Dict) -> Option<(&'static str, char)> { } /// `/Font` resources naming `/Helv`: standard Helvetica with WinAnsiEncoding. -fn helvetica_resource() -> Object { +pub(crate) fn helvetica_resource() -> Object { let mut font = Dict::new(); font.set(b"Type".to_vec(), Object::name("Font")); font.set(b"Subtype".to_vec(), Object::name("Type1")); @@ -379,7 +452,9 @@ struct AppearanceContext<'a> { subtype: &'a [u8], rect: [f64; 4], cloudy: bool, + /// The line's opacity (`parts.line`). opacity: f64, + parts: PartOpacity, stroke: Option, width: f64, } @@ -396,7 +471,7 @@ impl AppearanceDrawing { fn new(context: &AppearanceContext<'_>) -> Self { let mut drawing = Self::default(); if let Some(gs) = ext_gstate( - context.opacity, + context.parts, context.subtype == b"Highlight" || (context.subtype == b"Ink" && context.dict.name(b"IT") == Some(b"FreehandHighlight")), ) { drawing.resources.set(b"ExtGState".to_vec(), Object::Dict(gs)); @@ -443,7 +518,8 @@ fn appearance_context(d: &Dict) -> Option> { subtype, rect, cloudy, - opacity: d.get(b"CA").and_then(|o| o.as_f64()).unwrap_or(1.0).clamp(0.0, 1.0), + opacity: part_opacity(d).line, + parts: part_opacity(d), stroke: color(d, b"C")?, width: border_width(d), }) @@ -786,15 +862,12 @@ fn draw_stamp(context: &AppearanceContext<'_>, mut drawing: AppearanceDrawing) - // the standard counter-rotation as its form `/Matrix`: viewers fit the turned // bounding box to `/Rect` (ISO 32000-2 §12.5.5), so it reads upright as displayed. if let Some(pic) = d.get(b"PCPicture").and_then(Object::as_ref) { - let [x0, y0, x1, y1] = rect; - let (w, h) = (x1 - x0, y1 - y0); let image = matches!(d.get(b"PCPictureImage"), Some(Object::Bool(true))); - let place = if image { - format!("{} 0 0 {} {} {} cm", n(w), n(h), n(x0), n(y0)) - } else { - let size = nums(d, b"PCPictureSize").filter(|s| s.len() == 2 && s[0] > 0.0 && s[1] > 0.0)?; - format!("{} 0 0 {} {} {} cm", n(w / size[0]), n(h / size[1]), n(x0), n(y0)) - }; + let size = nums(d, b"PCPictureSize").and_then(|s| match s.as_slice() { + [w, h] if *w > 0.0 && *h > 0.0 => Some((*w, *h)), + _ => None, + }); + let place = picture_place(rect, image, size)?; c.push_str(&format!("q {place} /Pic Do Q\n")); let mut xo = Dict::new(); xo.set(b"Pic".to_vec(), Object::Ref(pic)); @@ -836,46 +909,25 @@ fn draw_stamp(context: &AppearanceContext<'_>, mut drawing: AppearanceDrawing) - let [x0, y0, x1, y1] = draw; let (w, h) = (x1 - x0, y1 - y0); let lw = w.min(h) * 0.12; + // The strokes come from `FillMark::strokes`, which a copy's outline draws too. + let path = |mark: crate::FillMark| { + let ops: Vec = mark + .strokes(draw) + .iter() + .flat_map(|s| s.iter().enumerate().map(|(i, p)| format!("{} {} {}", n(p[0]), n(p[1]), if i == 0 { "m" } else { "l" }))) + .collect(); + format!("{} S\n", ops.join(" ")) + }; match name { - b"PCCheck" => c.push_str(&format!( - "{}{} w 1 J 1 j\n{} {} m {} {} l {} {} l S\n", - rg_stroke(col), - n(lw), - n(x0 + w * 0.15), - n(y0 + h * 0.5), - n(x0 + w * 0.4), - n(y0 + h * 0.2), - n(x0 + w * 0.88), - n(y0 + h * 0.85) - )), - b"PCCross" => c.push_str(&format!( - "{}{} w 1 J\n{} {} m {} {} l {} {} m {} {} l S\n", - rg_stroke(col), - n(lw), - n(x0 + w * 0.18), - n(y0 + h * 0.18), - n(x1 - w * 0.18), - n(y1 - h * 0.18), - n(x0 + w * 0.18), - n(y1 - h * 0.18), - n(x1 - w * 0.18), - n(y0 + h * 0.18) - )), + b"PCCheck" => c.push_str(&format!("{}{} w 1 J 1 j\n{}", rg_stroke(col), n(lw), path(crate::FillMark::Check))), + b"PCCross" => c.push_str(&format!("{}{} w 1 J\n{}", rg_stroke(col), n(lw), path(crate::FillMark::Cross))), b"PCDot" => { let r = w.min(h) * 0.3; c.push_str(&rg(col)); c.push_str(&ellipse(x0 + w / 2.0 - r, y0 + h / 2.0 - r, x0 + w / 2.0 + r, y0 + h / 2.0 + r)); c.push_str("f\n"); } - b"PCLine" => c.push_str(&format!( - "{}{} w 1 J\n{} {} m {} {} l S\n", - rg_stroke(col), - n(h.clamp(0.5, 2.0)), - n(x0), - n(y0 + h / 2.0), - n(x1), - n(y0 + h / 2.0) - )), + b"PCLine" => c.push_str(&format!("{}{} w 1 J\n{}", rg_stroke(col), n(h.clamp(0.5, 2.0)), path(crate::FillMark::Line))), other => { // Only stamps PdfCraft made: others keep their own artwork. if !matches!(d.get(b"PCStamp"), Some(Object::Bool(true))) { @@ -893,14 +945,59 @@ fn draw_stamp(context: &AppearanceContext<'_>, mut drawing: AppearanceDrawing) - Some(form(rect, c.as_bytes(), res)) } +/// The oval or circle (`shape`) around a text box whose whole box is `full`: its background +/// and border go into `c`; returns the text box inside it. +#[allow(clippy::too_many_arguments)] +fn shaped_border( + d: &Dict, + shape: crate::BorderShape, + full: [f64; 4], + turn: i64, + bw: f64, + bg: Option, + line_color: Rgb, + c: &mut String, +) -> Option<[f64; 4]> { + // Laid out as displayed: a quarter turn swaps the sides' differences. + let rect = match inset(d) { + Some(rd) => { + let (dx, dy) = if matches!(turn, 90 | 270) { (rd[1], rd[0]) } else { (rd[0], rd[1]) }; + [full[0] + dx, full[1] + dy, full[2] - dx, full[3] - dy] + } + None => shape.inside(full, bw, full), + }; + if rect[2] - rect[0] < 1.0 || rect[3] - rect[1] < 1.0 { + return None; + } + let h = bw / 2.0; + let [x0, y0, x1, y1] = [full[0] + h, full[1] + h, full[2] - h, full[3] - h]; + let path = match shape { + crate::BorderShape::Circle => { + let (cx, cy, r) = ((x0 + x1) / 2.0, (y0 + y1) / 2.0, (x1 - x0).min(y1 - y0) / 2.0); + ellipse(cx - r, cy - r, cx + r, cy + r) + } + _ => ellipse(x0, y0, x1, y1), + }; + if let Some(bg) = bg { + c.push_str(&format!("{}{path}f\n", rg(bg))); + } + if bw > 0.0 { + c.push_str(&format!("{}{} w\n{}{path}S\n[] 0 d\n", rg_stroke(line_color), n(bw), dash(d))); + } + Some(rect) +} + fn draw_free_text(context: &AppearanceContext<'_>, mut drawing: AppearanceDrawing) -> Option { let d = context.dict; + let parts = context.parts; let rect = context.rect; let stroke = context.stroke; let mut c = std::mem::take(&mut drawing.content); let mut res = std::mem::take(&mut drawing.resources); let _tail = std::mem::take(&mut drawing.tail); let (text_color, size) = parse_da(d); + // The box's border and a callout's leader line: their own colour, else the text's. + let line_color = da_border(d).unwrap_or(text_color); let bg = stroke; let bw = if d.contains(b"BS") || d.contains(b"Border") { border_width(d) } else { 0.0 }; // A callout: the leader line (arrowhead at its first point), and the text box inside @@ -939,7 +1036,7 @@ fn draw_free_text(context: &AppearanceContext<'_>, mut drawing: AppearanceDrawin if pts.len() < 2 || pts.iter().any(|p| !p.0.is_finite() || !p.1.is_finite()) { return None; } - c.push_str(&format!("{}{}{} w 1 J 1 j\n", rg_stroke(text_color), rg(bg.unwrap_or([1.0; 3])), n(lw))); + c.push_str(&format!("{}{}{} w 1 J 1 j\n", rg_stroke(line_color), rg(bg.unwrap_or([1.0; 3])), n(lw))); for (i, p) in pts.iter().enumerate() { c.push_str(&format!("{} {} {}\n", n(p.0), n(p.1), if i == 0 { "m" } else { "l" })); } @@ -948,26 +1045,40 @@ fn draw_free_text(context: &AppearanceContext<'_>, mut drawing: AppearanceDrawin return None; } } - if let Some(bg) = bg { - c.push_str(&format!("{}{} {} {} {} re f\n", rg(bg), n(rect[0]), n(rect[1]), n(rect[2] - rect[0]), n(rect[3] - rect[1]))); - } - if bw > 0.0 { - let h = bw / 2.0; - c.push_str(&format!( - "{}{} w\n{}{} {} {} {} re S\n[] 0 d\n", - rg_stroke(text_color), - n(bw), - dash(d), - n(rect[0] + h), - n(rect[1] + h), - n(rect[2] - rect[0] - bw), - n(rect[3] - rect[1] - bw) - )); + // An oval or a circle (`/PCBorderShape`) around the text box, which is `/Rect` inset by + // `/RD` and keeps its place: background and border take the shape, the text is laid out + // in the text box as in a rectangle. + let shape = crate::BorderShape::of(d); + if shape != crate::BorderShape::Rectangle { + rect = shaped_border(d, shape, full, turn, bw, bg, line_color, &mut c)?; + } else { + if let Some(bg) = bg { + c.push_str(&format!("{}{} {} {} {} re f\n", rg(bg), n(rect[0]), n(rect[1]), n(rect[2] - rect[0]), n(rect[3] - rect[1]))); + } + if bw > 0.0 { + let h = bw / 2.0; + c.push_str(&format!( + "{}{} w\n{}{} {} {} {} re S\n[] 0 d\n", + rg_stroke(line_color), + n(bw), + dash(d), + n(rect[0] + h), + n(rect[1] + h), + n(rect[2] - rect[0] - bw), + n(rect[3] - rect[1] - bw) + )); + } } let text = d.get(b"Contents").and_then(|o| o.as_string()).map(PdfString::to_text).unwrap_or_default(); let pad = 2.0 + bw; let width = (rect[2] - rect[0] - 2.0 * pad).max(1.0); - let q = d.int(b"Q").unwrap_or(0); + // In an oval or a circle the lines are centred in the text box, up and down too. + let round = shape != crate::BorderShape::Rectangle; + let q = if round { 1 } else { d.int(b"Q").unwrap_or(0) }; + // The text at its own opacity, where it has one. + if parts.text != parts.fill { + c.push_str("/GS1 gs\n"); + } c.push_str(&format!( "{} {} {} {} re W n\nBT\n/Helv {} Tf\n{}", n(rect[0]), @@ -978,8 +1089,12 @@ fn draw_free_text(context: &AppearanceContext<'_>, mut drawing: AppearanceDrawin rg(text_color) )); let mut out = c.into_bytes(); + let lines = wrap(&text, size, width); let mut y = rect[3] - pad - size * 0.9; - for line in wrap(&text, size, width) { + if round { + y -= ((rect[3] - rect[1] - 2.0 * pad - lines.len() as f64 * size * 1.2) / 2.0).max(0.0); + } + for line in lines { if y < rect[1] - size { break; } diff --git a/crates/annot/src/appearance/symbols.rs b/crates/annot/src/appearance/symbols.rs index 3b33662524..3178f78e94 100644 --- a/crates/annot/src/appearance/symbols.rs +++ b/crates/annot/src/appearance/symbols.rs @@ -1,4 +1,4 @@ -use super::{Dict, Object, PdfString, Rgb, Stream, color, ext_gstate, form, literal, n, nums, rg, rg_stroke, text_width, win_ansi}; +use super::{Dict, Object, PartOpacity, PdfString, Rgb, Stream, color, ext_gstate, form, literal, n, nums, rg, rg_stroke, text_width, win_ansi}; /// The label and colour of a standard rubber stamp name (ISO 32000-2 Table 184), for a stamp /// another program wrote without an appearance. Names are matched byte for byte, as PDF names @@ -51,7 +51,7 @@ pub(crate) fn display_stamp(d: &Dict) -> Option { // A malformed `/C` falls back to the stamp's own colour rather than hiding the stamp. let stroke = color(d, b"C").flatten(); let mut res = Dict::new(); - if let Some(gs) = ext_gstate(opacity, false) { + if let Some(gs) = ext_gstate(PartOpacity { line: opacity, fill: opacity, text: opacity }, false) { res.set(b"ExtGState".to_vec(), Object::Dict(gs)); } // A stamp whose intent (`/IT`, PDF 2.0) is an image or a snapshot has no `/Name` to default. diff --git a/crates/annot/src/create.rs b/crates/annot/src/create.rs index 1904745e95..6376655ce9 100644 --- a/crates/annot/src/create.rs +++ b/crates/annot/src/create.rs @@ -1,6 +1,6 @@ use super::draw::draw_appearance; use super::lookup::{annot_dict, annot_ref, annots, fill_sign_shape, page_ref, set_annots}; -use super::modify::{n, rich_text}; +use super::modify::{free_text_da, n, rich_text, set_part_keys, shape_text_box}; use super::types::{ FLAG_NO_ROTATE, FLAG_NO_ZOOM, FLAG_PRINT, LineEnding, Markup, Meta, NOTE_SIZE, NewAnnotation, OverlayLook, Rgb, Shape, StampGroup, Style, }; @@ -214,8 +214,13 @@ pub fn add_annotation(doc: &mut Document, new: &NewAnnotation, meta: &Meta) -> R let rect = rect_for(&new.shape, style)?; let mut d = base_dict(new.shape.subtype(), rect, page, &new.contents, &new.author, meta); d.set(b"Subj".to_vec(), PdfString::text(subject(&new.shape))); - let opacity = style.opacity.clamp(0.0, 1.0); - set_non_opaque_style(&mut d, opacity); + if style.fill_opacity.into_iter().chain(style.text_opacity).any(|o| !o.is_finite()) { + return Err(AnnotError::Invalid("invalid opacity".into())); + } + let line = style.opacity.clamp(0.0, 1.0); + let fill = style.fill_opacity.map_or(line, |o| o.clamp(0.0, 1.0)); + let text = style.text_opacity.map_or(fill, |o| o.clamp(0.0, 1.0)); + set_part_keys(&mut d, appearance::PartOpacity { line, fill, text }); let border = |d: &mut Dict| { let mut bs = Dict::new(); bs.set(b"W".to_vec(), Object::Real(style.width.max(0.0))); @@ -380,20 +385,22 @@ pub fn add_annotation(doc: &mut Document, new: &NewAnnotation, meta: &Meta) -> R d.set(b"IT".to_vec(), Object::name("FreeTextTypeWriter")); } let size = if font_size.is_finite() && *font_size > 0.0 { font_size.clamp(1.0, 400.0) } else { 12.0 }; - let [r, g, b] = style.color.map(|x| x.clamp(0.0, 1.0)); - d.set(b"DA".to_vec(), PdfString::literal(format!("{} {} {} rg /Helv {} Tf", n(r), n(g), n(b), n(size)).into_bytes())); + d.set(b"DA".to_vec(), free_text_da(style.color, style.border_color, size)); d.set(b"Q".to_vec(), Object::Int(0)); if let Some(f) = style.fill { d.set(b"C".to_vec(), rgb(f)); } border(&mut d); rich_text(&mut d); - // The text is laid out for the page as displayed, like a stamp's label. if !matches!(new.shape, Shape::Callout { .. }) { + // The text is laid out for the page as displayed, like a stamp's label. let turn = pdfcraft_model::pages(doc).get(new.page).map_or(0, |p| p.rotation(doc)); if matches!(turn, 90 | 180 | 270) { d.set(b"PCTextRotate".to_vec(), Object::Int(turn)); } + // An oval or a circle goes around the text box, its text wrapped as displayed: + // after `/PCTextRotate`, which swaps the box's sides on a quarter-turned page. + shape_text_box(&mut d, rect, style.border_shape, style.width); } } _ => {} @@ -421,7 +428,7 @@ pub fn add_annotation(doc: &mut Document, new: &NewAnnotation, meta: &Meta) -> R } fn validate_style(style: &Style) -> Result<(), AnnotError> { - if !finite(&style.color) || !style.opacity.is_finite() || !style.width.is_finite() { + if !finite(&style.color) || style.border_color.is_some_and(|c| !finite(&c)) || !style.opacity.is_finite() || !style.width.is_finite() { return Err(AnnotError::Invalid("invalid style".into())); } Ok(()) @@ -443,12 +450,6 @@ fn set_redaction_properties(dict: &mut Dict, style: &Style, quads: &[[f64; 8]], } } -fn set_non_opaque_style(dict: &mut Dict, opacity: f64) { - if opacity < 1.0 { - dict.set(b"CA".to_vec(), Object::Real(opacity)); - } -} - fn add_popup(doc: &mut Document, parent: ObjRef, rect: Option<[f64; 4]>, annots: &mut Vec) -> Result<(), AnnotError> { let Some(rect) = rect else { return Ok(()) }; let mut popup = Dict::new(); diff --git a/crates/annot/src/info.rs b/crates/annot/src/info.rs index 1a54dbc077..ba181d7504 100644 --- a/crates/annot/src/info.rs +++ b/crates/annot/src/info.rs @@ -2,7 +2,7 @@ use super::create::finite; use super::draw::set_appearance; use super::lookup::{annot_dict, annot_ref, annotation_flags, annots, drawn_upright, is_fill_sign, page_ref, page_refs}; use super::modify::{fillable, touch, unlocked}; -use super::types::{FLAG_LOCKED, LineEnding, Meta, NoteIcon, Rgb}; +use super::types::{BorderShape, FLAG_LOCKED, LineEnding, Meta, NoteIcon, Rgb}; use super::{AnnotError, Dict, Document, ObjRef, Object, PdfString, appearance}; // ── reading ───────────────────────────────────────────────────────────────────────────────── @@ -217,6 +217,15 @@ pub struct Props { pub fill: Option, /// Text boxes, typewriter text and callouts: the `/DA` font size in points. pub font_size: Option, + /// Text boxes, typewriter text and callouts: the border's own colour (`None`: the text's). + pub border_color: Option, + /// Text boxes and typewriter text: the border's shape (callouts have none to choose). + pub border_shape: Option, + /// How opaque the fill is drawn, for a comment with a fill control (`opacity` is then its + /// line's or border's). + pub fill_opacity: Option, + /// How opaque a text box's, typewriter text's or callout's text is drawn. + pub text_opacity: Option, } /// The current properties of the comment at `(page, index)`. @@ -239,7 +248,7 @@ pub fn props(doc: &Document, page: usize, index: usize) -> Option { author: text_value(doc, d, b"T").unwrap_or_default(), subject: text_value(doc, d, b"Subj").unwrap_or_default(), color, - opacity: d.get(b"CA").and_then(|o| o.as_f64()).unwrap_or(1.0), + opacity: appearance::part_opacity(d).line, width, icon: (subtype == "Text").then(|| d.name(b"Name").and_then(|n| NoteIcon::from_name(&String::from_utf8_lossy(n))).unwrap_or(NoteIcon::Note)), modified: text_value(doc, d, b"M"), @@ -248,6 +257,10 @@ pub fn props(doc: &Document, page: usize, index: usize) -> Option { fillable: fillable(&subtype), fill: fill_of(d), font_size: (subtype == "FreeText").then(|| appearance::parse_da(d).1), + border_color: (subtype == "FreeText").then(|| appearance::da_border(d)).flatten(), + border_shape: (subtype == "FreeText" && !d.contains(b"CL")).then(|| BorderShape::of(d)), + fill_opacity: fillable(&subtype).then(|| appearance::part_opacity(d).fill), + text_opacity: (subtype == "FreeText").then(|| appearance::part_opacity(d).text), subtype, endings: line_endings_of(d), }) diff --git a/crates/annot/src/lib.rs b/crates/annot/src/lib.rs index 7bd0894c66..52c0aad74f 100644 --- a/crates/annot/src/lib.rs +++ b/crates/annot/src/lib.rs @@ -22,7 +22,9 @@ mod create; mod draw; mod info; mod lookup; +mod model3d; mod modify; +mod threed; mod types; #[cfg(test)] @@ -36,14 +38,17 @@ use lookup::{annots, page_ref}; pub use lookup::{ has_visible_fill_sign, is_fill_sign, isolate_annotation, isolate_annotations, page_refs, picture_rotation, set_annots, signature_image, }; +pub use model3d::{Model3d, Model3dKind, model_3d, model_3d_annotations}; pub(crate) use modify::n; pub use modify::{ - add_reply, add_text_replacement, deletable, delete_annotation, erase_ink, fillable, goes_with, lift_annotation, move_annotation, set_contents, - set_fill, set_font_size, set_locked, set_marked, set_points, set_rect, set_review_state, set_style, with_replies_and_popups, + add_reply, add_text_replacement, deletable, delete_annotation, erase_ink, fillable, goes_with, lift_annotation, move_annotation, + set_border_color, set_border_shape, set_contents, set_fill, set_font_size, set_locked, set_marked, set_part_opacity, set_points, set_rect, + set_review_state, set_style, with_replies_and_popups, }; +pub use threed::{Activation as Activation3d, MAX_MODEL_BYTES, ModelFormat, NewModel3d, PosterPicture, add_3d_model, check_model}; pub use types::{ - AttachIcon, FillMark, LineEnding, Markup, Meta, NOTE_SIZE, NewAnnotation, NoteIcon, OverlayFont, OverlayLook, ReviewState, Rgb, Shape, - StampGroup, StampKind, Style, StyleChange, TextReplacement, rect_quad, + AttachIcon, BorderShape, FillMark, LineEnding, Markup, Meta, NOTE_SIZE, NewAnnotation, NoteIcon, OverlayFont, OverlayLook, ReviewState, Rgb, + Shape, StampGroup, StampKind, Style, StyleChange, TextReplacement, rect_quad, }; pub mod appearance; @@ -70,6 +75,12 @@ pub enum AnnotError { /// with a font that has those characters. The message names the character. #[error("{0}")] Undrawable(String), + /// A 3D model file that is neither U3D (starts with `U3D` and a zero byte) nor PRC (starts `PRC`). + #[error("not a 3D model: the file must be U3D (ECMA-363) or PRC (ISO 14739-1)")] + NotAModel, + /// A 3D model larger than `MAX_MODEL_BYTES`. + #[error("the 3D model is {size} bytes; the largest accepted is {max}")] + ModelTooLarge { size: usize, max: usize }, #[error("{0}")] Cos(#[from] pdfcraft_cos::CosError), } diff --git a/crates/annot/src/model3d.rs b/crates/annot/src/model3d.rs new file mode 100644 index 0000000000..b50342898e --- /dev/null +++ b/crates/annot/src/model3d.rs @@ -0,0 +1,77 @@ +//! Reading the model a 3D annotation shows (ISO 32000-2 §13.6.3 and §13.6.4): its `/3DD` is a +//! 3D stream, or a 3D reference dictionary naming the stream another annotation shares. + +use super::lookup::{annots, page_ref}; +use super::threed::MAX_MODEL_BYTES; +use super::{AnnotError, Dict, Document, Object}; + +/// The format of a 3D stream, from its `/Subtype`. +#[derive(Debug, Clone, Copy, PartialEq, Eq)] +pub enum Model3dKind { + /// Product Representation Compact (ISO 14739-1). + Prc, + /// Universal 3D (ECMA-363). + U3d, +} + +/// A 3D annotation's model: its format and the decoded stream. +#[derive(Debug, Clone, PartialEq)] +pub struct Model3d { + pub kind: Model3dKind, + pub data: Vec, +} + +/// The `/Annots` positions of the 3D annotations on 0-based `page`, in order. +/// +/// # Errors +/// [`AnnotError::NoPageTree`] or [`AnnotError::NoSuchPage`]. +pub fn model_3d_annotations(doc: &Document, page: usize) -> Result, AnnotError> { + let list = annots(doc, page_ref(doc, page)?); + Ok(list.iter().enumerate().filter(|(_, a)| doc.dict(a).is_some_and(|d| d.name(b"Subtype") == Some(b"3D"))).map(|(i, _)| i).collect()) +} + +/// The model of the 3D annotation at `/Annots` position `index` on 0-based `page`, decoded. +/// +/// # Errors +/// [`AnnotError::NoSuchAnnotation`] when there is no annotation there; [`AnnotError::Invalid`] +/// when it is not a 3D annotation, has no readable 3D stream, names a format other than PRC or +/// U3D, or decodes past [`MAX_MODEL_BYTES`]. +pub fn model_3d(doc: &Document, page: usize, index: usize) -> Result { + let list = annots(doc, page_ref(doc, page)?); + let entry = list.get(index).ok_or(AnnotError::NoSuchAnnotation { page, index })?; + let annot = doc.dict(entry).ok_or(AnnotError::NoSuchAnnotation { page, index })?; + let at = || format!("annotation {} on page {}", index.saturating_add(1), page.saturating_add(1)); + if annot.name(b"Subtype") != Some(b"3D") { + return Err(AnnotError::Invalid(format!("{} is not a 3D annotation", at()))); + } + let stream = stream_of(doc, &annot).ok_or_else(|| AnnotError::Invalid(format!("{} has no 3D stream", at())))?; + let Object::Stream(s) = stream.as_ref() else { + return Err(AnnotError::Invalid(format!("{} has no 3D stream", at()))); + }; + let kind = match s.dict.name(b"Subtype") { + Some(b"PRC") => Model3dKind::Prc, + Some(b"U3D") => Model3dKind::U3d, + other => { + let name = other.map_or_else(|| "none".to_owned(), |n| String::from_utf8_lossy(n).into_owned()); + return Err(AnnotError::Invalid(format!("{}'s 3D stream is of an unknown format ({name})", at()))); + } + }; + let data = doc + .with_direct_parms(s) + .decoded_within(MAX_MODEL_BYTES) + .map_err(|e| AnnotError::Invalid(format!("{}'s 3D stream could not be decoded: {e}", at())))?; + Ok(Model3d { kind, data }) +} + +/// What an annotation's `/3DD` names: the 3D stream itself, or the `/3D` entry of a 3D reference +/// dictionary. A reference is followed once; the caller refuses anything that is not a stream. +fn stream_of(doc: &Document, annot: &Dict) -> Option> { + let obj = doc.resolve(annot.get(b"3DD")?); + match obj.as_ref() { + Object::Dict(d) if d.name(b"Type") == Some(b"3DRef") => Some(doc.resolve(d.get(b"3D")?)), + _ => Some(obj), + } +} + +#[cfg(test)] +mod tests; diff --git a/crates/annot/src/model3d/tests.rs b/crates/annot/src/model3d/tests.rs new file mode 100644 index 0000000000..90addb1bb6 --- /dev/null +++ b/crates/annot/src/model3d/tests.rs @@ -0,0 +1,88 @@ +use std::sync::Arc; + +use super::*; + +/// A one-page document whose page has `annots` as its `/Annots`, plus `extra` objects from 4 on. +fn doc(annots: &str, extra: &[Vec]) -> Document { + let mut objs: Vec> = vec![ + b"<< /Type /Catalog /Pages 2 0 R >>".to_vec(), + b"<< /Type /Pages /Kids [3 0 R] /Count 1 /MediaBox [0 0 612 792] >>".to_vec(), + format!("<< /Type /Page /Parent 2 0 R /Annots {annots} >>").into_bytes(), + ]; + objs.extend(extra.iter().cloned()); + let mut out = b"%PDF-1.7\n".to_vec(); + let mut offsets = Vec::new(); + for (i, o) in objs.iter().enumerate() { + offsets.push(out.len()); + out.extend_from_slice(format!("{} 0 obj\n", i + 1).as_bytes()); + out.extend_from_slice(o); + out.extend_from_slice(b"\nendobj\n"); + } + let xref = out.len(); + out.extend_from_slice(format!("xref\n0 {}\n0000000000 65535 f \n", objs.len() + 1).as_bytes()); + for o in offsets { + out.extend_from_slice(format!("{o:010} 00000 n \n").as_bytes()); + } + out.extend_from_slice(format!("trailer\n<< /Size {} /Root 1 0 R >>\nstartxref\n{xref}\n%%EOF\n", objs.len() + 1).as_bytes()); + Document::open(Arc::new(out)).expect("opens") +} + +fn stream(dict: &str, body: &[u8]) -> Vec { + let mut s = format!("<< {dict} /Length {} >>\nstream\n", body.len()).into_bytes(); + s.extend_from_slice(body); + s.extend_from_slice(b"\nendstream"); + s +} + +fn annot(dd: &str) -> Vec { + format!("<< /Type /Annot /Subtype /3D /Rect [0 0 100 100] /3DD {dd} >>").into_bytes() +} + +#[test] +fn a_plain_3d_stream_is_read_through_its_filters() { + let d = doc("[4 0 R]", &[annot("5 0 R"), stream("/Type /3D /Subtype /PRC /Filter /ASCIIHexDecode", b"505243>")]); + assert_eq!(model_3d(&d, 0, 0), Ok(Model3d { kind: Model3dKind::Prc, data: b"PRC".to_vec() })); +} + +#[test] +fn a_3dref_leads_to_the_stream_it_names() { + let d = doc("[4 0 R]", &[annot("<< /Type /3DRef /3D 5 0 R >>"), stream("/Type /3D /Subtype /PRC", b"PRC")]); + assert_eq!(model_3d(&d, 0, 0).map(|m| m.data), Ok(b"PRC".to_vec())); +} + +#[test] +fn a_u3d_stream_is_read_and_says_so() { + let d = doc("[4 0 R]", &[annot("5 0 R"), stream("/Type /3D /Subtype /U3D", b"U3D")]); + assert_eq!(model_3d(&d, 0, 0).map(|m| m.kind), Ok(Model3dKind::U3d)); +} + +#[test] +fn a_3d_annotation_without_a_stream_is_invalid() { + let d = doc("[<< /Type /Annot /Subtype /3D /Rect [0 0 1 1] >>]", &[]); + assert!(matches!(model_3d(&d, 0, 0), Err(AnnotError::Invalid(m)) if m.contains("no 3D stream"))); +} + +#[test] +fn an_unknown_format_is_invalid() { + let d = doc("[4 0 R]", &[annot("5 0 R"), stream("/Type /3D /Subtype /STL", b"x")]); + assert!(matches!(model_3d(&d, 0, 0), Err(AnnotError::Invalid(m)) if m.contains("unknown format (STL)"))); +} + +#[test] +fn a_3dref_is_followed_once_and_only_to_a_stream() { + // A reference naming another reference, or one that names itself, is damage: no stream. + let d = doc("[4 0 R]", &[annot("5 0 R"), b"<< /Type /3DRef /3D 6 0 R >>".to_vec(), b"<< /Type /3DRef /3D 5 0 R >>".to_vec()]); + assert!(matches!(model_3d(&d, 0, 0), Err(AnnotError::Invalid(m)) if m.contains("no 3D stream"))); + // The stream is under `/3D`; a `/3DD` inside the reference is not the standard's key. + let d = doc("[4 0 R]", &[annot("<< /Type /3DRef /3DD 5 0 R >>"), stream("/Type /3D /Subtype /PRC", b"PRC")]); + assert!(matches!(model_3d(&d, 0, 0), Err(AnnotError::Invalid(m)) if m.contains("no 3D stream"))); +} + +#[test] +fn another_annotation_is_not_a_3d_one() { + let d = doc("[<< /Type /Annot /Subtype /Square /Rect [0 0 1 1] >> 4 0 R]", &[annot("5 0 R"), stream("/Subtype /PRC", b"PRC")]); + assert!(matches!(model_3d(&d, 0, 0), Err(AnnotError::Invalid(m)) if m.contains("not a 3D annotation"))); + assert_eq!(model_3d(&d, 0, 2), Err(AnnotError::NoSuchAnnotation { page: 0, index: 2 })); + assert_eq!(model_3d_annotations(&d, 0), Ok(vec![1])); + assert_eq!(model_3d_annotations(&d, 1), Err(AnnotError::NoSuchPage(1))); +} diff --git a/crates/annot/src/modify.rs b/crates/annot/src/modify.rs index a615b8e013..c7d87be090 100644 --- a/crates/annot/src/modify.rs +++ b/crates/annot/src/modify.rs @@ -5,8 +5,8 @@ use super::draw::{draw_appearance, redraw_or_restore, set_appearance}; use super::info::{line_endings_of, text_value}; use super::lookup::{annot_dict, annot_ref, annotation_flags, annots, page_ref, set_annots}; use super::types::{ - FLAG_HIDDEN, FLAG_LOCKED, FLAG_NO_ROTATE, FLAG_NO_ZOOM, FLAG_PRINT, LineEnding, Markup, Meta, NewAnnotation, ReviewState, Rgb, Shape, Style, - StyleChange, TextReplacement, + BorderShape, FLAG_HIDDEN, FLAG_LOCKED, FLAG_NO_ROTATE, FLAG_NO_ZOOM, FLAG_PRINT, LineEnding, Markup, Meta, NewAnnotation, ReviewState, Rgb, + Shape, Style, StyleChange, TextReplacement, }; use super::{AnnotError, Dict, Document, HashMap, HashSet, ObjRef, Object, PdfString, appearance}; @@ -159,10 +159,13 @@ pub fn with_replies_and_popups(doc: &Document, list: &[Object], doomed: impl Int } /// Change a comment's text. Text boxes are redrawn to show it, and their rectangle follows the -/// new text: the wrap width and top edge stay, the height fits the wrapped lines. +/// new text: the wrap width and top edge stay, the height fits the wrapped lines. A circle snug +/// around its text stays snug around the new text, centred where it was. pub fn set_contents(doc: &mut Document, page: usize, index: usize, text: &str, meta: &Meta) -> Result<(), AnnotError> { let (_, r) = annot_ref(doc, page, index)?; - let free_text = annot_dict(doc, r).name(b"Subtype") == Some(b"FreeText"); + let d = annot_dict(doc, r); + let free_text = d.name(b"Subtype") == Some(b"FreeText"); + let snug = free_text && is_snug(&d); doc.update_dict(r, |d| { d.set(b"Contents".to_vec(), PdfString::text(text)); // The rich-text version follows the new plain text (text boxes), or would contradict it. @@ -177,8 +180,16 @@ pub fn set_contents(doc: &mut Document, page: usize, index: usize, text: &str, m // The box grows and shrinks with the text instead of clipping it. A locked box keeps its // text editable (as in Acrobat), so the re-fit ignores the lock. New text is drawn even // where the standard fonts can't draw some of it (marked and reported, see - // `draw_appearance`). - match fitted_box(doc, r, text) { + // `draw_appearance`). A circle snug around its text stays snug around the new text; one + // made larger by hand keeps its size, only growing for more lines. + let fit = if snug { + snug_refit(doc, r) + } else { + let d = annot_dict(doc, r); + let circle = text_box_of(&d).filter(|_| BorderShape::of(&d) == BorderShape::Circle); + fitted_box(doc, r, text).map(|fit| circle.map_or(fit, |old| [fit[0], fit[1].min(old[1]), fit[2], fit[3]])) + }; + match fit { Some(rect) => apply_text_box(doc, r, rect, meta, true)?, None => draw_appearance(doc, r)?, } @@ -197,11 +208,12 @@ fn fitted_box(doc: &Document, r: ObjRef, text: &str) -> Option<[f64; 4]> { let rect = nums(b"Rect").filter(|v| v.len() == 4 && v.iter().all(|x| x.is_finite()))?; let rect = [rect[0], rect[1], rect[2], rect[3]]; // A callout re-fits its text box (`/Rect` inset by `/RD`); the leader line keeps its place. + // So does a text box inside an oval or a circle. let tb = if d.contains(b"CL") { let rd = nums(b"RD").filter(|v| v.len() == 4 && v.iter().all(|x| *x >= 0.0))?; [rect[0] + rd[0], rect[1] + rd[1], rect[2] - rd[2], rect[3] - rd[3]] } else { - rect + text_box_of(&d).unwrap_or(rect) }; let (x0, top) = (tb[0].min(tb[2]), tb[1].max(tb[3])); let w = tb[2] - tb[0]; @@ -388,9 +400,134 @@ pub fn set_rect(doc: &mut Document, page: usize, index: usize, rect: [f64; 4], m if !matches!(subtype.as_str(), "Square" | "Circle" | "FreeText") { return Err(AnnotError::Invalid(format!("{subtype} comments can't be resized"))); } + // An oval's or a circle's frame is resized: its text box follows inside. + let shape = if subtype == "FreeText" { BorderShape::of(&d) } else { BorderShape::Rectangle }; + let rect = match (shape, text_box_of(&d)) { + (BorderShape::Rectangle, _) | (_, None) => rect, + // A circle's text box keeps its proportions and reaches the circle with its longer sides; + // the circle then goes around it again, never closer than its text allows. + (BorderShape::Circle, Some(was)) => { + let (o, w) = (normalize(rect), normalize(was)); + let (cx, cy, r) = ((o[0] + o[2]) / 2.0, (o[1] + o[3]) / 2.0, (o[2] - o[0]).min(o[3] - o[1]) / 2.0); + let (wx, wy) = ((w[2] - w[0]) / 2.0, (w[3] - w[1]) / 2.0); + let k = r / wx.max(wy).max(1e-6); + let (ix, iy) = ((wx * k).max(0.5), (wy * k).max(0.5)); + [cx - ix, cy - iy, cx + ix, cy + iy] + } + (shape, Some(was)) => shape.inside(rect, border_width_of(&d), was), + }; apply_text_box(doc, r, rect, meta, false) } +/// A FreeText's text box: `/Rect` inset by `/RD` for a callout or a text box inside an oval or a +/// circle, else `/Rect`. `None` when they can't be read. +pub(crate) fn text_box_of(d: &Dict) -> Option<[f64; 4]> { + let nums = |key: &[u8]| -> Option> { d.get(key)?.as_array()?.iter().map(|o| o.as_f64()).collect() }; + let rect = normalize(four(&nums(b"Rect")?)?); + if !d.contains(b"CL") && BorderShape::of(d) == BorderShape::Rectangle { + return Some(rect); + } + let rd = appearance::inset(d)?; + let tb = [rect[0] + rd[0], rect[1] + rd[1], rect[2] - rd[2], rect[3] - rd[3]]; + (tb[2] - tb[0] >= 1.0 && tb[3] - tb[1] >= 1.0).then_some(tb) +} + +/// Give text box or typewriter text `d`, whose text box is `tb`, the border `shape` (`width` +/// wide): `/Rect` around it, `/RD` and `/PCBorderShape`, or for a rectangle `/Rect` the text +/// box itself again. +pub(crate) fn shape_text_box(d: &mut Dict, tb: [f64; 4], shape: BorderShape, width: f64) { + // A circle goes around the text as the box lays it out. + let text = (shape == BorderShape::Circle).then(|| text_extent(d, displayed_sides(d, tb).0 - 2.0 * (2.0 + width.max(0.0)))); + let outer = shape.around(tb, width, text); + d.set(b"Rect".to_vec(), num_array(&outer)); + if shape == BorderShape::Rectangle { + d.remove(b"PCBorderShape"); + d.remove(b"RD"); + } else { + d.set(b"PCBorderShape".to_vec(), Object::name(shape.name())); + // A circle touching the text box's sides would put them a rounding error outside it. + let tb = normalize(tb); + let rd = [tb[0] - outer[0], tb[1] - outer[1], outer[2] - tb[2], outer[3] - tb[3]].map(|x| x.max(0.0)); + d.set(b"RD".to_vec(), num_array(&rd)); + } +} + +/// The width of the widest line and the height of the lines of FreeText `d`'s text, as its +/// appearance lays the text out `wrap` wide. +fn text_extent(d: &Dict, wrap: f64) -> [f64; 2] { + let text = d.get(b"Contents").and_then(|o| o.as_string()).map(PdfString::to_text).unwrap_or_default(); + let (_, size) = appearance::parse_da(d); + let lines = appearance::wrap(&text, size, wrap.max(1.0)); + let widest = lines.iter().map(|l| appearance::text_width(l, size)).fold(0.0, f64::max); + [widest, lines.len().max(1) as f64 * size * 1.2] +} + +/// The width and height of FreeText `d`'s text box `tb` as displayed: swapped on a page turned by +/// a quarter, where its text is laid out upright. +fn displayed_sides(d: &Dict, tb: [f64; 4]) -> (f64, f64) { + let tb = normalize(tb); + let (w, h) = (tb[2] - tb[0], tb[3] - tb[1]); + if matches!(appearance::text_turn(d), 90 | 270) { (h, w) } else { (w, h) } +} + +/// FreeText `d`'s text box `tb`, with a border `width` wide, made as large as its text wrapped +/// `wrap` wide needs (its widest line and its lines, inside the padding a box sized to its text +/// has) and centred where it was: a circle goes around this, not around a box wider than its text. +fn snug_text_box(d: &Dict, tb: [f64; 4], width: f64, wrap: f64) -> [f64; 4] { + let tb = normalize(tb); + let pad = 2.0 + width.max(0.0); + let [w, h] = text_extent(d, wrap); + let (hx, hy) = (w / 2.0 + pad + 2.0, h / 2.0 + pad + 1.0); + let (hx, hy) = if matches!(appearance::text_turn(d), 90 | 270) { (hy, hx) } else { (hx, hy) }; + let (cx, cy) = ((tb[0] + tb[2]) / 2.0, (tb[1] + tb[3]) / 2.0); + [cx - hx, cy - hy, cx + hx, cy + hy] +} + +/// Whether FreeText `d` is a circle whose text box is snug around its text ([`snug_text_box`]): +/// it then follows its text as that changes, where one made larger by hand keeps its size. +fn is_snug(d: &Dict) -> bool { + let Some(tb) = text_box_of(d).filter(|_| BorderShape::of(d) == BorderShape::Circle) else { return false }; + let width = border_width_of(d); + let s = snug_text_box(d, tb, width, displayed_sides(d, tb).0 - 2.0 * (2.0 + width)); + let tb = normalize(tb); + ((s[2] - s[0]) - (tb[2] - tb[0])).abs() < 0.5 && ((s[3] - s[1]) - (tb[3] - tb[1])).abs() < 0.5 +} + +/// The text box of a circle snug around its text for the text and size it has now: its lines up to +/// 300 pt wide, centred where it was. +fn snug_refit(doc: &Document, r: ObjRef) -> Option<[f64; 4]> { + let d = annot_dict(doc, r); + let width = border_width_of(&d); + Some(snug_text_box(&d, text_box_of(&d)?, width, 300.0 - 2.0 * (2.0 + width))) +} + +/// Give a text box or typewriter text an oval, a circle or a rectangle for its border, and redraw +/// it. Its text box keeps its place; the oval or circle goes around it. For a circle the text box +/// first shrinks to its text, its lines wrapped as they are. +pub fn set_border_shape(doc: &mut Document, page: usize, index: usize, shape: BorderShape, meta: &Meta) -> Result<(), AnnotError> { + let (_, r) = annot_ref(doc, page, index)?; + unlocked(doc, r)?; + let d = annot_dict(doc, r); + if d.name(b"Subtype") != Some(b"FreeText") || d.contains(b"CL") { + return Err(AnnotError::Invalid("only text boxes and typewriter text have a border shape".into())); + } + let tb = text_box_of(&d).ok_or_else(|| AnnotError::Invalid("the text box can't be read".into()))?; + let tb = match shape { + BorderShape::Circle => { + let width = border_width_of(&d); + snug_text_box(&d, tb, width, displayed_sides(&d, tb).0 - 2.0 * (2.0 + width)) + } + _ => tb, + }; + refuse_foreign_text(doc, &d)?; + let before = d; + doc.update_dict(r, |d| { + shape_text_box(d, tb, shape, border_width_of(d)); + touch(d, meta); + })?; + redraw_or_restore(doc, r, before) +} + /// At most this many points reshape a comment (a drawn polygon has a handful). const MAX_POINTS: usize = 100_000; @@ -503,6 +640,8 @@ fn apply_text_box(doc: &mut Document, r: ObjRef, rect: [f64; 4], meta: &Meta, ne } None => (rect, None), }; + let (shape, width) = (BorderShape::of(&d), border_width_of(&d)); + let free_text = d.name(b"Subtype") == Some(b"FreeText"); doc.update_dict(r, |d| { d.set(b"Rect".to_vec(), num_array(&outer)); match &cl { @@ -510,6 +649,8 @@ fn apply_text_box(doc: &mut Document, r: ObjRef, rect: [f64; 4], meta: &Meta, ne d.set(b"CL".to_vec(), num_array(l)); d.set(b"RD".to_vec(), num_array(&[rect[0] - outer[0], rect[1] - outer[1], outer[2] - rect[2], outer[3] - rect[3]])); } + // An oval or a circle goes around the text box again. + None if free_text => shape_text_box(d, rect, shape, width), None => { d.remove(b"RD"); } @@ -528,6 +669,8 @@ fn apply_text_box(doc: &mut Document, r: ObjRef, rect: [f64; 4], meta: &Meta, ne } } +/// The border width as the appearance draws it: `/BS /W`, else the third number of `/Border` +/// (another app's), else 1. fn border_width_of(d: &Dict) -> f64 { appearance::border_width(d) } @@ -558,7 +701,8 @@ fn ink_rect_for_width(d: &Dict, old_w: f64, w: f64) -> Option<[f64; 4]> { Some([inner[0] - base[0] - new_half, inner[1] - base[1] - new_half, inner[2] + base[2] + new_half, inner[3] + base[3] + new_half]) } -/// Change a comment's colour, opacity, line width and/or line endings, and redraw it. +/// Change a comment's colour, opacity (of the whole comment), line width and/or line endings, and +/// redraw it. pub fn set_style(doc: &mut Document, page: usize, index: usize, change: StyleChange, meta: &Meta) -> Result<(), AnnotError> { let StyleChange { color, opacity, width, endings } = change; let (_, r) = annot_ref(doc, page, index)?; @@ -581,25 +725,34 @@ pub fn set_style(doc: &mut Document, page: usize, index: usize, change: StyleCha } refuse_foreign_text(doc, &d)?; let free_text = subtype == "FreeText"; + // Typewriter text has no box of its own: a border goes around its text, so the box grows + // (or shrinks) by the change on every side and the text keeps its place and its lines. + // So does a text box inside an oval or a circle, which goes around it again. + let shape = if free_text { BorderShape::of(&d) } else { BorderShape::Rectangle }; + let typewriter = free_text && (d.name(b"IT") == Some(b"FreeTextTypeWriter") || shape != BorderShape::Rectangle) && !d.contains(b"CL"); + let regrow = width.filter(|_| typewriter).and_then(|w| { + let by = w.max(0.0) - if d.contains(b"BS") || d.contains(b"Border") { border_width_of(&d) } else { 0.0 }; + let rect = grow(text_box_of(&d)?, by); + (by != 0.0 && finite(&rect) && rect[2] - rect[0] >= 1.0 && rect[3] - rect[1] >= 1.0).then_some((rect, w.max(0.0))) + }); let before = d; doc.update_dict(r, |d| { + if let Some((rect, w)) = regrow { + shape_text_box(d, rect, shape, w); + } if let Some(c) = color { if free_text { let (_, size) = appearance::parse_da(d); - let [r, g, b] = c.map(|x| x.clamp(0.0, 1.0)); - d.set(b"DA".to_vec(), PdfString::literal(format!("{} {} {} rg /Helv {} Tf", n(r), n(g), n(b), n(size)).into_bytes())); + d.set(b"DA".to_vec(), free_text_da(c, appearance::da_border(d), size)); rich_text(d); } else { d.set(b"C".to_vec(), rgb(c)); } } + // The whole comment: its fill and text drop opacities of their own. if let Some(o) = opacity { let o = o.clamp(0.0, 1.0); - if o < 1.0 { - d.set(b"CA".to_vec(), Object::Real(o)); - } else { - d.remove(b"CA"); - } + set_part_keys(d, appearance::PartOpacity { line: o, fill: o, text: o }); } if let Some(w) = width { let old_w = border_width_of(d); @@ -629,6 +782,37 @@ pub fn set_style(doc: &mut Document, page: usize, index: usize, change: StyleCha redraw_or_restore(doc, r, before) } +/// A FreeText's `/DA`: the text colour, the border's own colour when it has one (the stroking +/// colour, `RG`), Helvetica at `size`. +pub(crate) fn free_text_da(text: Rgb, border: Option, size: f64) -> PdfString { + let [r, g, b] = text.map(|x| x.clamp(0.0, 1.0)); + let stroke = border.map(|c| c.map(|x| x.clamp(0.0, 1.0))).map(|[r, g, b]| format!("{} {} {} RG ", n(r), n(g), n(b))).unwrap_or_default(); + PdfString::literal(format!("{} {} {} rg {stroke}/Helv {} Tf", n(r), n(g), n(b), n(size)).into_bytes()) +} + +/// Give a text box, typewriter text or callout a border colour of its own (`None`: the text's +/// again), and redraw it. +pub fn set_border_color(doc: &mut Document, page: usize, index: usize, color: Option, meta: &Meta) -> Result<(), AnnotError> { + let (_, r) = annot_ref(doc, page, index)?; + unlocked(doc, r)?; + let d = annot_dict(doc, r); + if d.name(b"Subtype") != Some(b"FreeText") { + return Err(AnnotError::Invalid("only text boxes, typewriter text and callouts have a border colour".into())); + } + if color.is_some_and(|c| !finite(&c)) { + return Err(AnnotError::Invalid("invalid colour".into())); + } + refuse_foreign_text(doc, &d)?; + let before = d; + doc.update_dict(r, |d| { + let (text, size) = appearance::parse_da(d); + d.set(b"DA".to_vec(), free_text_da(text, color, size)); + rich_text(d); + touch(d, meta); + })?; + redraw_or_restore(doc, r, before) +} + /// Change the font size of a text box, typewriter text or callout (`/DA` keeps its colour; the /// rich-text strings follow) and redraw it. The box grows to hold the wrapped text at the new /// size and never shrinks, so a box the user drew large stays large. Sizes are 1 to 400 pt. @@ -643,26 +827,27 @@ pub fn set_font_size(doc: &mut Document, page: usize, index: usize, size: f64, m return Err(AnnotError::Invalid("font size must be 1\u{2013}400 pt".into())); } refuse_foreign_text(doc, &d)?; + let snug = is_snug(&d); let before = d; doc.update_dict(r, |d| { let (col, _) = appearance::parse_da(d); - d.set(b"DA".to_vec(), PdfString::literal(format!("{} {} {} rg /Helv {} Tf", n(col[0]), n(col[1]), n(col[2]), n(size)).into_bytes())); + d.set(b"DA".to_vec(), free_text_da(col, appearance::da_border(d), size)); rich_text(d); touch(d, meta); })?; let text = text_value(doc, &annot_dict(doc, r), b"Contents").unwrap_or_default(); - let result = match fitted_box(doc, r, &text) { - Some(fit) => { + let result = match (snug, if snug { snug_refit(doc, r) } else { fitted_box(doc, r, &text) }) { + // A circle snug around its text stays snug around it at the new size. + (true, Some(fit)) => apply_text_box(doc, r, fit, meta, false), + (false, Some(fit)) => { let cur = annot_dict(doc, r); - let nums: Vec = - cur.get(b"Rect").and_then(|o| o.as_array()).map(|a| a.iter().filter_map(|x| x.as_f64()).collect()).unwrap_or_default(); // Keep the top and sides; only the bottom edge moves, and only downwards. - match four(&nums).map(normalize) { + match text_box_of(&cur) { Some(old) if !cur.contains(b"CL") => apply_text_box(doc, r, [fit[0], fit[1].min(old[1]), fit[2], fit[3]], meta, false), _ => apply_text_box(doc, r, fit, meta, false), } } - None => redraw_or_restore(doc, r, before.clone()), + (_, None) => redraw_or_restore(doc, r, before.clone()), }; if result.is_err() { doc.set(r, Object::Dict(before)); @@ -707,6 +892,63 @@ pub fn set_fill(doc: &mut Document, page: usize, index: usize, fill: Option redraw_or_restore(doc, r, before) } +/// Draw a rectangle's, oval's, polygon's or text box's fill, or a text box's text, at an +/// opacity of its own (0–1), its line or border keeping the comment's (`/CA`), and redraw it. +/// `None` leaves that part as it is drawn now. The fill goes in `/ca` (PDF 2.0's opacity for +/// what is filled), the text in `/PCTextOpacity`, each only where it differs from the one it +/// would otherwise follow. +pub fn set_part_opacity(doc: &mut Document, page: usize, index: usize, fill: Option, text: Option, meta: &Meta) -> Result<(), AnnotError> { + let (_, r) = annot_ref(doc, page, index)?; + unlocked(doc, r)?; + let d = annot_dict(doc, r); + let subtype = String::from_utf8_lossy(d.name(b"Subtype").unwrap_or_default()).into_owned(); + if fill.is_some() && !fillable(&subtype) { + return Err(AnnotError::Invalid("only rectangles, ovals, polygons and text boxes have a fill of their own".into())); + } + if text.is_some() && subtype != "FreeText" { + return Err(AnnotError::Invalid("only text boxes, typewriter text and callouts have text of their own".into())); + } + if fill.into_iter().chain(text).any(|o| !o.is_finite()) { + return Err(AnnotError::Invalid("invalid opacity".into())); + } + // Check before changing anything: a stale appearance would contradict the new opacity. + if appearance::build(&d).is_none() { + return Err(AnnotError::Unsupported(subtype)); + } + refuse_foreign_text(doc, &d)?; + let before = d; + doc.update_dict(r, |d| { + let mut parts = appearance::part_opacity(d); + if let Some(f) = fill { + parts.fill = f.clamp(0.0, 1.0); + } + if let Some(t) = text { + parts.text = t.clamp(0.0, 1.0); + } + set_part_keys(d, parts); + touch(d, meta); + })?; + redraw_or_restore(doc, r, before) +} + +/// Write a comment's parts' opacities (see [`appearance::PartOpacity`]): the line's as `/CA`, the +/// fill's as `/ca` and the text's as `/PCTextOpacity`, each only where it differs from the one +/// it would otherwise follow (the fill the line, the text the fill), so a comment drawn at one +/// opacity keeps just `/CA`. +pub(crate) fn set_part_keys(d: &mut Dict, parts: appearance::PartOpacity) { + let appearance::PartOpacity { line, fill, text } = parts; + let mut put = |key: &[u8], value: f64, unless: f64| { + if value == unless { + d.remove(key); + } else { + d.set(key.to_vec(), Object::Real(value)); + } + }; + put(b"CA", line, 1.0); + put(b"ca", fill, line); + put(b"PCTextOpacity", text, fill); +} + /// Grow a line or polyline's rectangle so a non-`None` ending is not clipped. Idempotent. fn ensure_ending_room(d: &mut Dict) { let subtype = d.name(b"Subtype").unwrap_or_default(); diff --git a/crates/annot/src/tests.rs b/crates/annot/src/tests.rs index 32d28fc6ec..ff7f8925f0 100644 --- a/crates/annot/src/tests.rs +++ b/crates/annot/src/tests.rs @@ -2,6 +2,7 @@ use std::sync::Arc; use pdfcraft_cos::{Document, ObjRef, Object, SaveOptions, write_incremental}; +use super::modify::text_box_of; use super::*; /// Two pages. Page 1 has an inline Square annotation in a direct /Annots array; page 2 has an @@ -40,7 +41,7 @@ fn new(page: usize, shape: Shape) -> NewAnnotation { NewAnnotation { page, shape, - style: Style { color: [0.0, 0.4, 1.0], opacity: 1.0, width: 2.0, fill: None }, + style: Style { color: [0.0, 0.4, 1.0], opacity: 1.0, width: 2.0, fill: None, ..Style::default() }, contents: "hello".into(), author: "Ada".into(), } @@ -356,6 +357,296 @@ fn restyle_and_resize_redraw_the_appearance() { assert!(matches!(set_rect(&mut doc, 0, l, [0.0, 0.0, 5.0, 5.0], &meta("")), Err(AnnotError::Invalid(_)))); } +/// Typed text or a text box takes an oval or a circle for its border: the text box keeps its +/// place and the shape goes around it (`/Rect` grows, `/RD` holds the text box, `/PCBorderShape` +/// the shape); back to a rectangle, it is the text box it was. A resized frame takes its text box +/// along, a thicker border leaves it where it is. +#[test] +fn text_boxes_take_an_oval_or_a_circle_around_their_text() { + let close = |a: &[f64], b: &[f64]| a.len() == b.len() && a.iter().zip(b).all(|(x, y)| (x - y).abs() < 0.01); + let nums = |d: &Dict, k: &[u8]| -> Vec { + d.get(k).and_then(|o| o.as_array()).map(|a| a.iter().filter_map(|x| x.as_f64()).collect()).unwrap_or_default() + }; + // Where the text starts: its first line's text matrix. + let text_at = |doc: &Document, d: &Dict| ap_content(doc, d).lines().find(|l| l.ends_with(" Tm") || l.contains(" Tm ")).map(str::to_owned); + let mut doc = fixture(); + let mut tw = new(0, Shape::Typewriter { rect: [100.0, 100.0, 200.0, 120.0], font_size: 10.0 }); + tw.style.width = 1.0; + let t = add_annotation(&mut doc, &tw, &meta("t")).unwrap(); + // In an oval or a circle the text is centred in its text box (left-aligned in a box): one + // line of "hello" at 10 pt in the 100 × 20 text box centred on (150, 110), 3 pt from its top. + let centred = format!("1 0 0 1 {} {} Tm (hello) Tj", n(150.0 - appearance::text_width("hello", 10.0) / 2.0), n(107.0)); + assert!(text_at(&doc, &list(&doc, 0)[t]).is_some_and(|l| l.starts_with("1 0 0 1 103 ")), "a box: left-aligned"); + set_border_shape(&mut doc, 0, t, BorderShape::Oval, &meta("")).unwrap(); + let doc2 = reopen(&doc); + let d = &list(&doc2, 0)[t]; + // Around the 100 × 20 text box, centred: √2 its size, two points more for the 1 pt border. + let (ax, ay) = (50.0 * std::f64::consts::SQRT_2 + 2.0, 10.0 * std::f64::consts::SQRT_2 + 2.0); + assert!(close(&rect(d), &[150.0 - ax, 110.0 - ay, 150.0 + ax, 110.0 + ay]), "{:?}", rect(d)); + assert!(close(&nums(d, b"RD"), &[ax - 50.0, ay - 10.0, ax - 50.0, ay - 10.0]), "{:?}", nums(d, b"RD")); + assert_eq!(d.name(b"PCBorderShape"), Some(&b"Oval"[..])); + let ap = ap_content(&doc2, d); + assert!(ap.contains(" c\n") && ap.contains("S\n") && !ap.contains("re S"), "an oval, not a box: {ap}"); + assert_eq!(text_at(&doc2, d).as_deref(), Some(centred.as_str()), "the text keeps its place"); + assert_eq!(props(&doc2, 0, t).unwrap().border_shape, Some(BorderShape::Oval)); + // A circle around the text: the text box shrinks to "hello" with its padding, centred where it + // was, and the circle touches its sides (the text a point clear of the border inside). + set_border_shape(&mut doc, 0, t, BorderShape::Circle, &meta("")).unwrap(); + let d = &list(&doc, 0)[t]; + let hw = appearance::text_width("hello", 10.0) / 2.0; + let r = (hw + 5.0).max(hw.hypot(6.0) + 2.0); + assert!(close(&rect(d), &[150.0 - r, 110.0 - r, 150.0 + r, 110.0 + r]), "{:?}", rect(d)); + assert!(close(&nums(d, b"RD"), &[r - hw - 5.0, r - 10.0, r - hw - 5.0, r - 10.0]), "{:?}", nums(d, b"RD")); + assert_eq!(text_at(&doc, d).as_deref(), Some(centred.as_str())); + // A thicker border: the typed text's box grows by the change on every side, so its text keeps + // its place inside the thicker line (as with a rectangle), and the circle goes around that. + set_style(&mut doc, 0, t, StyleChange::new(None, None, Some(3.0), None), &meta("")).unwrap(); + let d = &list(&doc, 0)[t]; + assert_eq!(text_at(&doc, d).as_deref(), Some(centred.as_str())); + let r3 = (hw + 7.0).max(hw.hypot(6.0) + 4.0); + assert!(close(&rect(d), &[150.0 - r3, 110.0 - r3, 150.0 + r3, 110.0 + r3]), "{:?}", rect(d)); + // Its frame resized: the text box follows inside, keeping its proportions, its longer sides + // reaching the circle. + set_rect(&mut doc, 0, t, [150.0 - 2.0 * r3, 110.0 - 2.0 * r3, 150.0 + 2.0 * r3, 110.0 + 2.0 * r3], &meta("")).unwrap(); + let d = &list(&doc, 0)[t]; + let rd = nums(d, b"RD"); + let tb = [rect(d)[0] + rd[0], rect(d)[1] + rd[1], rect(d)[2] - rd[2], rect(d)[3] - rd[3]]; + assert!(((tb[2] - tb[0]) / (tb[3] - tb[1]) - (hw + 7.0) / 12.0).abs() < 0.01, "the text box keeps its proportions: {tb:?}"); + assert!(close(&rect(d), &[150.0 - 2.0 * r3, 110.0 - 2.0 * r3, 150.0 + 2.0 * r3, 110.0 + 2.0 * r3]), "{:?}", rect(d)); + // A rectangle again: the text box it has, nothing around it. + set_border_shape(&mut doc, 0, t, BorderShape::Rectangle, &meta("")).unwrap(); + let d = &list(&doc, 0)[t]; + assert!(close(&rect(d), &tb), "{:?} vs {tb:?}", rect(d)); + assert!(!d.contains(b"RD") && !d.contains(b"PCBorderShape")); + assert!(ap_content(&doc, d).contains("re S")); + + // A filled text box made with an oval: the background takes the shape too. + let mut tb = new(0, Shape::TextBox { rect: [300.0, 500.0, 400.0, 540.0], font_size: 12.0 }); + (tb.style.fill, tb.style.width, tb.style.border_shape) = (Some([1.0, 1.0, 0.6]), 1.0, BorderShape::Oval); + let b = add_annotation(&mut doc, &tb, &meta("b")).unwrap(); + let d = &list(&doc, 0)[b]; + assert_eq!(d.name(b"PCBorderShape"), Some(&b"Oval"[..])); + let ap = ap_content(&doc, d); + assert!(ap.contains("1 1 0.6 rg") && ap.contains(" c\nh\nf\n") && !ap.contains(" re f"), "{ap}"); + // A callout's border stays a rectangle. + let c = add_annotation( + &mut doc, + &new( + 0, + Shape::Callout { + rect: [300.0, 300.0, 400.0, 340.0], + knee: [280.0, 320.0], + point: [250.0, 280.0], + ending: LineEnding::OpenArrow, + font_size: 10.0, + }, + ), + &meta("c"), + ) + .unwrap(); + assert!(matches!(set_border_shape(&mut doc, 0, c, BorderShape::Oval, &meta("")), Err(AnnotError::Invalid(_)))); + assert_eq!(props(&doc, 0, c).unwrap().border_shape, None); + assert_eq!(BorderShape::parse("CIRCLE"), Some(BorderShape::Circle)); +} + +/// A circle goes snug around short text, as a number on a drawing is circled: the text box is no +/// wider than its text needs and the circle touches its sides, where one through the box's corners +/// stood well clear. It follows the text and its size as they change; one made larger by hand +/// keeps its size. +#[test] +fn a_circle_goes_snug_around_its_text_and_follows_it() { + let close = |a: (f64, f64), b: (f64, f64)| (a.0 - b.0).abs() < 0.01 && (a.1 - b.1).abs() < 0.01; + let mut doc = fixture(); + // "3" at 10 pt with a 1 pt border, in the box Fill & Sign sizes to it: the digit with 2 pt of + // padding and 4 pt to spare across, a line and 6 pt down, the border around that. + let w3 = appearance::text_width("3", 10.0); + let mut typed = new(0, Shape::Typewriter { rect: [100.0, 680.0, 110.0 + w3, 700.0], font_size: 10.0 }); + (typed.contents, typed.style.width, typed.style.border_shape) = ("3".into(), 1.0, BorderShape::Circle); + let t = add_annotation(&mut doc, &typed, &meta("t")).unwrap(); + let frame = |doc: &Document| { + let r = rect(&list(doc, 0)[t]); + ((r[2] - r[0], r[3] - r[1]), ((r[0] + r[2]) / 2.0, (r[1] + r[3]) / 2.0)) + }; + let centre = (105.0 + w3 / 2.0, 690.0); + // As tall as the box, 20 pt across: through the corners it was 29. + assert!(close(frame(&doc).0, (20.0, 20.0)) && close(frame(&doc).1, centre), "{:?}", frame(&doc)); + // The text box Fill & Sign gives a box to type into, 40 pt wide: a circle goes snug around its + // text there too. + let mut wide = typed.clone(); + wide.shape = Shape::Typewriter { rect: [300.0, 680.0, 342.0, 700.0], font_size: 10.0 }; + wide.style.border_shape = BorderShape::Rectangle; + let b = add_annotation(&mut doc, &wide, &meta("b")).unwrap(); + set_border_shape(&mut doc, 0, b, BorderShape::Circle, &meta("")).unwrap(); + let r = rect(&list(&doc, 0)[b]); + assert!(close((r[2] - r[0], r[3] - r[1]), (20.0, 20.0)), "{r:?}"); + // More text: the circle grows with it, about the same centre; less, and it shrinks back. + set_contents(&mut doc, 0, t, "123", &meta("")).unwrap(); + let w = appearance::text_width("123", 10.0) / 2.0; + let d = 2.0 * (w + 5.0).max(10.0).max(w.hypot(6.0) + 2.0); + assert!(close(frame(&doc).0, (d, d)) && close(frame(&doc).1, centre), "{:?}", frame(&doc)); + set_contents(&mut doc, 0, t, "3", &meta("")).unwrap(); + assert!(close(frame(&doc).0, (20.0, 20.0)) && close(frame(&doc).1, centre), "{:?}", frame(&doc)); + // A larger font: a larger circle, still snug. + set_font_size(&mut doc, 0, t, 20.0, &meta("")).unwrap(); + let w = appearance::text_width("3", 20.0) / 2.0; + let d = 2.0 * (w + 5.0).max(16.0).max(w.hypot(12.0) + 2.0); + assert!(close(frame(&doc).0, (d, d)) && close(frame(&doc).1, centre), "{:?}", frame(&doc)); + set_font_size(&mut doc, 0, t, 10.0, &meta("")).unwrap(); + assert!(close(frame(&doc).0, (20.0, 20.0)), "{:?}", frame(&doc)); + // Made larger by hand, it keeps its size as its text changes. + set_rect(&mut doc, 0, t, [centre.0 - 20.0, centre.1 - 20.0, centre.0 + 20.0, centre.1 + 20.0], &meta("")).unwrap(); + assert!(close(frame(&doc).0, (40.0, 40.0)), "{:?}", frame(&doc)); + set_contents(&mut doc, 0, t, "4", &meta("")).unwrap(); + assert!(close(frame(&doc).0, (40.0, 40.0)) && close(frame(&doc).1, centre), "{:?}", frame(&doc)); +} + +/// A circle around a line wider than it is tall touches its text box's sides: `/RD` is 0 there, +/// not a rounding error below it, and a hair below 0 still reads. (A text box that couldn't be +/// read stopped following its text, and its text wrapped in a narrower one.) The circle keeps its +/// centre as its text changes. +#[test] +fn a_circle_touching_its_text_box_keeps_following_its_text() { + let mut doc = fixture(); + // Sized to its text: 2 pt of padding and half a point to spare across, a line down. + let (w, h) = (appearance::text_width("Testtext", 12.0) + 4.5, 12.0 * 1.11 + 4.0); + let mut b = new(0, Shape::TextBox { rect: [210.79, 700.0 - h, 210.79 + w, 700.0], font_size: 12.0 }); + (b.contents, b.style.width) = ("Testtext".into(), 0.0); + let t = add_annotation(&mut doc, &b, &meta("t")).unwrap(); + // Circle, then a 1 pt border, which the circle grows by. + set_border_shape(&mut doc, 0, t, BorderShape::Circle, &meta("")).unwrap(); + set_style(&mut doc, 0, t, StyleChange::new(None, None, Some(1.0), None), &meta("")).unwrap(); + let lines = |doc: &Document| ap_content(doc, &list(doc, 0)[t]).matches(" Tj").count(); + let frame = |doc: &Document| { + let r = rect(&list(doc, 0)[t]); + ((r[2] - r[0], r[3] - r[1]), ((r[0] + r[2]) / 2.0, (r[1] + r[3]) / 2.0)) + }; + let rd = |doc: &Document| -> Vec { list(doc, 0)[t].get(b"RD").unwrap().as_array().unwrap().iter().map(|o| o.as_f64().unwrap()).collect() }; + assert!(rd(&doc).iter().all(|x| *x >= 0.0) && rd(&doc)[0] < 0.01, "touching the sides: {:?}", rd(&doc)); + assert_eq!(lines(&doc), 1, "Testtext on one line"); + let (wide, centre) = frame(&doc); + set_contents(&mut doc, 0, t, "KRS", &meta("")).unwrap(); + let (small, at) = frame(&doc); + assert!(small.0 < wide.0 - 10.0 && (small.0 - small.1).abs() < 0.01, "a smaller circle: {small:?} from {wide:?}"); + assert!((at.0 - centre.0).abs() < 0.01 && (at.1 - centre.1).abs() < 0.01, "about its centre: {at:?} {centre:?}"); + assert_eq!(lines(&doc), 1); + // A hair below 0, as rounding left it: still read, so the circle follows its text again. + let (_, r) = annot_ref(&mut doc, 0, t).unwrap(); + let mut was = rd(&doc); + was[0] = -2.8e-14; + doc.update_dict(r, |d| d.set(b"RD".to_vec(), num_array(&was))).unwrap(); + assert!(text_box_of(&list(&doc, 0)[t]).is_some()); + set_contents(&mut doc, 0, t, "Testtext", &meta("")).unwrap(); + assert!((frame(&doc).0.0 - wide.0).abs() < 0.01 && lines(&doc) == 1, "{:?}", frame(&doc)); + assert!(rd(&doc).iter().all(|x| *x >= 0.0), "{:?}", rd(&doc)); +} + +/// On a page turned a quarter, typed text is laid out as the page is shown (`/PCTextRotate`), and a +/// circle made with it goes around the text as shown: as large as around the same text on an +/// upright page. (Made before the turn was set, its text wrapped at the box's displayed height and +/// the circle went around several lines.) +#[test] +fn a_circle_made_on_a_turned_page_is_sized_as_on_an_upright_one() { + let w = appearance::text_width("12345", 10.0); + let typed = |rect: [f64; 4]| { + let mut t = new(0, Shape::Typewriter { rect, font_size: 10.0 }); + (t.contents, t.style.width, t.style.border_shape) = ("12345".into(), 1.0, BorderShape::Circle); + t + }; + let size = |doc: &Document, i: usize| { + let r = rect(&list(doc, 0)[i]); + (r[2] - r[0], r[3] - r[1]) + }; + let mut upright = fixture(); + let u = add_annotation(&mut upright, &typed([100.0, 680.0, 110.0 + w, 700.0]), &meta("u")).unwrap(); + for turn in [90, 270] { + let mut doc = fixture(); + let page = page_refs(&doc).unwrap()[0]; + doc.update_dict(page, |d| d.set(b"Rotate".to_vec(), Object::Int(turn))).unwrap(); + // The same box as shown: its sides swapped in user space. + let t = add_annotation(&mut doc, &typed([100.0, 680.0, 120.0, 690.0 + w]), &meta("t")).unwrap(); + assert_eq!(list(&doc, 0)[t].get(b"PCTextRotate").and_then(Object::as_int), Some(turn)); + let (turned, straight) = (size(&doc, t), size(&upright, u)); + assert!((turned.0 - straight.0).abs() < 0.01 && (turned.1 - straight.1).abs() < 0.01, "{turn}°: {turned:?}, upright {straight:?}"); + } +} + +/// A rectangle's fill, and a text box's border, background and text, each drawn at an opacity of +/// its own: the line's in `/CA`, the fill's in `/ca` (PDF 2.0), the text's in `/PCTextOpacity`, and +/// in the appearance `/GS0` (strokes at the line's, fills at the fill's) and `/GS1` for the text. +#[test] +fn parts_are_drawn_at_opacities_of_their_own() { + let num = |d: &Dict, k: &[u8]| d.get(k).and_then(|o| o.as_f64()); + let gstate = |doc: &Document, d: &Dict, name: &[u8]| -> Option<(f64, f64)> { + let n = d.get(b"AP").map(|a| doc.resolve(a)).and_then(|a| a.as_dict().and_then(|a| a.reference(b"N")))?; + let obj = doc.get(n); + let Object::Stream(s) = &*obj else { return None }; + let res = s.dict.get(b"Resources").map(|r| doc.resolve(r))?; + let gs = res.as_dict()?.get(b"ExtGState").map(|g| doc.resolve(g))?; + let gs = gs.as_dict()?.get(name).map(|g| doc.resolve(g))?; + let gs = gs.as_dict()?; + Some((gs.get(b"CA")?.as_f64()?, gs.get(b"ca")?.as_f64()?)) + }; + let mut doc = fixture(); + let r = add_annotation(&mut doc, &new(0, Shape::Rectangle { rect: [10.0, 10.0, 110.0, 60.0] }), &meta("r")).unwrap(); + set_fill(&mut doc, 0, r, Some([1.0, 1.0, 0.0]), &meta("")).unwrap(); + set_part_opacity(&mut doc, 0, r, Some(0.3), None, &meta("")).unwrap(); + let doc2 = reopen(&doc); + let d = &list(&doc2, 0)[r]; + assert_eq!((num(d, b"CA"), num(d, b"ca")), (None, Some(0.3)), "an opaque line, a fill of its own"); + assert_eq!(gstate(&doc2, d, b"GS0"), Some((1.0, 0.3))); + let ap = ap_content(&doc2, d); + assert!(ap.starts_with("/GS0 gs") && ap.contains("\nB\n"), "filled and stroked in one go: {ap}"); + let p = props(&doc2, 0, r).unwrap(); + assert_eq!((p.opacity, p.fill_opacity, p.text_opacity), (1.0, Some(0.3), None)); + // The whole comment's opacity (StyleAnnotation, the quick bar): the fill goes along again. + set_style(&mut doc, 0, r, StyleChange::new(None, Some(0.6), None, None), &meta("")).unwrap(); + let d = &list(&doc, 0)[r]; + assert_eq!((num(d, b"CA"), num(d, b"ca")), (Some(0.6), None)); + assert_eq!(props(&doc, 0, r).unwrap().fill_opacity, Some(0.6)); + + // A text box with its border opaque, its background at 20 % and its text opaque, as made. + let mut tb = new(0, Shape::TextBox { rect: [200.0, 600.0, 400.0, 650.0], font_size: 12.0 }); + tb.style.fill = Some([1.0, 1.0, 0.6]); + tb.style.width = 1.0; + (tb.style.fill_opacity, tb.style.text_opacity) = (Some(0.2), Some(1.0)); + let t = add_annotation(&mut doc, &tb, &meta("t")).unwrap(); + let doc2 = reopen(&doc); + let d = &list(&doc2, 0)[t]; + assert_eq!((num(d, b"CA"), num(d, b"ca"), num(d, b"PCTextOpacity")), (None, Some(0.2), Some(1.0))); + assert_eq!((gstate(&doc2, d, b"GS0"), gstate(&doc2, d, b"GS1")), (Some((1.0, 0.2)), Some((1.0, 1.0)))); + let ap = ap_content(&doc2, d); + let (gs1, text) = (ap.find("/GS1 gs\n").expect("the text's own state, an operator of its own"), ap.find("BT").expect("text")); + assert!(ap.find(" re f").is_some_and(|fill| fill < gs1) && gs1 < text, "background at 20 %, then the text at its own: {ap}"); + // A new fill opacity leaves the text as it is drawn. + set_part_opacity(&mut doc, 0, t, Some(0.5), None, &meta("")).unwrap(); + let d = &list(&doc, 0)[t]; + assert_eq!((num(d, b"ca"), num(d, b"PCTextOpacity")), (Some(0.5), Some(1.0))); + let p = props(&doc, 0, t).unwrap(); + assert_eq!((p.opacity, p.fill_opacity, p.text_opacity), (1.0, Some(0.5), Some(1.0))); + // All at one opacity again: only /CA stays. + set_part_opacity(&mut doc, 0, t, Some(1.0), Some(1.0), &meta("")).unwrap(); + let d = &list(&doc, 0)[t]; + assert_eq!((num(d, b"CA"), num(d, b"ca"), num(d, b"PCTextOpacity")), (None, None, None)); + + // Only the parts a comment has, and finite values; nothing changes when refused. + let l = add_annotation( + &mut doc, + &new(0, Shape::Line { from: [0.0, 0.0], to: [10.0, 10.0], start: LineEnding::None, end: LineEnding::None }), + &meta("l"), + ) + .unwrap(); + assert!(matches!(set_part_opacity(&mut doc, 0, l, Some(0.5), None, &meta("")), Err(AnnotError::Invalid(_))), "a line has no fill"); + assert!(matches!(set_part_opacity(&mut doc, 0, r, None, Some(0.5), &meta("")), Err(AnnotError::Invalid(_))), "a rectangle has no text"); + let before = list(&doc, 0)[r].clone(); + assert!(matches!(set_part_opacity(&mut doc, 0, r, Some(f64::NAN), None, &meta("")), Err(AnnotError::Invalid(_)))); + assert_eq!(list(&doc, 0)[r], before); + // Another app's comment with /CA alone is drawn at it throughout, as before. + assert_eq!( + appearance::part_opacity(&Dict::from_iter([(b"CA".to_vec(), Object::Real(0.4))])), + appearance::PartOpacity { line: 0.4, fill: 0.4, text: 0.4 } + ); +} + #[test] fn shapes_can_be_filled_and_unfilled() { let mut doc = fixture(); @@ -591,6 +882,85 @@ fn fill_and_sign_items_are_drawn() { assert!(set_style(&mut doc, 1, 0, StyleChange::new(Some([0.0; 3]), None, None, None), &meta("")).is_err()); } +/// A border chosen for typewriter text goes around it in the text's colour. Changing it grows or +/// shrinks the box on every side, so the text stays where it was typed; a text box keeps its box. +#[test] +fn a_typewriter_border_goes_around_its_text_which_stays_put() { + let mut doc = fixture(); + let shape = Shape::Typewriter { rect: [100.0, 700.0, 200.0, 716.0], font_size: 11.0 }; + let style = Style { width: 1.0, ..Style::default_for(&shape) }; + let t = add_annotation(&mut doc, &NewAnnotation { page: 1, shape, style, contents: "Ada".into(), author: "Ada".into() }, &meta("x")).unwrap(); + // The text starts 2 pt inside the border: 100 + 1 + 2 across, 716 - 1 - 2 - 0.9 em down. + let at = "1 0 0 1 103 703.1 Tm"; + let d = &list(&doc, 1)[t]; + let ap = ap_content(&doc, d); + assert!(ap.contains("1 w\n") && ap.contains(" re S"), "a 1 pt border: {ap}"); + assert!(ap.contains(at), "{ap}"); + for (width, expected) in [(2.0, [99.0, 699.0, 201.0, 717.0]), (0.0, [101.0, 701.0, 199.0, 715.0])] { + set_style(&mut doc, 1, t, StyleChange::new(None, None, Some(width), None), &meta("")).unwrap(); + let d = &list(&doc, 1)[t]; + assert_eq!(rect(d), expected, "{width} pt border"); + let ap = ap_content(&doc, d); + assert!(ap.contains(at), "{width} pt: the text keeps its place: {ap}"); + assert_eq!(ap.contains(" re S"), width > 0.0, "{width} pt: {ap}"); + } + let shape = Shape::TextBox { rect: [100.0, 600.0, 200.0, 650.0], font_size: 11.0 }; + let b = add_annotation(&mut doc, &new(1, shape), &meta("b")).unwrap(); + set_style(&mut doc, 1, b, StyleChange::new(None, None, Some(3.0), None), &meta("")).unwrap(); + assert_eq!(rect(&list(&doc, 1)[b]), [100.0, 600.0, 200.0, 650.0], "a text box keeps the box it was drawn with"); + // Another app's typed text with its width in /Border (no /BS): 4 pt, so 2 pt shrinks the box + // by 2 on every side, as it is drawn. + let shape = Shape::Typewriter { rect: [100.0, 500.0, 200.0, 530.0], font_size: 11.0 }; + let o = add_annotation( + &mut doc, + &NewAnnotation { page: 1, style: Style::default_for(&shape), shape: shape.clone(), contents: "Bo".into(), author: "Ada".into() }, + &meta("o"), + ) + .unwrap(); + let (_, r) = annot_ref(&mut doc, 1, o).unwrap(); + doc.update_dict(r, |d| { + d.remove(b"BS"); + d.set(b"Border".to_vec(), num_array(&[0.0, 0.0, 4.0])); + }) + .unwrap(); + set_style(&mut doc, 1, o, StyleChange::new(None, None, Some(2.0), None), &meta("")).unwrap(); + assert_eq!(rect(&list(&doc, 1)[o]), [102.0, 502.0, 198.0, 528.0], "the imported 4 pt border is read from /Border"); + // A border colour that isn't a colour is refused before anything is written. + let mut bad = NewAnnotation { page: 1, style: Style::default_for(&shape), shape, contents: "x".into(), author: "Ada".into() }; + bad.style.border_color = Some([f64::NAN, 0.0, 0.0]); + let before = list(&doc, 1).len(); + assert!(matches!(add_annotation(&mut doc, &bad, &meta("n")), Err(AnnotError::Invalid(_)))); + assert_eq!(list(&doc, 1).len(), before); +} + +/// Text keeps its colour and the border gets its own: the stroking colour in `/DA`, which a new +/// text colour or font size keeps, and which `None` gives back to the text's. +#[test] +fn a_text_border_can_have_its_own_colour() { + let mut doc = fixture(); + let shape = Shape::Typewriter { rect: [100.0, 700.0, 200.0, 716.0], font_size: 11.0 }; + let style = Style { color: [0.0, 0.5, 0.0], width: 1.0, border_color: Some([1.0, 0.0, 0.0]), ..Style::default_for(&shape) }; + let t = add_annotation(&mut doc, &NewAnnotation { page: 1, shape, style, contents: "Ada".into(), author: "Ada".into() }, &meta("x")).unwrap(); + let da = |doc: &Document| text(&list(doc, 1)[t], b"DA"); + assert_eq!(da(&doc), "0 0.5 0 rg 1 0 0 RG /Helv 11 Tf"); + let ap = ap_content(&doc, &list(&doc, 1)[t]); + assert!(ap.contains("1 0 0 RG\n1 w"), "the border in red: {ap}"); + assert!(ap.contains("0 0.5 0 rg"), "the text in green: {ap}"); + assert_eq!(props(&doc, 1, t).unwrap().border_color, Some([1.0, 0.0, 0.0])); + set_style(&mut doc, 1, t, StyleChange::new(Some([0.0, 0.0, 1.0]), None, None, None), &meta("")).unwrap(); + assert_eq!(da(&doc), "0 0 1 rg 1 0 0 RG /Helv 11 Tf", "a new text colour keeps the border's"); + set_font_size(&mut doc, 1, t, 14.0, &meta("")).unwrap(); + assert_eq!(da(&doc), "0 0 1 rg 1 0 0 RG /Helv 14 Tf", "so does a new size"); + set_border_color(&mut doc, 1, t, Some([0.5, 0.5, 0.5]), &meta("")).unwrap(); + assert!(ap_content(&doc, &list(&doc, 1)[t]).contains("0.5 0.5 0.5 RG\n1 w")); + set_border_color(&mut doc, 1, t, None, &meta("")).unwrap(); + assert_eq!(da(&doc), "0 0 1 rg /Helv 14 Tf"); + assert!(ap_content(&doc, &list(&doc, 1)[t]).contains("0 0 1 RG\n1 w"), "the text's colour again"); + assert_eq!(props(&doc, 1, t).unwrap().border_color, None); + let m = add_annotation(&mut doc, &new(0, Shape::Rectangle { rect: [10.0, 10.0, 50.0, 50.0] }), &meta("m")).unwrap(); + assert!(set_border_color(&mut doc, 0, m, Some([1.0, 0.0, 0.0]), &meta("")).is_err(), "a rectangle's colour is its border"); +} + #[test] fn author_subject_and_note_icon_change() { let mut doc = fixture(); @@ -873,8 +1243,8 @@ fn cloud_ys(path: &str) -> Vec { fn replace_text_groups_a_strikeout_and_a_caret() { let mut doc = fixture(); let quads = [[100.0, 720.0, 200.0, 720.0, 100.0, 706.0, 200.0, 706.0]]; - let red = Style { color: [0.89, 0.13, 0.13], opacity: 1.0, width: 1.0, fill: None }; - let blue = Style { color: [0.0, 0.47, 0.84], opacity: 1.0, width: 1.0, fill: None }; + let red = Style { color: [0.89, 0.13, 0.13], opacity: 1.0, width: 1.0, fill: None, ..Style::default() }; + let blue = Style { color: [0.0, 0.47, 0.84], opacity: 1.0, width: 1.0, fill: None, ..Style::default() }; let i = add_text_replacement( &mut doc, TextReplacement { @@ -2251,6 +2621,141 @@ fn freehand_highlight_is_compatible_ink_with_marker_blending() { } } +/// A 3D annotation is PDF 1.6 and a PRC model 1.7 (ISO 32000-2 §13.6): an older document's +/// catalog says so, and a newer one is left as it is. +#[test] +fn a_3d_model_raises_an_older_document_to_its_version() { + let catalog_version = |doc: &Document| { + let root = doc.root().unwrap(); + doc.get(root).as_dict().unwrap().name(b"Version").map(<[u8]>::to_vec) + }; + let add = |doc: &mut Document, model: &[u8]| { + let new = NewModel3d { page: 0, rect: [100.0, 100.0, 300.0, 300.0], model, activation: Activation3d::default(), poster: None, contents: "" }; + add_3d_model(doc, &new, &meta("m")).unwrap(); + }; + let mut doc = fixture(); + add(&mut doc, b"U3D\0rest"); + assert_eq!(catalog_version(&reopen(&doc)), None, "a 1.7 file needs nothing for U3D"); + for (model, version) in [(&b"U3D\0rest"[..], &b"1.6"[..]), (b"PRC rest", b"1.7")] { + let mut doc = fixture(); + doc.set_version("1.4"); + add(&mut doc, model); + assert_eq!(catalog_version(&reopen(&doc)).as_deref(), Some(version)); + } + let mut doc = fixture(); + doc.set_version("2.0"); + add(&mut doc, b"PRC rest"); + assert_eq!(catalog_version(&reopen(&doc)), None, "a 2.0 file takes PRC as it is"); +} + +/// Before 2.0, PRC is an Adobe extension to PDF 1.7 (§7.12): a PRC model declares it, keeping +/// other extensions and a later ADBE level; U3D and a 2.0 file need nothing. +#[test] +fn a_prc_model_declares_the_adobe_extension_before_pdf_2() { + let adbe = |doc: &Document| { + let root = doc.root().unwrap(); + let ext = doc.get(root).as_dict().unwrap().get(b"Extensions").map(|e| doc.resolve(e).as_dict().unwrap().clone()); + ext.map(|e| { + let adbe = e.get(b"ADBE").map(|a| a.as_dict().unwrap().clone()); + (e.get(b"XYZ").is_some(), adbe.map(|a| (a.name(b"BaseVersion").map(<[u8]>::to_vec), a.get(b"ExtensionLevel").and_then(Object::as_int)))) + }) + }; + let add = |doc: &mut Document, model: &[u8]| { + let new = NewModel3d { page: 0, rect: [100.0, 100.0, 300.0, 300.0], model, activation: Activation3d::default(), poster: None, contents: "" }; + add_3d_model(doc, &new, &meta("m")).unwrap(); + }; + let mut doc = fixture(); + add(&mut doc, b"PRC rest"); + assert_eq!(adbe(&reopen(&doc)), Some((false, Some((Some(b"1.7".to_vec()), Some(1)))))); + let mut doc = fixture(); + add(&mut doc, b"U3Drest"); + assert_eq!(adbe(&reopen(&doc)), None, "U3D is in ISO 32000-1"); + let mut doc = fixture(); + doc.set_version("2.0"); + add(&mut doc, b"PRC rest"); + assert_eq!(adbe(&reopen(&doc)), None, "PRC is in ISO 32000-2"); + + let mut doc = fixture(); + let root = doc.root().unwrap(); + let mut level3 = Dict::new(); + level3.set(b"BaseVersion".to_vec(), Object::name("1.7")); + level3.set(b"ExtensionLevel".to_vec(), Object::Int(3)); + let mut ext = Dict::new(); + ext.set(b"ADBE".to_vec(), Object::Dict(level3)); + ext.set(b"XYZ".to_vec(), Object::Dict(Dict::new())); + doc.update_dict(root, |c| c.set(b"Extensions".to_vec(), Object::Dict(ext.clone()))).unwrap(); + add(&mut doc, b"PRC rest"); + assert_eq!(adbe(&reopen(&doc)), Some((true, Some((Some(b"1.7".to_vec()), Some(3))))), "a later level is kept"); + ext.remove(b"ADBE"); + let mut doc = fixture(); + doc.update_dict(root, |c| c.set(b"Extensions".to_vec(), Object::Dict(ext))).unwrap(); + add(&mut doc, b"PRC rest"); + assert_eq!(adbe(&reopen(&doc)), Some((true, Some((Some(b"1.7".to_vec()), Some(1))))), "other extensions are kept"); + + let mut doc = fixture(); + let mut ext = Dict::new(); + ext.set(b"ADBE".to_vec(), Object::Array(vec![Object::Dict(Dict::new())])); + let ext = doc.add(Object::Dict(ext)); + doc.update_dict(root, |c| c.set(b"Extensions".to_vec(), Object::Ref(ext))).unwrap(); + add(&mut doc, b"PRC rest"); + let doc = reopen(&doc); + let ext = doc.resolve(doc.get(doc.root().unwrap()).as_dict().unwrap().get(b"Extensions").unwrap()); + assert!(ext.as_dict().unwrap().get(b"ADBE").unwrap().as_array().is_some(), "an array of ADBE extensions is left alone"); +} + +#[test] +fn a_picture_is_placed_by_its_box_and_a_form_needs_a_positive_size() { + let at = [10.0, 20.0, 110.0, 70.0]; + assert_eq!(appearance::picture_place(at, true, None).as_deref(), Some("100 0 0 50 10 20 cm"), "an image fills its unit square"); + assert_eq!(appearance::picture_place(at, false, Some((200.0, 25.0))).as_deref(), Some("0.5 0 0 2 10 20 cm")); + for size in [None, Some((0.0, 25.0)), Some((200.0, -1.0))] { + assert_eq!(appearance::picture_place(at, false, size), None, "{size:?}"); + } +} + +/// The poster's box must lie in the rectangle, and a form poster needs a size to scale by. +#[test] +fn a_3d_poster_outside_its_rectangle_or_without_a_size_is_refused() { + let picture = |at: [f64; 4], size: (f64, f64)| PosterPicture { xobject: ObjRef { num: 6, generation: 0 }, image: false, size, at }; + for (poster, why) in [ + (picture([50.0, 100.0, 250.0, 300.0], (10.0, 10.0)), "starts left of the rectangle"), + (picture([100.0, 100.0, 300.0, 301.0], (10.0, 10.0)), "ends above it"), + (picture([100.0, 100.0, 300.0, 300.0], (0.0, 10.0)), "has no width"), + (picture([100.0 - 2e-7, 150.0, 100.0 - 1e-7, 200.0], (10.0, 10.0)), "lies outside within the rounding slack"), + ] { + let mut doc = fixture(); + let new = NewModel3d { + page: 0, + rect: [100.0, 100.0, 300.0, 300.0], + model: b"PRC rest", + activation: Activation3d::default(), + poster: Some(poster), + contents: "", + }; + assert!(matches!(add_3d_model(&mut doc, &new, &meta("m")), Err(AnnotError::Invalid(_))), "{why}"); + assert_eq!(list(&doc, 0).len(), 1, "{why}: nothing added"); + } +} + +/// A poster box fitted to the rectangle in floating point ends an ulp outside it: a 7×3 picture +/// centred in [0 100 3.6 110.8] spans -2.2e-16 to 3.6000000000000005. It is taken, inside the +/// rectangle. +#[test] +fn a_3d_poster_fitted_to_its_rectangle_with_rounding_is_taken() { + let mut doc = fixture(); + let at = [-2.220446049250313e-16, 104.62857142857143, 3.6000000000000005, 106.17142857142858]; + let poster = PosterPicture { xobject: ObjRef { num: 6, generation: 0 }, image: false, size: (7.0, 3.0), at }; + let new = NewModel3d { + page: 0, + rect: [0.0, 100.0, 3.6, 110.8], + model: b"PRC rest", + activation: Activation3d::default(), + poster: Some(poster), + contents: "", + }; + assert_eq!(add_3d_model(&mut doc, &new, &meta("m")), Ok(1)); +} + #[test] fn indirect_typewriter_names_keep_outer_whitespace_in_summaries() { let mut doc = fixture(); diff --git a/crates/annot/src/threed.rs b/crates/annot/src/threed.rs new file mode 100644 index 0000000000..cf828388c0 --- /dev/null +++ b/crates/annot/src/threed.rs @@ -0,0 +1,292 @@ +//! 3D annotations (ISO 32000-2 §13.6.2): a U3D or PRC model shown in a rectangle on a page, +//! with a poster (`/AP /N`) for printing, for readers without 3D, and until it is activated. + +use super::create::{finite, normalize, num_array}; +use super::lookup::{annots, page_ref, set_annots}; +use super::types::{FLAG_PRINT, Meta}; +use super::{AnnotError, Dict, Document, ObjRef, Object, PdfString, Stream, appearance, n}; + +/// The largest model accepted, in bytes: well past real U3D/PRC files, and bounded because the +/// model is held and compressed in memory. +pub const MAX_MODEL_BYTES: usize = 256 << 20; + +/// What the poster says when there is no picture. +const PLACEHOLDER_LABEL: &str = "3D model"; + +/// A 3D stream's data format (§13.6.3): ECMA-363 U3D or ISO 14739-1 PRC. +#[derive(Clone, Copy, Debug, PartialEq, Eq)] +pub enum ModelFormat { + U3d, + Prc, +} + +impl ModelFormat { + /// The format a model file's first bytes declare: U3D files start `U3D\0` (ECMA-363 file + /// header block type), PRC files start `PRC`. + /// + /// # Errors + /// + /// [`AnnotError::NotAModel`] for anything else. + pub fn detect(bytes: &[u8]) -> Result { + if bytes.starts_with(b"U3D\0") { + Ok(Self::U3d) + } else if bytes.starts_with(b"PRC") { + Ok(Self::Prc) + } else { + Err(AnnotError::NotAModel) + } + } + + /// The 3D stream's `/Subtype`. + pub fn subtype(self) -> &'static str { + match self { + Self::U3d => "U3D", + Self::Prc => "PRC", + } + } + + /// The PDF version a document holding it needs (§13.6): a 3D annotation is 1.6, a PRC stream + /// 1.7 under Adobe's extension, declared by [`add_3d_model`] (PDF 2.0 takes PRC as standard). + pub fn version(self) -> &'static str { + match self { + Self::U3d => "1.6", + Self::Prc => "1.7", + } + } +} + +/// When a reader turns the poster into the live model (§13.6.2, the activation dictionary's +/// `/A`). The standard's default is [`Self::Click`]. +#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)] +pub enum Activation { + /// `/XA`: when the user clicks it; it prints as its poster. + #[default] + Click, + /// `/PO`: when the page opens. + PageOpen, + /// `/PV`: when the page becomes visible. + PageVisible, +} + +impl Activation { + /// Every activation, with the word a tool argument names it by. + pub const ALL: [(Self, &'static str); 3] = [(Self::Click, "click"), (Self::PageOpen, "page_open"), (Self::PageVisible, "page_visible")]; + + /// The `/A` name. + pub fn name(self) -> &'static str { + match self { + Self::Click => "XA", + Self::PageOpen => "PO", + Self::PageVisible => "PV", + } + } +} + +/// A picture already in the document (an image XObject, or a form XObject whose `/Matrix` maps it +/// to `size` points) drawn into `at`, a box inside the annotation's rectangle. +#[derive(Clone, Copy, Debug, PartialEq)] +pub struct PosterPicture { + pub xobject: ObjRef, + pub image: bool, + pub size: (f64, f64), + pub at: [f64; 4], +} + +/// A 3D model to add on `page` (0-based) in `rect` (user space). +#[derive(Clone, Debug, PartialEq)] +pub struct NewModel3d<'a> { + pub page: usize, + pub rect: [f64; 4], + pub model: &'a [u8], + pub activation: Activation, + /// The poster's picture; `None` draws a labelled frame. + pub poster: Option, + /// `/Contents`: the annotation's alternate description (§12.5.2). + pub contents: &'a str, +} + +/// Check a model before anything is added: its size and its format. +/// +/// # Errors +/// +/// [`AnnotError::ModelTooLarge`] past [`MAX_MODEL_BYTES`], else as [`ModelFormat::detect`]. +pub fn check_model(model: &[u8]) -> Result { + check_model_size(model, MAX_MODEL_BYTES)?; + ModelFormat::detect(model) +} + +fn check_model_size(model: &[u8], max: usize) -> Result<(), AnnotError> { + if model.len() > max { + return Err(AnnotError::ModelTooLarge { size: model.len(), max }); + } + Ok(()) +} + +/// Add a 3D annotation: the model in a Flate-compressed 3D stream referenced from `/3DD` (so the +/// annotation has its own instance, §13.6.2), `/3DA` stating only `/A`, and the poster as `/AP /N`. +/// No view (`/3DV`, `/VA`) is written: readers open on the model's own default view. A document +/// older than [`ModelFormat::version`] is raised to it. Returns its index in the page's `/Annots`. +/// +/// # Errors +/// +/// As [`check_model`]; [`AnnotError::NoSuchPage`]; [`AnnotError::Invalid`] for a rectangle with no +/// area or a poster box outside it. +pub fn add_3d_model(doc: &mut Document, new: &NewModel3d<'_>, meta: &Meta) -> Result { + let format = check_model(new.model)?; + let page = page_ref(doc, new.page)?; + let rect = normalize(new.rect); + if !finite(&new.rect) || rect[2] - rect[0] < 1.0 || rect[3] - rect[1] < 1.0 { + return Err(AnnotError::Invalid("invalid rectangle for the 3D model (each side at least 1 pt)".into())); + } + let poster = match &new.poster { + Some(p) => picture_poster(rect, p)?, + None => placeholder_poster(rect), + }; + let mut stream = Dict::new(); + stream.set(b"Type".to_vec(), Object::name("3D")); + stream.set(b"Subtype".to_vec(), Object::name(format.subtype())); + let stream = doc.add(Object::Stream(Stream::flate(stream, new.model))); + let poster = doc.add(Object::Stream(poster)); + let d = annotation_dict(new, rect, page, stream, poster, meta); + let r = doc.add(Object::Dict(d)); + let mut list = annots(doc, page); + let index = list.len(); + list.push(Object::Ref(r)); + set_annots(doc, page, list)?; + at_least_version(doc, format.version())?; + if format == ModelFormat::Prc && !at_least(doc, "2.0") { + declare_prc_extension(doc)?; + } + Ok(index) +} + +/// Whether the header or the catalog `/Version` says `min` or later (§7.7.2). +fn at_least(doc: &Document, min: &str) -> bool { + let reached = |v: &[u8]| v.len() >= 3 && v >= min.as_bytes(); + let catalog = doc.root().and_then(|root| doc.get(root).as_dict().and_then(|c| c.get(b"Version").cloned())); + let catalog = catalog.and_then(|v| doc.resolve(&v).as_name().map(<[u8]>::to_vec)); + reached(doc.version().as_bytes()) || catalog.as_deref().is_some_and(reached) +} + +/// PRC came before PDF 2.0 as an Adobe extension to 1.7: such a file declares +/// `/Extensions /ADBE` with ExtensionLevel 1 or later (§7.12). Other extensions and an ADBE entry +/// that already has a level are kept; one that is not a dictionary (an array of them) is left alone. +fn declare_prc_extension(doc: &mut Document) -> Result<(), AnnotError> { + let Some(root) = doc.root() else { return Ok(()) }; + let catalog = doc.get(root); + // `None` when present but not a dictionary; an empty one when absent. + let dict_or_new = |o: Option<&Object>| match o.map(|o| doc.resolve(o)) { + None => Some(Dict::new()), + Some(o) => o.as_dict().cloned(), + }; + let Some(mut extensions) = dict_or_new(catalog.as_dict().and_then(|c| c.get(b"Extensions"))) else { return Ok(()) }; + let Some(mut adbe) = dict_or_new(extensions.get(b"ADBE")) else { return Ok(()) }; + if adbe.get(b"ExtensionLevel").map(|l| doc.resolve(l)).and_then(|l| l.as_int()).is_some_and(|l| l >= 1) { + return Ok(()); + } + adbe.set(b"BaseVersion".to_vec(), Object::name("1.7")); + adbe.set(b"ExtensionLevel".to_vec(), Object::Int(1)); + extensions.set(b"ADBE".to_vec(), Object::Dict(adbe)); + doc.update_dict(root, |c| c.set(b"Extensions".to_vec(), Object::Dict(extensions)))?; + Ok(()) +} + +/// A document whose header and catalog `/Version` both say less than `min` gets `/Version min` in +/// its catalog (§7.7.2: the later of the two applies). +fn at_least_version(doc: &mut Document, min: &str) -> Result<(), AnnotError> { + let Some(root) = doc.root() else { return Ok(()) }; + if !at_least(doc, min) { + doc.update_dict(root, |c| c.set(b"Version".to_vec(), Object::name(min)))?; + } + Ok(()) +} + +fn annotation_dict(new: &NewModel3d<'_>, rect: [f64; 4], page: ObjRef, stream: ObjRef, poster: ObjRef, meta: &Meta) -> Dict { + let mut d = Dict::new(); + d.set(b"Type".to_vec(), Object::name("Annot")); + d.set(b"Subtype".to_vec(), Object::name("3D")); + d.set(b"Rect".to_vec(), num_array(&rect)); + d.set(b"P".to_vec(), Object::Ref(page)); + d.set(b"F".to_vec(), Object::Int(FLAG_PRINT)); + if !new.contents.is_empty() { + d.set(b"Contents".to_vec(), PdfString::text(new.contents)); + } + if !meta.id.is_empty() { + d.set(b"NM".to_vec(), PdfString::text(&meta.id)); + } + if let Some(date) = &meta.date { + d.set(b"M".to_vec(), PdfString::literal(date.as_bytes().to_vec())); + } + d.set(b"3DD".to_vec(), Object::Ref(stream)); + let mut activation = Dict::new(); + activation.set(b"A".to_vec(), Object::name(new.activation.name())); + d.set(b"3DA".to_vec(), Object::Dict(activation)); + let mut ap = Dict::new(); + ap.set(b"N".to_vec(), Object::Ref(poster)); + d.set(b"AP".to_vec(), Object::Dict(ap)); + d +} + +/// The poster drawing `p` into its box, which must lie within `rect`. +fn picture_poster(rect: [f64; 4], p: &PosterPicture) -> Result { + let at = normalize(p.at); + // A box fitted to `rect` in floating point can end a few ulps outside it: allow that, then clamp. + let slack = rect.iter().fold(1.0_f64, |m, v| m.max(v.abs())) * 1e-9; + let near = finite(&p.at) && at[0] >= rect[0] - slack && at[1] >= rect[1] - slack && at[2] <= rect[2] + slack && at[3] <= rect[3] + slack; + let at = [at[0].max(rect[0]), at[1].max(rect[1]), at[2].min(rect[2]), at[3].min(rect[3])]; + if !(near && at[2] > at[0] && at[3] > at[1]) { + return Err(AnnotError::Invalid("the poster must fit inside the 3D model's rectangle".into())); + } + let place = appearance::picture_place(at, p.image, Some(p.size)).ok_or_else(|| AnnotError::Invalid("the poster picture has no size".into()))?; + let mut xobjects = Dict::new(); + xobjects.set(b"Pic".to_vec(), Object::Ref(p.xobject)); + let mut resources = Dict::new(); + resources.set(b"XObject".to_vec(), Object::Dict(xobjects)); + Ok(appearance::form(rect, format!("q {place} /Pic Do Q\n").as_bytes(), resources)) +} + +/// A light frame with "3D model" centred in it, shrunk to fit (left out when even 2 pt won't). +fn placeholder_poster(rect: [f64; 4]) -> Stream { + let [x0, y0, x1, y1] = rect; + let (w, h) = (x1 - x0, y1 - y0); + let mut c = + format!("0.94 g {} {} {} {} re f\n0.45 G 1 w {} {} {} {} re S\n", n(x0), n(y0), n(w), n(h), n(x0 + 0.5), n(y0 + 0.5), n(w - 1.0), n(h - 1.0)); + let mut resources = Dict::new(); + let unit = appearance::text_width(PLACEHOLDER_LABEL, 1.0); + let size = if unit > 0.0 { (w * 0.8 / unit).min(h * 0.4).min(14.0) } else { 0.0 }; + if size >= 2.0 { + let (tx, ty) = (x0 + (w - unit * size) / 2.0, y0 + (h - size * 0.7) / 2.0); + c.push_str(&format!("BT 0.3 g /Helv {} Tf {} {} Td ({PLACEHOLDER_LABEL}) Tj ET\n", n(size), n(tx), n(ty))); + resources.set(b"Font".to_vec(), appearance::helvetica_resource()); + } + appearance::form(rect, c.as_bytes(), resources) +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn model_size_is_capped() { + assert_eq!(check_model_size(&[0; 8], 8), Ok(())); + assert_eq!(check_model_size(&[0; 9], 8), Err(AnnotError::ModelTooLarge { size: 9, max: 8 })); + } + + #[test] + fn the_format_is_read_from_the_header() { + assert_eq!(ModelFormat::detect(b"U3D\0rest"), Ok(ModelFormat::U3d)); + assert_eq!(ModelFormat::detect(b"PRC8800"), Ok(ModelFormat::Prc)); + for junk in [&b""[..], b"U3D", b"U3DX", b"PR", b"%PDF-1.7", b"prc"] { + assert_eq!(ModelFormat::detect(junk), Err(AnnotError::NotAModel), "{junk:?}"); + } + } + + #[test] + fn the_placeholder_label_shrinks_to_fit_and_is_dropped_when_it_cannot() { + let wide = String::from_utf8(placeholder_poster([0.0, 0.0, 300.0, 200.0]).decoded().unwrap().to_vec()).unwrap(); + assert!(wide.contains("/Helv 14 Tf") && wide.contains("(3D model) Tj"), "{wide}"); + let tiny = placeholder_poster([0.0, 0.0, 4.0, 4.0]); + assert!(!String::from_utf8(tiny.decoded().unwrap().to_vec()).unwrap().contains("Tj")); + assert!(tiny.dict.get(b"Resources").and_then(Object::as_dict).is_none_or(|r| r.get(b"Font").is_none())); + } +} diff --git a/crates/annot/src/types.rs b/crates/annot/src/types.rs index 522995b9f9..f6bcc68a7e 100644 --- a/crates/annot/src/types.rs +++ b/crates/annot/src/types.rs @@ -1,5 +1,5 @@ use super::create::normalize; -use super::{ObjRef, appearance}; +use super::{Dict, ObjRef, appearance}; pub type Rgb = [f64; 3]; @@ -124,6 +124,28 @@ impl FillMark { let (w, h) = if matches!(rotation.rem_euclid(360), 90 | 270) { (h, w) } else { (w, h) }; [at[0] - w / 2.0, at[1] - h / 2.0, at[0] + w / 2.0, at[1] + h / 2.0] } + + /// The mark whose `/Name` this is. + pub fn from_name(n: &[u8]) -> Option { + [FillMark::Check, FillMark::Cross, FillMark::Dot, FillMark::Line].into_iter().find(|m| m.name().as_bytes() == n) + } + + /// The pen strokes the mark is drawn with in `rect` (user space): what its appearance strokes, + /// and what a copy of it shows while it is being placed. The dot is a filled circle (radius + /// 0.3 × the shorter side, centred) and has none. + pub fn strokes(self, rect: [f64; 4]) -> Vec> { + let [x0, y0, x1, y1] = rect; + let (w, h) = (x1 - x0, y1 - y0); + match self { + FillMark::Check => vec![vec![[x0 + w * 0.15, y0 + h * 0.5], [x0 + w * 0.4, y0 + h * 0.2], [x0 + w * 0.88, y0 + h * 0.85]]], + FillMark::Cross => vec![ + vec![[x0 + w * 0.18, y0 + h * 0.18], [x1 - w * 0.18, y1 - h * 0.18]], + vec![[x0 + w * 0.18, y1 - h * 0.18], [x1 - w * 0.18, y0 + h * 0.18]], + ], + FillMark::Dot => Vec::new(), + FillMark::Line => vec![vec![[x0, y0 + h / 2.0], [x1, y0 + h / 2.0]]], + } + } } /// The stamps of Acrobat's stamp palette (drawn in PdfCraft's own style). @@ -566,6 +588,16 @@ pub struct Style { pub width: f64, /// Interior fill for rectangles and ovals (`/IC`), background for text boxes. pub fill: Option, + /// A text box's, typewriter text's or callout's border colour where it differs from the + /// text's (`None`: the text colour), the stroking colour in its `/DA`. + pub border_color: Option, + /// A text box's or typewriter text's border shape (a callout's is always a rectangle). + pub border_shape: BorderShape, + /// The fill's own opacity, 0–1 (`/ca`), where it differs from `opacity` (which is then the + /// line's or border's); `None`: as `opacity`. + pub fill_opacity: Option, + /// A text box's text's own opacity, 0–1 (`/PCTextOpacity`); `None`: as the fill's. + pub text_opacity: Option, } /// The style properties changed by one annotation edit. `None` leaves a property unchanged. @@ -597,7 +629,102 @@ pub struct TextReplacement<'a> { impl Default for Style { fn default() -> Self { - Self { color: [1.0, 0.82, 0.0], opacity: 1.0, width: 1.0, fill: None } + Self { + color: [1.0, 0.82, 0.0], + opacity: 1.0, + width: 1.0, + fill: None, + fill_opacity: None, + text_opacity: None, + border_color: None, + border_shape: BorderShape::Rectangle, + } + } +} + +/// The shape of a text box's or typewriter text's border (PdfCraft's `/PCBorderShape`, absent for +/// a rectangle). An oval or a circle goes around the text box, which keeps its place: `/Rect` +/// grows to hold it, centred on the text box, and `/RD` says where the text box is inside +/// (§12.5.6.6), as for a callout. Other apps that redraw the comment draw a rectangle. +#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)] +pub enum BorderShape { + #[default] + Rectangle, + Oval, + Circle, +} + +impl BorderShape { + pub const ALL: [BorderShape; 3] = [BorderShape::Rectangle, BorderShape::Oval, BorderShape::Circle]; + + pub fn name(self) -> &'static str { + match self { + BorderShape::Rectangle => "Rectangle", + BorderShape::Oval => "Oval", + BorderShape::Circle => "Circle", + } + } + + /// `rectangle`, `oval` or `circle`, in any case. + pub fn parse(s: &str) -> Option { + Self::ALL.into_iter().find(|b| b.name().eq_ignore_ascii_case(s)) + } + + /// The border shape of FreeText `d`: always a rectangle for a callout. + pub fn of(d: &Dict) -> Self { + if d.contains(b"CL") { + return BorderShape::Rectangle; + } + match d.name(b"PCBorderShape") { + Some(b"Oval") => BorderShape::Oval, + Some(b"Circle") => BorderShape::Circle, + _ => BorderShape::Rectangle, + } + } + + /// The rectangle this shape, with a border `width` wide, takes around text box `tb`, centred on + /// it: an oval through the text box's corners (`√2` its size), a point clear of the border; a + /// circle around the text's lines (`text`: the widest line's width and the lines' height, when + /// known), a point clear of the border and touching the text box's longer sides at least + /// (its corners, padding only, may lie outside), else through the box's corners; the text box + /// itself for a rectangle. + pub fn around(self, tb: [f64; 4], width: f64, text: Option<[f64; 2]>) -> [f64; 4] { + let tb = normalize(tb); + let (cx, cy, hx, hy) = ((tb[0] + tb[2]) / 2.0, (tb[1] + tb[3]) / 2.0, (tb[2] - tb[0]) / 2.0, (tb[3] - tb[1]) / 2.0); + let gap = width.max(0.0) + 1.0; + let (ax, ay) = match self { + BorderShape::Rectangle => return tb, + BorderShape::Oval => (hx * std::f64::consts::SQRT_2 + gap, hy * std::f64::consts::SQRT_2 + gap), + BorderShape::Circle => { + let r = match text { + Some([w, h]) => ((w / 2.0).hypot(h / 2.0) + gap).max(hx).max(hy), + None => hx.hypot(hy) + gap, + }; + (r, r) + } + }; + [cx - ax, cy - ay, cx + ax, cy + ay] + } + + /// The text box inside `outer` for this shape when none is recorded: the inverse of + /// [`Self::around`] for an oval, a box through a circle's corners for a circle, keeping the + /// proportions of `was` (the one it had). + pub(crate) fn inside(self, outer: [f64; 4], width: f64, was: [f64; 4]) -> [f64; 4] { + let o = normalize(outer); + let (cx, cy, hx, hy) = ((o[0] + o[2]) / 2.0, (o[1] + o[3]) / 2.0, (o[2] - o[0]) / 2.0, (o[3] - o[1]) / 2.0); + let gap = width.max(0.0) + 1.0; + let (ix, iy) = match self { + BorderShape::Rectangle => return o, + BorderShape::Oval => ((hx - gap) / std::f64::consts::SQRT_2, (hy - gap) / std::f64::consts::SQRT_2), + BorderShape::Circle => { + let w = normalize(was); + let (wx, wy) = ((w[2] - w[0]) / 2.0, (w[3] - w[1]) / 2.0); + let k = (hx.min(hy) - gap) / wx.hypot(wy).max(1e-6); + (wx * k, wy * k) + } + }; + let (ix, iy) = (ix.max(0.5), iy.max(0.5)); + [cx - ix, cy - iy, cx + ix, cy + iy] } } @@ -615,7 +742,7 @@ impl Style { } Shape::Callout { .. } => ([0.0, 0.0, 0.0], 1.0), Shape::Caret { .. } | Shape::Attachment { .. } => ([0.0, 0.47, 0.84], 1.0), - Shape::FreehandHighlight { .. } => return Self { color: [1.0, 0.94, 0.0], opacity: 0.35, width: 14.0, fill: None }, + Shape::FreehandHighlight { .. } => return Self { color: [1.0, 0.94, 0.0], opacity: 0.35, width: 14.0, ..Self::default() }, Shape::Ink { .. } => ([0.0, 0.4, 0.87], 2.0), Shape::TextBox { .. } | Shape::Typewriter { .. } => ([0.0, 0.0, 0.0], 0.0), Shape::Mark { .. } => ([0.0, 0.0, 0.0], 1.5), @@ -624,9 +751,9 @@ impl Style { Shape::CustomStamp { .. } => ([0.0, 0.0, 0.0], 0.0), Shape::TypedSignature { .. } => ([0.0, 0.0, 0.0], 0.0), // Red outline while marked; a black box once applied. - Shape::Redact { .. } => return Self { color: [0.89, 0.13, 0.13], opacity: 1.0, width: 1.0, fill: Some([0.0, 0.0, 0.0]) }, + Shape::Redact { .. } => return Self { color: [0.89, 0.13, 0.13], fill: Some([0.0, 0.0, 0.0]), ..Self::default() }, }; - Self { color, opacity: 1.0, width, fill: None } + Self { color, width, ..Self::default() } } } diff --git a/crates/automation/Cargo.toml b/crates/automation/Cargo.toml index a6e43de1f0..a825b1fe52 100644 --- a/crates/automation/Cargo.toml +++ b/crates/automation/Cargo.toml @@ -14,6 +14,8 @@ pdfcraft-cos = { workspace = true } pdfcraft-render = { workspace = true } pdfcraft-organize = { workspace = true } pdfcraft-platform = { workspace = true } +pdfcraft-prc = { workspace = true } +pdfcraft-prc-render = { workspace = true } serde_json = { workspace = true } sha2 = "0.11" tempfile = "3.27" diff --git a/crates/automation/README.md b/crates/automation/README.md index 8ffa2d0b3e..db5b988394 100644 --- a/crates/automation/README.md +++ b/crates/automation/README.md @@ -250,6 +250,15 @@ choices and values over 4096 UTF-8 bytes return an error. Example: `form_set_props doc=1 field=agree export_value=Accepted widget=1`. +## Rendering a 3D model + +`model_render` draws a page's 3D model (a 3D annotation's PRC stream, directly or through a 3D +reference) to a PNG, without the viewer. `model` picks the page's Nth 3D model (default 1); +`width` and `height` are 1 to 4096 (default 512); `direction` is the view direction from the eye +(default `[-1, 1, -1]`, with Z up unless the view looks along Z); `perspective` defaults to true. +The model is framed to fill the image and shaded by a headlight on white. U3D models are refused +with an error, and the annotation's own views, lighting and render modes are not applied yet. + ## Viewer navigation history A running PdfCraft instance launched with `--control FILE` exposes its per-tab history through @@ -272,3 +281,26 @@ navigation-history tool; use the opt-in UI control channel to navigate the displ (`#RRGGBB`, default black) for `text` and `date`. The colour is saved in the FreeText appearance and survives reopening. Other mark/signature types reject `color`; their appearance controls are separate. + +## Reviewing OCR headlessly + +`ocr_words {doc, retain_review: true}` reads words without editing the document and returns +`review`, plus pages and word boxes/confidences. Page and word positions are 1-based. +`region` can restrict recognition to one named page, using displayed-page points. Without +`retain_review`, word inspection remains stateless. + +Apply reviewed text with `ocr_review_apply {doc, review, corrections: [{page: 1, word: 1, +text: "Corrected"}]}`. Omitted words keep their text; empty text removes a word. Geometry +comes from the retained recognition, never caller input. The captured document guard rejects +stale/foreign results. Successful application consumes the handle and is one undoable step. +Use `output_mode: "editable", out: "reviewed.pdf"` to write a separate editable document, +or `discard: true` to release the result without applying it. Failed validation preserves the +review for correction or discard. A new retained recognition replaces that document's prior +handle; closing the document releases it. At most eight documents may retain reviews, with +65,536-word/4-MiB input bounds and a 65,536-page cap. + +`ocr_dictionary {path: "ocr-dictionary-en.txt", add: ["Quartz"], remove: ["old"]}` atomically +updates the same one-word-per-UTF-8-line format OCR Verify uses. Omit add/remove to read it. +The explicit path honors `--root`, including symlink confinement; choose the appropriate +per-language file used by the UI. Entries are case-insensitive and duplicate-free. Invalid +or oversized inputs fail before writing (65,536 words, 4 MiB, at most 1,024 bytes per word). diff --git a/crates/automation/src/a11y/ocr.rs b/crates/automation/src/a11y/ocr.rs index 82e521ddd5..2846aa44a6 100644 --- a/crates/automation/src/a11y/ocr.rs +++ b/crates/automation/src/a11y/ocr.rs @@ -1,98 +1,534 @@ use super::{Args, Automation, Result, ToolError, Value, child, failed, json, write_atomic}; impl Automation { - /// Recognition settings from the arguments: `dpi` (clamped to 72–600), `language` (refused - /// unless the models read it) and `skip_text_pages`, each left at its default when absent. + /// The settings every OCR tool shares: dpi, language, the engine/strategy choice, the four + /// preprocessing flags, the skip flags and the output mode — each named exactly as its tool + /// schema says. An unknown engine, strategy or mode is an honest error, never a silent + /// fallback: a tool run must not quietly do something other than what it was asked. pub(crate) fn ocr_settings(&self, a: &Args) -> Result { + use pdfcraft_engine::ocr::{EngineChoice, MergeStrategy, OutputMode}; let mut settings = pdfcraft_engine::ocr::OcrSettings::default(); if let Some(d) = a.opt_num("dpi")? { settings.dpi = d.clamp(72.0, 600.0) as f32; } if let Some(l) = a.opt_str("language")? { if !pdfcraft_engine::ocr::LANGUAGES.iter().any(|x| x.0 == l) { - return Err(ToolError::InvalidArgs(format!("unsupported language {l:?}"))); + return Err(ToolError::InvalidArgs(format!( + "unsupported language {l:?} (supported: {})", + pdfcraft_engine::ocr::LANGUAGES.iter().map(|(c, _)| *c).collect::>().join(", ") + ))); } settings.language = l.into(); } - if let Some(s) = a.opt_bool("skip_text_pages")? { - settings.skip_text_pages = s; + if let Some(e) = a.opt_str("engine")? { + settings.engine = match e.trim().to_ascii_lowercase().as_str() { + "ocrs" => EngineChoice::Ocrs, + "tesseract" => EngineChoice::Tesseract, + "auto" | "automatic" | "ensemble" => EngineChoice::Auto, + other => return Err(ToolError::InvalidArgs(format!("unsupported engine {other:?} (ocrs, tesseract or auto)"))), + }; + } + settings.strategy = match a.opt_str("strategy")? { + Some(s) => match s.trim().to_ascii_lowercase().as_str() { + "primary" | "primary-only" | "primary_only" => MergeStrategy::PrimaryOnly, + "confidence" | "confidence-weighted" | "confidence_weighted" => MergeStrategy::ConfidenceWeighted, + "rover" | "rover-vote" | "rover_vote" | "vote" => MergeStrategy::RoverVote, + other => return Err(ToolError::InvalidArgs(format!("unsupported strategy {other:?} (primary, confidence or rover)"))), + }, + // Unset: the engine choice decides (auto brings in the second engine when the + // first reading is weak; a named engine reads alone). + None => settings.engine.default_strategy(), + }; + if let Some(m) = a.opt_str("output_mode")? { + settings.output_mode = match m.trim().to_ascii_lowercase().as_str() { + "searchable" | "searchable-image" | "searchable_image" => OutputMode::Searchable, + "editable" | "editable-text" | "editable_text" => OutputMode::EditableText, + other => return Err(ToolError::InvalidArgs(format!("unsupported output mode {other:?} (searchable or editable)"))), + }; + } + for (key, flag) in [ + ("auto_rotate", &mut settings.auto_rotate), + ("deskew", &mut settings.deskew), + ("denoise", &mut settings.denoise), + ("binarize", &mut settings.binarize), + ] { + if let Some(on) = a.opt_bool(key)? { + *flag = on; + } + } + for (key, flag) in [ + ("skip_text_pages", &mut settings.skip_text_pages), + ("skip_text_files", &mut settings.skip_text_files), + ("force_ocr", &mut settings.force_ocr), + ] { + if let Some(on) = a.opt_bool(key)? { + *flag = on; + } } Ok(settings) } - /// Recognizes text in each of `paths` and writes the result under the same file name in - /// `folder`, reporting per file its output, word count and skipped pages, or its error. A - /// path outside the root refuses the whole call before any work starts. + /// The optional `region` argument ([x0, y0, x1, y1] in points from the top-left of the + /// displayed page) as the job's pixel crop, at the resolution that page is really read at. + /// A region reads exactly one page. + fn ocr_region( + &self, + a: &Args, + id: pdfcraft_engine::DocId, + page: usize, + settings: &pdfcraft_engine::ocr::OcrSettings, + ) -> Result> { + const BAD: &str = "region must be [x0, y0, x1, y1] in points"; + let Some(v) = a.get("region") else { return Ok(None) }; + let arr = v.as_array().ok_or_else(|| ToolError::InvalidArgs(BAD.into()))?; + if arr.len() != 4 { + return Err(ToolError::InvalidArgs(BAD.into())); + } + let mut pts = [0.0f64; 4]; + for (i, x) in arr.iter().enumerate() { + pts[i] = x.as_f64().filter(|f| f.is_finite()).ok_or_else(|| ToolError::InvalidArgs(BAD.into()))?; + } + let scale = f64::from( + self.session.ocr_job(id, &[page], settings.clone()).ok_or_else(|| failed("the document could not be read"))?.render_scale(page), + ); + let px: Vec = pts.iter().map(|p| scale * p).map(|p| if p.is_finite() { p } else { 0.0 }).map(|p| p as f32).collect(); + Ok(Some([px[0], px[1], px[2], px[3]])) + } + + /// The per-page result entries the OCR tools return: recognised text or why the page was + /// passed over. The text itself is only echoed when the document allows copying; the + /// counts and metadata always are. + fn ocr_pages_json(found: &[pdfcraft_engine::ocr::OcrPage], include_text: bool) -> Vec { + found + .iter() + .map(|p| match &p.skipped { + Some(why) => json!({ "page": p.page + 1, "skipped": why }), + None => { + let mut page = json!({ "page": p.page + 1, "words": p.words.len(), "already_on_page": p.already_on_page }); + // The editable output's substitution warnings (characters no font could + // show) are part of the result: reported per page, never dropped. + if !p.notes.is_empty() { + page["notes"] = json!(p.notes); + } + if include_text { + page["text"] = json!(p.text()); + } + page + } + }) + .collect() + } + pub(crate) fn ocr_recognize_files(&mut self, a: &Args) -> Result { let settings = self.ocr_settings(a)?; let folder = self.resolve(a.str("folder")?, true)?; - // A path that leaves the root refuses the whole call before any work starts, like the - // other tools (the write check passes missing files, which stay per-file errors below). + // A path that leaves the read root refuses the whole call before any work starts; + // a file that is merely missing stays a per-file error below. let paths = a.strs("paths")?; for p in &paths { self.resolve_for(p, crate::file_access::Access::Read, true)?; } - // Two inputs that would write the same output file overwrite each other while both report - // success, so the whole batch is refused here: before the folder is created, anything is - // read, or the OCR engines load. + // Two inputs of one file name (from different folders) would write the same output: + // the second would replace the first while both report success. The whole batch is + // refused up front — before the folder is created, anything is read or the engines + // load — naming the file, the way the default filesystem compares names. let targets: Vec = paths.iter().map(|p| child(&folder, &input_file_name(p))).collect(); let shared = shared_file_names(&targets); if !shared.is_empty() { return Err(ToolError::InvalidArgs(format!( - "more than one input would write to the same output file, so each would replace the one before it. Colliding file names: {}. Give each input a different file name, or run them in separate batches with different folders.", + "several paths share a file name, so their outputs in folder would overwrite each other (rename or run them separately): {}", shared.join(", ") ))); } std::fs::create_dir_all(&folder).map_err(|e| failed(e.to_string()))?; - let ocr = pdfcraft_engine::ocr::engine().map_err(failed)?; + let names: Vec = paths.iter().map(|p| input_file_name(p)).collect(); + // Collision pre-flight: a searchable copy never silently overwrites. Every target is + // checked before the first page is read; a collision aborts the whole batch with the + // names in the error. + let mut collisions = std::collections::BTreeSet::new(); + for name in names { + if child(&folder, &name).exists() { + collisions.insert(name); + } + } + if !collisions.is_empty() { + let mut list: Vec = collisions.iter().take(5).cloned().collect(); + if collisions.len() > list.len() { + list.push(format!("and {} more", collisions.len() - list.len())); + } + return Err(failed(format!( + "the output folder already holds {} (delete or move them, or choose another folder): {}", + if collisions.len() == 1 { "a file of that name" } else { "files of those names" }, + list.join(", ") + ))); + } + // The engines are built (and named if they cannot be) before the first file is read. + let recognizers = pdfcraft_engine::ocr::recognizers(&settings).map_err(failed)?; let mut out = Vec::new(); + let (mut ok, mut skipped_files, mut words) = (0usize, 0usize, 0usize); for p in paths { let name = input_file_name(p); let result = self.resolve(p, false).map_err(|e| e.to_string()).and_then(|src| crate::read_input_bytes(&src).map_err(|e| e.to_string())).and_then( - |bytes| pdfcraft_engine::ocr::recognize_file(&name, std::sync::Arc::new(bytes), None, settings.clone(), &ocr, |_, _| true), + |bytes| { + pdfcraft_engine::ocr::recognize_file(&name, std::sync::Arc::new(bytes), None, settings.clone(), &recognizers, |_, _| true) + }, ); - out.push(match result { + match result { Ok(r) => { + let skipped_pages: Vec = r.pages.iter().filter(|p| p.skipped.is_some()).map(|p| p.page + 1).collect(); + if r.skipped() { + // A file passed over by the skip-files option is its own bucket: its + // original is left alone and nothing is written. + skipped_files += 1; + out.push(json!({ "path": p, "skipped": r.file_skipped, "skipped_pages": skipped_pages })); + continue; + } let target = child(&folder, &name); write_atomic(&target, &r.bytes)?; - let skipped: Vec = r.pages.iter().filter(|p| p.skipped.is_some()).map(|p| p.page + 1).collect(); - json!({ "path": p, "output": target.to_string_lossy(), "words": r.words(), "skipped_pages": skipped }) + ok += 1; + words += r.words(); + let mut entry = json!({ "path": p, "output": target.to_string_lossy(), "words": r.words(), "skipped_pages": skipped_pages }); + // The batch's low-confidence note: at or under the settings' threshold (60 + // by default), on 1-based sorted pages. + let threshold = settings.low_confidence_threshold; + if let Some((count, pages)) = pdfcraft_engine::ocr::low_confidence_pages(&r.pages, threshold) { + entry["low_confidence"] = json!({ "count": count, "pages": pages, "threshold": threshold }); + } + out.push(entry); } - Err(e) => json!({ "path": p, "error": e }), - }); + Err(e) => out.push(json!({ "path": p, "error": e })), + } } - Ok(json!({ "files": out })) + Ok(json!({ "files": out, "recognized": ok, "words": words, "skipped_files": skipped_files })) } - /// Recognizes text on the open document's `pages` (every page when absent) and adds it, - /// reporting the total word count and, per page, why it was skipped or what it found. pub(crate) fn ocr_recognize(&mut self, a: &Args) -> Result { - let settings = self.ocr_settings(a)?; + use pdfcraft_engine::ocr::OutputMode; + let mut settings = self.ocr_settings(a)?; let pages = if a.opt_ints("pages")?.is_some() { self.pages(a, "pages")? } else { Vec::new() }; - let doc = self.doc(a)?; // Recognizing needs only permission to change the document; the words it found are // returned only when copying is allowed too. - let (id, copy) = (doc.id, doc.allows_copying()); - let found = self.session.recognize_text(id, &pages, settings).map_err(failed)?; + let (id, copy) = { + let doc = self.doc(a)?; + (doc.id, doc.allows_copying()) + }; + // The optional region reads exactly one page. + if a.get("region").is_some() { + if pages.len() != 1 { + return Err(ToolError::InvalidArgs("region reads one page: name it in pages".into())); + } + settings.region = self.ocr_region(a, id, pages[0], &settings)?; + } + // The editable output's destination is checked before anything runs: a NEW file, never + // the open source and never an existing file replaced silently. + let out = match settings.output_mode { + OutputMode::Searchable => { + // `out` belongs to the editable output; a searchable run writes into the open + // document, so an `out` here would be silently ignored (and never checked + // against the root). + if a.get("out").is_some() { + return Err(ToolError::InvalidArgs("out is only used with output_mode editable; remove it, or choose editable".into())); + } + None + } + OutputMode::EditableText => { + // The editable output writes the recognized text out of the document: a + // document that forbids copying refuses here, before anything runs. + self.doc_to_copy(a)?; + // A write: the file does not exist yet, and write_atomic makes its folder. + let out = self.resolve(a.str("out")?, true)?; + if let Some(source) = self.session.get(id).and_then(|d| d.path.clone()) + && let (Ok(a), Ok(b)) = (std::fs::canonicalize(&out), std::fs::canonicalize(&source)) + && a == b + { + return Err(failed("out is the source document; the editable output is a new file, so choose another path")); + } + if out.exists() { + return Err(failed(format!( + "out already exists ({}); the editable output is a new file, so choose a path that does not exist yet", + out.to_string_lossy() + ))); + } + Some(out) + } + }; + // Honest up front: the engines are built (and named if they cannot be) before anything + // is read or written. + let recognizers = pdfcraft_engine::ocr::recognizers(&settings).map_err(failed)?; + match out { + None => { + let found = self.session.recognize_text(id, &pages, settings).map_err(failed)?; + let words = found.iter().map(|p| p.words.len()).sum::(); + Ok(json!({ "words": words, "pages": Self::ocr_pages_json(&found, copy) })) + } + Some(out) => { + let job = self.session.ocr_job(id, &pages, settings).ok_or_else(|| failed("the document could not be read"))?; + let found = job.run(&recognizers, |_, _| true); + let bytes = self.session.editable_ocr_bytes(id, &found).map_err(failed)?; + // recognition can take minutes: a file created meanwhile is refused, not replaced + if let Some(folder) = out.parent() { + std::fs::create_dir_all(folder).map_err(failed)?; + } + crate::bates::publish_new(&out, &bytes)?; + Ok(json!({ + "out": out.to_string_lossy(), + "words": found.iter().map(|p| p.words.len()).sum::(), + "pages": Self::ocr_pages_json(&found, copy), + })) + } + } + } + + /// The words of a recognition, for reviewing without the GUI: text, box (points from the + /// top-left of the displayed page), confidence, band, and which words are suspects. + /// Changes nothing; needs the engines (ocr_status). + pub(crate) fn ocr_words(&mut self, a: &Args) -> Result { + let mut settings = self.ocr_settings(a)?; + let pages = if a.opt_ints("pages")?.is_some() { self.pages(a, "pages")? } else { Vec::new() }; + // The word list is the document's text, word by word: refused when the document + // forbids copying it out. + let id = self.doc_to_copy(a)?.id; + if a.get("region").is_some() { + if pages.len() != 1 { + return Err(ToolError::InvalidArgs("region reads one page: name it in pages".into())); + } + settings.region = self.ocr_region(a, id, pages[0], &settings)?; + } + let retain = a.opt_bool("retain_review")?.unwrap_or(false); + if retain && !self.ocr_reviews.contains_key(&id) && self.ocr_reviews.len() >= 8 { + return Err(failed("at most 8 OCR reviews may be retained; discard one first")); + } + let recognizers = pdfcraft_engine::ocr::recognizers(&settings).map_err(failed)?; + let job = self.session.ocr_job(id, &pages, settings.clone()).ok_or_else(|| failed("the document could not be read"))?; + let infos: std::collections::HashMap = + self.session.get(id).map(|d| d.info.pages.clone()).unwrap_or_default().into_iter().enumerate().collect(); + let guard = job.guard.clone(); + let found = job.run(&recognizers, |_, _| true); + if found.len() > 65536 { + return Err(failed("at most 65536 OCR review pages are allowed")); + } + pdfcraft_engine::ocr::validate_ocr_words(found.iter().flat_map(|p| &p.words)).map_err(failed)?; + let band_name = |band: pdfcraft_engine::ocr::ConfidenceBand| match band { + pdfcraft_engine::ocr::ConfidenceBand::High => "high", + pdfcraft_engine::ocr::ConfidenceBand::Medium => "medium", + pdfcraft_engine::ocr::ConfidenceBand::Low => "low", + }; + let mut suspects = 0usize; let list: Vec = found .iter() - .map(|p| match &p.skipped { - Some(why) => json!({ "page": p.page + 1, "skipped": why }), - None if copy => json!({ "page": p.page + 1, "words": p.words.len(), "already_on_page": p.already_on_page, "text": p.text() }), - None => json!({ "page": p.page + 1, "words": p.words.len(), "already_on_page": p.already_on_page }), + .map(|p| { + let info = infos.get(&p.page); + let words: Vec = p + .words + .iter() + .map(|w| { + let suspect = pdfcraft_engine::ocr::is_suspect(w.confidence); + if suspect { + suspects += 1; + } + // The word's corners in view space, as the box [x, y, w, h] in points + // from the top-left of the displayed page. + let view = |pt: [f64; 2]| info.map(|i| i.user_to_view(pt[0] as f32, pt[1] as f32)); + let corners = [ + view(w.origin), + view([w.origin[0] + w.across[0], w.origin[1] + w.across[1]]), + view([w.origin[0] + w.up[0], w.origin[1] + w.up[1]]), + view([w.origin[0] + w.across[0] + w.up[0], w.origin[1] + w.across[1] + w.up[1]]), + ]; + let xs: Vec = corners.iter().flatten().map(|v| v[0]).collect(); + let ys: Vec = corners.iter().flatten().map(|v| v[1]).collect(); + let (min_x, min_y) = (xs.iter().cloned().fold(f32::INFINITY, f32::min), ys.iter().cloned().fold(f32::INFINITY, f32::min)); + let (w_pt, h_pt) = ( + xs.iter().cloned().fold(f32::NEG_INFINITY, f32::max) - min_x, + ys.iter().cloned().fold(f32::NEG_INFINITY, f32::max) - min_y, + ); + json!({ + "text": w.text, + "box": [min_x, min_y, w_pt, h_pt], + "confidence": w.confidence.map(|c| (c * 10.0).round() / 10.0), + "band": w.confidence.map(|c| band_name(pdfcraft_engine::ocr::band_for(c))), + "suspect": suspect, + "source": w.source, + }) + }) + .collect(); + let mut page = + json!({ "page": p.page + 1, "already_on_page": p.already_on_page, "words": words, "mean_confidence": p.mean_confidence().map(|c| (c * 10.0).round() / 10.0) }); + if let Some(why) = &p.skipped { + page["skipped"] = json!(why); + } + page }) .collect(); - Ok(json!({ "words": found.iter().map(|p| p.words.len()).sum::(), "pages": list })) + let mut out = json!({ "words": found.iter().map(|p| p.words.len()).sum::(), "suspects": suspects, "pages": list }); + if retain { + self.next_ocr_review = + self.next_ocr_review.checked_add(1).filter(|n| *n <= i64::MAX as u64).ok_or_else(|| failed("OCR review handles exhausted"))?; + out["review"] = json!(self.next_ocr_review); + self.ocr_reviews.insert(id, crate::OcrReview { handle: self.next_ocr_review, guard, pages: found }); + } + Ok(out) + } + + /// Apply only caller corrections to retained original boxes, through the stored guard. + pub(crate) fn ocr_review_apply(&mut self, a: &Args) -> Result { + let id = self.doc(a)?.id; + let handle = a.int("review")?; + let review = self + .ocr_reviews + .get(&id) + .filter(|r| i64::try_from(r.handle).ok() == Some(handle)) + .ok_or_else(|| failed("unknown OCR review for this document; call ocr_words with retain_review"))?; + if a.opt_bool("discard")?.unwrap_or(false) { + if a.get("corrections").is_some() { + return Err(ToolError::InvalidArgs("discard cannot include corrections".into())); + } + self.ocr_reviews.remove(&id); + return Ok(json!({ "discarded": handle })); + } + if let Some(why) = self.session.stale_reason(&review.guard) { + return Err(failed(why)); + } + let corrections = a.get("corrections").map(|v| v.as_array().ok_or_else(|| Args::wrong("corrections", "an array"))).transpose()?; + if corrections.is_some_and(|c| c.len() > 65536) { + return Err(failed("at most 65536 OCR corrections are allowed")); + } + // Prevalidate and cap every replacement before cloning the retained result. + let mut edits = Vec::new(); + let mut bytes = 0usize; + let mut seen = std::collections::BTreeSet::new(); + let page_indices: std::collections::BTreeMap<_, _> = review.pages.iter().enumerate().map(|(i, p)| (p.page, i)).collect(); + for item in corrections.into_iter().flatten() { + if !item.is_object() || item.as_object().is_some_and(|o| o.keys().any(|k| !matches!(k.as_str(), "page" | "word" | "text"))) { + return Err(ToolError::InvalidArgs("corrections accept only page, word and text".into())); + } + let c = Args(item); + let page = + usize::try_from(c.int("page")?).ok().and_then(|p| p.checked_sub(1)).ok_or_else(|| Args::wrong("page", "a 1-based position"))?; + let word = + usize::try_from(c.int("word")?).ok().and_then(|w| w.checked_sub(1)).ok_or_else(|| Args::wrong("word", "a 1-based position"))?; + let text = c.str("text")?; + bytes = bytes.checked_add(text.len()).filter(|n| *n <= 4 * 1024 * 1024).ok_or_else(|| failed("OCR corrections exceed 4 MiB"))?; + if !seen.insert((page, word)) { + return Err(failed("duplicate OCR word correction")); + } + let pi = page_indices.get(&page).copied().ok_or_else(|| failed("page is not in this OCR review"))?; + if review.pages.get(pi).and_then(|p| p.words.get(word)).is_none() { + return Err(failed("word is not in this OCR review")); + } + edits.push((pi, word, text.to_owned())); + } + let mut pages = review.pages.clone(); + for (pi, wi, text) in edits { + if let Some(w) = pages.get_mut(pi).and_then(|p| p.words.get_mut(wi)) { + w.text = text; + } + } + pdfcraft_engine::ocr::validate_ocr_words(pages.iter().flat_map(|p| &p.words)).map_err(failed)?; + for page in &mut pages { + page.words.retain(|w| !w.text.trim().is_empty()); + } + if pages.iter().all(|p| p.words.is_empty()) { + return Err(failed("the OCR review has no words to apply")); + } + let out = match a.opt_str("output_mode")?.unwrap_or("searchable") { + "searchable" => { + if a.get("out").is_some() { + return Err(ToolError::InvalidArgs("out is only for editable output".into())); + } + let count = self.session.apply_ocr_checked(id, &pages, &review.guard).map_err(failed)?; + json!({ "applied": count, "review": handle }) + } + "editable" => { + // The editable output writes the review's words out of the document: a + // document that forbids copying refuses here (the review is kept). + self.doc_to_copy(a)?; + let target = self.resolve(a.str("out")?, true)?; + if self + .session + .get(id) + .and_then(|d| d.path.as_deref()) + .and_then(|p| std::fs::canonicalize(p).ok()) + .as_ref() + .is_some_and(|source| std::fs::canonicalize(&target).as_ref().ok() == Some(source)) + { + return Err(failed("out is the source document; choose a new file")); + } + // A NEW file: an existing one is never replaced silently (the review is kept). + if target.exists() { + return Err(failed(format!("out already exists ({}); choose a path that does not exist yet", target.to_string_lossy()))); + } + let bytes = self.session.editable_ocr_bytes(id, &pages).map_err(failed)?; + write_atomic(&target, &bytes)?; + json!({ "out": target.to_string_lossy(), "review": handle, "words": pages.iter().map(|p| p.words.len()).sum::() }) + } + _ => return Err(ToolError::InvalidArgs("output_mode must be searchable or editable".into())), + }; + self.ocr_reviews.remove(&id); + Ok(out) + } + + /// Read/update the same per-language dictionary file used by OCR Verify. + pub(crate) fn ocr_dictionary(&mut self, a: &Args) -> Result { + use pdfcraft_engine::ocr::dictionary; + use std::io::Read as _; + for key in ["add", "remove"] { + if let Some(value) = a.get(key) { + let values = value.as_array().ok_or_else(|| Args::wrong(key, "an array of strings"))?; + if values.len() > 65536 { + return Err(failed("at most 65536 dictionary words are allowed")); + } + } + } + let add = if a.get("add").is_some() { a.strs("add")? } else { Vec::new() }; + let remove = if a.get("remove").is_some() { a.strs("remove")? } else { Vec::new() }; + let writing = a.get("add").is_some() || a.get("remove").is_some(); + // Updating sorts, lower-cases and rewrites the file, so this tool names dictionary files only: + // pointed at `.env` or a README it would corrupt it. + let file = std::path::Path::new(a.str("path")?).file_name().and_then(|n| n.to_str()).unwrap_or(""); + let plain = file.chars().all(|c| c.is_ascii_alphanumeric() || matches!(c, '-' | '_' | '.')); + if file.len() > 64 || !plain || !file.starts_with("ocr-dictionary") || !file.ends_with(".txt") { + return Err(failed("path must name a dictionary file: ocr-dictionary-.txt (letters, digits, '-', '_' and '.' only)")); + } + let path = self.resolve(a.str("path")?, writing)?; + let mut input = String::new(); + match std::fs::File::open(&path) { + Ok(file) => { + file.take((dictionary::MAX_BYTES + 1) as u64).read_to_string(&mut input).map_err(|e| failed(format!("OCR dictionary: {e}")))?; + } + Err(e) if e.kind() == std::io::ErrorKind::NotFound && writing => {} + Err(e) => return Err(failed(format!("OCR dictionary: {e}"))), + } + let text = dictionary::update(&input, &add, &remove).map_err(failed)?; + if writing { + write_atomic(&path, text.as_bytes())?; + } + let words = dictionary::words(&text).map_err(failed)?; + Ok(json!({ "path": path.to_string_lossy(), "count": words.len(), "words": words })) } - /// Whether text recognition's models are installed, the directories searched for them (in - /// order), and the languages it reads, each as a code and a name. pub(crate) fn ocr_status(&self) -> Result { use pdfcraft_engine::ocr; let dirs: Vec = ocr::Models::search_dirs().iter().map(|d| d.to_string_lossy().into_owned()).collect(); let langs: Vec = ocr::LANGUAGES.iter().map(|(c, n)| json!({ "code": c, "name": n })).collect(); - Ok(json!({ "available": ocr::available(), "search_dirs": dirs, "languages": langs })) + let engines: Vec = ocr::engines_status() + .iter() + .map(|e| { + let mut v = json!({ "id": e.id, "available": e.available }); + if let Some(p) = &e.program { + v["path"] = json!(p); + } + if let Some(vv) = &e.version { + v["version"] = json!(vv); + } + if let Some(why) = &e.why_not { + v["unavailable"] = json!(why); + } + v + }) + .collect(); + Ok(json!({ "available": ocr::available(), "search_dirs": dirs, "languages": langs, "engines": engines })) } } diff --git a/crates/automation/src/batch.rs b/crates/automation/src/batch.rs index decb18d2b7..dd5ca6aec0 100644 --- a/crates/automation/src/batch.rs +++ b/crates/automation/src/batch.rs @@ -84,8 +84,8 @@ impl<'a> Recipe<'a> { match *step { "convert" => bytes = session.convert_to_pdf(name, &bytes)?.1, "ocr" => { - let engine = ocr::engine().map_err(failed)?; - let result = ocr::recognize_file(name, bytes, None, self.ocr.clone(), &engine, |_, _| true).map_err(failed)?; + let recognizers = ocr::recognizers(&self.ocr).map_err(failed)?; + let result = ocr::recognize_file(name, bytes, None, self.ocr.clone(), &recognizers, |_, _| true).map_err(failed)?; log.push( json!({"step":"ocr","words":result.words(),"skipped_pages":result.pages.iter().filter(|p| p.skipped.is_some()).count()}), ); diff --git a/crates/automation/src/bates.rs b/crates/automation/src/bates.rs index 3cd1d7849e..5267716e85 100644 --- a/crates/automation/src/bates.rs +++ b/crates/automation/src/bates.rs @@ -199,7 +199,7 @@ impl Automation { } /// Publish a PDF atomically while refusing a target created by another process after preflight. -fn publish_new(path: &Path, bytes: &[u8]) -> Result<()> { +pub(super) fn publish_new(path: &Path, bytes: &[u8]) -> Result<()> { let parent = path.parent().ok_or_else(|| failed("output has no parent directory"))?; let mut temp = tempfile::NamedTempFile::new_in(parent).map_err(failed)?; temp.write_all(bytes).map_err(failed)?; diff --git a/crates/automation/src/bookmark_io.rs b/crates/automation/src/bookmark_io.rs new file mode 100644 index 0000000000..2a06d912f4 --- /dev/null +++ b/crates/automation/src/bookmark_io.rs @@ -0,0 +1,209 @@ +//! Bookmark trees in and out of documents: `bookmark_export`, `bookmark_import`, `bookmark_copy`. +//! +//! The portable form is JSON: `{"bookmarks": [node, …]}`, where a node is `title`, `page` +//! (1-based, or `null` for one that goes nowhere), `view` (`{"type", "values"}`: a destination +//! type and its values in PDF user space, as the file stores them), `color`, `bold`, `italic`, +//! `open` and `children`. What export writes, import takes back unchanged. A file also carries +//! `"truncated": true` when the export left out bookmarks nested past the listing limit. +use crate::{Args, Automation, Result, ToolError, failed}; +use pdfcraft_engine::{DocId, Edit, OutlineNode}; +use serde_json::{Value, json}; + +/// Deepest nesting a bookmark tree may have (the listing reaches 33 levels). +const MAX_DEPTH: usize = 33; + +/// The keys a node may have: the ones [`node_json`] writes. +const NODE_KEYS: [&str; 8] = ["title", "page", "view", "color", "bold", "italic", "open", "children"]; + +impl Automation { + pub(crate) fn bookmark_export(&self, a: &Args) -> Result { + let id = self.doc(a)?.id; + let path = a.opt_path("path")?.unwrap_or_default(); + let tree = self.outline_of(id, &path)?; + let mut out = json!({ "bookmarks": tree.nodes.iter().map(node_json).collect::>(), "truncated": tree.truncated }); + if let Some(file) = a.opt_str("out")? { + let target = self.resolve(file, true)?; + let mut saved = json!({ "bookmarks": out["bookmarks"] }); + if tree.truncated { + saved["truncated"] = json!(true); + } + let text = serde_json::to_vec_pretty(&saved).map_err(failed)?; + crate::write_atomic(&target, &text)?; + out["path"] = json!(target); + } + Ok(out) + } + + pub(crate) fn bookmark_import(&mut self, a: &Args) -> Result { + let (nodes, truncated) = match (a.get("bookmarks"), a.opt_str("file")?) { + (Some(list), None) => (nodes_of(list)?, false), + (None, Some(file)) => { + let path = self.resolve(file, false)?; + let value: Value = serde_json::from_slice(&crate::read_input(&path)?).map_err(|e| failed(format!("not a bookmark file: {e}")))?; + let list = value.get("bookmarks").ok_or_else(|| failed("not a bookmark file: it has no \"bookmarks\" list"))?; + (nodes_of(list)?, value.get("truncated") == Some(&Value::Bool(true))) + } + _ => return Err(ToolError::InvalidArgs("give either bookmarks or file".into())), + }; + let mut out = self.write_tree(a, nodes)?; + out["truncated"] = json!(truncated); + Ok(out) + } + + pub(crate) fn bookmark_copy(&mut self, a: &Args) -> Result { + let from = a.int("from")?; + let from = DocId(u64::try_from(from).map_err(|_| ToolError::InvalidArgs("from must be positive".into()))?); + let path = a.opt_path("path")?.unwrap_or_default(); + let tree = self.outline_of(from, &path)?; + let mut out = self.write_tree(a, tree.nodes)?; + out["truncated"] = json!(tree.truncated); + Ok(out) + } + + fn outline_of(&self, id: DocId, path: &[usize]) -> Result { + match self.session.outline_tree(id, path) { + // The engine names the path 0-based; the caller gave it 1-based. + Some(tree) => tree.map_err(|e| match e { + pdfcraft_organize::OutlineError::NoSuchBookmark(_) => failed(format!("there is no bookmark at {:?}", one_based(path))), + e => failed(e), + }), + None => Err(failed(format!("document {} is not open, or its bookmarks can't be read for editing", id.0))), + } + } + + /// Write `nodes` into `doc` under `parent` at `position` (or in place of the outline with + /// `replace`), and say which of them went nowhere because the document lacks their page. + fn write_tree(&mut self, a: &Args, bookmarks: Vec) -> Result { + let parent = a.opt_path("parent")?.unwrap_or_default(); + let index = a.opt_int("position")?.map_or(usize::MAX, |p| usize::try_from(p.max(1) - 1).unwrap_or(usize::MAX)); + let replace = a.opt_bool("replace")?.unwrap_or(false); + let count = self.doc(a)?.info.pages.len(); + let unmapped = pdfcraft_engine::outline_unmapped(&bookmarks, count); + let mut out = self.apply(a, Edit::ImportBookmarks { parent, index, bookmarks, replace })?; + out["unmapped"] = json!(unmapped.iter().map(|p| one_based(p)).collect::>()); + Ok(out) + } +} + +fn one_based(path: &[usize]) -> Vec { + path.iter().map(|i| i.saturating_add(1)).collect() +} + +/// A node as JSON, leaving out what is at its default (no view, black, regular, no children). +fn node_json(n: &OutlineNode) -> Value { + let mut v = json!({ "title": n.title, "page": n.page.map(|p| p.saturating_add(1)), "open": n.open }); + if let Some((kind, values)) = &n.view { + v["view"] = json!({ "type": String::from_utf8_lossy(kind), "values": values }); + } + if let Some(c) = n.color { + v["color"] = json!(c); + } + if n.bold { + v["bold"] = json!(true); + } + if n.italic { + v["italic"] = json!(true); + } + if !n.children.is_empty() { + v["children"] = json!(n.children.iter().map(node_json).collect::>()); + } + v +} + +fn nodes_of(list: &Value) -> Result> { + let mut count = 0; + list_of(list, &[], &mut count) +} + +/// The nodes of one level; `at` is the 1-based path of their parent, for messages. +fn list_of(list: &Value, at: &[usize], count: &mut usize) -> Result> { + let items = list.as_array().ok_or_else(|| bad(at, "children must be a list of bookmarks"))?; + if at.len() >= MAX_DEPTH && !items.is_empty() { + return Err(bad(at, &format!("bookmarks may nest at most {MAX_DEPTH} levels deep"))); + } + *count = count.saturating_add(items.len()); + if *count > pdfcraft_engine::MAX_IMPORT_BOOKMARKS { + return Err(ToolError::InvalidArgs(format!("at most {} bookmarks can be imported at once", pdfcraft_engine::MAX_IMPORT_BOOKMARKS))); + } + let mut out = Vec::with_capacity(items.len()); + for (i, item) in items.iter().enumerate() { + let mut path = at.to_vec(); + path.push(i.saturating_add(1)); + out.push(node_of(item, &path, count)?); + } + Ok(out) +} + +/// One node and, through [`list_of`], its children (recursion bounded by `MAX_DEPTH`). +fn node_of(v: &Value, at: &[usize], count: &mut usize) -> Result { + let o = v.as_object().ok_or_else(|| bad(at, "must be an object"))?; + if let Some(k) = o.keys().find(|k| !NODE_KEYS.contains(&k.as_str())) { + return Err(bad(at, &format!("unknown key {k:?} (a bookmark has {})", NODE_KEYS.join(", ")))); + } + let field = |k: &str| o.get(k).filter(|x| !x.is_null()); + // An empty title is kept: files have them, and export writes them as "". + let title = field("title").and_then(Value::as_str).ok_or_else(|| bad(at, "needs a title"))?.to_string(); + let page = match field("page") { + None => None, + Some(p) => Some(p.as_u64().filter(|p| *p >= 1).and_then(|p| usize::try_from(p - 1).ok()).ok_or_else(|| bad(at, "page must be 1 or more"))?), + }; + let flag = |k: &str| field(k).map_or(Ok(None), |x| x.as_bool().map(Some).ok_or_else(|| bad(at, &format!("{k} must be true or false")))); + let color = match field("color") { + None => None, + Some(c) => { + let rgb: Option> = c.as_array().and_then(|c| c.iter().map(|x| x.as_f64().map(|x| x as f32)).collect()); + match rgb.as_deref() { + Some(&[r, g, b]) if [r, g, b].iter().all(|c| (0.0..=1.0).contains(c)) => Some([r, g, b]), + _ => return Err(bad(at, "color must be [r, g, b], each from 0 to 1")), + } + } + }; + let children = match field("children") { + None => Vec::new(), + Some(list) => list_of(list, at, count)?, + }; + Ok(OutlineNode { + title, + page, + view: field("view").map(|x| view_of(x, at)).transpose()?, + color, + italic: flag("italic")?.unwrap_or(false), + bold: flag("bold")?.unwrap_or(false), + open: flag("open")?.unwrap_or(true), + children, + }) +} + +/// `{"type": "XYZ", "values": [left, top, zoom]}`: a destination type with exactly its values. +fn view_of(v: &Value, at: &[usize]) -> Result<(Vec, Vec>)> { + let kind = v.get("type").and_then(Value::as_str).ok_or_else(|| bad(at, "view needs a type such as XYZ or Fit"))?; + let values = match v.get("values").filter(|x| !x.is_null()) { + None => Vec::new(), + Some(list) => { + let list = list.as_array().filter(|l| l.len() <= 4).ok_or_else(|| bad(at, "view values must be a list of at most 4 numbers or nulls"))?; + list.iter() + .map(|x| match x { + Value::Null => Ok(None), + x => x.as_f64().map(Some).ok_or_else(|| bad(at, "view values must be numbers or null")), + }) + .collect::>()? + } + }; + let arity = pdfcraft_engine::outline_view_arity(kind.as_bytes()).ok_or_else(|| bad(at, VIEW_TYPES))?; + if values.len() != arity { + return Err(bad(at, &format!("a {kind} view takes {arity} values (numbers, or null to keep the reader's)"))); + } + if kind == "FitR" && values.iter().any(Option::is_none) { + return Err(bad(at, "a FitR view needs all four values: left, bottom, right, top")); + } + Ok((kind.as_bytes().to_vec(), values)) +} + +const VIEW_TYPES: &str = "view type must be XYZ (3 values: left, top, zoom), Fit, FitB (none), FitH, FitV, FitBH, FitBV (1) or FitR (4)"; + +fn bad(at: &[usize], what: &str) -> ToolError { + if at.is_empty() { + return ToolError::InvalidArgs(format!("bookmarks: {what}")); + } + ToolError::InvalidArgs(format!("bookmark {at:?}: {what}")) +} diff --git a/crates/automation/src/comments.rs b/crates/automation/src/comments.rs index ae46be87d0..78f4cefbe0 100644 --- a/crates/automation/src/comments.rs +++ b/crates/automation/src/comments.rs @@ -5,7 +5,9 @@ use std::collections::HashMap; -use pdfcraft_engine::{Edit, LineEnding, Markup, NewAnnotation, NoteIcon, ReviewState, Rgb, Shape, StampGroup, StampKind, Style, SummarySort}; +use pdfcraft_engine::{ + BorderShape, Edit, LineEnding, Markup, NewAnnotation, NoteIcon, ReviewState, Rgb, Shape, StampGroup, StampKind, Style, SummarySort, +}; use pdfcraft_render::{Annotation, PageInfo}; use serde_json::{Value, json}; @@ -75,7 +77,7 @@ fn to_user(p: &PageInfo, x: f64, y: f64) -> [f64; 2] { [ux as f64, uy as f64] } -fn rect_to_user(p: &PageInfo, r: [f64; 4]) -> [f64; 4] { +pub(crate) fn rect_to_user(p: &PageInfo, r: [f64; 4]) -> [f64; 4] { let (a, b) = (to_user(p, r[0], r[1]), to_user(p, r[2], r[3])); [a[0].min(b[0]), a[1].min(b[1]), a[0].max(b[0]), a[1].max(b[1])] } @@ -190,6 +192,25 @@ impl Automation { "rect": crate::view_rect(&doc.info, c.page, c.shown_rect(doc.info.pages.get(c.page).map_or(0, |p| p.rotation))), "color": c.color.map(hex), }); + // A text box's or typed text's oval or circle (a rectangle is not reported). + if c.subtype == "FreeText" + && let Some(shape) = doc.comment_props(c.page, c.index).and_then(|p| p.border_shape).filter(|s| *s != BorderShape::Rectangle) + { + v["border_shape"] = json!(shape.name().to_ascii_lowercase()); + } + // How opaque it is drawn: as a whole, or its fill and text each at their own. + // Not for replies, which are drawn with the comment they answer. + if !c.is_reply() + && let Some(p) = doc.comment_props(c.page, c.index) + { + v["opacity"] = json!(p.opacity); + if let Some(f) = p.fill_opacity { + v["fill_opacity"] = json!(f); + } + if let Some(t) = p.text_opacity { + v["text_opacity"] = json!(t); + } + } // A line's ends, or the corners of connected lines, a polygon or a cloud, as // comment_edit's `points` takes them. // Unchanged on a page the viewer could not read, as `rect` is. @@ -253,6 +274,12 @@ impl Automation { if endings.is_some() && !matches!(kind, "line" | "arrow" | "polyline" | "callout") { return Err(ToolError::InvalidArgs("endings apply to a line, arrow, polyline or callout".into())); } + // Shift while drawing: each end or corner held level or plumb with the one before. + let ortho = a.opt_bool("ortho")?.unwrap_or(false); + if ortho && !matches!(kind, "line" | "arrow" | "polygon" | "cloud" | "polyline") { + return Err(ToolError::InvalidArgs("ortho applies to a line, arrow, polygon, cloud or polyline".into())); + } + let held = |last: [f64; 2], p: [f64; 2]| if ortho { pdfcraft_engine::measure::snap::ortho(last, p) } else { p }; let markup = match kind { "highlight" => Some(Markup::Highlight), "underline" => Some(Markup::Underline), @@ -325,7 +352,8 @@ impl Automation { Shape::TextBox { rect: rect_to_user(&info, a.need::<4>("rect", "a text box")?), font_size } } "line" | "arrow" => { - let (f, t) = (a.need::<2>("from", "a line")?, a.need::<2>("to", "a line")?); + let f = a.need::<2>("from", "a line")?; + let t = held(f, a.need::<2>("to", "a line")?); let [start, end] = ending_pair(kind, endings.as_deref())?; Shape::Line { from: to_user(&info, f[0], f[1]), to: to_user(&info, t[0], t[1]), start, end } } @@ -352,10 +380,15 @@ impl Automation { "polygon" | "cloud" | "polyline" => { let wrong = || ToolError::InvalidArgs(format!("{kind} needs `points`: an array of [x, y] points")); let pts = a.get("points").and_then(|p| p.as_array()).ok_or_else(wrong)?; + let mut last = None; let vertices = pts .iter() .map(|p| match p.as_array().map(|p| p.iter().map(|v| v.as_f64()).collect::>()).as_deref() { - Some([Some(x), Some(y)]) => Ok(to_user(&info, *x, *y)), + Some([Some(x), Some(y)]) => { + let p = last.map_or([*x, *y], |l| held(l, [*x, *y])); + last = Some(p); + Ok(to_user(&info, p[0], p[1])) + } _ => Err(wrong()), }) .collect::>>()?; @@ -425,9 +458,28 @@ impl Automation { if let Some(o) = a.opt_num("opacity")? { style.opacity = o.clamp(0.0, 1.0); } + // The fill's and a text box's text's own opacities (the line keeps `opacity`). + let (fill_opacity, text_opacity) = (a.opt_num("fill_opacity")?, a.opt_num("text_opacity")?); + let texts = matches!(shape, Shape::TextBox { .. } | Shape::Typewriter { .. } | Shape::Callout { .. }); + if fill_opacity.is_some() && !(texts || matches!(shape, Shape::Rectangle { .. } | Shape::Oval { .. } | Shape::Polygon { .. })) { + return Err(ToolError::InvalidArgs( + "fill_opacity applies to a rectangle, oval, polygon, cloud, text box, typewriter text or callout".into(), + )); + } + if text_opacity.is_some() && !texts { + return Err(ToolError::InvalidArgs("text_opacity applies to a text box, typewriter text or callout".into())); + } + style.fill_opacity = fill_opacity.map(|o| o.clamp(0.0, 1.0)); + style.text_opacity = text_opacity.map(|o| o.clamp(0.0, 1.0)); if let Some(w) = a.opt_num("width")? { style.width = w.max(0.0); } + if let Some(b) = border_shape(a)? { + if !matches!(shape, Shape::TextBox { .. } | Shape::Typewriter { .. }) { + return Err(ToolError::InvalidArgs("border_shape is for a text box or typewriter text".into())); + } + style.border_shape = b; + } let contents = a.opt_str("contents")?.unwrap_or_default().to_string(); let author = a.opt_str("author")?.unwrap_or(DEFAULT_AUTHOR).to_string(); // Replace Text: the struck-out text and a grouped caret holding `contents`. @@ -488,6 +540,7 @@ impl Automation { } pub(crate) fn fill_sign_add(&mut self, a: &Args) -> Result { + let border_shape = border_shape(a)?; use pdfcraft_engine::FillMark; let page = self.page(a)?; let info = self.doc(a)?.info.pages[page].clone(); @@ -515,6 +568,25 @@ impl Automation { if !(4.0..=300.0).contains(&size) { return Err(ToolError::InvalidArgs("font_size must be 4–300 pt".into())); } + let requested_border = a.opt_num("border")?; + if requested_border.is_some() && !matches!(kind, "text" | "date") { + return Err(ToolError::InvalidArgs("border is only for type text or date".into())); + } + let border = requested_border.unwrap_or(0.0); + if !(0.0..=12.0).contains(&border) { + return Err(ToolError::InvalidArgs("border must be 0–12 pt".into())); + } + let border_color = a.opt_str("border_color")?.map(parse_color).transpose()?; + if border_color.is_some() && border <= 0.0 { + return Err(ToolError::InvalidArgs("border_color needs a border".into())); + } + if border_shape.is_some() && !matches!(kind, "text" | "date") { + return Err(ToolError::InvalidArgs("border_shape is only for type text or date".into())); + } + // Without a border an oval or a circle would grow the box around nothing drawn. + if border_shape.is_some_and(|s| s != BorderShape::Rectangle) && border <= 0.0 { + return Err(ToolError::InvalidArgs("border_shape needs a border".into())); + } if let Some(path) = a.opt_str("path")? { if !matches!(kind, "signature" | "initials") || a.opt_str("text")?.is_some() { return Err(ToolError::InvalidArgs("path is only for an image signature or initials; pass either path or text".into())); @@ -538,13 +610,16 @@ impl Automation { let edit = image.edit(page, &info, at, kind == "initials", &author).ok_or_else(|| ToolError::InvalidArgs("at must be finite".into()))?; return self.apply(a, edit); } + // A circle goes around the text, so its box is no wider than the text needs. + let min_width = if border_shape == Some(BorderShape::Circle) { 0.0 } else { 20.0 }; let text_at = |t: &str| { let longest = t.lines().map(|line| pdfcraft_engine::annot_text::text_width(line, size)).fold(0.0, f64::max); - let w = (longest + 8.0).clamp(20.0, 600.0); + let w = (longest + 8.0).clamp(min_width, 600.0); let rows = pdfcraft_engine::annot_text::wrap(t, size, w - 4.0).len().max(1); let h = rows as f64 * size * 1.2 + 6.0; - let rect = pdfcraft_engine::text_box_at(i64::from(info.rotation), at, w, h).unwrap_or([at[0], at[1] - h, at[0] + w, at[1]]); - Shape::Typewriter { rect, font_size: size } + let [x0, y0, x1, y1] = pdfcraft_engine::text_box_at(i64::from(info.rotation), at, w, h).unwrap_or([at[0], at[1] - h, at[0] + w, at[1]]); + // A border goes around the box, so the text sits where it would without one. + Shape::Typewriter { rect: [x0 - border, y0 - border, x1 + border, y1 + border], font_size: size } }; let (shape, contents) = match kind { "text" => { @@ -590,6 +665,11 @@ impl Automation { if let Some(color) = requested_color { style.color = color; } + if matches!(shape, Shape::Typewriter { .. }) { + style.width = border; + style.border_color = border_color; + style.border_shape = border_shape.unwrap_or_default(); + } self.apply(a, Edit::AddAnnotation(NewAnnotation { page, shape, style, contents, author })) } @@ -656,6 +736,21 @@ impl Automation { let fill = if f.eq_ignore_ascii_case("none") { None } else { Some(parse_color(f)?) }; edits.push(Edit::FillAnnotation { page, index, fill }); } + if let Some(c) = a.opt_str("border_color")? { + let color = if c.eq_ignore_ascii_case("text") { None } else { Some(parse_color(c)?) }; + edits.push(Edit::BorderColorAnnotation { page, index, color }); + } + if let Some(shape) = border_shape(a)? { + edits.push(Edit::BorderShapeAnnotation { page, index, shape }); + } + // After `opacity`, which sets the whole comment's. + let (fill_opacity, text_opacity) = (a.opt_num("fill_opacity")?, a.opt_num("text_opacity")?); + if fill_opacity.into_iter().chain(text_opacity).any(|o| !(0.0..=1.0).contains(&o)) { + return Err(ToolError::InvalidArgs("fill_opacity and text_opacity are 0–1".into())); + } + if fill_opacity.is_some() || text_opacity.is_some() { + edits.push(Edit::PartOpacity { page, index, fill: fill_opacity, text: text_opacity }); + } if let Some(size) = a.opt_num("font_size")? { if !(1.0..=400.0).contains(&size) { return Err(ToolError::InvalidArgs("font_size must be 1–400 pt".into())); @@ -689,7 +784,7 @@ impl Automation { } if edits.is_empty() { return Err(ToolError::InvalidArgs( - "nothing to change: pass contents, color, opacity, width, fill, font_size, rect, move or points".into(), + "nothing to change: pass contents, color, opacity, fill_opacity, text_opacity, width, fill, border_color, border_shape, font_size, rect, move or points".into(), )); } let edit = if edits.len() == 1 { edits.remove(0) } else { Edit::Batch { label: "Edit comment".into(), edits } }; @@ -700,6 +795,49 @@ impl Automation { let (page, index) = self.comment_target(a)?; self.apply(a, Edit::DeleteAnnotation { page, index }) } + + /// The XFDF of the listed comments and their replies (`comments`: 1-based [page, index] pairs). + pub(crate) fn comments_copy(&self, a: &Args) -> Result { + let id = self.doc(a)?.id; + let all = self.comments(a)?; + let shape = || ToolError::InvalidArgs("comments: [[page, index], …] (1-based, as comment_list reports)".into()); + let list = a.get("comments").and_then(Value::as_array).ok_or_else(shape)?; + let mut which = Vec::with_capacity(list.len()); + for item in list { + let pair = item.as_array().filter(|p| p.len() == 2).and_then(|p| Some((p.first()?.as_u64()?, p.get(1)?.as_u64()?))); + let Some((page, index)) = pair.filter(|(p, i)| *p >= 1 && *i >= 1) else { return Err(shape()) }; + let (page, index) = (usize::try_from(page - 1).map_err(|_| shape())?, usize::try_from(index - 1).map_err(|_| shape())?); + if !all.iter().any(|c| c.page == page && c.index == index) { + return Err(failed(format!("there is no comment {} on page {} (see comment_list)", index + 1, page + 1))); + } + which.push((page, index)); + } + let xfdf = self.session.export_comments(id, &which).map_err(failed)?; + Ok(json!({ "xfdf": xfdf, "comments": which.len() })) + } + + /// Add the comments of `xfdf` (from `comments_copy`) to `page` as new comments. + pub(crate) fn comments_paste(&mut self, a: &Args) -> Result { + let page = self.page(a)?; + let xfdf = a.str("xfdf")?.to_owned(); + if !pdfcraft_engine::is_xfdf_comments(&xfdf) { + return Err(ToolError::InvalidArgs("xfdf: the text comments_copy returned (XFDF with )".into())); + } + let offset = match a.nums::<2>("at")? { + Some([x, y]) => { + let info = self.doc(a)?.info.pages.get(page).cloned().ok_or_else(|| failed(format!("no page {}", page + 1)))?; + let [cx, cy] = to_user(&info, x, y); + let b = + pdfcraft_engine::comments_bounds(&xfdf).ok_or_else(|| ToolError::InvalidArgs("xfdf: no comment with a rect to place".into()))?; + Some([cx - (b[0] + b[2]) / 2.0, cy - (b[1] + b[3]) / 2.0]) + } + None => None, + }; + let before = self.comments(a)?.len(); + let mut out = self.apply(a, Edit::PasteComments { xfdf, page, offset })?; + out["comments"] = json!(self.comments(a)?.len().saturating_sub(before)); + Ok(out) + } } impl Automation { @@ -744,6 +882,13 @@ impl Automation { } } +/// `border_shape`: `rectangle`, `oval` or `circle`. +fn border_shape(a: &Args) -> Result> { + a.opt_str("border_shape")? + .map(|s| BorderShape::parse(s).ok_or_else(|| ToolError::InvalidArgs(format!("unknown border_shape {s:?}: rectangle, oval or circle")))) + .transpose() +} + #[cfg(test)] mod tests { use super::*; diff --git a/crates/automation/src/dispatch.rs b/crates/automation/src/dispatch.rs index f3c51d22e0..ebe2a93526 100644 --- a/crates/automation/src/dispatch.rs +++ b/crates/automation/src/dispatch.rs @@ -69,6 +69,7 @@ impl Automation { self.apply(&a, edit)? } "page_render" => return self.page_render(&a), + "model_render" => return self.model_render(&a), "comment_image_preview" => return self.comment_image_preview(&a), "text_extract" => self.text_extract(&a)?, "text_word_count" => self.text_word_count(&a)?, @@ -136,6 +137,7 @@ impl Automation { out["detected"] = json!(given.is_none()); out } + "page_enhance_scan" => self.page_enhance_scan(&a)?, "page_delete" => { let pages = self.pages(&a, "pages")?; self.apply(&a, Edit::DeletePages { pages })? @@ -196,6 +198,9 @@ impl Automation { out } "bookmark_list" => self.bookmark_listing(&a)?, + "bookmark_export" => self.bookmark_export(&a)?, + "bookmark_import" => self.bookmark_import(&a)?, + "bookmark_copy" => self.bookmark_copy(&a)?, "bookmark_add" => { let page = self.page(&a)?; let parent = a.opt_path("parent")?.unwrap_or_default(); @@ -769,6 +774,7 @@ impl Automation { "comment_list" => self.comment_list(&a)?, "comment_add" => self.comment_add(&a)?, "stamp_custom" => self.stamp_custom(&a)?, + "model_add" => self.model_add(&a)?, "comment_reply" => self.comment_reply(&a)?, "comment_set_status" => self.comment_set_status(&a)?, "sign_list" => self.sign_list(&a)?, @@ -806,6 +812,9 @@ impl Automation { "form_set_actions" => self.form_set_actions(&a)?, "ocr_status" => self.ocr_status()?, "ocr_recognize_files" => self.ocr_recognize_files(&a)?, + "ocr_words" => self.ocr_words(&a)?, + "ocr_review_apply" => self.ocr_review_apply(&a)?, + "ocr_dictionary" => self.ocr_dictionary(&a)?, "accessibility_report" => self.a11y_report(&a)?, "accessibility_review_record" => self.a11y_review_record(&a)?, "accessibility_review_save" => self.a11y_review_save(&a)?, @@ -861,6 +870,8 @@ impl Automation { "comments_summarize" => self.comments_summarize(&a)?, "comment_edit" => self.comment_edit(&a)?, "comment_delete" => self.comment_delete(&a)?, + "comments_copy" => self.comments_copy(&a)?, + "comments_paste" => self.comments_paste(&a)?, other => return Err(ToolError::UnknownTool(other.into())), }; Ok(vec![Content::Json(out)]) @@ -914,3 +925,35 @@ impl Automation { Ok(out) } } + +impl Automation { + /// `page_enhance_scan`: Enhance ▸ Scanned document on the chosen pages (all by default). + fn page_enhance_scan(&mut self, a: &Args) -> Result { + let pages = match a.opt_ints("pages")? { + Some(_) => self.pages(a, "pages")?, + None => Vec::new(), + }; + let level = match a.opt_int("level")? { + None => pdfcraft_engine::scan_clean::ScanCleanSettings::default().level, + Some(l @ 0..=100) => l as u8, + Some(l) => return Err(ToolError::InvalidArgs(format!("level must be between 0 and 100, not {l}"))), + }; + let mut settings = pdfcraft_engine::scan_clean::ScanCleanSettings::at(level); + if let Some(on) = a.opt_bool("edge_shadows")? { + settings.edge_shadows = on; + } + if let Some(on) = a.opt_bool("despeckle")? { + settings.despeckle = on; + } + let id = self.doc(a)?.id; + let report = self.session.enhance_scanned_pages(id, &pages, settings).map_err(failed)?; + let numbered = |p: &[usize]| p.iter().map(|p| p + 1).collect::>(); + let skipped: Vec = report.skipped.iter().map(|k| json!({ "pages": numbered(&k.pages), "name": k.name, "reason": k.reason })).collect(); + let mut out = summary(self.doc(a)?); + out["pages_cleaned"] = json!(numbered(&report.pages)); + out["images"] = json!(report.images); + out["not_scanned"] = json!(numbered(&report.not_scanned)); + out["skipped"] = json!(skipped); + Ok(out) + } +} diff --git a/crates/automation/src/documents.rs b/crates/automation/src/documents.rs index 979b041a73..f8c7988bbc 100644 --- a/crates/automation/src/documents.rs +++ b/crates/automation/src/documents.rs @@ -120,6 +120,7 @@ impl Automation { self.texts.remove(&id); self.listed.retain(|(d, _), _| *d != id); self.evicted.remove(&id); + self.ocr_reviews.remove(&id); Ok(json!({ "closed": id.0 })) } diff --git a/crates/automation/src/file_access.rs b/crates/automation/src/file_access.rs index 067a593888..dab2760f5d 100644 --- a/crates/automation/src/file_access.rs +++ b/crates/automation/src/file_access.rs @@ -394,8 +394,18 @@ mod tests { // Helpers that are only called with a resolved path. `doc_batch` (batch.rs) resolves its // inputs and out_dir, walks only inside them (skipping links) and resolves every // discovered file again before `process` reads it or `publish` writes it. - const RESOLVED_BY_CALLER: &[&str] = - &["write_atomic", "write_atomic_with", "read_input", "read_input_bytes", "read_review_sidecar", "process", "linked", "walk", "publish"]; + const RESOLVED_BY_CALLER: &[&str] = &[ + "write_atomic", + "write_atomic_with", + "read_input", + "read_input_bytes", + "read_model", + "read_review_sidecar", + "process", + "linked", + "walk", + "publish", + ]; let mut dirs = vec![Path::new(env!("CARGO_MANIFEST_DIR")).join("src")]; let mut files = vec![]; while let Some(dir) = dirs.pop() { diff --git a/crates/automation/src/lib.rs b/crates/automation/src/lib.rs index 5bcb032fce..7e4443d244 100644 --- a/crates/automation/src/lib.rs +++ b/crates/automation/src/lib.rs @@ -27,6 +27,8 @@ mod links; #[cfg(feature = "mcp")] pub mod mcp; mod measure; +mod model3d; +mod model_render; mod printing; mod reading; mod redact; @@ -50,6 +52,7 @@ use file_access::foreign_share; use text_cache::TextCache; mod assemble; +mod bookmark_io; mod dispatch; mod documents; mod error; @@ -73,6 +76,12 @@ pub(crate) use paths::{read_input, read_input_bytes}; #[cfg(test)] use paths::{write_atomic_synced, write_atomic_with}; +struct OcrReview { + handle: u64, + guard: pdfcraft_engine::ocr::OcrGuard, + pages: Vec, +} + /// A headless PdfCraft session driven by tool calls. pub struct Automation { session: Session, @@ -89,6 +98,9 @@ pub struct Automation { /// Documents that lost a listing to the cap. A page of one with no listing is refused rather /// than read as "a number is a position", which the page may no longer match. evicted: HashSet, + /// Review sessions held for `ocr_words` (a guard and the pages read), by document. + ocr_reviews: HashMap, + next_ocr_review: u64, } impl Default for Automation { @@ -107,6 +119,8 @@ impl Automation { texts: TextCache::default(), listed: HashMap::new(), evicted: HashSet::new(), + ocr_reviews: HashMap::new(), + next_ocr_review: 0, } } diff --git a/crates/automation/src/mcp.rs b/crates/automation/src/mcp.rs index 84cc21b258..b033871867 100644 --- a/crates/automation/src/mcp.rs +++ b/crates/automation/src/mcp.rs @@ -557,7 +557,7 @@ fn tool_json(t: &crate::ToolDef) -> Value { "title": t.title, "description": t.description, "inputSchema": t.input_schema, - "annotations": { "title": t.title, "readOnlyHint": t.read_only, "destructiveHint": t.destructive, "idempotentHint": t.read_only || matches!(t.name, "doc_save" | "doc_optimize" | "doc_export_data" | "measure_export" | "doc_export_images" | "doc_export_office" | "doc_export_all_images" | "doc_export_text" | "image_save"), "openWorldHint": false }, + "annotations": { "title": t.title, "readOnlyHint": t.read_only, "destructiveHint": t.destructive, "idempotentHint": t.read_only || matches!(t.name, "doc_save" | "doc_optimize" | "doc_export_data" | "measure_export" | "doc_export_images" | "doc_export_office" | "doc_export_all_images" | "doc_export_text" | "image_save" | "bookmark_export"), "openWorldHint": false }, }) } diff --git a/crates/automation/src/model3d.rs b/crates/automation/src/model3d.rs new file mode 100644 index 0000000000..2d89e4ae60 --- /dev/null +++ b/crates/automation/src/model3d.rs @@ -0,0 +1,114 @@ +//! 3D model tools (Acrobat: Rich Media ▸ Add 3D, AGENTS.md §3). + +use std::path::Path; + +use pdfcraft_engine::{Activation3d, Edit, MAX_3D_MODEL_BYTES, MarkFile, Model3dFormat}; +use serde_json::Value; + +use crate::comments::rect_to_user; +use crate::{Args, Automation, Result, ToolError, failed}; + +/// The largest side of a 3D model's box, in points: the largest page PDF allows (ISO 32000-2 Annex C). +const MAX_MODEL_SIDE: f64 = 14_400.0; + +impl Automation { + /// A U3D or PRC model in `rect` on `page`, its poster the picture at `poster` (a PDF page or an + /// image) fitted inside the box, or a labelled frame. + pub(crate) fn model_add(&mut self, a: &Args) -> Result { + let page = self.page(a)?; + let info = self.doc(a)?.info.pages.get(page).cloned().ok_or_else(|| ToolError::InvalidArgs(format!("page {} does not exist", page + 1)))?; + let r = a.nums::<4>("rect")?.ok_or_else(|| ToolError::InvalidArgs("missing argument rect".into()))?; + let (w, h) = ((r[2] - r[0]).abs(), (r[3] - r[1]).abs()); + if !((1.0..=MAX_MODEL_SIDE).contains(&w) && (1.0..=MAX_MODEL_SIDE).contains(&h)) { + return Err(ToolError::InvalidArgs(format!( + "rect must be [x0, y0, x1, y1] with each side at least 1 pt and at most {MAX_MODEL_SIDE} pt" + ))); + } + let activation = match a.opt_str("activation")? { + None => Activation3d::default(), + Some(word) => Activation3d::ALL + .iter() + .find(|(_, w)| *w == word) + .map(|(act, _)| *act) + .ok_or_else(|| ToolError::InvalidArgs(format!("activation must be one of: {}", Activation3d::ALL.map(|(_, w)| w).join(", "))))?, + }; + let path = self.resolve(a.str("path")?, false)?; + let model = read_model(&path, MAX_3D_MODEL_BYTES)?; + let file = path.file_name().map(|n| n.to_string_lossy().into_owned()).unwrap_or_default(); + let name = a.opt_str("name")?.map(str::to_owned).unwrap_or_else(|| file.rsplit_once('.').map_or(file.clone(), |(s, _)| s.to_owned())); + let poster = match a.opt_str("poster")? { + None => None, + Some(p) => { + let path = self.resolve(p, false)?; + let bytes = crate::read_input(&path)?; + let name = path.file_name().map(|n| n.to_string_lossy().into_owned()).unwrap_or_default(); + let page = usize::try_from(a.opt_int("poster_page")?.unwrap_or(1).max(1)).unwrap_or(1).saturating_sub(1); + Some(MarkFile { name, bytes: std::sync::Arc::new(bytes), page }) + } + }; + let edit = Edit::AddModel3d { page, rect: rect_to_user(&info, r), model: std::sync::Arc::new(model), poster, activation, name }; + self.apply(a, edit) + } +} + +/// Read a resolved model file of at most `max` bytes. Its size (from the file system) and its first +/// bytes are checked before the rest is read, so a large file that is not a model is never loaded. +fn read_model(path: &Path, max: usize) -> Result> { + use std::io::Read as _; + let io = |e: std::io::Error| failed(format!("{}: {e}", path.display())); + let too_large = |size: u64| failed(format!("the 3D model is {size} bytes; the largest accepted is {max}")); + let cap = u64::try_from(max).unwrap_or(u64::MAX); + let mut file = std::fs::File::open(path).map_err(io)?; + if let Ok(meta) = file.metadata() + && meta.is_file() + && meta.len() > cap + { + return Err(too_large(meta.len())); + } + let mut bytes = Vec::new(); + (&mut file).take(4).read_to_end(&mut bytes).map_err(io)?; + Model3dFormat::detect(&bytes).map_err(failed)?; + let rest = cap.saturating_add(1).saturating_sub(bytes.len() as u64); + file.take(rest).read_to_end(&mut bytes).map_err(io)?; + if bytes.len() > max { + return Err(failed(format!("the 3D model is more than {max} bytes, the largest accepted"))); + } + Ok(bytes) +} + +#[cfg(test)] +mod tests { + use super::*; + + fn file(dir: &Path, name: &str, bytes: &[u8]) -> std::path::PathBuf { + let path = dir.join(name); + std::fs::write(&path, bytes).unwrap(); + path + } + + #[test] + fn a_model_is_read_whole_when_it_fits() { + let dir = tempfile::tempdir().unwrap(); + let mut prc = b"PRC".to_vec(); + prc.extend_from_slice(&[7; 40]); + assert_eq!(read_model(&file(dir.path(), "m.prc", &prc), 64).unwrap(), prc); + assert_eq!(read_model(&file(dir.path(), "m.u3d", b"U3D\0x"), 6).unwrap(), b"U3D\0x"); + } + + #[test] + fn a_model_past_the_cap_is_refused_by_its_size_before_it_is_read() { + let dir = tempfile::tempdir().unwrap(); + let path = file(dir.path(), "big.prc", &[b'P', b'R', b'C', 0, 0, 0, 0, 0, 0, 0]); + let err = read_model(&path, 4).unwrap_err().to_string(); + assert!(err.contains("is 10 bytes") && err.contains("largest accepted is 4"), "the file's own size, from the file system: {err}"); + } + + #[test] + fn a_file_that_is_not_a_model_is_refused_by_its_header() { + let dir = tempfile::tempdir().unwrap(); + for junk in [&b"%PDF-1.7 a page"[..], b"PR", b"", b"U3DX and more"] { + let err = read_model(&file(dir.path(), "junk.bin", junk), 64).unwrap_err().to_string(); + assert!(err.contains("not a 3D model"), "{junk:?}: {err}"); + } + } +} diff --git a/crates/automation/src/model_render.rs b/crates/automation/src/model_render.rs new file mode 100644 index 0000000000..dbfabc6708 --- /dev/null +++ b/crates/automation/src/model_render.rs @@ -0,0 +1,104 @@ +//! `model_render`: a page's 3D model (ISO 32000-2 §13.6) drawn to a PNG, without a viewer. + +use pdfcraft_engine::model3d::Model3dKind; +use pdfcraft_prc::PrcFile; +use pdfcraft_prc_render::{Camera, RenderOptions, assemble, render}; + +use super::Automation; +use super::dispatch::Content; +use super::error::{Result, ToolError, failed}; +use super::json::Args; +use super::pages::encode_png; + +/// Default and largest side of the image, in pixels. +pub(crate) const DEFAULT_SIDE: u32 = 512; +pub(crate) const MAX_SIDE: u32 = 4096; + +/// The default view: from the front, right and above, as an isometric view looks. +const DEFAULT_DIRECTION: [f64; 3] = [-1.0, 1.0, -1.0]; + +impl Automation { + pub(crate) fn model_render(&mut self, a: &Args) -> Result> { + let page = self.page(a)?; + let width = side(a, "width")?; + let height = side(a, "height")?; + let direction = direction(a)?; + let perspective = a.opt_bool("perspective")?.unwrap_or(true); + let doc = self.doc(a)?; + let found = doc.model_3d_annotations(page).map_err(failed)?; + let index = match a.opt_int("model")? { + None => *found.first().ok_or_else(|| failed(format!("page {} has no 3D model", page.saturating_add(1))))?, + Some(n) => { + let pos = usize::try_from(n).ok().and_then(|n| n.checked_sub(1)).ok_or_else(|| Args::wrong("model", "a positive integer"))?; + *found + .get(pos) + .ok_or_else(|| ToolError::InvalidArgs(format!("page {} has {} 3D models, not {n}", page.saturating_add(1), found.len())))? + } + }; + let model = doc.model_3d(page, index).map_err(failed)?; + if model.kind == Model3dKind::U3d { + return Err(failed("this 3D model is in U3D, which can't be drawn yet; only PRC models can")); + } + let prc = PrcFile::parse(&model.data).map_err(|e| failed(format!("the 3D model could not be read: {e}")))?; + let meshes = assemble(&prc).map_err(failed)?; + let aspect = f64::from(width) / f64::from(height); + let camera = Camera::fit_meshes(&meshes, direction, up_for(direction), perspective, aspect).map_err(failed)?; + let image = render(&meshes, &camera, &RenderOptions { width, height, ..RenderOptions::default() }).map_err(failed)?; + let data = encode_png(image.width, image.height, &premultiplied(image.rgba))?; + Ok(vec![Content::Png { data, width: image.width, height: image.height }]) + } +} + +fn side(a: &Args, key: &str) -> Result { + match a.opt_int(key)? { + None => Ok(DEFAULT_SIDE), + Some(n) => u32::try_from(n).ok().filter(|n| (1..=MAX_SIDE).contains(n)).ok_or_else(|| Args::wrong(key, &format!("between 1 and {MAX_SIDE}"))), + } +} + +/// `direction`: three finite numbers, not all zero, from the eye towards the model. +fn direction(a: &Args) -> Result<[f64; 3]> { + let Some(v) = a.get("direction") else { return Ok(DEFAULT_DIRECTION) }; + let wrong = || Args::wrong("direction", "three numbers [x, y, z], not all zero"); + let items = v.as_array().filter(|x| x.len() == 3).ok_or_else(wrong)?; + let mut out = [0.0; 3]; + for (o, x) in out.iter_mut().zip(items) { + *o = x.as_f64().filter(|f| f.is_finite()).ok_or_else(wrong)?; + } + if out.iter().all(|c| *c == 0.0) { + return Err(wrong()); + } + Ok(out) +} + +/// Z is up, as CAD models are drawn, unless the view looks along it. +fn up_for([x, y, z]: [f64; 3]) -> [f64; 3] { + if x.hypot(y) <= z.abs() * 1e-6 { [0.0, 1.0, 0.0] } else { [0.0, 0.0, 1.0] } +} + +/// Straight RGBA to the premultiplied RGBA `encode_png` takes. +fn premultiplied(mut rgba: Vec) -> Vec { + for [r, g, b, alpha] in rgba.as_chunks_mut::<4>().0 { + let a = u16::from(*alpha); + for c in [r, g, b] { + *c = u8::try_from((u16::from(*c) * a + 127) / 255).unwrap_or(u8::MAX); + } + } + rgba +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn premultiplying_scales_colour_by_alpha() { + assert_eq!(premultiplied(vec![200, 100, 0, 255, 200, 100, 0, 128, 9, 9, 9, 0]), vec![200, 100, 0, 255, 100, 50, 0, 128, 0, 0, 0, 0]); + } + + #[test] + fn a_view_along_z_keeps_y_up() { + assert_eq!(up_for([0.0, 0.0, -1.0]), [0.0, 1.0, 0.0]); + assert_eq!(up_for(DEFAULT_DIRECTION), [0.0, 0.0, 1.0]); + } +} diff --git a/crates/automation/src/tools.rs b/crates/automation/src/tools.rs index 1b4b6ac4ba..af5146377a 100644 --- a/crates/automation/src/tools.rs +++ b/crates/automation/src/tools.rs @@ -5,6 +5,8 @@ use serde_json::{Value, json}; +use pdfcraft_engine::REDACT_PATTERNS; + use crate::ToolError; #[derive(Clone, Debug)] @@ -30,6 +32,12 @@ fn pages(what: &str) -> Value { json!({ "type": "array", "items": { "type": "integer", "minimum": 1 }, "minItems": 1, "description": format!("1-based page numbers {what}.") }) } +/// The Search & Redact pattern ids, from the engine's registry, so the advertised schema cannot +/// drift from the patterns the runtime accepts. +fn pattern_enum() -> Value { + json!({ "type": "string", "enum": REDACT_PATTERNS.iter().map(|p| p.id()).collect::>() }) +} + fn path(what: &str) -> Value { json!({ "type": "array", "items": { "type": "integer", "minimum": 1 }, "description": format!("{what}: 1-based positions from the top level, e.g. [2, 1] = the first child of the second bookmark.") }) } @@ -77,6 +85,12 @@ fn schema(props: Value, required: &[&str]) -> Value { json!({ "type": "object", "properties": props, "required": required, "additionalProperties": false }) } +/// The language codes the engines read, as the `language` enum — from the engine's own table, +/// so the schema and the validator can never disagree. +fn language_enum() -> Value { + Value::Array(pdfcraft_engine::ocr::LANGUAGES.iter().map(|(code, _)| json!(code)).collect()) +} + struct T { name: &'static str, title: &'static str, @@ -181,6 +195,13 @@ pub fn tools() -> Vec { "hide_text_block": { "type": "integer", "minimum": 1, "description": "Render an inline editor background without this paragraph (number from text_paragraphs, resolved as text_edit does: refused once an edit merged that paragraph into a neighbour). Does not modify the document." }, "hide_added_text": { "type": "integer", "minimum": 1, "description": "Render a retyping background without this added text item (added index from object_list). Mutually exclusive with hide_text_block. Does not modify the document." } }), &["doc", "page"])), + t("model_render", "Render a 3D model", "Draw a page's 3D model (a 3D annotation's PRC stream) to a PNG, framed to fill the image, with headlight shading on a white background. model is its 1-based position among the page's 3D models (default 1). direction is the view direction [x, y, z] from the eye towards the model (default [-1, 1, -1]); Z is up unless the view looks along Z. perspective defaults to true. U3D models can't be drawn yet and give an error. Does not modify the document.") + .ro() + .with(schema(json!({ "doc": doc(), "page": { "type": "integer", "minimum": 1 }, "model": { "type": "integer", "minimum": 1 }, + "width": { "type": "integer", "minimum": 1, "maximum": 4096 }, "height": { "type": "integer", "minimum": 1, "maximum": 4096 }, + "direction": { "type": "array", "items": { "type": "number" }, "minItems": 3, "maxItems": 3 }, + "perspective": { "type": "boolean" } + }), &["doc", "page"])), t("object_list", "List editable objects", "List existing paragraphs, Image/Form artwork and added content once, with source kind/index references, displayed rectangles and document generation. Indexes change when content changes; use generation in object_move to reject stale references.") .ro().with(schema(json!({"doc":doc(),"page":{"type":"integer","minimum":1}}), &["doc","page"])), t("object_move", "Move objects together", "Atomically translate 1–1000 mixed references from object_list by [dx, dy] in displayed points (right/down). Preserves glyph codes, fonts, image data and relative spacing; success is one undo step. Text used as a clipping path and vertical text are refused. A bad/stale reference refuses the entire move.") @@ -233,6 +254,18 @@ pub fn tools() -> Vec { }), &["doc", "page"], )), + t("page_enhance_scan", "Enhance scanned file", "Clean up scanned pages in place (Scan & OCR ▸ Enhance ▸ Scanned document). A page is a scan when its pictures cover at least half of it; the pictures covering at least a tenth of it are cleaned and written back into the same image objects at the same size and colour space (a JPEG stays a JPEG at quality 90, other grey/RGB images are stored losslessly, a 1-bit image stays 1-bit), so the page's invisible OCR text, comments, links and other content stay as they were, and an image drawn on several pages is cleaned once. level (0–100, default 50) is background removal: the paper's illumination is evened out so a grey, yellowed or unevenly lit background turns white, then the contrast is raised so faint strokes darken; colour ink stays colour. edge_shadows (default: on unless level is 0) lightens shadows along the scan's edges (book gutters, a copier's dark border). despeckle (default: from level 60) turns isolated dark specks white; it is the only change a 1-bit scan takes. Returns pages_cleaned (1-based), images (rewritten), not_scanned (pages that are no scan) and skipped ([{pages, name, reason}]: CMYK, indexed, JPEG 2000, JBIG2 or CCITT images, a scan drawn inside grouped Form artwork, broken data, or nothing to change). One undoable step; none when nothing changed.") + .cmd("ocr.enhance") + .with(schema( + json!({ + "doc": doc(), + "pages": pages("to clean up (default: all)"), + "level": { "type": "integer", "minimum": 0, "maximum": 100 }, + "edge_shadows": { "type": "boolean" }, + "despeckle": { "type": "boolean" }, + }), + &["doc"], + )), t("page_delete", "Delete pages", "Delete pages. Undoable until saved.") .destructive() .cmd("page.delete") @@ -323,6 +356,50 @@ pub fn tools() -> Vec { }), &["doc", "title", "page"], )), + t( + "bookmark_export", + "Export bookmarks", + "The bookmark tree, or the subtree at path, as a list of nodes bookmark_import takes back unchanged. A node has title, page (1-based, null when it goes nowhere), open, and when set view ({type, values}: a destination type from ISO 32000-2 §12.3.2.2 Table 149 and its values in PDF user space, as the file stores them), color ([r, g, b], 0-1), bold, italic and children. truncated is true when bookmarks nested deeper than 33 levels were left out. With out, the tree is also written there as a JSON file {\"bookmarks\": [...]}, with \"truncated\": true when that is so; out is overwritten.", + ) + .destructive() + .with(schema( + json!({ "doc": doc(), "path": path("The bookmark whose subtree to export (omit for the whole outline)"), "out": path_arg() }), + &["doc"], + )), + t( + "bookmark_import", + "Import bookmarks", + "Add a tree of bookmarks, as bookmark_export gives it (bookmarks) or as a JSON file it wrote (file), under a parent bookmark (or at the top level) at a position, or with replace in place of the whole outline. A bookmark whose page this document lacks is kept without a destination and listed in unmapped (1-based paths into the given tree). A view takes exactly its type's values (XYZ 3, FitR 4, FitH/FitV/FitBH/FitBV 1, Fit/FitB none). Give exactly one of bookmarks and file. A title may be empty but not missing; a key a node does not have is refused. Every node is checked before anything is written: at most 100000 bookmarks, nested at most 33 levels counting the parent's. truncated is true when the file says its export left bookmarks out. Undoable.", + ) + .destructive() + .with(schema( + json!({ + "doc": doc(), + "bookmarks": { "type": "array", "items": { "type": "object" }, "minItems": 1, "description": "Nodes as bookmark_export returns them." }, + "file": path_arg(), + "parent": path("Parent bookmark (omit for the top level)"), + "position": { "type": "integer", "minimum": 1, "description": "1-based position among the parent's children (default: last)." }, + "replace": { "type": "boolean", "description": "Replace the document's bookmarks instead of adding to them (default false; leave parent out)." } + }), + &["doc"], + )), + t( + "bookmark_copy", + "Copy bookmarks from another document", + "Copy the bookmarks of document from (or the subtree at path) into doc, by page number: under a parent bookmark (or at the top level) at a position, or with replace in place of doc's outline. A bookmark whose page doc lacks is kept without a destination and listed in unmapped (1-based paths into the copied tree). truncated is true when bookmarks nested deeper than 33 levels were left out of the copy. Undoable.", + ) + .destructive() + .with(schema( + json!({ + "doc": doc(), + "from": { "type": "integer", "minimum": 1, "description": "Id of the document to copy from." }, + "path": path("The bookmark whose subtree to copy (omit for the whole outline)"), + "parent": path("Parent bookmark in doc (omit for the top level)"), + "position": { "type": "integer", "minimum": 1, "description": "1-based position among the parent's children (default: last)." }, + "replace": { "type": "boolean", "description": "Replace doc's bookmarks instead of adding to them (default false; leave parent out)." } + }), + &["doc", "from"], + )), t("bookmark_rename", "Rename a bookmark", "Change a bookmark's title. Undoable.") .with(schema(json!({ "doc": doc(), "path": path("The bookmark"), "title": { "type": "string", "minLength": 1 } }), &["doc", "path", "title"])), t("bookmark_delete", "Delete a bookmark", "Delete a bookmark and the bookmarks under it. Undoable.") @@ -469,6 +546,20 @@ pub fn tools() -> Vec { }), &["doc", "page", "path"], )), + t("model_add", "Add a 3D model", "Add a 3D model (Acrobat: Rich Media ▸ Add 3D): the U3D or PRC file at path, recognised by its first bytes, as a 3D annotation in rect [x0, y0, x1, y1] (points from the top-left of the displayed page; each side 1–14400 pt; at most 256 MiB). Its poster, what prints and what readers show until the model is activated, is the picture at poster (a PDF page, poster_page default 1, or an image) fitted inside rect keeping its shape and centred; without poster, a frame labelled \"3D model\". activation: click (default), page_open or page_visible. name is its description (default: the file name). No views are added: readers open the model on its own default view. Undoable.") + .with(schema( + json!({ + "doc": doc(), + "page": { "type": "integer", "minimum": 1 }, + "path": { "type": "string" }, + "rect": { "type": "array", "items": { "type": "number" }, "minItems": 4, "maxItems": 4 }, + "poster": { "type": "string" }, + "poster_page": { "type": "integer", "minimum": 1 }, + "activation": { "type": "string", "enum": ["click", "page_open", "page_visible"] }, + "name": { "type": "string" }, + }), + &["doc", "page", "path", "rect"], + )), t("form_set_image", "Set an image field's picture", "Show an image file (PNG, JPEG, TIFF, GIF or BMP) in an image field or button, scaled to fit and centred (what clicking an image field does). On an XFA form the picture is kept in its data too (base64, as Designer saves it) and the form shows it from there. Undoable.") .with(schema(json!({ "doc": doc(), "field": { "type": "string" }, "path": { "type": "string" } }), &["doc", "field", "path"])), t( @@ -528,7 +619,7 @@ pub fn tools() -> Vec { t( "redact_mark", "Mark for redaction", - "Mark content for redaction (nothing is removed until redact_apply). One of: rect [x0, y0, x1, y1] (points from the top-left of the displayed page) with page; find (text, every match); words (a list of words or phrases, every match of each; the result's matched_words says how many marks each made); pattern (phone, email, credit-card, ssn, date: every match, Acrobat's Search & Redact patterns); whole_pages: true. find and words accept case_sensitive and whole_words (both default false). Select one page with page or multiple with pages (default all); do not pass both. rect requires page. overlay: text shown on the box once applied, or code_set (foia: U.S. FOIA (b)(1)(A)…(b)(9); privacy-act: U.S. Privacy Act (d)(5), (j)(1)…(k)(7)) with codes, shown as \"(b)(6), (b)(7)(C)\"; fill: box colour (default black). Undoable.", + "Mark content for redaction (nothing is removed until redact_apply). One of: rect [x0, y0, x1, y1] (points from the top-left of the displayed page) with page; find (text, every match); words (a list of words or phrases, every match of each; the result's matched_words says how many marks each made); pattern (phone, email, credit-card, ssn, date, iban: every match, Acrobat's Search & Redact patterns); whole_pages: true. find and words accept case_sensitive and whole_words (both default false). Select one page with page or multiple with pages (default all); do not pass both. rect requires page. overlay: text shown on the box once applied, or code_set (foia: U.S. FOIA (b)(1)(A)…(b)(9); privacy-act: U.S. Privacy Act (d)(5), (j)(1)…(k)(7)) with codes, shown as \"(b)(6), (b)(7)(C)\"; fill: box colour (default black). Undoable.", ) .with(schema( json!({ @@ -539,7 +630,7 @@ pub fn tools() -> Vec { "words": { "type": "array", "items": { "type": "string", "minLength": 1 }, "minItems": 1, "maxItems": 1000 }, "case_sensitive": { "type": "boolean", "description": "Match letter case for find or words (default false)." }, "whole_words": { "type": "boolean", "description": "Match whole words for find or words (default false)." }, - "pattern": { "type": "string", "enum": ["phone", "email", "credit-card", "ssn", "date"] }, + "pattern": pattern_enum(), "whole_pages": { "type": "boolean" }, "pages": pages("to search or mark (default: all); mutually exclusive with page, and not valid with rect"), "overlay": { "type": "string" }, @@ -781,9 +872,13 @@ pub fn tools() -> Vec { "icon": { "type": "string", "description": "note: Comment, Note, Help, Insert, Key, NewParagraph, Paragraph (the standard names), Check, Circle, Cross, CrossHairs, Star, RightArrow, UpArrow; attachment: PushPin, Paperclip, Graph, Tag." }, "color": color(), "fill": color(), - "opacity": { "type": "number", "minimum": 0, "maximum": 1 }, + "opacity": { "type": "number", "minimum": 0, "maximum": 1, "description": "The whole comment's opacity (its line or border's, where its fill or text has its own)." }, + "fill_opacity": { "type": "number", "minimum": 0, "maximum": 1, "description": "The fill's own opacity: rectangle, oval, polygon, cloud, text box, typewriter text or callout." }, + "text_opacity": { "type": "number", "minimum": 0, "maximum": 1, "description": "The text's own opacity: text box, typewriter text or callout." }, "width": { "type": "number", "minimum": 0, "description": "Line width in points." }, + "border_shape": { "type": "string", "enum": ["rectangle", "oval", "circle"], "description": "A text box's or typewriter text's border shape: an oval or a circle goes around the text box." }, "endings": { "type": "array", "items": { "type": "string", "enum": ["None", "Square", "Circle", "Diamond", "OpenArrow", "ClosedArrow", "Butt", "ROpenArrow", "RClosedArrow", "Slash"] }, "minItems": 1, "maxItems": 2, "description": "Line endings. Two names (start, end) for a line, arrow or polyline; one name for a callout." }, + "ortho": { "type": "boolean", "description": "line/arrow/polygon/cloud/polyline: hold each end or corner level or plumb with the one before, whichever is nearer (Shift while drawing)." }, "font_size": { "type": "number", "exclusiveMinimum": 0 }, }), &["doc", "page", "type"], @@ -914,13 +1009,17 @@ pub fn tools() -> Vec { json!({ "doc": doc(), "sort": { "type": "string", "enum": ["page", "author", "date", "type"] }, "out": save_out, "open": open }), &["doc"], )), - t("comment_edit", "Edit a comment", "Change a comment's text, colour, opacity, line width, font size (text box/typewriter/callout), fill (rectangle/oval/polygon/text box/callout; \"none\" removes it), rectangle (rectangle/oval/text box/stamp), position (`move` [dx, dy] in points) or shape (`points`: a line's two ends, or the corners of connected lines, a polygon or a cloud). Stamps keep their original appearance when resized. One undo step.").with(schema( + t("comment_edit", "Edit a comment", "Change a comment's text, colour, opacity (the whole comment's, or its fill's and text's own), line width, font size (text box/typewriter/callout), fill (rectangle/oval/polygon/text box/callout; \"none\" removes it), border colour (`border_color`, text box/typewriter/callout; \"text\" gives it the text's colour again), rectangle (rectangle/oval/text box/stamp), position (`move` [dx, dy] in points) or shape (`points`: a line's two ends, or the corners of connected lines, a polygon or a cloud). Stamps keep their original appearance when resized. One undo step.").with(schema( comment_ref(json!({ "contents": { "type": "string" }, "color": color(), - "opacity": { "type": "number", "minimum": 0, "maximum": 1 }, + "opacity": { "type": "number", "minimum": 0, "maximum": 1, "description": "The whole comment's opacity: its fill and text drop opacities of their own." }, + "fill_opacity": { "type": "number", "minimum": 0, "maximum": 1, "description": "The fill's own opacity (rectangle, oval, polygon, cloud, text box, typewriter text, callout); the line or border keeps the comment's." }, + "text_opacity": { "type": "number", "minimum": 0, "maximum": 1, "description": "The text's own opacity (text box, typewriter text, callout)." }, "width": { "type": "number", "minimum": 0 }, "fill": { "type": "string", "description": "Interior colour of a rectangle, oval, polygon, text box or callout: #RRGGBB, a colour name, or \"none\" for no fill." }, + "border_color": { "type": "string", "description": "A text box's, typewriter text's or callout's border colour: #RRGGBB, a colour name, or \"text\" for the text's colour." }, + "border_shape": { "type": "string", "enum": ["rectangle", "oval", "circle"], "description": "A text box's or typewriter text's border shape: an oval or a circle goes around the text, which keeps its place." }, "font_size": { "type": "number", "minimum": 1, "maximum": 400, "description": "Font size in points of a text box, typewriter text or callout; the box grows to hold the text." }, "rect": { "type": "array", "items": { "type": "number" }, "minItems": 4, "maxItems": 4 }, "move": point(), @@ -931,6 +1030,25 @@ pub fn tools() -> Vec { t("comment_image_preview", "Preview image signature layers", "Return a PNG of the page without the selected image signature/initials (layer=background, default), or its embedded image with alpha (layer=image). Includes the displayed rectangle, document rotation, image_rotation (how far clockwise the image is shown turned) and annotation opacity. Cache these layers for live placement/resizing; commit once with comment_edit. Does not change the document or undo history.") .ro().with(schema(comment_ref(json!({ "layer": { "type": "string", "enum": ["background", "image"], "default": "background" }, "dpi": { "type": "number", "minimum": 1, "maximum": 600, "default": 96 } })), &["doc"])), t("comment_delete", "Delete a comment", "Delete a comment with its pop-up and replies, or one reply (its id, or page and index, from comment_list) with its pop-up and the replies to it; the comment and its other replies stay. Undoable.").destructive().with(schema(comment_ref(json!({})), &["doc"])), + t( + "comments_copy", + "Copy comments", + "The XFDF of these comments, each with its replies: what Copy puts on the clipboard. Paste it into this or another document with comments_paste. `comments`: [[page, index], …], 1-based as comment_list reports them.", + ) + .ro() + .with(schema( + json!({ + "doc": doc(), + "comments": { "type": "array", "items": { "type": "array", "items": { "type": "integer", "minimum": 1 }, "minItems": 2, "maxItems": 2 }, "minItems": 1 } + }), + &["doc", "comments"], + )), + t( + "comments_paste", + "Paste comments", + "Add the comments of an XFDF text (from comments_copy) to `page` as new comments: copies under names of their own, so nothing in the document is replaced and pasting twice gives two sets; replies keep their copied parent. `at` [x, y] (points, origin at the top-left of the displayed page, y down, as comment_add takes them) puts the centre of the copies' bounding box there; without it they keep their positions. Undoable.", + ) + .with(schema(json!({ "doc": doc(), "page": { "type": "integer", "minimum": 1 }, "xfdf": { "type": "string" }, "at": point() }), &["doc", "page", "xfdf"])), t( "doc_protect", "Protect with passwords", @@ -1261,31 +1379,79 @@ pub fn tools() -> Vec { )), t("js_enabled", "JavaScript on or off", "Preferences ▸ JavaScript ▸ Enable Acrobat JavaScript: set it with `enabled`, or read it. With JavaScript off, field scripts other than Acrobat's AF calls don't run.") .with(schema(json!({ "enabled": { "type": "boolean" } }), &[])), - t("ocr_recognize", "Recognize text (OCR)", "Scan & OCR ▸ Recognize text: render pages, read the words in them and add them as invisible text over the page image, so scanned pages become searchable and selectable (a searchable image; the image is not changed). Pages that already have text are skipped unless skip_text_pages is false; then pictures on those pages are read too, and recognised words that the page's own text already has (an earlier OCR layer included) are left out and counted as already_on_page. Returns each page's recognised text (left out when the document's security doesn't allow copying), word count or why it was skipped. Needs the OCR models (ocr_status). Undoable as one step.") + t("ocr_recognize", "Recognize text (OCR)", "Scan & OCR ▸ Recognize text: render pages, read the words in them and add them as invisible text over the page image, so scanned pages become searchable and selectable (a searchable image; the image is not changed). Pages that already have text are skipped unless skip_text_pages or force_ocr say otherwise; words overlapping existing text are omitted and counted as already_on_page. With output_mode \"editable\" nothing is applied: a NEW document of visible text at the recognised boxes is written to out (which must not be the source; empty pages stay blank). engine and strategy choose the readers (an engine that cannot run is an error named up front); the four preprocessing flags clean the raster first. region reads just that part of one page ([x0, y0, x1, y1] in points from the top-left of the displayed page). Returns each page's recognised text, word count or why it was skipped. Needs the engines (ocr_status). Searchable output is undoable as one step.") .cmd("ocr.recognize") .with(schema( json!({ "doc": doc(), "pages": pages("to recognise (default: all)"), "dpi": { "type": "number", "minimum": 72, "maximum": 600, "description": "Resolution pages are read at (default 300; a scanned page is read at most at its own resolution)." }, - "language": { "type": "string", "enum": ["en"], "description": "Document language (default en)." }, - "skip_text_pages": { "type": "boolean", "description": "Leave pages that already contain text alone (default true). false also reads pictures on such pages, without adding their text again." }, + "language": { "type": "string", "enum": language_enum(), "description": "Document language (default en)." }, + "engine": { "type": "string", "enum": ["auto", "ocrs", "tesseract"], "description": "Which engine reads the pages (default ocrs). auto: ocrs first and tesseract as the second engine, combined by strategy (only the engines that can run take part)." }, + "strategy": { "type": "string", "enum": ["primary", "confidence", "rover"], "description": "How two engines' readings combine: primary = the first engine alone; confidence = the second engine joins when the first reading's mean confidence is under 70 or unknown; rover = both always read, merged. Default: confidence with engine auto, primary otherwise." }, + "output_mode": { "type": "string", "enum": ["searchable", "editable"], "description": "searchable (default): invisible text over the pages of the open document. editable: a NEW _ocr-style file of visible text written to out; the source is not touched." }, + "out": { "type": "string", "description": "Where the editable output is written (required with output_mode editable; a new file: never the source path, never an existing file)." }, + "region": { "type": "array", "items": { "type": "number" }, "minItems": 4, "maxItems": 4, "description": "[x0, y0, x1, y1] in points from the top-left of the displayed page: read only that part. Needs exactly one page." }, + "skip_text_pages": { "type": "boolean", "description": "Leave pages that already contain text alone (default true)." }, + "skip_text_files": { "type": "boolean", "description": "Batch: leave whole files that already contain text alone (they come back skipped, not read)." }, + "force_ocr": { "type": "boolean", "description": "Read pages and files even when the skip options would leave them alone (default false)." }, + "auto_rotate": { "type": "boolean", "description": "Turn pages scanned sideways or upside down upright before reading (default false)." }, + "deskew": { "type": "boolean", "description": "Straighten skewed text lines before reading (default false)." }, + "denoise": { "type": "boolean", "description": "A 3×3 median filter against scan speckle (default false)." }, + "binarize": { "type": "boolean", "description": "Sauvola binarization: ink to black, paper to white (default false)." }, }), &["doc"], )), - t("ocr_recognize_files", "Recognize text in multiple files", "Scan & OCR ▸ Recognize text ▸ In multiple files: read every page of each PDF in paths and write the searchable result into folder under the same name (pages that already have text are left alone unless skip_text_pages is false, as in ocr_recognize). Two inputs that would write the same output file are refused before anything is read or written. Returns, per file, the output path, word count and skipped pages, or the error.") + t("ocr_words", "Words of a recognition (for review)", "The words a recognition finds, for inspection without the GUI: per page, each word's text, its box ([x, y, w, h] in points from the top-left of the displayed page), its confidence and band (high >= 90, medium 70–89, low < 70), which engine read it, and whether it is a suspect (low band). suspects counts the words a reviewer should look at first. Changes nothing on the document. retain_review returns a session-local handle for ocr_review_apply (one per document, at most eight; a new retained recognition replaces the old one). Needs the engines (ocr_status).") + .ro() + .with(schema( + json!({ + "doc": doc(), + "retain_review": { "type": "boolean" }, + "region": { "type": "array", "items": { "type": "number" }, "minItems": 4, "maxItems": 4 }, + "pages": pages("to read (default: all)"), + "dpi": { "type": "number", "minimum": 72, "maximum": 600 }, + "language": { "type": "string", "enum": language_enum() }, + "engine": { "type": "string", "enum": ["auto", "ocrs", "tesseract"] }, + "strategy": { "type": "string", "enum": ["primary", "confidence", "rover"] }, + "auto_rotate": { "type": "boolean" }, + "deskew": { "type": "boolean" }, + "denoise": { "type": "boolean" }, + "binarize": { "type": "boolean" }, + "skip_text_pages": { "type": "boolean", "description": "Leave pages that already contain text alone (default true)." }, + "force_ocr": { "type": "boolean", "description": "Read pages even when the skip options would leave them alone (default false)." }, + }), + &["doc"], + )), + t("ocr_review_apply", "Apply or discard reviewed OCR words", "Accept corrected recognition from ocr_words retain_review. Corrections use 1-based page/word positions and text only; empty text removes a word, omitted words keep their original text and boxes. The stored document guard refuses stale, foreign or consumed reviews. Searchable output is one undoable step; editable requires out and writes a new document (out must not exist yet; the review is kept when it does). discard releases the review without applying it. Input is capped at 65536 words and 4 MiB, never truncated.") + .cmd("ocr.correct") + .with(schema(json!({ "doc": doc(), "review": { "type": "integer", "minimum": 1 }, "discard": { "type": "boolean" }, "output_mode": { "type": "string", "enum": ["searchable", "editable"] }, "out": { "type": "string" }, + "corrections": { "type": "array", "maxItems": 65536, "items": { "type": "object", "additionalProperties": false, "properties": { "page": { "type": "integer", "minimum": 1 }, "word": { "type": "integer", "minimum": 1 }, "text": { "type": "string" } }, "required": ["page", "word", "text"] } } }), &["doc", "review"])), + t("ocr_dictionary", "Read or update an OCR review dictionary", "Read or add/remove words in an OCR Verify dictionary file (one case-insensitive word per UTF-8 line). Choose its per-language ocr-dictionary-.txt path. Paths honor root; updates are atomic. Refuses invalid words, over 65536 entries or over 4 MiB. No document changes.") + .destructive() + .cmd("ocr.correct") + .with(schema(json!({ "path": { "type": "string" }, "add": { "type": "array", "maxItems": 65536, "items": { "type": "string" } }, "remove": { "type": "array", "maxItems": 65536, "items": { "type": "string" } } }), &["path"])), + t("ocr_recognize_files", "Recognize text in multiple files", "Scan & OCR ▸ Recognize text ▸ In multiple files: read every page of each PDF in paths and write the searchable result into folder under the same name. skip_text_files leaves whole files that already contain text alone (they are their own skipped bucket, nothing is written); skip_text_pages leaves single text pages alone; force_ocr reads regardless of both. Every output name is checked before the first page is read: two paths sharing a file name (from different folders) are refused, and a file of that name already in folder aborts the batch, with the names in the error. Each written file carries a low_confidence note when words at or under 60 confidence sit on it (1-based sorted pages). Returns, per file, the output path, word count, skipped pages or the error, plus recognized/words/skipped_files totals. Needs the engines (ocr_status).") .cmd("ocr.recognize_batch") .with(schema( json!({ "paths": { "type": "array", "items": { "type": "string" }, "minItems": 1 }, "folder": { "type": "string" }, "dpi": { "type": "number", "minimum": 72, "maximum": 600 }, - "language": { "type": "string", "enum": ["en"] }, - "skip_text_pages": { "type": "boolean", "description": "Leave pages that already contain text alone (default true); see ocr_recognize." }, + "language": { "type": "string", "enum": language_enum() }, + "engine": { "type": "string", "enum": ["auto", "ocrs", "tesseract"] }, + "strategy": { "type": "string", "enum": ["primary", "confidence", "rover"] }, + "skip_text_pages": { "type": "boolean", "description": "Leave pages that already contain text alone (default true)." }, + "skip_text_files": { "type": "boolean", "description": "Leave whole files that already contain text alone (default false)." }, + "force_ocr": { "type": "boolean", "description": "Read even where the skip options would leave things alone (default false)." }, + "auto_rotate": { "type": "boolean" }, + "deskew": { "type": "boolean" }, + "denoise": { "type": "boolean" }, + "binarize": { "type": "boolean" }, }), &["paths", "folder"], )), - t("ocr_status", "OCR status", "Whether text recognition is available (its models are installed: release packages include them; a source build fetches them with `cargo xtask models`; PDFCRAFT_MODELS overrides), where it looks for them, and the languages it reads.") + t("ocr_status", "OCR status", "Whether text recognition is available, where it looks for models (release packages include them; a source build fetches them with `cargo xtask models`; PDFCRAFT_MODELS overrides), its languages, and the engines: ocrs (built in, models) and tesseract (an installed program, with its path).") .ro() .cmd("ocr.recognize") .with(schema(json!({}), &[])), @@ -1306,7 +1472,7 @@ pub fn tools() -> Vec { t( "fill_sign_add", "Fill & Sign: type text or place a mark", - "Fill in a form that has no fields, as Acrobat's Fill & Sign does: type text (`text`, default 10 pt; `font_size` 4–300 pt; optional `color` as #RRGGBB for text/date, default black), place a check, cross, dot or line, today's date (in Preferences ▸ Date format, see fill_sign_date_format, or `format`; `font_size` also applies), or a signature or initials (either `text` drawn in a script font, or `path` to a local PNG/JPEG image up to 4 MiB and 32 megapixels, scaled down to 4 megapixels, preserving transparency; with `scan: true`, `path` is instead a photo of a signature on paper, within the same limits: the signature is found, and the paper, shadows and table around it are removed, with `sensitivity` 0–100, default 50, picking up fainter ink when higher). `at` [x, y] is in points from the top-left of the page: text's top-left, mark's centre, signature's left edge centred vertically. Image signatures fit within 150 pt wide and 32 pt tall (24 pt for initials). Creates movable, undoable annotations. Typed text the standard fonts can't draw is saved and drawn as `?` (`warnings` in the result says so); a typed signature neither the script font nor its fallback can draw is refused (pass `path` instead).", + "Fill in a form that has no fields, as Acrobat's Fill & Sign does: type text (`text`, default 10 pt; `font_size` 4–300 pt; optional `color` as #RRGGBB for text/date, default black; `border` 0–12 pt around it, default none, in its colour or `border_color`), place a check, cross, dot or line, today's date (in Preferences ▸ Date format, see fill_sign_date_format, or `format`; `font_size` also applies), or a signature or initials (either `text` drawn in a script font, or `path` to a local PNG/JPEG image up to 4 MiB and 32 megapixels, scaled down to 4 megapixels, preserving transparency; with `scan: true`, `path` is instead a photo of a signature on paper, within the same limits: the signature is found, and the paper, shadows and table around it are removed, with `sensitivity` 0–100, default 50, picking up fainter ink when higher). `at` [x, y] is in points from the top-left of the page: text's top-left, mark's centre, signature's left edge centred vertically. Image signatures fit within 150 pt wide and 32 pt tall (24 pt for initials). Creates movable, undoable annotations. Typed text the standard fonts can't draw is saved and drawn as `?` (`warnings` in the result says so); a typed signature neither the script font nor its fallback can draw is refused (pass `path` instead).", ) .cmd("sign.fill.text") .with(schema( @@ -1318,6 +1484,9 @@ pub fn tools() -> Vec { "text": { "type": "string", "minLength": 1 }, "font_size": { "type": "number", "minimum": 4, "maximum": 300, "description": "For type text or date (default 10 pt)." }, "color": color(), + "border": { "type": "number", "minimum": 0, "maximum": 12, "description": "For type text or date: a border around it in points, in the text's colour (default 0, none)." }, + "border_color": { "type": "string", "description": "With border: its own colour, #RRGGBB or a colour name." }, + "border_shape": { "type": "string", "enum": ["rectangle", "oval", "circle"], "description": "With border, for type text or date: the border's shape (default rectangle); an oval or a circle goes around the text." }, "path": { "type": "string", "description": "PNG or JPEG for signature/initials; pass either path or text." }, "scan": { "type": "boolean", "description": "With path: scan a photo of a signature on paper (default false)." }, "sensitivity": { "type": "integer", "minimum": 0, "maximum": 100, "description": "With scan: how faint ink may be, 0–100 (default 50)." }, diff --git a/crates/automation/tests/it/action_failures.rs b/crates/automation/tests/it/action_failures.rs new file mode 100644 index 0000000000..b487103a21 --- /dev/null +++ b/crates/automation/tests/it/action_failures.rs @@ -0,0 +1,66 @@ +//! A step that fails fails its file: the file is reported with an error and nothing is written +//! for it. A step that finds nothing to do is still not a failure. + +use std::sync::Arc; + +use pdfcraft_automation::{Automation, Content}; +use pdfcraft_engine::Session; +use pdfcraft_engine::actions::{Action, Step, run_on}; +use serde_json::json; + +fn pdf() -> Vec { + Session::new().create_from_text("Original", "Original text").unwrap().as_ref().clone() +} + +fn one_step(step: Step) -> Action { + Action { name: "Custom".into(), description: String::new(), steps: vec![step], builtin: false } +} + +/// The standard fonts can't draw the character, so the watermark step fails and no file is produced. +#[test] +fn failing_watermark_step_fails_the_file_and_publishes_nothing() { + let r = run_on(&one_step(Step::AddWatermark("中".into())), "one.pdf", Arc::new(pdf()), |_, _| {}); + assert!(r.is_err(), "a failed step must fail its file: {r:?}"); +} + +/// A comment step on a document with no comments has nothing to do, which is not a failure. +#[test] +fn nothing_to_do_step_still_succeeds() { + let r = run_on(&one_step(Step::FlattenComments), "one.pdf", Arc::new(pdf()), |_, _| {}); + assert!(r.is_ok(), "a step with nothing to do is not a failure: {r:?}"); +} + +/// Through the tool: the failing file gets an error entry and no output, and the folder stays empty. +#[test] +fn action_run_marks_the_failing_file_as_error_without_output() { + let dir = tempfile::tempdir().unwrap(); + let input = dir.path().join("in"); + std::fs::create_dir_all(&input).unwrap(); + let in_file = input.join("one.pdf"); + std::fs::write(&in_file, pdf()).unwrap(); + let out = dir.path().join("out"); + let mut auto = Automation::new().with_root(dir.path()).unwrap(); + let reply = auto.call( + "action_run", + &json!({ + "paths": [in_file.to_string_lossy()], + "folder": out.to_string_lossy(), + "steps": [{ "step": "add_watermark", "arg": "中" }], + }), + ); + let report = match reply.unwrap().remove(0) { + Content::Json(value) => value, + other => panic!("expected a report: {other:?}"), + }; + let entry = &report["files"][0]; + assert!(entry["error"].is_string(), "{report}"); + assert!(entry.get("output").is_none(), "{report}"); + assert_eq!(std::fs::read_dir(&out).unwrap().count(), 0, "nothing is published for a failed file"); +} + +/// A removal step with nothing to remove keeps its existing note and succeeds. +#[test] +fn removal_step_with_nothing_to_remove_still_succeeds() { + let r = run_on(&one_step(Step::RemoveHiddenInformation), "one.pdf", Arc::new(pdf()), |_, _| {}); + assert!(r.is_ok(), "nothing to remove is not a failure: {r:?}"); +} diff --git a/crates/automation/tests/it/automation.rs b/crates/automation/tests/it/automation.rs index 12a40530ad..037b9026dc 100644 --- a/crates/automation/tests/it/automation.rs +++ b/crates/automation/tests/it/automation.rs @@ -843,10 +843,11 @@ fn arabic_ligature_actual_text_is_one_editable_word_through_save_reopen() { } /// Each of these writes to its required `path` and replaces an existing file there, so the MCP -/// annotations must not tell clients the call is read-only or harmless (#130). +/// annotations must not tell clients the call is read-only or harmless (#130). The OCR +/// dictionary's updates rewrite the file in place (lower-cased and sorted), so it counts too. #[test] fn file_writing_tools_are_not_read_only() { - for name in ["doc_export_data", "accessibility_report", "image_save"] { + for name in ["doc_export_data", "accessibility_report", "image_save", "ocr_dictionary"] { let t = tools().into_iter().find(|t| t.name == name).unwrap(); assert!(!t.read_only, "{name} writes a file but advertises read-only"); assert!(t.destructive, "{name} overwrites its path but advertises non-destructive"); @@ -1715,6 +1716,25 @@ fn mcp_default_lists_every_tool_and_has_no_meta_tools() { assert_eq!(rpc(&mut s, 3, "tools/call", json!({ "name": "tool_search" }))["error"]["code"], -32602); } +#[cfg(feature = "mcp")] +#[test] +/// `tools/list` advertises exactly the Search & Redact patterns the registry supports, in `pattern` +/// and by name in the description, so a schema-obeying client can select every one of them. +fn the_advertised_redact_mark_schema_offers_every_pattern() { + use pdfcraft_engine::REDACT_PATTERNS; + let registry: Vec<&str> = REDACT_PATTERNS.iter().map(|p| p.id()).collect(); + let mut s = McpServer::new(Automation::new()); + let list = rpc(&mut s, 1, "tools/list", json!({})); + let mark = list["result"]["tools"].as_array().unwrap().iter().find(|t| t["name"] == "redact_mark").expect("redact_mark is listed"); + let advertised: Vec<&str> = + mark["inputSchema"]["properties"]["pattern"]["enum"].as_array().unwrap().iter().map(|v| v.as_str().unwrap()).collect(); + assert_eq!(advertised, registry, "the advertised pattern enum must be the registry's ids"); + let description = mark["description"].as_str().unwrap(); + for id in ®istry { + assert!(description.contains(id), "the description must name {id}: {description}"); + } +} + #[cfg(feature = "mcp")] #[test] fn mcp_compact_lists_the_core_tools_and_two_meta_tools() { @@ -2135,6 +2155,104 @@ fn numbering_pages_through_tools() { assert!(matches!(a.call("page_number", &json!({ "doc": doc, "from": 1, "to": 1, "style": "klingon" })), Err(ToolError::InvalidArgs(_)))); } +/// Typed text and text boxes take an oval or a circle for their border through the tools; +/// comment_list reports it (a rectangle is not reported). Only typed text, dates and text boxes +/// have one, and only the three shapes. +#[test] +fn text_borders_take_an_oval_or_a_circle_through_the_tools() { + let dir = workdir("border-shape"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + ok( + &mut a, + "fill_sign_add", + json!({ "doc": doc, "page": 1, "type": "text", "at": [20, 40], "text": "12", "border": 1, "border_shape": "circle" }), + ); + ok( + &mut a, + "comment_add", + json!({ "doc": doc, "page": 1, "type": "textbox", "rect": [10, 200, 190, 240], "contents": "Hello", "width": 1, "border_shape": "Oval" }), + ); + let list = |a: &mut Automation| ok(a, "comment_list", json!({ "doc": doc, "page": 1 }))["comments"].as_array().unwrap().clone(); + let shapes: Vec = list(&mut a).iter().map(|c| c["border_shape"].clone()).collect(); + assert_eq!(shapes, [json!("circle"), json!("oval")]); + let typed = list(&mut a)[0].clone(); + // Snug around "12": the box no wider than the text needs, the circle as wide as that box. + let r: Vec = typed["rect"].as_array().unwrap().iter().map(|v| v.as_f64().unwrap()).collect(); + let w = pdfcraft_engine::annot_text::text_width("12", 10.0) + 10.0; + assert!((r[2] - r[0] - w).abs() < 0.05 && (r[3] - r[1] - w).abs() < 0.05, "{r:?}"); + ok(&mut a, "comment_edit", json!({ "doc": doc, "page": typed["page"], "index": typed["index"], "border_shape": "oval" })); + assert_eq!(list(&mut a)[0]["border_shape"], "oval"); + ok(&mut a, "comment_edit", json!({ "doc": doc, "page": typed["page"], "index": typed["index"], "border_shape": "rectangle" })); + assert!(list(&mut a)[0].get("border_shape").is_none(), "a rectangle is not reported"); + for (tool, args) in [ + ("fill_sign_add", json!({ "doc": doc, "page": 1, "type": "check", "at": [60, 60], "border_shape": "oval" })), + ("fill_sign_add", json!({ "doc": doc, "page": 1, "type": "text", "text": "7", "at": [60, 60], "border_shape": "circle" })), + ("comment_add", json!({ "doc": doc, "page": 1, "type": "rectangle", "rect": [80, 10, 120, 40], "border_shape": "oval" })), + ("comment_edit", json!({ "doc": doc, "page": typed["page"], "index": typed["index"], "border_shape": "star" })), + ] { + assert!(matches!(a.call(tool, &args), Err(ToolError::InvalidArgs(_))), "{args}"); + } + let callout = ok( + &mut a, + "comment_add", + json!({ "doc": doc, "page": 1, "type": "callout", "rect": [110, 100, 190, 140], "to": [40, 60], "contents": "Look" }), + ); + assert!(matches!( + a.call("comment_edit", &json!({ "doc": doc, "page": 1, "index": callout["comment"]["index"], "border_shape": "oval" })), + Err(ToolError::Failed(_)) + )); +} + +/// A rectangle's fill and a text box's text each take an opacity of their own through the tools +/// (the line or border keeps the comment's `opacity`); `opacity` sets the whole comment's again. +#[test] +fn comment_parts_take_opacities_of_their_own_through_the_tools() { + let dir = workdir("part-opacity"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + ok( + &mut a, + "comment_add", + json!({ "doc": doc, "page": 1, "type": "rectangle", "rect": [10, 10, 60, 40], "fill": "yellow", "fill_opacity": 0.25 }), + ); + ok( + &mut a, + "comment_add", + json!({ "doc": doc, "page": 1, "type": "textbox", "rect": [10, 200, 190, 240], "contents": "Hello", "fill": "#FFFF99", "width": 1, "fill_opacity": 0.2, "text_opacity": 1 }), + ); + let find = |a: &mut Automation, kind: &str| { + let list = ok(a, "comment_list", json!({ "doc": doc, "page": 1 })); + list["comments"].as_array().unwrap().iter().find(|c| c["type"] == kind).unwrap().clone() + }; + let rect = find(&mut a, "Square"); + assert_eq!((rect["opacity"].as_f64(), rect["fill_opacity"].as_f64(), rect.get("text_opacity")), (Some(1.0), Some(0.25), None)); + let tb = find(&mut a, "FreeText"); + assert_eq!((tb["opacity"].as_f64(), tb["fill_opacity"].as_f64(), tb["text_opacity"].as_f64()), (Some(1.0), Some(0.2), Some(1.0))); + // The fill changes, the text stays; an undo step. + ok(&mut a, "comment_edit", json!({ "doc": doc, "page": tb["page"], "index": tb["index"], "fill_opacity": 0.5 })); + let tb = find(&mut a, "FreeText"); + assert_eq!((tb["fill_opacity"].as_f64(), tb["text_opacity"].as_f64()), (Some(0.5), Some(1.0))); + assert_eq!(ok(&mut a, "edit_undo", json!({ "doc": doc }))["undone"], "Change comment properties"); + ok(&mut a, "edit_redo", json!({ "doc": doc })); + // The whole comment's opacity: everything at it again. + ok(&mut a, "comment_edit", json!({ "doc": doc, "page": rect["page"], "index": rect["index"], "opacity": 0.6 })); + let rect = find(&mut a, "Square"); + assert_eq!((rect["opacity"].as_f64(), rect["fill_opacity"].as_f64()), (Some(0.6), Some(0.6))); + // Only the parts a comment has, at 0–1. + for (tool, args) in [ + ("comment_add", json!({ "doc": doc, "page": 1, "type": "line", "from": [0, 0], "to": [50, 50], "fill_opacity": 0.5 })), + ("comment_add", json!({ "doc": doc, "page": 1, "type": "rectangle", "rect": [80, 10, 120, 40], "text_opacity": 0.5 })), + ("comment_edit", json!({ "doc": doc, "page": rect["page"], "index": rect["index"], "fill_opacity": 2 })), + ] { + assert!(matches!(a.call(tool, &args), Err(ToolError::InvalidArgs(_))), "{args}"); + } + assert!(matches!( + a.call("comment_edit", &json!({ "doc": doc, "page": rect["page"], "index": rect["index"], "text_opacity": 0.5 })), + Err(ToolError::Failed(_)) + )); +} + #[test] fn comments_through_tools() { let dir = workdir("comments"); @@ -2463,6 +2581,16 @@ fn copy_restricted_text_stays_in_the_document() { } let restyled = ok(&mut a, "text_edit", json!({ "doc": re, "page": 1, "paragraph": 1, "size": 30 })); assert!(!restyled.to_string().contains("Page"), "{restyled}"); + // A path that would carry the page's words OUT of the document — the editable output, or + // the review word list — is refused on a copy-restricted document (no engines needed). + match a.call("ocr_recognize", &json!({ "doc": re, "output_mode": "editable", "out": "copy.pdf" })) { + Err(ToolError::Failed(m)) => assert!(m.contains("don't allow copying"), "{m}"), + other => panic!("the editable output is refused: {other:?}"), + } + match a.call("ocr_words", &json!({ "doc": re })) { + Err(ToolError::Failed(m)) => assert!(m.contains("don't allow copying"), "{m}"), + other => panic!("ocr_words is refused: {other:?}"), + } if ok(&mut a, "ocr_status", json!({}))["available"] != true { eprintln!("skipped the OCR part: OCR models not installed"); return; @@ -3786,6 +3914,62 @@ fn enhance_camera_image_through_tools() { )); } +#[test] +fn enhance_scanned_file_cleans_the_scan_in_place_through_tools() { + let dir = workdir("enhance-scan"); + // A 300 × 400 grey scan: dull paper, a dark gutter at the left, text bars and a speck. + let img = image::GrayImage::from_fn(300, 400, |x, y| { + let light = 0.55 + 0.45 * (x as f32 / 60.0).min(1.0); + let ink = if ((60..360).contains(&y) && (y - 60) % 40 < 6 && (60..240).contains(&x)) || (x, y) == (150, 25) { 0.12 } else { 0.78 }; + image::Luma([(ink * light * 255.0) as u8]) + }); + img.save(dir.join("scan.jpg")).unwrap(); + let mut a = auto(&dir); + let scan = ok(&mut a, "doc_create", json!({ "from": "images", "paths": ["scan.jpg"] }))["doc"].as_u64().unwrap(); + ok(&mut a, "image_save", json!({ "doc": scan, "page": 1, "image": 1, "path": "before.jpg" })); + + let r = ok(&mut a, "page_enhance_scan", json!({ "doc": scan, "level": 100 })); + assert_eq!(r["undo"], "Enhance scanned file"); + assert_eq!((r["pages_cleaned"].clone(), r["images"].clone(), r["not_scanned"].clone()), (json!([1]), json!(1), json!([]))); + assert_eq!(r["skipped"], json!([])); + // In place: still one picture of the same size, still a JPEG, now with white paper. + let images = ok(&mut a, "page_images", json!({ "doc": scan, "page": 1 })); + assert_eq!((images["count"].clone(), images["images"][0]["pixels"].clone()), (json!(1), json!([300, 400]))); + let saved = ok(&mut a, "image_save", json!({ "doc": scan, "page": 1, "image": 1, "path": "after" })); + assert_eq!(saved["format"], "jpg"); + let (before, after) = (image::open(dir.join("before.jpg")).unwrap().to_luma8(), image::open(dir.join("after.jpg")).unwrap().to_luma8()); + for (x, y) in [(150, 40), (3, 40), (3, 380), (150, 25)] { + assert!( + before.get_pixel(x, y)[0] < 200 && after.get_pixel(x, y)[0] >= 230, + "({x}, {y}): {:?} → {:?}", + before.get_pixel(x, y), + after.get_pixel(x, y) + ); + } + assert!(after.get_pixel(150, 62)[0] <= 60, "a stroke stays dark"); + // Undo puts the scan back as it was. + ok(&mut a, "edit_undo", json!({ "doc": scan })); + ok(&mut a, "image_save", json!({ "doc": scan, "page": 1, "image": 1, "path": "undone.jpg" })); + assert_eq!(std::fs::read(dir.join("undone.jpg")).unwrap(), std::fs::read(dir.join("before.jpg")).unwrap()); + + // Switches: despeckle alone leaves everything but the speck. + let r = ok(&mut a, "page_enhance_scan", json!({ "doc": scan, "pages": [1], "level": 0, "despeckle": true, "edge_shadows": false })); + assert_eq!(r["images"], 1); + // A document of text pages has no scans: nothing changes, and it says so. + let text = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + let r = ok(&mut a, "page_enhance_scan", json!({ "doc": text })); + assert_eq!((r["images"].clone(), r["not_scanned"].clone(), r["undo"].clone()), (json!(0), json!([1, 2, 3]), Value::Null)); + // Bad arguments are refused before anything changes. + for bad in [ + json!({ "doc": scan, "level": 101 }), + json!({ "doc": scan, "pages": [9] }), + json!({ "doc": scan, "despeckle": "yes" }), + json!({ "doc": scan, "pages": [] }), + ] { + assert!(matches!(a.call("page_enhance_scan", &bad), Err(ToolError::InvalidArgs(_))), "{bad}"); + } +} + #[test] fn enhance_camera_image_finds_the_edges_by_itself() { let dir = workdir("enhance-camera-auto"); @@ -4014,6 +4198,48 @@ fn fill_sign_text_and_date_colours_are_saved_and_validated() { } } +#[test] +fn fill_sign_text_border_is_saved_and_validated() { + let dir = workdir("fill-sign-text-border"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + ok(&mut a, "fill_sign_add", json!({ "doc": doc, "page": 1, "type": "text", "at": [20, 40], "text": "Plain" })); + ok(&mut a, "fill_sign_add", json!({ "doc": doc, "page": 1, "type": "text", "at": [20, 80], "text": "Plain", "border": 1.5 })); + let comments = ok(&mut a, "comment_list", json!({ "doc": doc })); + let width = |k: usize| { + let rect = comments["comments"][k]["rect"].as_array().unwrap(); + rect[2].as_f64().unwrap() - rect[0].as_f64().unwrap() + }; + assert!((width(1) - width(0) - 3.0).abs() < 1e-6, "1.5 pt on each side: {comments}"); + ok(&mut a, "doc_save", json!({ "doc": doc, "path": "framed.pdf" })); + let saved = std::fs::read(dir.join("framed.pdf")).unwrap(); + let pdf = pdfcraft_cos::Document::open(std::sync::Arc::new(saved)).unwrap(); + let page = pdfcraft_model::pages(&pdf).remove(0); + let annotations = pdf.resolve(page.dict.get(b"Annots").unwrap()); + let annotation = pdf.resolve(annotations.as_array().unwrap().get(1).unwrap()); + let bs = annotation.as_dict().unwrap().get(b"BS").unwrap().as_dict().unwrap().get(b"W").unwrap().as_f64().unwrap(); + assert_eq!(bs, 1.5, "saved annotation has the requested border"); + // Its own colour, then the text's again through comment_edit. + ok( + &mut a, + "fill_sign_add", + json!({ "doc": doc, "page": 1, "type": "text", "at": [20, 160], "text": "Red", "border": 1, "border_color": "#ff0000" }), + ); + let list = ok(&mut a, "comment_list", json!({ "doc": doc, "page": 1 })); + let red = list["comments"].as_array().unwrap().iter().find(|c| c["contents"] == "Red").unwrap()["index"].clone(); + ok(&mut a, "comment_edit", json!({ "doc": doc, "page": 1, "index": red, "border_color": "text" })); + assert!( + a.call("fill_sign_add", &json!({ "doc": doc, "page": 1, "type": "text", "at": [20, 200], "text": "No", "border_color": "red" })).is_err(), + "no border" + ); + for (kind, border) in [("text", -1.0), ("text", 13.0), ("check", 1.0)] { + assert!(matches!( + a.call("fill_sign_add", &json!({ "doc": doc, "page": 1, "type": kind, "at": [20, 120], "text": "No", "border": border })), + Err(ToolError::InvalidArgs(_)) + )); + } +} + #[test] fn image_signatures_through_tools_preserve_transparency_and_survive_save() { let dir = workdir("image-signatures"); @@ -4658,6 +4884,35 @@ fn redacting_italian_international_slash_numbers_through_tools() { std::fs::remove_dir_all(dir).unwrap(); } +#[test] +/// A client obeying the advertised schema can select iban, and the mark made through the tool +/// covers the spaced number and nothing else. +fn redacting_an_iban_through_tools() { + let table = tools(); + let advertised: Vec<&str> = table.iter().find(|t| t.name == "redact_mark").unwrap().input_schema["properties"]["pattern"]["enum"] + .as_array() + .unwrap() + .iter() + .map(|v| v.as_str().unwrap()) + .collect(); + assert!(advertised.contains(&"iban"), "the advertised redact_mark schema must offer iban: {advertised:?}"); + let dir = workdir("redact-iban"); + std::fs::write(dir.join("invoice.txt"), "Pay DE89 3704 0044 0532 0130 00 today\nAccount 12345678 is ours\nTotal due").unwrap(); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_create", json!({ "from": "text", "path": "invoice.txt" }))["doc"].as_u64().unwrap(); + let marked = ok(&mut a, "redact_mark", json!({ "doc": doc, "pattern": "iban" })); + assert_eq!(marked["marked"], 1, "the spaced German sample IBAN needs one mark"); + assert!(page_text(&mut a, doc)[0].contains("DE89 3704 0044 0532 0130 00"), "marks alone remove nothing"); + assert_eq!(ok(&mut a, "redact_apply", json!({ "doc": doc }))["applied"], 1); + ok(&mut a, "doc_save", json!({ "doc": doc, "path": "redacted.pdf" })); + let reopened = ok(&mut a, "doc_open", json!({ "path": "redacted.pdf" }))["doc"].as_u64().unwrap(); + let text = &page_text(&mut a, reopened)[0]; + assert!(!text.contains("DE89") && !text.contains("0532"), "{text}"); + assert!(text.lines().any(|line| line.trim() == "Pay"), "only the label before the number remains: {text}"); + assert!(text.contains("Account 12345678") && text.contains("Total due"), "text without an IBAN remains: {text}"); + std::fs::remove_dir_all(dir).unwrap(); +} + /// One page with the text "Secret 1234" and a Redact annotation that has neither a usable /// QuadPoints array nor a Rect, so there is no area to redact. fn redact_pdf_with_unreadable_mark() -> Vec { @@ -6482,6 +6737,44 @@ fn drawing_comments_through_tools() { assert!(matches!(png[0], Content::Png { .. })); } +/// comment_add `ortho`, Shift's hold while drawing: a line's end is held level or plumb with its +/// start, and each corner of connected lines or a polygon with the corner before. +#[test] +fn lines_and_corners_are_held_level_or_plumb_through_the_tool() { + let dir = workdir("ortho"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + for args in [ + json!({ "type": "line", "from": [20, 40], "to": [120, 52], "ortho": true }), + json!({ "type": "arrow", "from": [180, 40], "to": [170, 140], "ortho": true }), + json!({ "type": "polyline", "points": [[20, 200], [90, 210], [95, 260], [130, 250]], "ortho": true }), + json!({ "type": "line", "from": [20, 280], "to": [120, 290] }), + ] { + let mut args = args; + args.as_object_mut().unwrap().extend([("doc".to_string(), json!(doc)), ("page".to_string(), json!(1))]); + ok(&mut a, "comment_add", args); + } + let list = ok(&mut a, "comment_list", json!({ "doc": doc, "page": 1 })); + // In the order added: the list isn't sorted by index. + let mut listed = list["comments"].as_array().unwrap().clone(); + listed.sort_by_key(|c| c["index"].as_u64()); + let points: Vec> = + listed.iter().map(|c| c["points"].as_array().unwrap().iter().map(|p| [p[0].as_f64().unwrap(), p[1].as_f64().unwrap()]).collect()).collect(); + let want: [&[[f64; 2]]; 4] = [ + &[[20.0, 40.0], [120.0, 40.0]], + &[[180.0, 40.0], [180.0, 140.0]], + &[[20.0, 200.0], [90.0, 200.0], [90.0, 260.0], [130.0, 260.0]], + &[[20.0, 280.0], [120.0, 290.0]], + ]; + assert_eq!(points.len(), want.len(), "{points:?}"); + for (got, want) in points.iter().zip(want) { + assert_eq!(got.len(), want.len(), "{got:?}"); + assert!(got.iter().zip(want).all(|(g, w)| (g[0] - w[0]).abs() < 0.01 && (g[1] - w[1]).abs() < 0.01), "{got:?} for {want:?}"); + } + let refused = a.call("comment_add", &json!({ "doc": doc, "page": 1, "type": "rectangle", "rect": [10, 10, 50, 50], "ortho": true })); + assert!(matches!(refused, Err(ToolError::InvalidArgs(_))), "ortho is for lines and corners: {refused:?}"); +} + #[test] fn signing_with_an_image_typed_or_drawn_graphic_through_tools() { let dir = workdir("signing-graphic"); @@ -7265,6 +7558,8 @@ fn ocr_tools_make_a_scan_searchable() { let mut a = auto(&dir); let status = ok(&mut a, "ocr_status", json!({})); assert_eq!(status["languages"][0]["code"], "en"); + assert_eq!(status["engines"][0]["id"], "ocrs", "{status}"); + assert!(status["engines"][1]["id"] == "tesseract" && status["engines"][1]["available"].is_boolean(), "{status}"); if status["available"] != true { eprintln!("skipped: OCR models not installed"); return; @@ -7279,6 +7574,64 @@ fn ocr_tools_make_a_scan_searchable() { let again = ok(&mut a, "ocr_recognize", json!({ "doc": scan, "pages": [1] })); assert!(again["pages"][0]["skipped"].is_string(), "{again}"); assert!(matches!(a.call("ocr_recognize", &json!({ "doc": scan, "language": "xx" })), Err(ToolError::InvalidArgs(_)))); + // An engine that cannot run is named up front, before anything is read. + if status["engines"][1]["available"] != true { + let why = a.call("ocr_recognize", &json!({ "doc": scan, "engine": "tesseract", "skip_text_pages": false })).unwrap_err(); + assert!(matches!(why, ToolError::Failed(_)), "{why}"); + assert!(why.to_string().contains("tesseract"), "{why}"); + } + // ocr_words: the recognition's words with confidence, band and suspects, changing nothing. + // ocrs reports no confidences, so `auto` + a confidence merge is asked for when the second + // engine is there to supply them; without it confidence and band are honestly null. + ok(&mut a, "doc_close", json!({ "doc": scan, "discard_changes": true })); + let scan = ok(&mut a, "doc_create", json!({ "from": "images", "paths": [png.to_string_lossy()] }))["doc"].as_u64().unwrap(); + let mut args = json!({ "doc": scan }); + if status["engines"][1]["available"] == true { + args["engine"] = json!("auto"); + args["strategy"] = json!("confidence"); + } + let words = ok(&mut a, "ocr_words", args); + assert!(words["words"].as_u64().unwrap() >= 4, "{words}"); + let first = &words["pages"][0]["words"][0]; + assert!(first["text"].is_string() && first["box"].as_array().is_some_and(|b| b.len() == 4), "{words}"); + if status["engines"][1]["available"] == true { + let with_confidence = words["pages"][0]["words"] + .as_array() + .expect("words") + .iter() + .filter(|w| w["confidence"].is_number() && w["band"].as_str().is_some()) + .count(); + assert!(with_confidence >= 4, "the second engine's confidences arrive: {words}"); + } else { + assert!(first["confidence"].is_null() && first["band"].is_null() && first["suspect"] == false, "{words}"); + } + assert!(a.call("edit_undo", &json!({ "doc": scan })).is_err(), "ocr_words changes nothing, so there is no undo"); +} + +/// An editable-text recognition writes a NEW file of visible text, never the source. +#[test] +fn ocr_recognize_editable_output_writes_visible_text() { + let dir = workdir("ocr-editable"); + let mut a = auto(&dir); + if ok(&mut a, "ocr_status", json!({}))["available"] != true { + eprintln!("skipped: OCR models not installed"); + return; + } + let text = ok(&mut a, "doc_create", json!({ "from": "text", "text": "Editable output words" }))["doc"].as_u64().unwrap(); + ok(&mut a, "doc_export_images", json!({ "doc": text, "folder": "scan", "dpi": 150, "pages": [1] })); + let png = std::fs::read_dir(dir.join("scan")).unwrap().next().unwrap().unwrap().path(); + let scan = ok(&mut a, "doc_create", json!({ "from": "images", "paths": [png.to_string_lossy()] }))["doc"].as_u64().unwrap(); + ok(&mut a, "doc_save", json!({ "doc": scan, "path": "src/scan.pdf" })); + // The output never lands on the source. + let why = a.call("ocr_recognize", &json!({ "doc": scan, "pages": [1], "output_mode": "editable", "out": "src/scan.pdf" })).unwrap_err(); + assert!(why.to_string().contains("source"), "{why}"); + let r = ok(&mut a, "ocr_recognize", json!({ "doc": scan, "pages": [1], "output_mode": "editable", "out": "out/ocr.pdf" })); + assert!(r["words"].as_u64().unwrap() >= 3, "{r}"); + let ocr = ok(&mut a, "doc_open", json!({ "path": "out/ocr.pdf" }))["doc"].as_u64().unwrap(); + // The text is visible (the engine may misread a letter at this dpi — look for a word that + // survives). + let found = ok(&mut a, "text_find", json!({ "doc": ocr, "query": "output" })); + assert!(found["count"].as_u64() >= Some(1), "the new file's text is visible: {found}"); } /// #1028: `skip_text_pages: false` reads a picture on a page that has its own text, and the @@ -7321,6 +7674,46 @@ fn ocr_recognize_files_refuses_paths_outside_the_root() { assert!(!dir.join("out").exists(), "nothing was started"); } +/// Two inputs of one file name from different folders would write the same output in the +/// batch folder: the batch is refused up front, with or without the OCR engines, naming the +/// file, and nothing is written. +#[test] +fn ocr_recognize_files_refuses_inputs_that_share_a_file_name() { + let dir = workdir("ocr-files-same-name"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + ok(&mut a, "doc_save", json!({ "doc": doc, "path": "one/scan.pdf" })); + ok(&mut a, "doc_save", json!({ "doc": doc, "path": "two/scan.pdf" })); + let r = a.call("ocr_recognize_files", &json!({ "paths": ["one/scan.pdf", "two/scan.pdf"], "folder": "out" })); + assert!(matches!(&r, Err(ToolError::InvalidArgs(m)) if m.contains("scan.pdf") && m.contains("overwrite")), "{r:?}"); + assert!(std::fs::read_dir(dir.join("out")).map_or(true, |mut d| d.next().is_none()), "nothing was written"); +} + +/// `out` belongs to the editable output. A searchable run (the default) writes into the open +/// document, so an `out` is refused rather than silently ignored, with or without the engines. +#[test] +fn ocr_recognize_refuses_an_out_that_nothing_would_use() { + let dir = workdir("ocr-out-ignored"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + let r = a.call("ocr_recognize", &json!({ "doc": doc, "out": "../outside.pdf" })); + assert!(matches!(&r, Err(ToolError::InvalidArgs(m)) if m.contains("editable")), "{r:?}"); + assert!(!dir.join("../outside.pdf").exists()); +} + +/// The editable output is a NEW file: an existing `out` is refused before anything runs (with +/// or without the engines), and the file is left as it was. +#[test] +fn ocr_recognize_editable_output_never_replaces_an_existing_file() { + let dir = workdir("ocr-editable-exists"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + std::fs::write(dir.join("taken.pdf"), b"keep me").unwrap(); + let r = a.call("ocr_recognize", &json!({ "doc": doc, "output_mode": "editable", "out": "taken.pdf" })); + assert!(matches!(&r, Err(e) if e.to_string().contains("already exists")), "{r:?}"); + assert_eq!(std::fs::read(dir.join("taken.pdf")).unwrap(), b"keep me"); +} + #[test] fn ocr_recognize_files_writes_searchable_copies() { let dir = workdir("ocr-files"); @@ -7334,9 +7727,22 @@ fn ocr_recognize_files_writes_searchable_copies() { let png = std::fs::read_dir(dir.join("scan")).unwrap().next().unwrap().unwrap().path(); let scan = ok(&mut a, "doc_create", json!({ "from": "images", "paths": [png.to_string_lossy()] }))["doc"].as_u64().unwrap(); ok(&mut a, "doc_save", json!({ "doc": scan, "path": "in/scan.pdf" })); + // A file that already has text: the skip-files option passes it over as its own bucket. + let typed = ok(&mut a, "doc_create", json!({ "from": "text", "text": "Already typed" }))["doc"].as_u64().unwrap(); + ok(&mut a, "doc_save", json!({ "doc": typed, "path": "in/typed.pdf" })); + // Collision pre-flight: the existing out/scan.pdf (written below) stops a rerun. let r = ok(&mut a, "ocr_recognize_files", json!({ "paths": ["in/scan.pdf", "missing.pdf"], "folder": "out" })); assert!(r["files"][0]["words"].as_u64().unwrap() >= 3, "{r}"); assert!(r["files"][1]["error"].is_string(), "{r}"); + let why = a.call("ocr_recognize_files", &json!({ "paths": ["in/scan.pdf"], "folder": "out" })).unwrap_err(); + assert!(matches!(why, ToolError::Failed(_)), "{why}"); + assert!(why.to_string().contains("scan.pdf"), "{why}"); + // The skip-files bucket and the low-confidence note. + let r = ok(&mut a, "ocr_recognize_files", json!({ "paths": ["in/typed.pdf", "in/scan.pdf"], "folder": "out2", "skip_text_files": true })); + assert_eq!(r["skipped_files"].as_u64(), Some(1), "{r}"); + assert!(r["files"][0]["skipped"].is_string(), "{r}"); + assert_eq!(r["recognized"].as_u64(), Some(1), "{r}"); + assert!(r["files"][1]["output"].is_string(), "{r}"); let out = ok(&mut a, "doc_open", json!({ "path": "out/scan.pdf" }))["doc"].as_u64().unwrap(); let found = ok(&mut a, "text_find", json!({ "doc": out, "query": "recognition" })); assert!(found.to_string().contains("\"page\""), "{found}"); @@ -7346,18 +7752,6 @@ fn ocr_recognize_files_writes_searchable_copies() { /// the second output replaces the first while both results report success. The batch is refused /// up front, naming the file, before the folder is created, anything is read or the OCR engines /// load, so the outcome is the same with or without the models installed. -#[test] -fn ocr_recognize_files_refuses_inputs_that_share_a_file_name() { - let dir = workdir("ocr-files-same-name"); - let mut a = auto(&dir); - let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); - ok(&mut a, "doc_save", json!({ "doc": doc, "path": "one/scan.pdf" })); - ok(&mut a, "doc_save", json!({ "doc": doc, "path": "two/scan.pdf" })); - let r = a.call("ocr_recognize_files", &json!({ "paths": ["one/scan.pdf", "two/scan.pdf"], "folder": "out" })); - assert!(matches!(&r, Err(ToolError::InvalidArgs(m)) if m.contains("scan.pdf")), "{r:?}"); - assert!(!dir.join("out").exists(), "nothing was started: the output folder was not created"); -} - /// File names are compared without regard to case on Windows and macOS (`scan.pdf` and `Scan.PDF` /// are one file there), and exactly elsewhere, where they are two files and the batch is not /// refused. @@ -10239,6 +10633,31 @@ fn changing_a_lists_options_through_tools_keeps_its_saved_selection_on_the_same_ assert_eq!(saved_selection_indices(&dir.join("letters.pdf"), "letters"), Some(vec![0, 1]), "B and C are now items 0 and 1"); } +/// Clearing a field's length limit through `form_set_props` (`"max_length": 0`) leaves it with +/// none, even when its parent field has one it would otherwise inherit. +#[test] +fn clearing_a_kids_length_limit_through_tools_does_not_bring_back_its_parents() { + let objs = [ + "<< /Type /Catalog /Pages 2 0 R /AcroForm 4 0 R >>", + "<< /Type /Pages /Kids [3 0 R] /Count 1 /MediaBox [0 0 400 300] >>", + "<< /Type /Page /Parent 2 0 R /Annots [6 0 R] >>", + "<< /Fields [5 0 R] /DA (/Helv 12 Tf 0 g) /DR << /Font << /Helv 7 0 R >> >> >>", + "<< /T (address) /FT /Tx /MaxLen 3 /Kids [6 0 R] >>", + "<< /Type /Annot /Subtype /Widget /T (city) /Parent 5 0 R /MaxLen 8 /Rect [10 250 150 268] /P 3 0 R >>", + "<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica /Encoding /WinAnsiEncoding >>", + ]; + let dir = workdir("kid-max-len"); + std::fs::write(dir.join("form.pdf"), pdf_of(&objs)).unwrap(); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "form.pdf" }))["doc"].as_u64().unwrap(); + ok(&mut a, "form_set_props", json!({ "doc": doc, "field": "address.city", "max_length": 0 })); + ok(&mut a, "form_fill", json!({ "doc": doc, "values": { "address.city": "Absalom" } })); + let fields = ok(&mut a, "form_fields", json!({ "doc": doc })); + let city = fields["fields"].as_array().unwrap().iter().find(|f| f["name"] == "address.city").unwrap().clone(); + assert!(city.get("max_length").is_none(), "no limit, not the parent's 3: {city}"); + assert_eq!(city["value"], "Absalom", "the value is not cut to the parent's 3 characters"); +} + /// Clearing a field's default through `form_set_props` (`"default": ""`) leaves it with none, /// even when its parent field has one it would otherwise inherit. #[test] @@ -10508,6 +10927,162 @@ fn simple_rule_scan_drawn(content: &[u8]) -> Vec { write_full(&doc, &SaveOptions::default()).unwrap() } +/// Reviewing headlessly: corrections keep the boxes, a stale or foreign review is refused, and +/// applying consumes the handle (the pages already accepted were applied once). +#[test] +fn ocr_review_corrections_keep_boxes_are_guarded_and_apply_once() { + fn render(a: &mut Automation, doc: u64) -> Vec { + let mut result = a.call("page_render", &json!({ "doc": doc, "page": 1, "dpi": 72 })).unwrap(); + match result.remove(0) { + Content::Png { data, .. } => data, + other => panic!("expected a rendered page, got {other:?}"), + } + } + let dir = workdir("ocr-review"); + let mut a = auto(&dir); + if ok(&mut a, "ocr_status", json!({}))["available"] != true { + eprintln!("skipped: OCR models not installed"); + return; + } + let text = ok(&mut a, "doc_create", json!({ "from": "text", "text": "Review these original words" }))["doc"].as_u64().unwrap(); + ok(&mut a, "doc_export_images", json!({ "doc": text, "folder": "scan", "dpi": 150, "pages": [1] })); + let png = std::fs::read_dir(dir.join("scan")).unwrap().next().unwrap().unwrap().path(); + let scan = ok(&mut a, "doc_create", json!({ "from": "images", "paths": [png.to_string_lossy()] }))["doc"].as_u64().unwrap(); + let before = render(&mut a, scan); + let found = ok(&mut a, "ocr_words", json!({ "doc": scan, "retain_review": true })); + assert!(found["words"].as_u64().unwrap() >= 3, "{found}"); + let review = found["review"].as_u64().unwrap(); + assert!(page_text(&mut a, scan).iter().all(String::is_empty)); + for args in [ + json!({ "doc": text, "review": review }), + json!({ "doc": scan, "review": review, "corrections": [{"page":1,"word":999999,"text":"bad"}] }), + json!({ "doc": scan, "review": review, "corrections": [{"page":1,"word":1,"text":"bad","origin":[0,0]}] }), + json!({ "doc": scan, "review": review, "corrections": [{"page":1,"word":1,"text":"a".repeat(4*1024*1024+1)}] }), + json!({ "doc": scan, "review": review, "corrections": [{"page":1,"word":1,"text":"a"},{"page":1,"word":1,"text":"b"}] }), + ] { + assert!(a.call("ocr_review_apply", &args).is_err()); + assert!(page_text(&mut a, scan).iter().all(String::is_empty)); + } + let result = ok( + &mut a, + "ocr_review_apply", + json!({ "doc": scan, "review": review, "corrections": [{"page":1,"word":1,"text":"日本語"},{"page":1,"word":2,"text":""}] }), + ); + assert!(result["applied"].as_u64().unwrap() >= 2); + let extracted = page_text(&mut a, scan).join("\n"); + assert!(extracted.contains("日本語"), "{extracted}"); + assert_eq!(render(&mut a, scan), before, "corrected text remains invisible at its original geometry"); + assert!(a.call("ocr_review_apply", &json!({ "doc": scan, "review": review })).is_err(), "consumed reviews cannot be replayed"); + ok(&mut a, "edit_undo", json!({ "doc": scan })); + assert!(page_text(&mut a, scan).iter().all(String::is_empty)); + assert!(a.call("edit_undo", &json!({ "doc": scan })).is_err(), "review acceptance is one undo step"); + + let found = ok(&mut a, "ocr_words", json!({ "doc": scan, "retain_review": true })); + let stale = found["review"].as_u64().unwrap(); + ok(&mut a, "page_rotate", json!({ "doc": scan, "pages": [1], "degrees": 90 })); + let error = a.call("ocr_review_apply", &json!({ "doc": scan, "review": stale })).unwrap_err().to_string(); + assert!(error.contains("changed since recognition"), "{error}"); + assert!(page_text(&mut a, scan).iter().all(String::is_empty)); + ok(&mut a, "ocr_review_apply", json!({ "doc": scan, "review": stale, "discard": true })); + assert!(a.call("ocr_review_apply", &json!({ "doc": scan, "review": stale })).is_err()); + ok(&mut a, "edit_undo", json!({ "doc": scan })); + + let first = ok(&mut a, "ocr_words", json!({ "doc": scan, "retain_review": true }))["review"].as_u64().unwrap(); + let second = ok(&mut a, "ocr_words", json!({ "doc": scan, "retain_review": true }))["review"].as_u64().unwrap(); + assert_ne!(first, second); + assert!(a.call("ocr_review_apply", &json!({ "doc": scan, "review": first })).is_err()); + ok( + &mut a, + "ocr_review_apply", + json!({ "doc": scan, "review": second, "output_mode": "editable", "out": "reviewed.pdf", "corrections": [{"page":1,"word":1,"text":"Corrected"}] }), + ); + assert!(page_text(&mut a, scan).iter().all(String::is_empty)); + let output = ok(&mut a, "doc_open", json!({ "path": "reviewed.pdf" }))["doc"].as_u64().unwrap(); + assert!(page_text(&mut a, output).join("\n").contains("Corrected")); + // The editable output is a new file: an existing one is refused and the review is kept. + let third = ok(&mut a, "ocr_words", json!({ "doc": scan, "retain_review": true }))["review"].as_u64().unwrap(); + let again = a.call("ocr_review_apply", &json!({ "doc": scan, "review": third, "output_mode": "editable", "out": "reviewed.pdf" })); + assert!(matches!(&again, Err(e) if e.to_string().contains("already exists")), "{again:?}"); + ok(&mut a, "ocr_review_apply", json!({ "doc": scan, "review": third, "discard": true })); +} + +#[test] +fn ocr_dictionary_updates_are_atomic_bounded_and_confined() { + let dir = workdir("ocr-dictionary"); + let mut a = auto(&dir); + let path = "ocr-dictionary-en.txt"; + let out = ok(&mut a, "ocr_dictionary", json!({ "path": path, "add": ["Quartz", "quartz", "日本語"] })); + assert_eq!(out["count"], 2); + assert_eq!(pdfcraft_engine::ocr::dictionary::words(&std::fs::read_to_string(dir.join(path)).unwrap()).unwrap().len(), 2); + let before = std::fs::read(dir.join(path)).unwrap(); + for args in [ + json!({ "path": path, "add": ["a\nb"] }), + json!({ "path": path, "add": ["x".repeat(1025)] }), + json!({ "path": path, "add": ["İ".repeat(512)] }), + json!({ "path": path, "add": vec!["x"; 65537] }), + json!({ "path": "../outside-ocr-dictionary.txt", "add": ["x"] }), + ] { + assert!(a.call("ocr_dictionary", &args).is_err()); + } + assert_eq!(std::fs::read(dir.join(path)).unwrap(), before); + assert_eq!(ok(&mut a, "ocr_dictionary", json!({ "path": path, "remove": ["QUARTZ"] }))["words"], json!(["日本語"])); + assert_eq!(ok(&mut a, "ocr_dictionary", json!({ "path": path }))["count"], 1); + std::fs::write(dir.join("ocr-dictionary-large.txt"), vec![b'a'; 4 * 1024 * 1024 + 1]).unwrap(); + let too_big = a.call("ocr_dictionary", &json!({ "path": "ocr-dictionary-large.txt", "add": ["x"] })).unwrap_err().to_string(); + assert!(!too_big.contains("must name a dictionary file"), "refused for its size, not its name: {too_big}"); + assert_eq!(std::fs::metadata(dir.join("ocr-dictionary-large.txt")).unwrap().len(), 4 * 1024 * 1024 + 1); +} + +/// The tool rewrites what it is pointed at (lower-cased, sorted), so it only takes dictionary file +/// names: another small text file under the root is neither read nor changed. +#[test] +fn ocr_dictionary_refuses_files_that_are_not_dictionaries() { + let dir = workdir("ocr-dictionary-names"); + let mut a = auto(&dir); + let env = b"TOKEN=Zebra +APPLE=1 +"; + for name in [".env", "notes.txt", "ocr-dictionary-en.md", "ocr-dictionary-e n.txt", "sub/readme.txt"] { + if let Some(parent) = std::path::Path::new(name).parent() { + std::fs::create_dir_all(dir.join(parent)).unwrap(); + } + std::fs::write(dir.join(name), env).unwrap(); + for args in [json!({ "path": name }), json!({ "path": name, "add": [] }), json!({ "path": name, "add": ["x"] })] { + let err = a.call("ocr_dictionary", &args).unwrap_err().to_string(); + assert!(err.contains("must name a dictionary file"), "{name} {args}: {err}"); + } + assert_eq!(std::fs::read(dir.join(name)).unwrap(), env, "{name} is untouched"); + } +} + +#[test] +fn ocr_retained_reviews_have_eight_document_cap_and_close_releases_slot() { + let dir = workdir("ocr-review-cap"); + let mut a = auto(&dir); + if ok(&mut a, "ocr_status", json!({}))["available"] != true { + eprintln!("skipped: OCR models not installed"); + return; + } + let mut docs = Vec::new(); + let mut handles = Vec::new(); + for _ in 0..9 { + let doc = ok(&mut a, "doc_create", json!({"from":"text", "text":"Already recognized text"}))["doc"].as_u64().unwrap(); + docs.push(doc); + } + for doc in docs.iter().take(8) { + handles.push(ok(&mut a, "ocr_words", json!({"doc":doc,"retain_review":true}))["review"].as_u64().unwrap()); + } + let error = a.call("ocr_words", &json!({"doc":docs[8],"retain_review":true})).unwrap_err().to_string(); + assert!(error.contains("at most 8"), "{error}"); + let replacement = ok(&mut a, "ocr_words", json!({"doc":docs[0],"retain_review":true}))["review"].as_u64().unwrap(); + assert_ne!(replacement, handles[0]); + assert!(a.call("ocr_review_apply", &json!({"doc":docs[0],"review":handles[0],"discard":true})).is_err()); + ok(&mut a, "doc_close", json!({"doc":docs[0],"discard_changes":true})); + assert!(a.call("ocr_review_apply", &json!({"doc":docs[0],"review":replacement})).is_err()); + let final_handle = ok(&mut a, "ocr_words", json!({"doc":docs[8],"retain_review":true}))["review"].as_u64().unwrap(); + ok(&mut a, "ocr_review_apply", json!({"doc":docs[8],"review":final_handle,"discard":true})); +} + #[test] fn editor_crop_group_move_and_arrange_survive_added_item_deletion_through_tools() { let dir = workdir("editor-crop-group-arrange"); @@ -10687,3 +11262,214 @@ fn multiline_fill_sign_engine_round_trip_preserves_lines_and_appearance() { assert_eq!(rendered, re.call("page_render", &json!({"doc":reopened,"page":1,"dpi":72})).unwrap(), "saved appearances are identical"); std::fs::remove_dir_all(dir).unwrap(); } + +/// Export, then import into another document and copy between open documents: the tree comes +/// back unchanged, a page the target lacks is reported, and the import undoes in one step. +#[test] +fn bookmarks_export_import_and_copy_between_documents() { + let dir = workdir("bookmark-io"); + let mut a = auto(&dir); + let src = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + ok(&mut a, "bookmark_add", json!({ "doc": src, "title": "Intro", "page": 1 })); + ok(&mut a, "bookmark_add", json!({ "doc": src, "title": "Detail", "page": 3, "parent": [1] })); + ok(&mut a, "bookmark_add", json!({ "doc": src, "title": "End", "page": 2 })); + ok(&mut a, "bookmark_set_style", json!({ "doc": src, "path": [1], "color": [1, 0, 0], "bold": true })); + let exported = ok(&mut a, "bookmark_export", json!({ "doc": src, "out": "marks.json" })); + assert_eq!(exported["truncated"], false); + let tree = exported["bookmarks"].clone(); + assert_eq!((tree[0]["title"].clone(), tree[0]["color"].clone(), tree[0]["bold"].clone()), (json!("Intro"), json!([1.0, 0.0, 0.0]), json!(true))); + assert_eq!((tree[0]["children"][0]["title"].clone(), tree[0]["children"][0]["page"].clone()), (json!("Detail"), json!(3))); + assert_eq!(tree[0]["view"]["type"], "XYZ"); + let subtree = ok(&mut a, "bookmark_export", json!({ "doc": src, "path": [1, 1] }))["bookmarks"].clone(); + assert_eq!(subtree, json!([tree[0]["children"][0]])); + + // b.pdf has two pages: Detail goes to the third, so it arrives without a destination. + let dst = ok(&mut a, "doc_open", json!({ "path": "b.pdf" }))["doc"].as_u64().unwrap(); + let imported = ok(&mut a, "bookmark_import", json!({ "doc": dst, "file": "marks.json" })); + assert_eq!(imported["unmapped"], json!([[1, 1]])); + let back = ok(&mut a, "bookmark_export", json!({ "doc": dst }))["bookmarks"].clone(); + assert_eq!(back[0]["children"][0]["page"], json!(null)); + assert_eq!((back[0]["title"].clone(), back[1].clone()), (tree[0]["title"].clone(), tree[1].clone())); + ok(&mut a, "edit_undo", json!({ "doc": dst })); + assert_eq!(ok(&mut a, "bookmark_list", json!({ "doc": dst }))["bookmarks"], json!([])); + + // Into a document with every page, the tree is unchanged, wherever it is copied from. + let same = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + ok(&mut a, "bookmark_add", json!({ "doc": same, "title": "Old", "page": 1 })); + let copied = ok(&mut a, "bookmark_copy", json!({ "doc": same, "from": src, "replace": true })); + assert_eq!(copied["unmapped"], json!([])); + assert_eq!(ok(&mut a, "bookmark_export", json!({ "doc": same }))["bookmarks"], tree); + ok(&mut a, "bookmark_import", json!({ "doc": same, "bookmarks": subtree, "parent": [2], "position": 1 })); + assert_eq!(ok(&mut a, "bookmark_list", json!({ "doc": same }))["bookmarks"][1]["children"][0]["title"], "Detail"); +} + +/// A one-page document whose outline has an untitled item, a /FitR destination with three of +/// its four numbers, and a chain nested `deep` levels below them. +fn odd_outline(deep: usize) -> Vec { + let first_deep = 7; + let mut objs = vec![ + "<< /Type /Catalog /Pages 2 0 R /Outlines 4 0 R >>".to_string(), + "<< /Type /Pages /Kids [3 0 R] /Count 1 /MediaBox [0 0 200 300] >>".into(), + "<< /Type /Page /Parent 2 0 R >>".into(), + "<< /Type /Outlines /First 5 0 R /Last 6 0 R >>".into(), + "<< /Title () /Parent 4 0 R /Next 6 0 R /Dest [3 0 R /Fit] >>".into(), + format!( + "<< /Title (Short) /Parent 4 0 R /Prev 5 0 R /Dest [3 0 R /FitR 0 0 10]{} >>", + if deep > 0 { format!(" /First {first_deep} 0 R /Last {first_deep} 0 R") } else { String::new() } + ), + ]; + for i in 0..deep { + let (me, parent) = (first_deep + i, if i == 0 { 6 } else { first_deep + i - 1 }); + let child = if i + 1 < deep { format!(" /First {0} 0 R /Last {0} 0 R", me + 1) } else { String::new() }; + objs.push(format!("<< /Title (Level) /Parent {parent} 0 R /Dest [3 0 R /Fit]{child} >>")); + } + pdf_of(&objs.iter().map(String::as_str).collect::>()) +} + +/// What a real file holds and the JSON form can't say exactly still copies and round-trips: an +/// empty title stays empty, a short /FitR keeps its page and drops the view, and bookmarks past +/// the listing depth are reported as left out at every step. +#[test] +fn odd_bookmarks_round_trip_and_a_cut_tree_says_so() { + let dir = workdir("bookmark-io-odd"); + std::fs::write(dir.join("odd.pdf"), odd_outline(0)).unwrap(); + std::fs::write(dir.join("deep.pdf"), odd_outline(40)).unwrap(); + let mut a = auto(&dir); + let src = ok(&mut a, "doc_open", json!({ "path": "odd.pdf" }))["doc"].as_u64().unwrap(); + let tree = ok(&mut a, "bookmark_export", json!({ "doc": src, "out": "odd.json" }))["bookmarks"].clone(); + assert_eq!((tree[0]["title"].clone(), tree[1]["page"].clone(), tree[1].get("view")), (json!(""), json!(1), None)); + let dst = ok(&mut a, "doc_open", json!({ "path": "b.pdf" }))["doc"].as_u64().unwrap(); + assert_eq!(ok(&mut a, "bookmark_import", json!({ "doc": dst, "file": "odd.json", "replace": true }))["truncated"], false); + // A destination with no view is written as XYZ with nulls: the same page, the reader's view. + let back = ok(&mut a, "bookmark_export", json!({ "doc": dst }))["bookmarks"].clone(); + assert_eq!((back[0].clone(), back[1]["title"].clone(), back[1]["page"].clone()), (tree[0].clone(), json!("Short"), json!(1))); + assert_eq!(back[1]["view"], json!({ "type": "XYZ", "values": [null, null, null] })); + let copied = ok(&mut a, "bookmark_copy", json!({ "doc": dst, "from": src, "replace": true })); + assert_eq!((copied["unmapped"].clone(), copied["truncated"].clone()), (json!([]), json!(false))); + match a.call("bookmark_export", &json!({ "doc": src, "path": [3] })) { + Err(ToolError::Failed(m)) => assert!(m.contains("no bookmark at [3]"), "{m}"), + other => panic!("expected no bookmark, got {other:?}"), + } + + let deep = ok(&mut a, "doc_open", json!({ "path": "deep.pdf" }))["doc"].as_u64().unwrap(); + assert_eq!(ok(&mut a, "bookmark_export", json!({ "doc": deep, "out": "deep.json" }))["truncated"], true); + let saved: Value = serde_json::from_slice(&std::fs::read(dir.join("deep.json")).unwrap()).unwrap(); + assert_eq!(saved["truncated"], true); + assert_eq!(ok(&mut a, "bookmark_import", json!({ "doc": dst, "file": "deep.json", "replace": true }))["truncated"], true); + assert_eq!(ok(&mut a, "bookmark_copy", json!({ "doc": dst, "from": deep, "replace": true }))["truncated"], true); + let export = tools().into_iter().find(|t| t.name == "bookmark_export").unwrap(); + assert!(export.destructive && !export.read_only, "bookmark_export overwrites out"); + std::fs::remove_dir_all(dir).unwrap(); +} + +#[test] +fn a_misspelt_bookmark_key_is_refused_not_dropped() { + let dir = workdir("bookmark-io-key"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + let tree = json!([{ "title": "Parent", "children": [{ "title": "Red", "page": 1, "colour": [1, 0, 0] }] }]); + match a.call("bookmark_import", &json!({ "doc": doc, "bookmarks": tree })) { + Err(ToolError::InvalidArgs(m)) => assert!(m.contains("[1, 1]") && m.contains("\"colour\""), "{m}"), + other => panic!("expected InvalidArgs, got {other:?}"), + } + assert_eq!(ok(&mut a, "bookmark_list", json!({ "doc": doc }))["bookmarks"], json!([])); + std::fs::remove_dir_all(dir).unwrap(); +} + +#[test] +fn a_bad_bookmark_tree_is_refused_whole_and_names_the_node() { + let dir = workdir("bookmark-io-bad"); + let mut a = auto(&dir); + let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + let refused = |a: &mut Automation, bookmarks: Value| match a.call("bookmark_import", &json!({ "doc": doc, "bookmarks": bookmarks })) { + Err(ToolError::InvalidArgs(m)) => m, + other => panic!("expected InvalidArgs, got {other:?}"), + }; + let fine = json!({ "title": "Fine", "page": 1 }); + let m = refused(&mut a, json!([fine, { "title": "Parent", "children": [{ "page": 1 }] }])); + assert!(m.contains("[2, 1]") && m.contains("title"), "{m}"); + let m = refused(&mut a, json!([{ "title": 5 }])); + assert!(m.contains("[1]") && m.contains("title"), "{m}"); + let m = refused(&mut a, json!([fine, { "title": "Bad", "page": 0 }])); + assert!(m.contains("[2]") && m.contains("page"), "{m}"); + let m = refused(&mut a, json!([{ "title": "Bad", "color": [2, 0, 0] }])); + assert!(m.contains("color"), "{m}"); + let m = refused(&mut a, json!([{ "title": "Bad", "view": { "type": "XYZ", "values": [1, 2, 3, 4, 5] } }])); + assert!(m.contains("view"), "{m}"); + let mut deep = json!({ "title": "Leaf" }); + for _ in 0..40 { + deep = json!({ "title": "Level", "children": [deep] }); + } + assert!(refused(&mut a, json!([deep])).contains("33 levels")); + let m = refused(&mut a, json!([fine, { "title": "Z", "page": 1, "view": { "type": "Zoom" } }])); + assert!(m.contains("[2]") && m.contains("XYZ (3 values"), "{m}"); + let m = refused(&mut a, json!([fine, { "title": "Z", "page": 1, "view": { "type": "FitH", "values": [1, 2] } }])); + assert!(m.contains("[2]") && m.contains("FitH view takes 1"), "{m}"); + let m = refused(&mut a, json!([{ "title": "R", "view": { "type": "FitR", "values": [0, null, 1, 1] } }])); + assert!(m.contains("all four"), "{m}"); + assert!(matches!(a.call("bookmark_import", &json!({ "doc": doc })), Err(ToolError::InvalidArgs(_)))); + assert!(matches!(a.call("bookmark_import", &json!({ "doc": doc, "bookmarks": [fine], "file": "x.json" })), Err(ToolError::InvalidArgs(_)))); + // A real bookmark file outside the root is refused for being outside, not for being missing. + let outside = dir.with_extension("outside.json"); + std::fs::write(&outside, br#"{"bookmarks": [{"title": "Outside", "page": 1}]}"#).unwrap(); + let name = format!("../{}", outside.file_name().unwrap().to_string_lossy()); + match a.call("bookmark_import", &json!({ "doc": doc, "file": name })) { + Err(ToolError::Failed(m)) => assert!(m.contains("outside the allowed directory"), "{m}"), + other => panic!("expected the root to refuse it, got {other:?}"), + } + let _ = std::fs::remove_file(&outside); + let out = dir.with_extension("out.json"); + let name = format!("../{}", out.file_name().unwrap().to_string_lossy()); + match a.call("bookmark_export", &json!({ "doc": doc, "out": name })) { + Err(ToolError::Failed(m)) => assert!(m.contains("outside the allowed directory"), "{m}"), + other => panic!("expected the root to refuse it, got {other:?}"), + } + assert!(!out.exists()); + assert_eq!(ok(&mut a, "bookmark_list", json!({ "doc": doc }))["bookmarks"], json!([])); +} + +/// Copy and paste comments between documents through the tools: the copies carry their replies, +/// land on the asked-for page under their own names (a second paste gives a second set), and a +/// plain text is refused. +#[test] +fn comments_copy_and_paste_between_documents_through_tools() { + let dir = workdir("comments-clipboard"); + let mut a = auto(&dir); + let src = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap(); + ok(&mut a, "comment_add", json!({ "doc": src, "page": 2, "type": "note", "at": [30, 30], "contents": "Please review", "author": "Ada" })); + ok(&mut a, "comment_add", json!({ "doc": src, "page": 2, "type": "rectangle", "rect": [20, 20, 80, 60], "contents": "Box" })); + ok(&mut a, "comment_reply", json!({ "doc": src, "page": 2, "index": 1, "text": "Done", "author": "Bob" })); + // Positions as comment_list reports them (a note's pop-up takes a position of its own). + let listed = ok(&mut a, "comment_list", json!({ "doc": src, "page": 2 })); + let pairs: Vec = listed["comments"].as_array().unwrap().iter().map(|c| json!([c["page"], c["index"]])).collect(); + assert_eq!(pairs.len(), 2, "{listed}"); + let copied = ok(&mut a, "comments_copy", json!({ "doc": src, "comments": pairs })); + assert_eq!(copied["comments"], 2); + let xfdf = copied["xfdf"].as_str().unwrap().to_owned(); + assert!(xfdf.contains("inreplyto=") && xfdf.contains("> = + placed["comments"].as_array().unwrap().iter().map(|c| c["rect"].as_array().unwrap().iter().map(|v| v.as_f64().unwrap()).collect()).collect(); + let (x0, y0, x1, y1) = + rects.iter().fold((f64::MAX, f64::MAX, f64::MIN, f64::MIN), |(a, b, c, d), r| (a.min(r[0]), b.min(r[1]), c.max(r[2]), d.max(r[3]))); + assert!(((x0 + x1) / 2.0 - 100.0).abs() < 1.0 && ((y0 + y1) / 2.0 - 150.0).abs() < 1.0, "centred at [100, 150]: {rects:?}"); + assert!(matches!(a.call("comments_paste", &json!({ "doc": dst, "page": 1, "xfdf": "hello" })), Err(ToolError::InvalidArgs(_)))); + assert!(a.call("comments_copy", &json!({ "doc": src, "comments": [[9, 9]] })).is_err(), "no such comment"); + assert!(matches!(a.call("comments_copy", &json!({ "doc": src, "comments": [[2]] })), Err(ToolError::InvalidArgs(_)))); + assert_eq!(ok(&mut a, "comment_list", json!({ "doc": src }))["comments"].as_array().unwrap().len(), 2, "the source is untouched"); +} diff --git a/crates/automation/tests/it/automation_roots.rs b/crates/automation/tests/it/automation_roots.rs index a96870ca02..e65c582a73 100644 --- a/crates/automation/tests/it/automation_roots.rs +++ b/crates/automation/tests/it/automation_roots.rs @@ -145,6 +145,8 @@ fn tools_that_read_a_path_are_held_to_the_read_root() { ok(&mut server, "doc_open", json!({"path": s(&base.join("in/a.pdf"))})); let (secret, pic, gone) = (s(&base.join("outside/secret.pdf")), s(&base.join("outside/pic.png")), s(&base.join("outside/gone.bin"))); let out = s(&base.join("out/result.pdf")); + std::fs::write(base.join("in/m.prc"), b"PRC model").unwrap(); + let model = s(&base.join("in/m.prc")); for (tool, args) in [ ("page_insert_file", json!({"doc": 1, "path": secret, "at": 1})), ("page_replace", json!({"doc": 1, "pages": [1], "path": secret})), @@ -153,6 +155,8 @@ fn tools_that_read_a_path_are_held_to_the_read_root() { ("doc_create", json!({"from": "images", "paths": [pic]})), ("stamp_custom", json!({"doc": 1, "page": 1, "path": secret, "at": [10, 10]})), ("stamp_custom", json!({"doc": 1, "page": 1, "path": secret, "rect": [10, 10, 493, 300]})), + ("model_add", json!({"doc": 1, "page": 1, "path": secret, "rect": [10, 10, 100, 100]})), + ("model_add", json!({"doc": 1, "page": 1, "path": model, "rect": [10, 10, 100, 100], "poster": pic})), ("page_add_image", json!({"doc": 1, "page": 1, "path": pic})), ("image_edit", json!({"doc": 1, "page": 1, "image": 1, "action": "replace", "path": pic})), ("comment_add", json!({"doc": 1, "page": 1, "type": "attachment", "at": [10, 10], "path": secret})), diff --git a/crates/automation/tests/it/batch_failures.rs b/crates/automation/tests/it/batch_failures.rs new file mode 100644 index 0000000000..f993d85eab --- /dev/null +++ b/crates/automation/tests/it/batch_failures.rs @@ -0,0 +1,89 @@ +//! A batch run that fails a file publishes only the completed outputs, and a +//! publish the file system refuses publishes nothing. + +use pdfcraft_automation::{Automation, Content}; +use pdfcraft_engine::Session; +use serde_json::{Value, json}; + +fn result(auto: &mut Automation, args: Value) -> Value { + match auto.call("doc_batch", &args).unwrap().remove(0) { + Content::Json(value) => value, + _ => panic!("expected a report"), + } +} + +fn pdf() -> Vec { + Session::new().create_from_text("Original", "Original text").unwrap().as_ref().clone() +} + +/// Sets a folder's permission bits. +#[cfg(unix)] +fn set_mode(path: &std::path::Path, mode: u32) { + use std::os::unix::fs::PermissionsExt; + std::fs::set_permissions(path, std::fs::Permissions::from_mode(mode)).unwrap(); +} + +/// Puts back the write permission the temporary folder needs before it is removed. +#[cfg(unix)] +struct RestoreMode(std::path::PathBuf); + +#[cfg(unix)] +impl Drop for RestoreMode { + fn drop(&mut self) { + use std::os::unix::fs::PermissionsExt; + let _ = std::fs::set_permissions(&self.0, std::fs::Permissions::from_mode(0o755)); + } +} + +#[test] +fn failed_inputs_publish_only_completed_outputs() { + let dir = tempfile::tempdir().unwrap(); + std::fs::create_dir(dir.path().join("input")).unwrap(); + std::fs::write(dir.path().join("input/good.pdf"), pdf()).unwrap(); + std::fs::write(dir.path().join("input/bad.pdf"), "not a pdf").unwrap(); + let mut auto = Automation::new().with_root(dir.path()).unwrap(); + let report = result(&mut auto, json!({"paths":["input"],"steps":["optimize"],"out_dir":"out","recursive":false})); + assert_eq!(report["completed"], 1, "{report}"); + assert_eq!(report["failed"], 1, "{report}"); + let files = report["files"].as_array().unwrap(); + assert_eq!(files.len(), 2, "{report}"); + let bad = files.iter().find(|file| file["path"].as_str().is_some_and(|path| path.ends_with("bad.pdf"))).unwrap(); + assert_eq!(bad["status"], "error", "{report}"); + assert!(bad["output"].is_null(), "a failed file reports no output: {report}"); + let names: Vec = + std::fs::read_dir(dir.path().join("out")).unwrap().map(|entry| entry.unwrap().file_name().to_string_lossy().into_owned()).collect(); + // Publish stages its bytes under a dot-prefixed name in the output folder, so + // no entry carrying that prefix may survive the run. + for name in &names { + assert!(!name.starts_with('.'), "temporary name left behind: {name} ({names:?})"); + } + assert_eq!(names, vec!["good.pdf".to_string()], "only the completed output is published"); +} + +#[cfg(unix)] +#[test] +fn read_only_output_folder_publishes_nothing() { + let dir = tempfile::tempdir().unwrap(); + std::fs::create_dir(dir.path().join("input")).unwrap(); + std::fs::write(dir.path().join("input/good.pdf"), pdf()).unwrap(); + let out = dir.path().join("out"); + std::fs::create_dir(&out).unwrap(); + set_mode(&out, 0o555); + let _restore = RestoreMode(out.clone()); + // A root process ignores the mode bits, and then the assertions below would + // pass without testing the refusal. Fail loudly instead of skipping. + assert!(std::fs::write(out.join("probe"), b"").is_err(), "run as a non-root user: the read-only output folder accepted a new file"); + let mut auto = Automation::new().with_root(dir.path()).unwrap(); + let report = result(&mut auto, json!({"paths":["input"],"steps":["optimize"],"out_dir":"out","recursive":false})); + assert_eq!(report["completed"], 0, "{report}"); + assert_eq!(report["failed"], 1, "{report}"); + let files = report["files"].as_array().unwrap(); + assert_eq!(files.len(), 1, "{report}"); + // Publish cannot create its staging file in a folder it may not write, so + // the entry carries the system's error and no output. + assert_eq!(files[0]["status"], "error", "{report}"); + assert!(files[0]["output"].is_null(), "a refused publish reports no output: {report}"); + let error = files[0]["error"].as_str().unwrap(); + assert!(!error.is_empty(), "{report}"); + assert_eq!(std::fs::read_dir(&out).unwrap().count(), 0, "the refused publish writes nothing into the read-only folder: {report}"); +} diff --git a/crates/automation/tests/it/batch_secrets.rs b/crates/automation/tests/it/batch_secrets.rs new file mode 100644 index 0000000000..08223f0955 --- /dev/null +++ b/crates/automation/tests/it/batch_secrets.rs @@ -0,0 +1,84 @@ +//! Tool results and errors from the folder tools carry no known test password. +//! +//! The passwords are set on the document and passed to the tool, and must never come back out: a +//! result that echoes one has leaked it into logs, transcripts and MCP callers. + +use pdfcraft_automation::{Automation, Content}; +use pdfcraft_engine::{Changes, Edit, Protection, Session}; +use serde_json::{Value, json}; + +/// The known test passwords; neither may appear in a result or an error. +const OPEN_PASSWORD: &str = "PLANT-OPEN-5f2c9a"; +const OWNER_PASSWORD: &str = "PLANT-OWNER-71de0b"; + +/// A one-page PDF locked with the known test passwords. +/// +/// The open password is the only way in. The batch bates tool rewrites page content, and +/// `Changes::None` forbids that, so this fixture permits every change except extracting pages; +/// the restrictions stay real, since the open password never lifts them. +fn protected_pdf() -> Vec { + let mut session = Session::new(); + let bytes = session.create_from_text("Batch secrets", "Planted password test").unwrap(); + let id = session.open("locked.pdf", None, bytes, None).unwrap(); + session + .apply( + id, + Edit::Protect(Protection { + open_password: Some(OPEN_PASSWORD.into()), + permissions_password: Some(OWNER_PASSWORD.into()), + changes: Changes::AnyExceptExtract, + ..Default::default() + }), + ) + .unwrap(); + session.save_bytes(id).unwrap().as_ref().clone() +} + +fn json_result(a: &mut Automation, tool: &str, args: Value) -> Value { + match a.call(tool, &args) { + Ok(mut content) => match content.remove(0) { + Content::Json(value) => value, + other => panic!("{tool}: expected JSON, got {other:?}"), + }, + Err(error) => panic!("{tool} {args}: {error}"), + } +} + +#[test] +fn results_and_errors_hold_no_planted_password_for_bates_or_batch() { + let dir = tempfile::tempdir().unwrap(); + std::fs::write(dir.path().join("locked.pdf"), protected_pdf()).unwrap(); + let mut a = Automation::new().with_root(dir.path()).unwrap(); + + // The right open password: numbering succeeds and reports the numbers only. + let numbered = + json_result(&mut a, "doc_bates_multi", json!({"files":[{"input":"locked.pdf","output":"locked-numbered.pdf","password":OPEN_PASSWORD}]})); + assert_eq!(numbered["completed"], 1, "{numbered}"); + assert_eq!(numbered["files"][0]["status"], "ok", "{numbered}"); + // The report carries the per-file entry the Bates tool builds, so a field added to that entry + // lands in the text checked below. + assert_eq!(numbered["files"][0]["first"], 1, "{numbered}"); + assert!(numbered["files"][0]["output"].as_str().is_some_and(|path| path.ends_with("locked-numbered.pdf")), "{numbered}"); + assert!(dir.path().join("locked-numbered.pdf").exists(), "the numbered copy is written"); + + // The wrong open password: the tool refuses the file with an error. + let refused = match a + .call("doc_bates_multi", &json!({"files":[{"input":"locked.pdf","output":"locked-refused.pdf","password":"not-the-open-password"}]})) + { + Err(error) => error.to_string(), + Ok(value) => panic!("a wrong open password must be refused: {value:?}"), + }; + + // The folder job refuses the encrypted file without a password, in its per-file report. + let batch = json_result(&mut a, "doc_batch", json!({"paths":["locked.pdf"],"steps":["optimize"],"out_dir":"out"})); + assert_eq!(batch["failed"], 1, "{batch}"); + assert_eq!(batch["files"][0]["status"], "error", "{batch}"); + assert!(batch["files"][0]["error"].as_str().is_some_and(|text| !text.is_empty()), "{batch}"); + assert!(!dir.path().join("out/locked.pdf").exists(), "a refused file publishes nothing"); + + for report in [serde_json::to_string(&numbered).unwrap(), refused, serde_json::to_string(&batch).unwrap()] { + for secret in [OPEN_PASSWORD, OWNER_PASSWORD] { + assert!(!report.contains(secret), "a result or error echoed a planted password: {report}"); + } + } +} diff --git a/crates/automation/tests/it/main.rs b/crates/automation/tests/it/main.rs index d36dbe93a9..13f24013cd 100644 --- a/crates/automation/tests/it/main.rs +++ b/crates/automation/tests/it/main.rs @@ -1,13 +1,18 @@ //! The crate's integration tests, linked into one test binary. +mod action_failures; mod attachment_key; mod automation; mod automation_roots; mod backdrop; mod batch; +mod batch_failures; +mod batch_secrets; mod button_caption; mod form_bulk; mod hostile_args; +mod model_3d; +mod model_render; mod optimize_content_boundaries; mod redact_after_stray_q_form; mod redact_text_after_q; diff --git a/crates/automation/tests/it/model_3d.rs b/crates/automation/tests/it/model_3d.rs new file mode 100644 index 0000000000..179192f421 --- /dev/null +++ b/crates/automation/tests/it/model_3d.rs @@ -0,0 +1,120 @@ +//! `model_add` embeds a U3D or PRC model as a 3D annotation (ISO 32000-2 §13.6.2) with a poster, +//! end to end: the tool, the saved file, undo, and the refusals. + +use pdfcraft_automation::{Automation, Content, ToolError}; +use pdfcraft_cos::{Dict, Document, Object, Stream}; +use serde_json::{Value, json}; + +fn tool(a: &mut Automation, name: &str, args: Value) -> Value { + match a.call(name, &args).unwrap().remove(0) { + Content::Json(value) => value, + other => panic!("expected JSON: {other:?}"), + } +} + +fn png(w: u32, h: u32) -> Vec { + let mut out = Vec::new(); + let mut encoder = png::Encoder::new(&mut out, w, h); + encoder.set_color(png::ColorType::Rgb); + encoder.set_depth(png::BitDepth::Eight); + encoder.write_header().unwrap().write_image_data(&vec![90; (w * h * 3) as usize]).unwrap(); + out +} + +fn stream(doc: &Document, obj: &Object) -> Stream { + match doc.resolve(obj).as_ref() { + Object::Stream(s) => s.clone(), + other => panic!("expected a stream: {other:?}"), + } +} + +/// The 3D annotations on the first page of the saved file. +fn saved_3d(dir: &std::path::Path) -> (Document, Vec) { + let doc = Document::open(std::sync::Arc::new(std::fs::read(dir.join("out.pdf")).unwrap())).unwrap(); + let page = &pdfcraft_model::pages(&doc)[0]; + let list = page.dict.get(b"Annots").map(|a| doc.resolve(a).as_array().cloned().unwrap()).unwrap_or_default(); + let found = list.iter().map(|o| doc.resolve(o).as_dict().unwrap().clone()).filter(|d| d.name(b"Subtype") == Some(&b"3D"[..])).collect(); + (doc, found) +} + +#[test] +fn model_add_embeds_the_model_with_a_fitted_poster() { + let dir = tempfile::tempdir().unwrap(); + let mut prc = b"PRC".to_vec(); + prc.extend_from_slice(&[7; 200]); + std::fs::write(dir.path().join("bracket.prc"), &prc).unwrap(); + std::fs::write(dir.path().join("poster.png"), png(80, 40)).unwrap(); + let mut a = Automation::new().with_root(dir.path()).unwrap(); + let doc = tool(&mut a, "doc_create", json!({ "from": "blank", "width": 600, "height": 800 }))["doc"].as_u64().unwrap(); + // A 300 × 300 box from the top-left of the page: user space [100, 400, 400, 700]. + let call = + json!({ "doc": doc, "page": 1, "path": "bracket.prc", "rect": [100, 100, 400, 400], "poster": "poster.png", "activation": "page_open" }); + tool(&mut a, "model_add", call); + tool(&mut a, "doc_save", json!({ "doc": doc, "path": "out.pdf" })); + + let (saved, found) = saved_3d(dir.path()); + let [annot] = found.as_slice() else { panic!("one 3D annotation: {found:?}") }; + let rect: Vec = annot.get(b"Rect").unwrap().as_array().unwrap().iter().filter_map(Object::as_f64).collect(); + assert_eq!(rect, [100.0, 400.0, 400.0, 700.0]); + assert_eq!(annot.get(b"Contents").and_then(Object::as_string).map(|s| s.to_text()).as_deref(), Some("bracket")); + let model = stream(&saved, annot.get(b"3DD").unwrap()); + assert_eq!(model.dict.name(b"Type"), Some(&b"3D"[..])); + assert_eq!(model.dict.name(b"Subtype"), Some(&b"PRC"[..])); + assert_eq!(model.dict.name(b"Filter"), Some(&b"FlateDecode"[..])); + assert_eq!(model.decoded().unwrap().to_vec(), prc, "the model is stored unchanged"); + let activation = saved.resolve(annot.get(b"3DA").unwrap()).as_dict().unwrap().clone(); + assert_eq!(activation.name(b"A"), Some(&b"PO"[..])); + // The 80 × 40 picture fitted in the 300 × 300 box: 300 × 150, centred top to bottom. + let ap = saved.resolve(annot.get(b"AP").unwrap()).as_dict().unwrap().clone(); + let poster = stream(&saved, ap.get(b"N").unwrap()); + let content = String::from_utf8(poster.decoded().unwrap().to_vec()).unwrap(); + assert!(content.contains("300 0 0 150 100 475 cm /Pic Do"), "{content}"); + + assert_eq!(tool(&mut a, "edit_undo", json!({ "doc": doc }))["undone"], "Add 3D model"); + tool(&mut a, "doc_save", json!({ "doc": doc, "path": "out.pdf" })); + assert!(saved_3d(dir.path()).1.is_empty(), "undo removes it"); +} + +#[test] +fn model_add_without_a_poster_draws_a_labelled_frame_and_activates_on_click() { + let dir = tempfile::tempdir().unwrap(); + let mut u3d = b"U3D\0".to_vec(); + u3d.extend_from_slice(&[0; 60]); + std::fs::write(dir.path().join("part.u3d"), &u3d).unwrap(); + let mut a = Automation::new().with_root(dir.path()).unwrap(); + let doc = tool(&mut a, "doc_create", json!({ "from": "blank", "width": 600, "height": 800 }))["doc"].as_u64().unwrap(); + tool(&mut a, "model_add", json!({ "doc": doc, "page": 1, "path": "part.u3d", "rect": [50, 50, 250, 200], "name": "Part" })); + tool(&mut a, "doc_save", json!({ "doc": doc, "path": "out.pdf" })); + + let (saved, found) = saved_3d(dir.path()); + let [annot] = found.as_slice() else { panic!("one 3D annotation: {found:?}") }; + assert_eq!(stream(&saved, annot.get(b"3DD").unwrap()).dict.name(b"Subtype"), Some(&b"U3D"[..])); + let activation = saved.resolve(annot.get(b"3DA").unwrap()).as_dict().unwrap().clone(); + assert_eq!(activation.name(b"A"), Some(&b"XA"[..]), "the standard's default, stated"); + let ap = saved.resolve(annot.get(b"AP").unwrap()).as_dict().unwrap().clone(); + let content = String::from_utf8(stream(&saved, ap.get(b"N").unwrap()).decoded().unwrap().to_vec()).unwrap(); + assert!(content.contains("re S") && content.contains("(3D model) Tj"), "{content}"); +} + +#[test] +fn model_add_refuses_what_is_not_a_model_and_adds_nothing() { + let dir = tempfile::tempdir().unwrap(); + std::fs::write(dir.path().join("notes.txt"), b"PR not a model").unwrap(); + std::fs::write(dir.path().join("part.prc"), b"PRC....").unwrap(); + let mut a = Automation::new().with_root(dir.path()).unwrap(); + let doc = tool(&mut a, "doc_create", json!({ "from": "blank", "width": 600, "height": 800 }))["doc"].as_u64().unwrap(); + let not_a_model = a.call("model_add", &json!({ "doc": doc, "page": 1, "path": "notes.txt", "rect": [10, 10, 100, 100] })); + assert!(matches!(¬_a_model, Err(ToolError::Failed(m)) if m.contains("U3D") && m.contains("PRC")), "{not_a_model:?}"); + for (args, why) in [ + (json!({ "rect": [10, 10, 10, 100] }), "zero width"), + (json!({ "rect": [0, 0, 20000, 100] }), "wider than a PDF page can be"), + (json!({}), "no rect"), + (json!({ "rect": [10, 10, 100, 100], "activation": "now" }), "an unknown activation"), + ] { + let mut call = json!({ "doc": doc, "page": 1, "path": "part.prc" }); + call.as_object_mut().unwrap().extend(args.as_object().unwrap().clone()); + assert!(matches!(a.call("model_add", &call), Err(ToolError::InvalidArgs(_))), "{why}"); + } + tool(&mut a, "doc_save", json!({ "doc": doc, "path": "out.pdf" })); + assert!(saved_3d(dir.path()).1.is_empty()); +} diff --git a/crates/automation/tests/it/model_render.rs b/crates/automation/tests/it/model_render.rs new file mode 100644 index 0000000000..615a5346a9 --- /dev/null +++ b/crates/automation/tests/it/model_render.rs @@ -0,0 +1,112 @@ +//! `model_render` draws a page's 3D model headlessly: a synthetic PDF holding a synthetic PRC. + +use std::path::PathBuf; + +use pdfcraft_automation::{Automation, Content}; +use serde_json::{Value, json}; + +/// One triangle at z = 0 facing +z, with corners (-1,-1), (1,-1) and (0,1), as a PRC file of one +/// file structure with no product tree, in hexadecimal for the stream's `/ASCIIHexDecode`. +const TRIANGLE_PRC_HEX: &str = "505243c91f0000c91f00000100000002000000030000000400000000000000000000000000000000000000010000000500000006000000070000000800000000000000060000006b0000009a000000a3000000ab000000cd000000d5000000dd000000e500000000000000505243c91f0000c91f0000050000000600000007000000080000000000000000000000000000000000000000000000789c63000000010001789c030000000001789c9b71c087758da2422b43084fe3140e038713464cd759381a162800006ed70828789c030000000001789c030000000001789c030000000001"; + +/// A one-page PDF whose page holds a 3D annotation over a stream of `/Subtype /{subtype}`. +fn pdf(subtype: &str, hex: &str) -> Vec { + let stream = format!("<< /Type /3D /Subtype /{subtype} /Filter /ASCIIHexDecode /Length {} >>\nstream\n{hex}>\nendstream", hex.len() + 1); + let objs: Vec = vec![ + "<< /Type /Catalog /Pages 2 0 R >>".into(), + "<< /Type /Pages /Kids [3 0 R] /Count 1 /MediaBox [0 0 200 200] >>".into(), + "<< /Type /Page /Parent 2 0 R /Annots [<< /Type /Annot /Subtype /Square /Rect [0 0 10 10] >> 4 0 R] >>".into(), + "<< /Type /Annot /Subtype /3D /Rect [20 20 180 180] /3DD 5 0 R >>".into(), + stream, + ]; + let mut out = b"%PDF-1.7\n".to_vec(); + let mut offsets = Vec::new(); + for (i, o) in objs.iter().enumerate() { + offsets.push(out.len()); + out.extend_from_slice(format!("{} 0 obj\n{o}\nendobj\n", i + 1).as_bytes()); + } + let xref = out.len(); + out.extend_from_slice(format!("xref\n0 {}\n0000000000 65535 f \n", objs.len() + 1).as_bytes()); + for o in offsets { + out.extend_from_slice(format!("{o:010} 00000 n \n").as_bytes()); + } + out.extend_from_slice(format!("trailer\n<< /Size {} /Root 1 0 R >>\nstartxref\n{xref}\n%%EOF\n", objs.len() + 1).as_bytes()); + out +} + +/// A fresh folder with `prc.pdf` and `u3d.pdf` in `in`, `outside.pdf` beside it, and an +/// automation rooted at `in`. +fn setup(test: &str) -> (PathBuf, Automation) { + let dir = std::env::temp_dir().join(format!("pdfcraft-model3d-{test}-{}", std::process::id())); + let _ = std::fs::remove_dir_all(&dir); + std::fs::create_dir_all(dir.join("in")).unwrap(); + std::fs::write(dir.join("in/prc.pdf"), pdf("PRC", TRIANGLE_PRC_HEX)).unwrap(); + std::fs::write(dir.join("in/u3d.pdf"), pdf("U3D", "553344")).unwrap(); + std::fs::write(dir.join("outside.pdf"), pdf("PRC", TRIANGLE_PRC_HEX)).unwrap(); + let a = Automation::new().with_root(dir.join("in")).unwrap(); + (dir, a) +} + +fn open(a: &mut Automation, path: &str) -> u64 { + match a.call("doc_open", &json!({ "path": path })).unwrap().remove(0) { + Content::Json(v) => v["doc"].as_u64().unwrap(), + other => panic!("expected JSON: {other:?}"), + } +} + +/// The RGB of pixel (x, y) of a PNG. +fn rgb(data: &[u8], x: usize, y: usize) -> [u8; 3] { + let mut reader = png::Decoder::new(std::io::Cursor::new(data)).read_info().unwrap(); + let mut pixels = vec![0; reader.output_buffer_size().unwrap()]; + let info = reader.next_frame(&mut pixels).unwrap(); + let channels = info.line_size / info.width as usize; + let at = y * info.line_size + x * channels; + [pixels[at], pixels[at + 1], pixels[at + 2]] +} + +#[test] +fn a_prc_model_is_drawn_over_the_background() { + let (_dir, mut a) = setup("prc"); + let doc = open(&mut a, "prc.pdf"); + let args = json!({ "doc": doc, "page": 1, "width": 64, "height": 48, "direction": [0, 0, -1], "perspective": false }); + let out = a.call("model_render", &args).unwrap(); + let Content::Png { data, width, height } = &out[0] else { panic!("expected a PNG") }; + assert_eq!((*width, *height), (64, 48)); + assert_eq!(rgb(data, 0, 0), [255, 255, 255], "the corner is background"); + assert_eq!(rgb(data, 32, 24), [190, 192, 200], "the centre is the triangle, lit head on"); + // In perspective the headlight meets the face at a slight angle, so only not-background holds. + let out = a.call("model_render", &json!({ "doc": doc, "page": 1, "width": 64, "height": 48, "direction": [0, 0, -1] })).unwrap(); + let Content::Png { data, .. } = &out[0] else { panic!("expected a PNG") }; + assert_ne!(rgb(data, 32, 24), [255, 255, 255], "the centre is drawn in perspective too"); + // The same model named explicitly, and the default isometric view, both draw. + assert!(a.call("model_render", &json!({ "doc": doc, "page": 1, "model": 1 })).is_ok()); +} + +#[test] +fn a_u3d_model_or_a_missing_one_is_a_clear_error() { + let (_dir, mut a) = setup("u3d"); + let doc = open(&mut a, "u3d.pdf"); + let err = a.call("model_render", &json!({ "doc": doc, "page": 1 })).unwrap_err().to_string(); + assert!(err.contains("U3D"), "{err}"); + let doc = open(&mut a, "prc.pdf"); + let err = a.call("model_render", &json!({ "doc": doc, "page": 1, "model": 2 })).unwrap_err().to_string(); + assert!(err.contains("has 1 3D models, not 2"), "{err}"); + for bad in [json!({ "direction": [0, 0, 0] }), json!({ "direction": [1, 2] }), json!({ "width": 0 }), json!({ "height": 5000 })] { + let mut args = json!({ "doc": doc, "page": 1 }); + args.as_object_mut().unwrap().extend(bad.as_object().unwrap().clone()); + assert!(a.call("model_render", &args).is_err(), "{bad}"); + } +} + +#[test] +fn a_model_outside_the_root_is_never_opened_to_be_drawn() { + let (dir, mut a) = setup("root"); + let outside = dir.join("outside.pdf"); + for path in [Value::from("../outside.pdf"), Value::from(outside.to_string_lossy().into_owned())] { + assert!(a.call("doc_open", &json!({ "path": path })).is_err(), "{path} lies outside the root"); + } + // The tool takes no path of its own: only a document opened inside the root reaches it. + assert!(a.call("model_render", &json!({ "doc": 1, "page": 1 })).is_err()); + let doc = open(&mut a, "prc.pdf"); + assert!(a.call("model_render", &json!({ "doc": doc, "page": 1 })).is_ok()); +} diff --git a/crates/cos/Cargo.toml b/crates/cos/Cargo.toml index a2fb3f4c4e..0bf2e894d2 100644 --- a/crates/cos/Cargo.toml +++ b/crates/cos/Cargo.toml @@ -15,6 +15,13 @@ thiserror = { workspace = true } # PDF date strings (`pdf_date`) go through chrono (no hand-rolled calendar code). chrono = { workspace = true } +# On the web, a rewrite's fresh file identifier takes its entropy from the browser +# (`crypto.getRandomValues`). Both crates are already in the dependency tree of the web shell +# (`js-sys` 0.3 / `wasm-bindgen` 0.2); native builds compile neither. +[target.'cfg(target_arch = "wasm32")'.dependencies] +js-sys = "0.3" +wasm-bindgen = "0.2" + [dev-dependencies] pdfcraft-oracle = { workspace = true } pdfcraft-testkit = { workspace = true } diff --git a/crates/cos/README.md b/crates/cos/README.md index 7b576f293f..9f5fb56725 100644 --- a/crates/cos/README.md +++ b/crates/cos/README.md @@ -53,7 +53,20 @@ another. Edits never touch the original bytes. - **Every cross-reference form:** classic tables with their quirky subsections, cross-reference streams, hybrid-reference files (`/XRefStm`), and object streams. - **Repair.** A file whose xref is wrong or missing is reconstructed by scanning for `obj` - headers. Repairs are *recorded*, never silent, so the UI can say the file was damaged. + headers. Repairs are *recorded*, never silent, so the UI can say the file was damaged. A live + in-file reference may match either its xref generation or a same-number object header at + that indexed offset; a missing, free, or neither-matching slot still resolves to null. + Header probes inspect at most 128 bytes at the indexed and header-relative offsets, without + parsing object bodies. On open, disagreements in retained xref revisions are recorded once + per object, up to 256 objects plus an omitted-details summary. This diagnostic scan does + not expand compact xref ranges; the winning xref entry remains authoritative for free + slots. Compressed objects require generation zero. Explicit `set` overlays accept only the + supplied generation; `set_existing` and `update_dict` retain both accepted generations of a + damaged live slot, including through snapshots, and require full rewrite. + `accepted_generations(num)` lists the generations a slot accepts, so code comparing two + revisions can tell whether a reference still reaches the same slot. Headers beyond + this bounded probe are not additionally recovered. Materialized primary and hybrid rows are + diagnosed before folding, so a newer row does not hide a retained disagreement. - **Tolerant, never fatal.** The lexer works on a byte slice and a position; it allocates nothing for skipped data, bounds-checks every index, and caps nesting depth. Malformed input yields `CosError`, never a panic — this crate is the first thing a hostile file meets. Exhausted xref entries are skipped with a repair log; exhausted /Size hints allocate unused numbers without overflowing or replacing existing objects. @@ -66,14 +79,25 @@ another. Edits never touch the original bytes. ## Writing -`write_incremental` appends only the edited objects, a cross-reference section **in the same -style as the file's last one**, and a trailer with `/Prev`. The original bytes are never modified, -which is what keeps a signature over an earlier revision valid. +`write_incremental` ordinarily appends only the edited objects, a cross-reference section **in +the same style as the file's last one**, and a trailer with `/Prev`, preserving the original +bytes and signatures over earlier revisions. It delegates to `write_full` when reconstruction, +encryption changes or `full_save_required()` demand a rewrite. Repair-aware edits to a live +xref/header generation disagreement use this fallback rather than append a contradictory row. `write_full` writes only what the trailer can reach, renumbered densely — the Save As and garbage-collecting path. By default (`SaveOptions::object_streams`) it packs non-stream objects into compressed object streams with a cross-reference stream (PDF 1.5+); turning that off writes a -classic table any reader accepts. +classic table any reader accepts. Both forms map accepted xref/header aliases to one output +object, retaining shared identity such as page-tree parent links; unrelated generations stay null. + +A full save keeps the file's `/ID` as it is, unless the document asked for a new identifier with +`Document::require_full_save_with_new_id` (redaction and sanitizing do). Then both strings are +regenerated, and `SaveOptions::id_entropy` can supply entropy for reproducible output (tests pin +output this way). Without it the identifier mixes in fresh entropy from the host — the browser's +`crypto.getRandomValues` on the web, the operating system's random hash seed on the desktop — so a +fresh copy of the program cannot reproduce it. An encrypted file keeps the first string, because its +key is derived from it. Dictionaries keep their key order and streams keep their *encoded* bytes, so rewritten objects stay recognisable in a diff and byte-stable when re-serialized. diff --git a/crates/cos/src/document.rs b/crates/cos/src/document.rs index e52aba2ce3..e5ce531f67 100644 --- a/crates/cos/src/document.rs +++ b/crates/cos/src/document.rs @@ -71,6 +71,9 @@ pub struct Document { encryption_changed: bool, /// Set by edits that must not leave earlier revisions in the file (redaction). full_save: bool, + /// The next full save also gives the file a new identifier (`/ID`), see + /// [`Document::require_full_save_with_new_id`]. + new_id: bool, /// The handler and `/Encrypt` object number that saves use, when encryption changed. out_security: Option>, out_encrypt_num: Option, diff --git a/crates/cos/src/document/objects.rs b/crates/cos/src/document/objects.rs index 2e211927ce..458840de3a 100644 --- a/crates/cos/src/document/objects.rs +++ b/crates/cos/src/document/objects.rs @@ -1,6 +1,8 @@ +use std::collections::HashSet; + use super::reader::find; use super::state::{LENGTHS, LoadGuard, ObjStm, ObjectReader, ReadState, Slot, SourceIdentity, held_bytes, objstm_member}; -use super::xref::{Revision, XrefEntry}; +use super::xref::{Revision, XrefEntry, XrefIndex}; use super::{ Arc, CosError, Cow, Dict, Document, Lexer, MAX_DECODED, MAX_JBIG2_GLOBALS, ObjRef, Object, Stream, is_whitespace, parse_indirect_shared, }; @@ -129,8 +131,8 @@ impl Document { pub fn xref_entry_is_exact(&self, num: u32) -> bool { match self.xref_entry(num) { None | Some(XrefEntry::Free { .. }) => true, - Some(XrefEntry::InFile { offset, .. }) => { - let at = |off: usize| matches!(parse_indirect_shared(&self.data, off, &|_| None, &self.endstreams), Ok((id, _)) if id.num == num); + Some(XrefEntry::InFile { offset, generation }) => { + let at = |off: usize| matches!(parse_indirect_shared(&self.data, off, &|_| None, &self.endstreams), Ok((id, _)) if id == ObjRef::new(num, generation)); // Offsets may count from a header that does not start the file. usize::try_from(offset).is_ok_and(|off| at(off) || (self.header_offset != 0 && at(off.saturating_add(self.header_offset)))) } @@ -151,9 +153,117 @@ impl Document { // ── object access ─────────────────────────────────────────────────────────────────────── - /// Fetch an object by reference. Missing objects are `Null` (ISO 32000-2 §7.3.10). + /// Fetch an object by reference. Undefined references are `Null` (ISO 32000-2 §7.3.10). + /// A live file slot also accepts its actual header generation when a damaged xref row + /// disagrees; the disagreement is recorded in [`Self::repair_log`]. Free slots never do. pub fn get(&self, r: ObjRef) -> Arc { - self.try_get(r.num).unwrap_or_else(|_| Arc::new(Object::Null)) + self.try_get_reference(r).unwrap_or_else(|_| Arc::new(Object::Null)) + } + + pub(super) fn reference_matches(&self, r: ObjRef) -> bool { + match self.overlay.get(&r.num) { + Some(Slot::Set(generation, _)) => r.generation == *generation, + Some(Slot::RepairedSet(generation, alias, _)) => r.generation == *generation || r.generation == *alias, + Some(Slot::Freed(_)) => false, + None => match self.entries.lookup(r.num, &self.data) { + Some(XrefEntry::InFile { offset, generation }) => { + r.generation == generation || self.indexed_header_reference(r.num, offset) == Some(r) + } + Some(XrefEntry::InStream { .. }) => r.generation == 0, + None | Some(XrefEntry::Free { .. }) => false, + }, + } + } + + /// The generations a reference to `num` may carry and still resolve (see [`Self::get`]), + /// sorted: none for a free or unknown slot, else the slot's generation, and for a damaged + /// file slot (or a repair-aware edit of one) also the generation its object header carries. + /// Code comparing revisions (signature validation) uses it to tell whether the references + /// that reached an object in one revision still reach the same slot in another. + pub fn accepted_generations(&self, num: u32) -> Vec { + // Mirrors `reference_matches`, which stays allocation-free for every lookup. + let mut generations = match self.overlay.get(&num) { + Some(Slot::Set(generation, _)) => vec![*generation], + Some(Slot::RepairedSet(generation, alias, _)) => vec![*generation, *alias], + Some(Slot::Freed(_)) => Vec::new(), + None => match self.entries.lookup(num, &self.data) { + Some(XrefEntry::InFile { offset, generation }) => { + std::iter::once(generation).chain(self.indexed_header_reference(num, offset).map(|h| h.generation)).collect() + } + Some(XrefEntry::InStream { .. }) => vec![0], + None | Some(XrefEntry::Free { .. }) => Vec::new(), + }, + }; + generations.sort_unstable(); + generations.dedup(); + generations + } + + pub(super) fn try_get_reference(&self, r: ObjRef) -> Result, CosError> { + if !self.reference_matches(r) { + return Ok(Arc::new(Object::Null)); + } + self.try_get(r.num) + } + + /// A repair probe, not a whole-file search: only the indexed header (or the same offset + /// relative to a shifted PDF header) can establish this additional generation. A small + /// window bounds hostile whitespace/tokens; ordinary xref-matching reads are unchanged. + pub(super) fn indexed_header_reference(&self, num: u32, offset: u64) -> Option { + let at = |off: usize| { + let bytes = self.data.get(off..off.saturating_add(128).min(self.data.len()))?; + let mut lexer = Lexer::new(bytes, 0); + lexer.skip_ws(); + let number = std::str::from_utf8(lexer.token()).ok()?.parse::().ok()?; + lexer.skip_ws(); + let generation = std::str::from_utf8(lexer.token()).ok()?.parse::().ok()?; + (number == num && lexer.eat_keyword(b"obj")).then_some(ObjRef::new(number, generation)) + }; + let off = usize::try_from(offset).ok()?; + at(off).or_else(|| if self.header_offset == 0 { None } else { off.checked_add(self.header_offset).and_then(at) }) + } + + /// Add at most one bounded diagnostic per object. False stops diagnostic probing only; + /// parsers must still read the remaining xref rows needed to build their index. + pub(super) fn inspect_xref_generation_repair(&self, num: u32, entry: XrefEntry, recorded: &mut HashSet, log: &mut Vec) -> bool { + let XrefEntry::InFile { offset, generation } = entry else { return true }; + if recorded.contains(&num) { + return true; + } + let Some(header) = self.indexed_header_reference(num, offset) else { return true }; + if header.generation == generation { + return true; + } + if recorded.len() == 256 { + log.push("additional xref/header generation disagreements omitted after 256 objects".into()); + return false; + } + recorded.insert(num); + log.push(format!("object {num}: xref generation {generation} disagrees with object header generation {}", header.generation)); + true + } + + /// Diagnose once at open without expanding compact ranges or decoding an xref twice. + /// Materialized readers retain bounded notes while they read accepted sections, before + /// newer rows hide older ones. Discarded parse attempts/rebuilt indexes carry no notes. + pub(super) fn note_xref_generation_repairs(&self, log: &mut Vec) { + match &*self.entries { + XrefIndex::Materialized(_, notes) => log.extend(notes.iter().cloned()), + XrefIndex::Compact { revisions, .. } => { + let mut recorded = HashSet::new(); + for revision in revisions { + for section in std::iter::once(&revision.primary).chain(revision.hybrid.iter()) { + for num in section.numbers() { + if let Some(entry) = section.lookup(num, &self.data) + && !self.inspect_xref_generation_repair(num, entry, &mut recorded, log) + { + return; + } + } + } + } + } + } } /// Follow a reference if `o` is one; otherwise return `o` itself. An object can be nothing but @@ -388,7 +498,7 @@ impl Document { pub fn try_get(&self, num: u32) -> Result, CosError> { match self.overlay.get(&num) { - Some(Slot::Set(_, o)) => return Ok(o.clone()), + Some(Slot::Set(_, o) | Slot::RepairedSet(_, _, o)) => return Ok(o.clone()), Some(Slot::Freed(_)) => return Ok(Arc::new(Object::Null)), None => {} } @@ -519,11 +629,11 @@ impl Document { None | Some(XrefEntry::Free { .. }) => Ok(Object::Null), Some(XrefEntry::InFile { offset, .. }) => { let resolve = |r: ObjRef| { - if let Some(known) = LENGTHS.with(|m| m.borrow().get(&r.num).copied()) { + if let Some(known) = LENGTHS.with(|m| m.borrow().get(&r).copied()) { return known; } - let length = self.try_get(r.num).ok().and_then(|o| o.as_int()); - LENGTHS.with(|m| m.borrow_mut().insert(r.num, length)); + let length = self.try_get_reference(r).ok().and_then(|o| o.as_int()); + LENGTHS.with(|m| m.borrow_mut().insert(r, length)); length }; let off = offset as usize; @@ -640,6 +750,38 @@ impl Document { } } + /// Replace an accepted existing reference without losing a damaged source slot's aliases. + /// Unlike [`Self::set`], this preserves both xref/header generations when they disagree. + /// Such an edit forces a full rewrite so the output has consistent references, rather than + /// appending a new contradictory xref row. Missing, free and stale references are errors. + pub fn set_existing(&mut self, r: ObjRef, obj: impl Into) -> Result<(), CosError> { + if !self.reference_matches(r) { + return Err(CosError::MissingObject(r.num)); + } + let alias = self.alternate_generation(r); + if let Some(alias) = alias { + let obj = self.adopt(obj.into()); + self.overlay.insert(r.num, Slot::RepairedSet(r.generation, alias, Arc::new(obj))); + } else { + self.set(r, obj); + } + Ok(()) + } + + /// Other accepted generation for a known-live reference to a damaged slot. + pub(super) fn alternate_generation(&self, r: ObjRef) -> Option { + match self.overlay.get(&r.num) { + Some(Slot::RepairedSet(generation, alias, _)) => Some(if r.generation == *generation { *alias } else { *generation }), + Some(_) => None, + None => match self.entries.lookup(r.num, &self.data) { + Some(XrefEntry::InFile { offset, generation }) => self.indexed_header_reference(r.num, offset).and_then(|header| { + (header.generation != generation).then_some(if r.generation == generation { header.generation } else { generation }) + }), + _ => None, + }, + } + } + /// Add a new indirect object and return its reference. pub fn add(&mut self, obj: impl Into) -> ObjRef { let r = ObjRef::new(self.allocate_num(), 0); @@ -679,8 +821,9 @@ impl Document { /// /// Used when the working snapshot was restored from an earlier revision and the next /// incremental save must still append to this file (the last saved revision). Objects - /// that already match are left untouched. Reads go through private readers so neither - /// document's editing cache is filled. + /// that already match are left untouched, except members of an object stream of this file + /// whose stream number `current` frees or reuses: those are written again on their own. + /// Reads go through private readers so neither document's editing cache is filled. pub fn overlay_changes_from(&mut self, current: &Document) -> Result<(), CosError> { let saved_nums = self.object_numbers(); let current_nums = current.object_numbers(); @@ -698,12 +841,29 @@ impl Document { let generation = current.generation(num); let same = saved_nums.binary_search(&num).is_ok() && self.generation(num) == generation - && saved_reader.try_get(ObjRef::new(num, 0))? == current_reader.try_get(ObjRef::new(num, 0))?; + && self.alternate_generation(ObjRef::new(num, generation)) == current.alternate_generation(ObjRef::new(num, generation)) + && saved_reader.try_get(ObjRef::new(num, generation))? == current_reader.try_get(ObjRef::new(num, generation))?; if !same { changed.push(num); } } } + // An object stream of this file that is freed or replaced takes its members with it, so + // a member that is otherwise unchanged is written again on its own. + let gone = |n: u32| freed.binary_search(&n).is_ok() || changed.binary_search(&n).is_ok(); + let orphaned: Vec = current_nums + .iter() + .copied() + .filter(|&num| changed.binary_search(&num).is_err()) + .filter(|&num| { + let packed_in = match self.xref_entry(num) { + Some(XrefEntry::InStream { stream, .. }) => Some(stream), + _ => None, + }; + packed_in.is_some_and(gone) + }) + .collect(); + changed.extend(orphaned); for num in freed { let generation = self.generation(num); self.free(ObjRef::new(num, generation)); @@ -711,8 +871,14 @@ impl Document { let mut current_reader = current.object_reader(); for num in changed { let generation = current.generation(num); - let object = current_reader.try_get(ObjRef::new(num, 0))?; - self.set(ObjRef::new(num, generation), (*object).clone()); + let object = current_reader.try_get(ObjRef::new(num, generation))?; + if let Some(alias) = current.alternate_generation(ObjRef::new(num, generation)) { + let object = self.adopt((*object).clone()); + self.overlay.insert(num, Slot::RepairedSet(generation, alias, Arc::new(object))); + self.next_num = self.next_num.max(num.saturating_add(1)); + } else { + self.set(ObjRef::new(num, generation), (*object).clone()); + } } *self.trailer_mut() = current.trailer().clone(); Ok(()) @@ -723,14 +889,13 @@ impl Document { let mut obj = (*self.get(r)).clone(); let d = obj.as_dict_mut().ok_or(CosError::NotADictionary(r))?; f(d); - self.set(r, obj); - Ok(()) + self.set_existing(r, obj) } /// Current generation of an object number. pub fn generation(&self, num: u32) -> u16 { match self.overlay.get(&num) { - Some(Slot::Set(g, _)) => *g, + Some(Slot::Set(g, _) | Slot::RepairedSet(g, _, _)) => *g, Some(Slot::Freed(g)) => *g, None => match self.entries.lookup(num, &self.data) { Some(XrefEntry::InFile { generation, .. }) => generation, @@ -742,7 +907,7 @@ impl Document { pub(crate) fn overlay_entries(&self) -> impl Iterator>)> + '_ { self.overlay.iter().map(|(n, s)| match s { - Slot::Set(g, o) => (*n, *g, Some(o.clone())), + Slot::Set(g, o) | Slot::RepairedSet(g, _, o) => (*n, *g, Some(o.clone())), Slot::Freed(g) => (*n, *g, None), }) } @@ -765,7 +930,7 @@ impl Document { if source.peek().is_some_and(|n| *n == num) { source.next(); } - if matches!(slot, Slot::Set(..)) { + if matches!(slot, Slot::Set(..) | Slot::RepairedSet(..)) { numbers.push(num); } } @@ -783,7 +948,7 @@ impl Document { pub(super) fn root_is_catalog(&self) -> bool { let Some(r) = self.root() else { return false }; - matches!(self.try_get(r.num).as_deref(), Ok(Object::Dict(d)) if d.name(b"Type") == Some(b"Catalog") || d.contains(b"Pages")) + matches!(self.try_get_reference(r).as_deref(), Ok(Object::Dict(d)) if d.name(b"Type") == Some(b"Catalog") || d.contains(b"Pages")) } } diff --git a/crates/cos/src/document/open.rs b/crates/cos/src/document/open.rs index 70bbcc313e..3bf6677e99 100644 --- a/crates/cos/src/document/open.rs +++ b/crates/cos/src/document/open.rs @@ -1,5 +1,5 @@ use super::reader::find; -use super::state::Credentials; +use super::state::{Credentials, Slot}; use super::xref::XrefIndex; use super::{Arc, BTreeMap, CosError, Dict, Document, HashSet, ObjRef, Object}; @@ -35,6 +35,7 @@ impl Document { unfound: Arc::default(), encryption_changed: false, full_save: false, + new_id: false, out_security: None, out_encrypt_num: None, }; @@ -121,6 +122,7 @@ impl Document { unfound: Arc::default(), encryption_changed: false, full_save: false, + new_id: false, out_security: None, out_encrypt_num: None, }; @@ -156,6 +158,9 @@ impl Document { } doc.reconstruct(&mut log)?; } + // Header generations are plaintext, including in encrypted files. Inspect only the + // bounded headers, not object bodies, and retain the compact xref representation. + doc.note_xref_generation_repairs(&mut log); doc.repair_log = Arc::new(log); // Reconstruction may have found a different trailer: (re)authenticate against it. if (!ok || !authenticated) @@ -287,9 +292,23 @@ impl Document { self.full_save = true; } + /// Like [`Document::require_full_save`], and the rewritten file also gets a new file identifier + /// (`/ID`), so that a redacted or sanitized copy never claims to be its source (ISO 32000-2 + /// §14.4). An encrypted file keeps the first string of its identifier, which its key depends on. + pub fn require_full_save_with_new_id(&mut self) { + self.full_save = true; + self.new_id = true; + } + /// The next save must rewrite the whole file (see [`Document::require_full_save`]). + /// Repair-aware edits retaining two accepted generations cannot use one incremental slot. pub fn full_save_required(&self) -> bool { - self.full_save + self.full_save || self.overlay.values().any(|slot| matches!(slot, Slot::RepairedSet(..))) + } + + /// The next full save gives the file a new identifier (see [`Document::require_full_save_with_new_id`]). + pub fn new_id_required(&self) -> bool { + self.new_id } /// Security was set or removed since the document was opened (the next save applies it). diff --git a/crates/cos/src/document/reader.rs b/crates/cos/src/document/reader.rs index d21bf9403f..e1691ace90 100644 --- a/crates/cos/src/document/reader.rs +++ b/crates/cos/src/document/reader.rs @@ -17,8 +17,8 @@ impl Document { return Ok((index, trailer, revisions, Vec::new())); } } - let (entries, trailer, revisions, oddities) = self.read_xref_chain_materialized(log)?; - Ok((XrefIndex::Materialized(Arc::new(entries)), trailer, revisions, oddities)) + let (entries, trailer, revisions, oddities, generation_repairs) = self.read_xref_chain_materialized(log)?; + Ok((XrefIndex::Materialized(Arc::new(entries), Arc::new(generation_repairs)), trailer, revisions, oddities)) } fn read_compact_xref_chain(&self, log: &mut Vec) -> Result)>, CosError> { @@ -323,7 +323,7 @@ impl Document { pub(super) fn read_xref_chain_materialized( &self, log: &mut Vec, - ) -> Result<(BTreeMap, Dict, Vec, Vec<(u64, u32)>), CosError> { + ) -> Result<(BTreeMap, Dict, Vec, Vec<(u64, u32)>, Vec), CosError> { let data = &self.data; let tail_start = data.len().saturating_sub(4096); let sx = rfind(data, b"startxref", tail_start).ok_or(CosError::Syntax { offset: data.len(), detail: "no startxref".into() })?; @@ -343,6 +343,9 @@ impl Document { // At most MAX_XREF_OBJECTS entries are kept in all; a stream past that is read again. let mut hybrids: HashMap>> = HashMap::new(); let (mut hybrid_rows, mut noted) = (0usize, HashSet::new()); + let mut generation_repairs = Vec::new(); + let mut recorded = HashSet::new(); + let mut diagnose = true; let mut next = Some(first); while let Some(off) = next.take() { if !visited.insert(off) || visited.len() > 10_000 { @@ -371,6 +374,12 @@ impl Document { } }; revisions.push(Revision { xref_offset: off, is_stream, header_relative }); + for (&num, &entry) in § { + if !diagnose { + break; + } + diagnose = self.inspect_xref_generation_repair(num, entry, &mut recorded, &mut generation_repairs); + } // Hybrid files (§7.5.8.4): the /XRefStm entries supplement the table and replace the // free entries it gives objects only a cross-reference stream can locate. let hybrid = sect_trailer.int(b"XRefStm").filter(|x| *x >= 0); @@ -390,6 +399,12 @@ impl Document { let mut stm = BTreeMap::new(); let read = match self.read_section(x as usize, &mut stm, &mut odd, log) { Ok(_) => { + for (&num, &entry) in &stm { + if !diagnose { + break; + } + diagnose = self.inspect_xref_generation_repair(num, entry, &mut recorded, &mut generation_repairs); + } for (&num, &e) in &stm { match sect.get(&num) { None | Some(XrefEntry::Free { .. }) => { @@ -427,7 +442,7 @@ impl Document { for k in [&b"Prev"[..], b"XRefStm", b"Type", b"W", b"Index", b"Filter", b"DecodeParms", b"Length"] { trailer.remove(k); } - Ok((entries, trailer, revisions, oddities)) + Ok((entries, trailer, revisions, oddities, generation_repairs)) } /// Read one section at `off`, adding entries not already known (newer sections win). `odd` @@ -606,7 +621,7 @@ impl Document { } if found > 0 { log.push(format!("the cross-reference data leaves out {found} objects the file contains; located them by their object headers")); - self.entries = Arc::new(XrefIndex::Materialized(Arc::new(filled))); + self.entries = Arc::new(XrefIndex::Materialized(Arc::new(filled), Arc::default())); } self.located = Arc::new(located); } @@ -631,7 +646,7 @@ impl Document { if entries.is_empty() { return Err(CosError::Syntax { offset: 0, detail: "no objects found".into() }); } - self.entries = Arc::new(XrefIndex::Materialized(Arc::new(entries.clone()))); + self.entries = Arc::new(XrefIndex::Materialized(Arc::new(entries.clone()), Arc::default())); self.cache.lock().map_err(|_| CosError::Poisoned)?.clear(); // Objects inside object streams. for num in objstms { @@ -648,7 +663,7 @@ impl Document { } } let object_count = entries.len(); - self.entries = Arc::new(XrefIndex::Materialized(Arc::new(entries))); + self.entries = Arc::new(XrefIndex::Materialized(Arc::new(entries), Arc::default())); self.cache.lock().map_err(|_| CosError::Poisoned)?.clear(); // Trailer: the last `trailer` dictionary or cross-reference stream dictionary with a // usable /Root (last in the file, whichever kind), else the catalog itself. One whose @@ -658,7 +673,7 @@ impl Document { // A file whose only trailers are cross-reference streams (and whose `startxref` is gone // or points nowhere) keeps its /Info, /ID and /Encrypt this way. let rank = |at: usize, d: &Dict| { - let root = self.try_get(d.reference(b"Root")?.num).ok()?; + let root = self.try_get_reference(d.reference(b"Root")?).ok()?; let Object::Dict(c) = &*root else { return None }; Some((c.name(b"Type") == Some(b"Catalog") || c.contains(b"Pages"), at)) }; diff --git a/crates/cos/src/document/state.rs b/crates/cos/src/document/state.rs index 96dde01c0f..f68a483290 100644 --- a/crates/cos/src/document/state.rs +++ b/crates/cos/src/document/state.rs @@ -7,7 +7,7 @@ thread_local! { /// Indirect `/Length` values resolved during the current outermost load: a damaged file /// retries a parse several ways, and each retry would resolve the same lengths again /// (exponential in the chain of streams whose lengths point at further streams). - pub(super) static LENGTHS: std::cell::RefCell>> = std::cell::RefCell::new(HashMap::new()); + pub(super) static LENGTHS: std::cell::RefCell>> = std::cell::RefCell::new(HashMap::new()); } /// Marks an object as being loaded on this thread. Loading can re-enter (indirect stream @@ -65,6 +65,9 @@ pub(super) fn objstm_member(num: i64, offset: i64, first: Option, len: us #[derive(Clone, Debug, PartialEq)] pub(super) enum Slot { Set(u16, Arc), + /// Both generations accepted from a damaged live source slot survive a repair-aware edit. + /// A full rewrite is required: a new xref slot cannot encode two generations. + RepairedSet(u16, u16, Arc), Freed(u16), } @@ -103,6 +106,7 @@ impl ReadState { a == b && match (x, y) { (Slot::Set(g, o), Slot::Set(h, p)) => g == h && Arc::ptr_eq(o, p), + (Slot::RepairedSet(g, a, o), Slot::RepairedSet(h, b, p)) => g == h && a == b && Arc::ptr_eq(o, p), (Slot::Freed(g), Slot::Freed(h)) => g == h, _ => false, } @@ -133,7 +137,7 @@ impl ObjectReader { } pub(crate) fn try_get(&mut self, reference: ObjRef) -> Result, CosError> { - let object = self.document.try_get(reference.num); + let object = self.document.try_get_reference(reference); // A missed eviction here only costs a re-parse, so never fail the read on a poisoned // lock. The editing-cache evictors propagate instead; they guard user edits. if let Ok(mut streams) = self.document.objstms.lock() { diff --git a/crates/cos/src/document/tests.rs b/crates/cos/src/document/tests.rs index 991323d607..ad8f86fdea 100644 --- a/crates/cos/src/document/tests.rs +++ b/crates/cos/src/document/tests.rs @@ -599,7 +599,7 @@ pub(crate) fn build(bodies: &[&str], trailer: &str) -> Vec { fn assert_compact_matches_materialized(doc: &Document) { assert!(matches!(doc.entries.as_ref(), XrefIndex::Compact { .. }), "valid xrefs should not retain a per-entry map"); - let (entries, _, revisions, _) = doc.read_xref_chain_materialized(&mut Vec::new()).unwrap(); + let (entries, _, revisions, _, _) = doc.read_xref_chain_materialized(&mut Vec::new()).unwrap(); assert_eq!(revisions.as_slice(), doc.revisions()); let max = entries.keys().next_back().copied().unwrap_or(0); assert_eq!(doc.entries.max_num(), max); @@ -687,7 +687,7 @@ fn malformed_size_uses_the_materialized_compatibility_path() { let at = bytes.windows(marker.len()).position(|window| window == marker).unwrap(); bytes[at..at + marker.len()].copy_from_slice(b"/Size 2"); let doc = Document::open(Arc::new(bytes)).unwrap(); - assert!(matches!(doc.entries.as_ref(), XrefIndex::Materialized(_))); + assert!(matches!(doc.entries.as_ref(), XrefIndex::Materialized(..))); assert_eq!(doc.next_num, 4); assert_eq!(doc.object_numbers(), vec![1, 2, 3]); } @@ -1181,15 +1181,18 @@ fn reads_classic_revision_and_edits_are_copy_on_write() { fn object_numbers_merge_source_and_overlay_in_order() { let mut doc = Document::new_empty(); doc.overlay.clear(); - doc.entries = Arc::new(XrefIndex::Materialized(Arc::new(BTreeMap::from([ - (0, XrefEntry::Free { next_generation: 65535 }), - (2, XrefEntry::InFile { offset: 10, generation: 7 }), - (4, XrefEntry::Free { next_generation: 9 }), - (6, XrefEntry::InStream { stream: 80, index: 1 }), - (8, XrefEntry::InFile { offset: 20, generation: 5 }), - (10, XrefEntry::InFile { offset: 30, generation: 3 }), - (u32::MAX, XrefEntry::InFile { offset: 40, generation: 1 }), - ])))); + doc.entries = Arc::new(XrefIndex::Materialized( + Arc::new(BTreeMap::from([ + (0, XrefEntry::Free { next_generation: 65535 }), + (2, XrefEntry::InFile { offset: 10, generation: 7 }), + (4, XrefEntry::Free { next_generation: 9 }), + (6, XrefEntry::InStream { stream: 80, index: 1 }), + (8, XrefEntry::InFile { offset: 20, generation: 5 }), + (10, XrefEntry::InFile { offset: 30, generation: 3 }), + (u32::MAX, XrefEntry::InFile { offset: 40, generation: 1 }), + ])), + Arc::default(), + )); doc.set(ObjRef::new(1, 2), Object::Int(1)); // before the first live source slot doc.set(ObjRef::new(2, 8), Object::Int(2)); // override, without duplication doc.set(ObjRef::new(4, 9), Object::Int(4)); // reuse a source free slot @@ -1846,6 +1849,32 @@ fn a_stream_whose_length_is_itself_and_never_ends_loads_quickly() { rx.recv_timeout(std::time::Duration::from_secs(20)).expect("loading must not retry exponentially"); } +#[test] +fn a_restored_revision_keeps_what_the_saved_file_held_in_object_streams() { + // The earlier revision: an empty document read back from a file whose catalog and pages + // sit in an object stream. + let earlier = Document::open(Arc::new(crate::write_full(&Document::new_empty(), &crate::SaveOptions::default()).unwrap())).unwrap(); + // The saved revision adds an object the catalog links to, so a full save numbers its objects + // anew; an undo then returns to the earlier graph. + let mut later = earlier.clone(); + let extra = later.add(Object::Int(7)); + let root = later.root().unwrap(); + later.update_dict(root, |c| c.set(b"Extra".to_vec(), Object::Ref(extra))).unwrap(); + let saved = crate::write_full(&later, &crate::SaveOptions::default()).unwrap(); + let mut base = Document::open(Arc::new(saved)).unwrap(); + assert!( + base.object_numbers().iter().any(|&n| matches!(base.xref_entry(n), Some(XrefEntry::InStream { .. }))), + "the saved file holds objects in a stream" + ); + base.overlay_changes_from(&earlier).unwrap(); + let file = crate::write_incremental(&base, &crate::SaveOptions::default()).unwrap(); + let reread = Document::open(Arc::new(file)).unwrap(); + let catalog = reread.get(reread.root().unwrap()).as_dict().unwrap().clone(); + assert!(catalog.get(b"Extra").is_none(), "the restored catalog"); + let pages = reread.resolve(catalog.get(b"Pages").unwrap()); + assert_eq!(pages.as_dict().and_then(|d| d.name(b"Type")), Some(&b"Pages"[..]), "the pages survive: {pages:?}"); +} + /// The cache keeps the sum of what its objects hold as they come and go, so neither reader has /// to measure the cache again on every read (#5873). #[test] diff --git a/crates/cos/src/document/xref.rs b/crates/cos/src/document/xref.rs index 02390ff7da..771df28b80 100644 --- a/crates/cos/src/document/xref.rs +++ b/crates/cos/src/document/xref.rs @@ -41,7 +41,7 @@ pub(super) struct XrefRevisionSection { #[derive(Clone, Debug)] pub(super) enum XrefIndex { Compact { revisions: Vec, max_num: u32, has_entries: bool }, - Materialized(Arc>), + Materialized(Arc>, Arc>), } pub(super) const MAX_COMPACT_XREF_ROWS: usize = 2_000_000; @@ -149,8 +149,8 @@ impl XrefSection { } } - pub(super) fn extend_numbers(&self, numbers: &mut Vec) { - for range in &self.ranges { + pub(super) fn numbers(&self) -> impl Iterator + '_ { + self.ranges.iter().flat_map(|range| { let available = match &self.data { XrefSectionData::Table => range.count as usize, XrefSectionData::Stream { bytes, widths } => { @@ -158,14 +158,12 @@ impl XrefSection { bytes.len().saturating_sub(range.offset).checked_div(row_width).unwrap_or(0) } }; - for i in 0..available.min(range.count as usize) { - if let Ok(i) = u32::try_from(i) - && let Some(num) = range.start.checked_add(i) - { - numbers.push(num); - } - } - } + (0..available.min(range.count as usize)).filter_map(move |i| u32::try_from(i).ok().and_then(|i| range.start.checked_add(i))) + }) + } + + pub(super) fn extend_numbers(&self, numbers: &mut Vec) { + numbers.extend(self.numbers()); } } @@ -189,20 +187,20 @@ impl XrefRevisionSection { impl XrefIndex { pub(super) fn empty() -> Self { - Self::Materialized(Arc::new(BTreeMap::new())) + Self::Materialized(Arc::new(BTreeMap::new()), Arc::default()) } pub(super) fn lookup(&self, num: u32, file: &[u8]) -> Option { match self { Self::Compact { revisions, .. } => revisions.iter().find_map(|revision| revision.lookup(num, file)), - Self::Materialized(entries) => entries.get(&num).copied(), + Self::Materialized(entries, _) => entries.get(&num).copied(), } } pub(super) fn is_empty(&self) -> bool { match self { Self::Compact { has_entries, .. } => !has_entries, - Self::Materialized(entries) => entries.is_empty(), + Self::Materialized(entries, _) => entries.is_empty(), } } @@ -211,14 +209,14 @@ impl XrefIndex { pub(super) fn remove_exhausted(&mut self) -> bool { match self { Self::Compact { .. } => false, - Self::Materialized(entries) => Arc::make_mut(entries).remove(&u32::MAX).is_some(), + Self::Materialized(entries, _) => Arc::make_mut(entries).remove(&u32::MAX).is_some(), } } /// Every entry, newest section first, as the materialized reader keeps them. pub(super) fn to_map(&self, file: &[u8]) -> BTreeMap { match self { - Self::Materialized(entries) => entries.as_ref().clone(), + Self::Materialized(entries, _) => entries.as_ref().clone(), Self::Compact { revisions, .. } => { let mut numbers = Vec::new(); for revision in revisions { @@ -232,13 +230,13 @@ impl XrefIndex { pub(super) fn max_num(&self) -> u32 { match self { Self::Compact { max_num, .. } => *max_num, - Self::Materialized(entries) => entries.keys().next_back().copied().unwrap_or(0), + Self::Materialized(entries, _) => entries.keys().next_back().copied().unwrap_or(0), } } pub(super) fn object_numbers(&self, file: &[u8]) -> Vec { match self { - Self::Materialized(entries) => { + Self::Materialized(entries, _) => { entries.iter().filter_map(|(&num, entry)| (!matches!(entry, XrefEntry::Free { .. })).then_some(num)).collect() } Self::Compact { revisions, .. } => { diff --git a/crates/cos/src/writer.rs b/crates/cos/src/writer.rs index dcfcb97997..e0762612dc 100644 --- a/crates/cos/src/writer.rs +++ b/crates/cos/src/writer.rs @@ -167,11 +167,17 @@ pub struct SaveOptions { /// before adding it, with [`Document::free_unlinked_additions`]. Full saves always leave /// these out. pub drop_unlinked: bool, + /// Entropy for the file identifier of a rewrite that asks for a new one (see + /// [`Document::require_full_save_with_new_id`]). Given, the identifier depends only on this and + /// on the written bytes, so the same save gives the same file; tests use it to pin output. + /// Without it, fresh entropy from the host (the operating system on native targets, the + /// browser on WebAssembly) is mixed in. + pub id_entropy: Option<[u8; 16]>, } impl Default for SaveOptions { fn default() -> Self { - Self { mod_date: None, id_seed: 0x5052_494E_5443_5241, object_streams: true, drop_unlinked: false } + Self { mod_date: None, id_seed: 0x5052_494E_5443_5241, object_streams: true, drop_unlinked: false, id_entropy: None } } } @@ -307,8 +313,17 @@ pub fn write_full(doc: &Document, opts: &SaveOptions) -> Result, CosErro if matches!(*o, Object::Null) { continue; // dangling reference: written as null by renumbering below } - map.insert(r, output_number(order.len())?); - order.push((r, matches!(*o, Object::Stream(_)))); + // Accepted xref/header aliases denote one slot, including page-tree cycles. + // Check for null first: an unrelated or freed generation must remain dangling. + let canonical = ObjRef::new(r.num, reader.document.generation(r.num)); + if let Some(&num) = map.get(&canonical) { + map.insert(r, num); + continue; + } + let num = output_number(order.len())?; + map.insert(r, num); + map.insert(canonical, num); + order.push((canonical, matches!(*o, Object::Stream(_)))); visit_refs(&o, &mut queue); } let version = reader.document.version().chars().take(3).collect::(); @@ -319,13 +334,15 @@ pub fn write_full(doc: &Document, opts: &SaveOptions) -> Result, CosErro out.extend_from_slice(&[b'%', 0xE2, 0xE3, 0xCF, 0xD3, b'\n']); // The /Encrypt dictionary and everything it references (e.g. indirect /CF crypt-filter // dictionaries) are needed before anything can be decrypted: never inside object streams. - let mut encryption_objects: std::collections::HashSet = std::collections::HashSet::new(); + let mut encryption_objects: std::collections::HashSet = std::collections::HashSet::new(); let mut pending: Vec = Vec::new(); if let Some(encrypt) = trailer_in.get(b"Encrypt") { collect_refs(encrypt, &mut |r| pending.push(r)); } while let Some(r) = pending.pop() { - if encryption_objects.insert(r) { + if let Some(&num) = map.get(&r) + && encryption_objects.insert(num) + { collect_refs(&reader.get(r), &mut |x| pending.push(x)); } } @@ -333,14 +350,15 @@ pub fn write_full(doc: &Document, opts: &SaveOptions) -> Result, CosErro // damaged and could not be repaired", before it even asks for the password), while other // readers accept it. qpdf keeps the catalog out of object streams too. Verified with // Acrobat DC on macOS (#774). - let root_outside = if reader.document.output_security().0.is_some() { trailer_in.reference(b"Root") } else { None }; + let root_outside = + if reader.document.output_security().0.is_some() { trailer_in.reference(b"Root").and_then(|r| map.get(&r)).copied() } else { None }; let mut rows: BTreeMap = BTreeMap::new(); // Keep only references, not renamed copies of the entire object graph. Standalone // objects still precede object streams, preserving the established output order. let mut packed: Vec<(u32, ObjRef)> = Vec::new(); for (i, (r, is_stream)) in order.iter().enumerate() { let num = output_number(i)?; - let in_stream = opts.object_streams && !is_stream && !encryption_objects.contains(r) && root_outside != Some(*r); + let in_stream = opts.object_streams && !is_stream && !encryption_objects.contains(&num) && root_outside != Some(num); if in_stream { packed.push((num, *r)); } else { @@ -371,26 +389,7 @@ pub fn write_full(doc: &Document, opts: &SaveOptions) -> Result, CosErro } } batch.flush(&reader.document, &mut next, &mut rows, &mut out)?; - let mut trailer = Dict::new(); - for key in [&b"Root"[..], b"Info", b"Encrypt"] { - if let Some(r) = trailer_in.reference(key).and_then(|r| map.get(&r)) { - trailer.set(key.to_vec(), Object::Ref(ObjRef::new(*r, 0))); - } - } - if let Some(encrypt @ Object::Dict(_)) = trailer_in.get(b"Encrypt") { - trailer.set(b"Encrypt".to_vec(), renumber(encrypt, &map)); - } - if let Some(id) = trailer_in.get(b"ID") { - trailer.set(b"ID".to_vec(), direct_id(&reader.document, id)); - } - for (key, value) in trailer_in.iter() { - if !trailer_key_rewritten(key) { - trailer.set(key.clone(), renumber(value, &map)); - } - } - if reader.document.output_security().0.is_none() { - ensure_id(&mut trailer, opts, reader.document.bytes()); - } + let mut trailer = full_save_trailer(&reader.document, &trailer_in, &map, opts, &out); if opts.object_streams { write_xref_stream(&mut out, 0, &mut trailer, rows, next); } else { @@ -409,6 +408,34 @@ pub fn write_full(doc: &Document, opts: &SaveOptions) -> Result, CosErro Ok(out) } +/// Assemble the full rewrite's trailer and apply its identifier policy. +fn full_save_trailer(doc: &Document, trailer_in: &Dict, map: &HashMap, opts: &SaveOptions, out: &[u8]) -> Dict { + let mut trailer = Dict::new(); + for key in [&b"Root"[..], b"Info", b"Encrypt"] { + if let Some(r) = trailer_in.reference(key).and_then(|r| map.get(&r)) { + trailer.set(key.to_vec(), Object::Ref(ObjRef::new(*r, 0))); + } + } + if let Some(encrypt @ Object::Dict(_)) = trailer_in.get(b"Encrypt") { + trailer.set(b"Encrypt".to_vec(), renumber(encrypt, map)); + } + if let Some(id) = trailer_in.get(b"ID") { + trailer.set(b"ID".to_vec(), direct_id(doc, id)); + } + for (key, value) in trailer_in.iter() { + if !trailer_key_rewritten(key) { + trailer.set(key.clone(), renumber(value, map)); + } + } + if doc.output_security().0.is_none() { + ensure_id(&mut trailer, opts, doc.bytes()); + } + if doc.new_id_required() { + regenerate_id(&mut trailer, opts, out, doc.output_security().0.is_some()); + } + trailer +} + /// Object `r` read again for writing. The traversal read it once and only kept its number, so /// a second read that fails, finds nothing or finds a different kind of object must fail the /// save rather than write `null` in its place. @@ -657,6 +684,136 @@ fn ensure_id(trailer: &mut Dict, opts: &SaveOptions, content: &[u8]) { trailer.set(b"ID".to_vec(), Object::Array(vec![s.clone(), s])); } +/// Give a rewritten file a new identifier (§14.4), so that a redacted or sanitized copy does not +/// claim to be its source. An encrypted file keeps the first string, which its key is derived from, +/// and gets a new second one. When there is no such string (no `/ID`, one entry, or a first entry +/// that is not a string) the reader takes the empty string as the key input (`Document::first_id`), +/// so the empty string is written first and the key stays valid. +fn regenerate_id(trailer: &mut Dict, opts: &SaveOptions, written: &[u8], encrypted: bool) { + let fresh = Object::String(PdfString { bytes: new_id_bytes(opts, written), hex: true }); + let pair = if encrypted { + let first = match trailer.get(b"ID") { + Some(Object::Array(a)) => match a.first() { + Some(Object::String(s)) => Object::String(s.clone()), + _ => Object::String(PdfString { bytes: Vec::new(), hex: true }), + }, + _ => Object::String(PdfString { bytes: Vec::new(), hex: true }), + }; + vec![first, fresh] + } else { + vec![fresh.clone(), fresh] + }; + trailer.set(b"ID".to_vec(), Object::Array(pair)); +} + +/// FNV-1a over 64 bits, with its published offset basis and prime. The identifier is built from it +/// rather than from `std`'s `DefaultHasher`, whose algorithm the standard library does not specify. +const FNV_OFFSET: u64 = 0xcbf2_9ce4_8422_2325; +const FNV_PRIME: u64 = 0x0000_0100_0000_01b3; + +fn fnv1a(state: u64, bytes: &[u8]) -> u64 { + bytes.iter().fold(state, |h, b| (h ^ u64::from(*b)).wrapping_mul(FNV_PRIME)) +} + +/// Sixteen bytes for a file identifier, one 8-byte half per lane. The bytes are a pure function of +/// `SaveOptions::id_entropy`, `id_seed` and the written file: the hash is FNV-1a (above) and every +/// integer is written as a fixed-width big-endian value, so the same input gives the same identifier +/// on every platform and toolchain. Without entropy only the salt changes (see `id_salt`). FNV is +/// not collision-resistant against chosen content; the identifier only has to be unique, not secret. +/// Every written byte is hashed, once, in a single traversal of the output. +fn new_id_bytes(opts: &SaveOptions, written: &[u8]) -> Vec { + let salt = id_salt(opts); + let length = u64::try_from(written.len()).unwrap_or(u64::MAX); + let mut halves = [0u8, 1].map(|lane| { + let mut h = fnv1a(FNV_OFFSET, &salt.to_be_bytes()); + h = fnv1a(h, &[lane]); + h = fnv1a(h, &opts.id_seed.to_be_bytes()); + fnv1a(h, &length.to_be_bytes()) + }); + for chunk in written.chunks(4096) { + for h in &mut halves { + *h = fnv1a(*h, chunk); + } + } + halves.iter().flat_map(|h| h.to_be_bytes()).collect() +} + +/// The salt of an identifier. With caller entropy it is a fixed hash of that entropy (deterministic, +/// for tests and callers who pin output). Without it the salt is fresh: on WebAssembly from the +/// browser ([`browser_salt`], so two fresh web apps that write the same bytes cannot share an +/// identifier), and elsewhere from the standard library's random hash keys — seeded by the operating +/// system, so separate processes differ — mixed with a process-wide counter, which separates +/// successive calls in one process. +fn id_salt(opts: &SaveOptions) -> u64 { + use std::hash::BuildHasher; + use std::sync::atomic::{AtomicU64, Ordering}; + static COUNTER: AtomicU64 = AtomicU64::new(0); + match &opts.id_entropy { + Some(entropy) => fnv1a(FNV_OFFSET, entropy), + None => { + #[cfg(target_arch = "wasm32")] + if let Some(salt) = browser_salt() { + return salt; + } + std::collections::hash_map::RandomState::new().hash_one(COUNTER.fetch_add(1, Ordering::Relaxed)) + } + } +} + +/// One byte from one `Math.random` draw: the draw (in [0, 1)) scaled to [0, 256), integer part +/// taken. The quotient is in [0, 256), so a saturating cast never wraps; a NaN or infinite draw +/// (which a conforming engine never returns) folds in as 0. No input can panic here. +#[cfg(any(target_arch = "wasm32", test))] +fn draw_byte(random: f64) -> u8 { + (random * 256.0).floor().rem_euclid(256.0) as u8 +} + +/// The `Math.random` fallback's mixing: one salt byte per draw (see [`draw_byte`]), folded through +/// FNV-1a exactly like the Web Crypto bytes, so all sixteen bytes of the fold carry entropy from +/// the draws. Pure and platform-independent, so the native test target can exercise it; on wasm +/// each draw comes from `Math.random`. +#[cfg(any(target_arch = "wasm32", test))] +fn salt_from_draws(draws: &[f64; 16]) -> u64 { + let mut bytes = [0u8; 16]; + for (byte, draw) in bytes.iter_mut().zip(draws) { + *byte = draw_byte(*draw); + } + fnv1a(FNV_OFFSET, &bytes) +} + +/// Fresh salt entropy from the browser, on WebAssembly. `crypto.getRandomValues` (Web Crypto, the +/// browser's secure generator, available in windows and workers alike) gives sixteen bytes, folded +/// into one salt. When that API is missing or refuses the call, `Math.random` is mixed instead +/// ([`salt_from_draws`]: sixteen draws, so the salt's full width carries entropy) — browsers seed +/// it afresh for every document, so a fresh instance still cannot reproduce another's salt. Only +/// where neither is reachable does the caller fall back to its `RandomState` salt, whose keys are +/// fixed at compile time on `wasm32-unknown-unknown` — the state before browser entropy was wired +/// in. +#[cfg(target_arch = "wasm32")] +fn browser_salt() -> Option { + use wasm_bindgen::JsCast; + let crypto = js_sys::Reflect::get(&js_sys::global(), &wasm_bindgen::JsValue::from_str("crypto")) + .ok() + .filter(|crypto| !crypto.is_undefined() && !crypto.is_null()); + if let Some(crypto) = crypto { + let fill = js_sys::Reflect::get(&crypto, &wasm_bindgen::JsValue::from_str("getRandomValues")) + .ok() + .and_then(|fill| fill.dyn_into::().ok()); + if let Some(fill) = fill { + let bytes = js_sys::Uint8Array::new_with_length(16); + if fill.call1(&crypto, &bytes).is_ok() { + return Some(fnv1a(FNV_OFFSET, &bytes.to_vec())); + } + } + } + // Web Crypto is missing or refused the call: degrade to `Math.random` rather than giving up. + let mut draws = [0.0; 16]; + for draw in &mut draws { + *draw = js_sys::Math::random(); + } + Some(salt_from_draws(&draws)) +} + /// A PDF date string for "now" given seconds since the Unix epoch (UTC). pub fn pdf_date(unix_secs: i64) -> String { let (days, secs) = (unix_secs.div_euclid(86_400), unix_secs.rem_euclid(86_400)); @@ -699,6 +856,56 @@ mod tests { use proptest::prelude::*; + #[test] + fn the_same_entropy_and_bytes_give_the_same_identifier_on_every_platform() { + // Computed from the formula outside this crate: fixed-width big-endian fields and FNV-1a. + let opts = super::SaveOptions { id_entropy: Some([0x42; 16]), id_seed: 7, ..super::SaveOptions::default() }; + let want = [0x96, 0xf8, 0xcb, 0x7d, 0x5b, 0xb0, 0x09, 0xc8, 0x92, 0xc9, 0xb7, 0x51, 0xdf, 0xe6, 0xab, 0xd1]; + assert_eq!(super::new_id_bytes(&opts, b"%PDF-1.7\n"), want); + } + + #[test] + fn fnv1a_is_the_published_64_bit_algorithm() { + // Published FNV-1a (64 bit) test vectors, so the identifier can be reproduced outside this crate. + assert_eq!(super::fnv1a(super::FNV_OFFSET, b""), 0xcbf2_9ce4_8422_2325); + assert_eq!(super::fnv1a(super::FNV_OFFSET, b"a"), 0xaf63_dc4c_8601_ec8c); + assert_eq!(super::fnv1a(super::FNV_OFFSET, b"foobar"), 0x8594_4171_f739_67e8); + } + + #[test] + fn math_random_draws_fill_the_whole_salt_width() { + // Draw i = (i + 1/2)/256 scales to the integer i, so the fold's byte i proves draw i landed + // there: all sixteen bytes carry their own draw's entropy (the fallback once shifted four + // draws into one word, leaving the other four bytes of the fold constant). + let draws: [f64; 16] = std::array::from_fn(|i| (i as f64 + 0.5) / 256.0); + let bytes: Vec = (0..16).collect(); + assert_eq!(super::salt_from_draws(&draws), super::fnv1a(super::FNV_OFFSET, &bytes)); + } + + #[test] + fn every_math_random_draw_reaches_the_salt() { + let base = [0.5; 16]; + let baseline = super::salt_from_draws(&base); + for i in 0..16 { + let mut draws = base; + draws[i] = 0.75; + assert_ne!(super::salt_from_draws(&draws), baseline, "draw {i} did not reach the salt"); + } + } + + #[test] + fn a_math_random_draw_outside_its_range_still_mixes_without_panicking() { + // A conforming `Math.random` stays in [0, 1), but the mixing must not panic outside it. + assert_eq!(super::draw_byte(0.0), 0); + assert_eq!(super::draw_byte(0.5), 128); + assert_eq!(super::draw_byte(255.5 / 256.0), 255); + assert_eq!(super::draw_byte(-0.5 / 256.0), 255); + assert_eq!(super::draw_byte(-1.0), 0); + assert_eq!(super::draw_byte(2.0), 0); + assert_eq!(super::draw_byte(f64::NAN), 0); + assert_eq!(super::draw_byte(f64::INFINITY), 0); + } + use super::*; use crate::Lexer; diff --git a/crates/cos/tests/file_identifier.rs b/crates/cos/tests/file_identifier.rs new file mode 100644 index 0000000000..313d56257b --- /dev/null +++ b/crates/cos/tests/file_identifier.rs @@ -0,0 +1,276 @@ +//! The file identifier (`/ID`, ISO 32000-2 §14.4) of a full save. An ordinary save keeps the +//! identifier the file has and writes the same bytes twice. A rewrite the document asked for (after +//! redaction or sanitizing) gives the file a new pair. An encrypted file keeps its first string, +//! because its file key is derived from it. + +use std::sync::Arc; + +use pdfcraft_cos::{Algorithm, Document, NewEncryption, Object, PdfString, SaveOptions, write_full, write_incremental}; +use pdfcraft_testkit::Pdf; + +const ID_HEX: &str = "/ID [<00112233445566778899AABBCCDDEEFF> <00112233445566778899AABBCCDDEEFF>]"; +const ID_BYTES: [u8; 16] = [0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77, 0x88, 0x99, 0xAA, 0xBB, 0xCC, 0xDD, 0xEE, 0xFF]; + +/// One page whose content holds a secret, with `entry` added to the trailer (for example an `/ID`). +fn source_with(entry: Option<&str>) -> Vec { + let pdf = Pdf::new() + .obj("<< /Type /Catalog /Pages 2 0 R >>") + .obj("<< /Type /Pages /Kids [3 0 R] /Count 1 /MediaBox [0 0 200 200] >>") + .obj("<< /Type /Page /Parent 2 0 R /Contents 4 0 R /Resources << /Font << /F1 5 0 R >> >> >>") + .stream("", "BT /F1 12 Tf 20 100 Td (TOPSECRET) Tj ET") + .obj("<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica >>"); + match entry { + Some(entry) => pdf.trailer_entry(entry).build(), + None => pdf.build(), + } +} + +/// [`source_with`] with the standard identifier, or none. +fn source(with_id: bool) -> Vec { + source_with(with_id.then_some(ID_HEX)) +} + +/// A one-page file whose 70,000-byte content stream ends in `last`. Two of these built with +/// different last bytes have the same length and differ only past the first 64 KiB. +fn large_source(last: u8) -> Vec { + let mut content = vec![b' '; 70_000]; + content.push(last); + Pdf::new() + .obj("<< /Type /Catalog /Pages 2 0 R >>") + .obj("<< /Type /Pages /Kids [3 0 R] /Count 1 /MediaBox [0 0 200 200] >>") + .obj("<< /Type /Page /Parent 2 0 R /Contents 4 0 R /Resources << /Font << /F1 5 0 R >> >> >>") + .stream("", content) + .obj("<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica >>") + .build() +} + +fn reopen(bytes: Vec) -> Document { + Document::open(Arc::new(bytes)).unwrap() +} + +/// The strings of the trailer's `/ID` (empty when there is none). +fn id_strings(doc: &Document) -> Vec> { + match doc.trailer().get(b"ID") { + Some(Object::Array(a)) => a.iter().map(|s| s.as_string().map(|s| s.bytes.clone()).unwrap_or_default()).collect(), + _ => Vec::new(), + } +} + +#[test] +fn an_ordinary_save_keeps_the_identifier_and_writes_the_same_bytes() { + let doc = reopen(source(true)); + let opts = SaveOptions::default(); + let saved = write_full(&doc, &opts).unwrap(); + assert_eq!(saved, write_full(&doc, &opts).unwrap(), "unchanged inputs produce identical bytes"); + assert_eq!(id_strings(&reopen(saved)), vec![ID_BYTES.to_vec(), ID_BYTES.to_vec()]); + let appended = reopen(write_incremental(&doc, &opts).unwrap()); + assert_eq!(id_strings(&appended), vec![ID_BYTES.to_vec(), ID_BYTES.to_vec()]); +} + +#[test] +fn an_ordinary_save_of_a_file_without_an_identifier_creates_the_same_one_every_time() { + let doc = reopen(source(false)); + let opts = SaveOptions::default(); + let saved = write_full(&doc, &opts).unwrap(); + assert_eq!(saved, write_full(&doc, &opts).unwrap(), "unchanged inputs produce identical bytes"); + let ids = id_strings(&reopen(saved)); + assert_eq!(ids.len(), 2); + assert!(ids.iter().all(|s| s.len() == 16), "{ids:?}"); +} + +#[test] +fn a_rewrite_the_document_asked_for_gets_a_new_identifier() { + let mut doc = reopen(source(true)); + doc.require_full_save_with_new_id(); + assert!(doc.full_save_required() && doc.new_id_required()); + let opts = SaveOptions::default(); + for saved in [write_incremental(&doc, &opts).unwrap(), write_full(&doc, &opts).unwrap()] { + let ids = id_strings(&reopen(saved)); + assert_eq!(ids.len(), 2); + for id in &ids { + assert_eq!(id.len(), 16); + assert_ne!(id.as_slice(), ID_BYTES.as_slice(), "the copy must not claim the source's identifier"); + } + } + // Every save is a new identity, not the same one again. + let first = id_strings(&reopen(write_full(&doc, &opts).unwrap())); + let second = id_strings(&reopen(write_full(&doc, &opts).unwrap())); + assert_ne!(first, second); +} + +#[test] +fn a_rewrite_without_an_identifier_still_gets_one() { + let mut doc = reopen(source(false)); + doc.require_full_save_with_new_id(); + let ids = id_strings(&reopen(write_full(&doc, &SaveOptions::default()).unwrap())); + assert_eq!(ids.len(), 2); + assert!(ids.iter().all(|s| s.len() == 16), "{ids:?}"); +} + +#[test] +fn caller_entropy_makes_the_new_identifier_reproducible() { + let mut doc = reopen(source(true)); + doc.require_full_save_with_new_id(); + let with = |entropy: u8| SaveOptions { id_entropy: Some([entropy; 16]), ..SaveOptions::default() }; + let first = write_full(&doc, &with(1)).unwrap(); + assert_eq!(first, write_full(&doc, &with(1)).unwrap(), "the same entropy gives the same bytes"); + let other = write_full(&doc, &with(2)).unwrap(); + assert_ne!(id_strings(&reopen(first)), id_strings(&reopen(other)), "different entropy gives a different identifier"); +} + +#[test] +fn an_encrypted_rewrite_keeps_the_first_identifier_string_its_key_depends_on() { + for algorithm in [Algorithm::Rc4_128, Algorithm::Aes128, Algorithm::Aes256] { + let mut doc = reopen(source(true)); + doc.set_encryption(&NewEncryption { + algorithm, + user_password: "user", + owner_password: "owner", + permissions: -1, + encrypt_metadata: true, + seed: [5; 32], + }) + .unwrap(); + let protected = write_full(&doc, &SaveOptions::default()).unwrap(); + let before = Document::open_with_password(Arc::new(protected), Some("user")).unwrap(); + + let mut doc = before.clone(); + doc.require_full_save_with_new_id(); + let out = write_incremental(&doc, &SaveOptions::default()).unwrap(); + // The output opens with the same password only if its first identifier string is unchanged. + let back = Document::open_with_password(Arc::new(out), Some("user")).unwrap(); + let (old, new) = (id_strings(&before), id_strings(&back)); + assert_eq!(new.len(), 2, "{algorithm:?}"); + assert_eq!(new.first(), old.first(), "{algorithm:?}: the key depends on the first string"); + assert_ne!(new.get(1), old.get(1), "{algorithm:?}: the second string is new"); + } +} + +/// The `/ID` shapes whose key the reader takes as the empty string: none at all, one entry, and a +/// first entry that is not a string. +fn unusable_ids() -> Vec<(&'static str, Option)> { + let empty = Object::String(PdfString::literal(Vec::new())); + vec![ + ("absent", None), + ("one entry", Some(Object::Array(vec![empty.clone()]))), + ("first entry not a string", Some(Object::Array(vec![Object::Int(7), empty]))), + ] +} + +/// An encrypted file whose key is derived from the empty string, with its trailer `/ID` replaced by `id`. +fn protected_with_id(algorithm: Algorithm, id: Option) -> Vec { + let mut doc = reopen(source_with(Some("/ID [<> <>]"))); + doc.set_encryption(&NewEncryption { + algorithm, + user_password: "user", + owner_password: "owner", + permissions: -1, + encrypt_metadata: true, + seed: [5; 32], + }) + .unwrap(); + match id { + Some(id) => doc.trailer_mut().set(b"ID".to_vec(), id), + None => { + doc.trailer_mut().remove(b"ID"); + } + } + write_full(&doc, &SaveOptions::default()).unwrap() +} + +#[test] +fn an_encrypted_rewrite_without_a_usable_identifier_still_gets_a_new_one() { + // The key is derived from the first string of /ID, and a missing /ID, a one-entry /ID and a + // first entry that is not a string all count as the empty string. The rewrite keeps that + // (empty) first string, so the file still opens with its password, and writes a new second one. + for algorithm in [Algorithm::Rc4_128, Algorithm::Aes128, Algorithm::Aes256] { + for (shape, id) in unusable_ids() { + let before = Document::open_with_password(Arc::new(protected_with_id(algorithm, id)), Some("user")).unwrap(); + let mut doc = before.clone(); + doc.require_full_save_with_new_id(); + let out = write_incremental(&doc, &SaveOptions::default()).unwrap(); + let back = Document::open_with_password(Arc::new(out), Some("user")).unwrap(); + let ids = id_strings(&back); + let previous = id_strings(&before); + assert_eq!(ids.len(), 2, "{algorithm:?}, {shape}: the rewrite writes a pair"); + assert!(ids.first().is_some_and(|s| s.is_empty()), "{algorithm:?}, {shape}: the first string is the one the key uses"); + assert_eq!(ids.get(1).map(Vec::len), Some(16), "{algorithm:?}, {shape}"); + assert_ne!(ids.get(1), previous.get(1), "{algorithm:?}, {shape}: the second string is new"); + } + } +} + +/// The standard fixture rewritten for a new identifier, with `id_entropy` of 0x42. +fn fixed_entropy_rewrite() -> Vec { + let mut doc = reopen(source(true)); + doc.require_full_save_with_new_id(); + write_full(&doc, &SaveOptions { id_entropy: Some([0x42; 16]), ..SaveOptions::default() }).unwrap() +} + +#[test] +fn a_fixed_entropy_gives_the_same_identifier_every_time() { + // The formula itself is pinned on fixed bytes in the writer's tests. The written file is not the + // same on every platform, so no identifier of a rewrite can be pinned here. + let first = id_strings(&reopen(fixed_entropy_rewrite())); + assert_eq!(first.len(), 2); + assert!(first.iter().all(|s| s.len() == 16), "{first:?}"); + assert_eq!(first, id_strings(&reopen(fixed_entropy_rewrite()))); +} + +#[test] +fn the_identifier_depends_on_the_whole_rewritten_file() { + // Same caller entropy and same length; the two rewrites differ only in one byte of a stream + // that lies past the first 64 KiB of the output. Their identifiers must still differ. + let opts = SaveOptions { id_entropy: Some([9; 16]), ..SaveOptions::default() }; + let rewrite = |last: u8| { + let mut doc = reopen(large_source(last)); + doc.require_full_save_with_new_id(); + write_full(&doc, &opts).unwrap() + }; + let (a, b) = (rewrite(b'A'), rewrite(b'B')); + assert_eq!(a.len(), b.len(), "the two rewrites have the same length"); + let first_difference = a.iter().zip(&b).position(|(x, y)| x != y).unwrap_or(usize::MAX); + assert!(first_difference > 64 * 1024, "the rewrites first differ at byte {first_difference}, inside the first 64 KiB"); + assert_ne!(id_strings(&reopen(a)), id_strings(&reopen(b)), "content past 64 KiB must change the identifier"); +} + +/// Where [`stand_in_for_a_fresh_process`] writes its rewrite. +const ID_PROBE: &str = "PDFCRAFT_TEST_ID_PROBE"; + +/// The child half of [`separate_fresh_processes_get_different_identifiers`]: it runs alone, in its +/// own process, and writes a default-path rewrite of the standard fixture to the file the parent +/// named. Without the variable it is a plain no-op. +#[test] +fn stand_in_for_a_fresh_process() { + let Some(path) = std::env::var_os(ID_PROBE) else { return }; + let mut doc = reopen(source(true)); + doc.require_full_save_with_new_id(); + std::fs::write(path, write_full(&doc, &SaveOptions::default()).unwrap()).unwrap(); +} + +#[test] +fn separate_fresh_processes_get_different_identifiers() { + // The default identifier is meant to differ between fresh instances of the whole save pipeline, + // not only between two calls in one process (which the counter already separates). On the native + // test target a fresh instance is a fresh process, whose hash keys the operating system seeds; + // the fix under test runs on WebAssembly, where no browser test harness exists — there the + // browser's `crypto.getRandomValues` stands in for this seeding, and this test cannot see it. + let exe = std::env::current_exe().unwrap(); + let probe = |name: &str| { + let path = std::env::temp_dir().join(format!("pdfcraft-id-probe-{}-{name}.pdf", std::process::id())); + let out = std::process::Command::new(&exe) + .args(["--exact", "stand_in_for_a_fresh_process", "--nocapture", "--"]) + .env(ID_PROBE, &path) + .output() + .expect("run this test's own binary as a child"); + assert!(out.status.success(), "{name}: the child failed: {}", String::from_utf8_lossy(&out.stderr)); + let saved = std::fs::read(&path).unwrap_or_else(|e| panic!("{name}: the child wrote its file: {e}")); + let ids = id_strings(&reopen(saved)); + assert_eq!(ids.len(), 2, "{name}"); + assert!(ids.iter().all(|s| s.len() == 16), "{name}: {ids:?}"); + let _ = std::fs::remove_file(&path); + ids + }; + let (first, second) = (probe("a"), probe("b")); + assert_ne!(first, second, "two fresh processes must not share an identifier"); +} diff --git a/crates/cos/tests/it/generations.rs b/crates/cos/tests/it/generations.rs new file mode 100644 index 0000000000..bbe2ff0b33 --- /dev/null +++ b/crates/cos/tests/it/generations.rs @@ -0,0 +1,510 @@ +//! An indirect reference identifies an object number AND its generation (ISO 32000-2 §7.3.10). + +use std::sync::Arc; + +use hayro_syntax::object::{Dict as HayroDict, ObjectIdentifier}; +use pdfcraft_cos::{Document, ObjRef, Object, SaveOptions, XrefEntry, write_full, write_incremental}; + +/// Object 3 was reused: only 3 1 R refers to the current value, never 3 0 R. +fn reused_object() -> Arc> { + object_file(&[ + (1, 0, "<< /Type /Catalog /Pages 2 0 R /Stale 3 0 R /Current 3 1 R >>"), + (2, 0, "<< /Type /Pages /Kids [] /Count 0 >>"), + (3, 1, "42"), + ]) +} + +fn object_file(objects: &[(u32, u16, &str)]) -> Arc> { + object_file_with_entry(objects, None) +} + +fn object_file_with_entry(objects: &[(u32, u16, &str)], changed: Option<(u32, u16, bool)>) -> Arc> { + let mut bytes = b"%PDF-1.7\n".to_vec(); + let mut offsets = Vec::new(); + for (num, generation, value) in objects { + offsets.push((*num, bytes.len(), *generation)); + bytes.extend_from_slice(format!("{num} {generation} obj\n{value}\nendobj\n").as_bytes()); + } + let xref = bytes.len(); + let size = objects.len() + 1; + bytes.extend_from_slice(format!("xref\n0 {size}\n0000000000 65535 f \n").as_bytes()); + for (num, offset, generation) in offsets { + let (generation, free) = changed.filter(|(n, _, _)| *n == num).map_or((generation, false), |(_, g, free)| (g, free)); + let (offset, kind) = if free { (0, 'f') } else { (offset, 'n') }; + bytes.extend_from_slice(format!("{offset:010} {generation:05} {kind} \n").as_bytes()); + } + bytes.extend_from_slice(format!("trailer\n<< /Size {size} /Root 1 0 R >>\nstartxref\n{xref}\n%%EOF\n").as_bytes()); + Arc::new(bytes) +} + +fn contradictory_entry(free: bool) -> Arc> { + object_file_with_entry( + &[(1, 0, "<< /Type /Catalog /Pages 2 0 R /Stale 3 0 R /Current 3 1 R >>"), (2, 0, "<< /Type /Pages /Kids [] /Count 0 >>"), (3, 1, "42")], + Some((3, if free { 2 } else { 0 }, free)), + ) +} + +#[test] +fn contradictory_xref_generation_recovers_the_matching_header_and_records_repair() { + for warm in [false, true] { + let doc = Document::open(contradictory_entry(false)).unwrap(); + if warm { + assert_eq!(*doc.try_get(3).unwrap(), Object::Int(42)); + } + assert_eq!(*doc.get(ObjRef::new(3, 1)), Object::Int(42)); + assert_eq!(*doc.resolve(&Object::Ref(ObjRef::new(3, 1))), Object::Int(42)); + // Preserve the existing xref-matching behavior, but never accept an unrelated generation. + assert_eq!(*doc.get(ObjRef::new(3, 0)), Object::Int(42)); + assert_eq!(*doc.get(ObjRef::new(3, 2)), Object::Null); + assert!(!doc.xref_entry_is_exact(3), "a contradictory generation is not an exact xref entry"); + assert!( + doc.repair_log() + .iter() + .any(|line| { line.contains("object 3") && line.contains("xref generation 0") && line.contains("header generation 1") }), + "{:?}", + doc.repair_log() + ); + let repairs = doc.repair_log().to_vec(); + for _ in 0..3 { + assert_eq!(*doc.get(ObjRef::new(3, 1)), Object::Int(42)); + } + assert_eq!(doc.repair_log(), repairs); + + let mut edited = doc.clone(); + edited.free(ObjRef::new(3, 1)); + for generation in [0, 1, 2] { + assert_eq!(*edited.get(ObjRef::new(3, generation)), Object::Null); + } + edited.set(ObjRef::new(3, 2), Object::Int(99)); + assert_eq!(*edited.get(ObjRef::new(3, 1)), Object::Null); + assert_eq!(*edited.get(ObjRef::new(3, 2)), Object::Int(99)); + assert_eq!(*doc.get(ObjRef::new(3, 1)), Object::Int(42)); + } +} + +/// A damaged row may name generation zero while the retained object and catalog use one. +fn contradictory_dictionary() -> Arc> { + object_file_with_entry( + &[(1, 0, "<< /Type /Catalog /Pages 2 0 R /Current 3 1 R >>"), (2, 0, "<< /Type /Pages /Kids [] /Count 0 >>"), (3, 1, "<< /Value 42 >>")], + Some((3, 0, false)), + ) +} + +fn assert_repaired_dictionary(bytes: Vec, value: i32) { + let bytes = Arc::new(bytes); + let oracle = hayro_syntax::Pdf::new(bytes.clone()).unwrap(); + let catalog = oracle.xref().get::>(ObjectIdentifier::new(1, 0)).unwrap(); + assert_eq!(catalog.get::>(b"Current").unwrap().get::(b"Value"), Some(value)); + let back = Document::open(bytes).unwrap(); + let catalog = back.get(back.root().unwrap()); + let current = back.resolve(catalog.as_dict().unwrap().get(b"Current").unwrap()); + assert_eq!(current.as_dict().unwrap().int(b"Value"), Some(i64::from(value))); +} + +#[test] +fn repaired_header_references_survive_full_saves_and_dictionary_edits() { + let original = contradictory_dictionary(); + let doc = Document::open(original.clone()).unwrap(); + for object_streams in [false, true] { + assert_repaired_dictionary(write_full(&doc, &SaveOptions { object_streams, ..SaveOptions::default() }).unwrap(), 42); + } + let mut edited = doc.clone(); + edited.update_dict(ObjRef::new(3, 1), |dict| dict.set(b"Value".to_vec(), Object::Int(77))).unwrap(); + assert_eq!(edited.generation(3), 1); + let incremental = write_incremental(&edited, &SaveOptions::default()).unwrap(); + assert!(edited.full_save_required(), "repair-aware dictionary edits normalize on full rewrite"); + assert_repaired_dictionary(incremental, 77); + for object_streams in [false, true] { + assert_repaired_dictionary(write_full(&edited, &SaveOptions { object_streams, ..SaveOptions::default() }).unwrap(), 77); + } + let mut restored = doc.clone(); + restored.overlay_changes_from(&edited).unwrap(); + assert_repaired_dictionary(write_incremental(&restored, &SaveOptions::default()).unwrap(), 77); +} + +#[test] +fn free_xref_slots_never_revive_retained_object_headers() { + for warm in [false, true] { + let doc = Document::open(contradictory_entry(true)).unwrap(); + assert!(matches!(doc.xref_entry(3), Some(XrefEntry::Free { next_generation: 2 }))); + if warm { + assert_eq!(*doc.try_get(3).unwrap(), Object::Null); + } + for generation in [0, 1, 2] { + assert_eq!(*doc.get(ObjRef::new(3, generation)), Object::Null); + } + assert_eq!(*doc.resolve(&Object::Ref(ObjRef::new(3, 1))), Object::Null); + assert!(doc.repair_log().is_empty(), "{:?}", doc.repair_log()); + } + // Control: the retained bytes really are a readable object when its slot is live. + let live = Document::open(contradictory_entry(false)).unwrap(); + assert_eq!(*live.try_get(3).unwrap(), Object::Int(42)); +} + +#[test] +fn incremental_generation_identity_is_visible_to_an_independent_reader() { + let original = reused_object(); + let mut doc = Document::open(original.clone()).unwrap(); + doc.set(ObjRef::new(3, 1), Object::Int(77)); + let bytes = write_incremental(&doc, &SaveOptions::default()).unwrap(); + assert!(bytes.starts_with(&original)); + let oracle = hayro_syntax::Pdf::new(Arc::new(bytes)).unwrap(); + let catalog = oracle.xref().get::>(ObjectIdentifier::new(1, 0)).unwrap(); + assert_eq!(catalog.get::(b"Current"), Some(77)); + assert_eq!(catalog.get::(b"Stale"), None); +} + +#[test] +fn generation_repair_diagnostics_are_bounded_without_disabling_recovery() { + let mut objects = vec![(1, 0, "<< /Type /Catalog /Pages 2 0 R >>"), (2, 0, "<< /Type /Pages /Kids [] /Count 0 >>")]; + objects.extend((3..303).map(|num| (num, 1, "42"))); + let bytes = object_file(&objects); + let bytes = String::from_utf8(bytes.as_ref().clone()).unwrap().replace("00001 n \n", "00000 n \n"); + for materialized in [false, true] { + let bytes = if materialized { bytes.replace("/Size 303", "/Size 302") } else { bytes.clone() }; + let doc = Document::open(Arc::new(bytes.into_bytes())).unwrap(); + assert_eq!(doc.repair_log().len(), 257); + assert!(doc.repair_log().last().unwrap().contains("omitted after 256 objects")); + assert_eq!(*doc.get(ObjRef::new(3, 1)), Object::Int(42)); + assert_eq!(*doc.get(ObjRef::new(302, 1)), Object::Int(42)); + assert_eq!(*doc.get(ObjRef::new(302, 2)), Object::Null); + assert_eq!(doc.repair_log().len(), 257); + } +} + +#[test] +fn references_require_the_current_generation_even_with_a_warm_cache() { + let bytes = reused_object(); + let oracle = hayro_syntax::Pdf::new(bytes.clone()).unwrap(); + assert_eq!(oracle.xref().get::(ObjectIdentifier::new(3, 0)), None); + assert_eq!(oracle.xref().get::(ObjectIdentifier::new(3, 1)), Some(42)); + + let doc = Document::open(bytes).unwrap(); + assert!(doc.repair_log().is_empty()); + assert_eq!(*doc.get(ObjRef::new(3, 0)), Object::Null); + assert_eq!(*doc.try_get(3).unwrap(), Object::Int(42)); + assert_eq!(*doc.get(ObjRef::new(3, 1)), Object::Int(42)); + assert_eq!(*doc.get(ObjRef::new(3, 0)), Object::Null); + assert_eq!(*doc.resolve(&Object::Ref(ObjRef::new(3, 0))), Object::Null); + assert_eq!(*doc.get(ObjRef::new(3, 2)), Object::Null); +} + +#[test] +fn overlay_generation_changes_do_not_retarget_old_references_or_clones() { + let mut doc = Document::open(reused_object()).unwrap(); + let snapshot = doc.clone(); + let old = ObjRef::new(3, 1); + let new = ObjRef::new(3, 2); + doc.free(old); + assert_eq!(*doc.get(old), Object::Null); + doc.set(new, Object::Int(99)); + assert_eq!(*doc.get(old), Object::Null); + assert_eq!(*doc.get(new), Object::Int(99)); + assert_eq!(*snapshot.get(old), Object::Int(42)); + assert_eq!(*snapshot.get(new), Object::Null); + + let saved = write_incremental(&doc, &SaveOptions::default()).unwrap(); + let back = Document::open(Arc::new(saved)).unwrap(); + assert!(back.repair_log().is_empty()); + assert_eq!(*back.get(old), Object::Null); + assert_eq!(*back.get(new), Object::Int(99)); +} + +#[test] +fn restoring_snapshot_preserves_changes_to_nonzero_generation_objects() { + for generation in [1, 2] { + let mut saved = Document::open(reused_object()).unwrap(); + let mut current = saved.clone(); + let reference = ObjRef::new(3, generation); + current.set(reference, Object::Int(99)); + current.update_dict(current.root().unwrap(), |catalog| catalog.set(b"Current".to_vec(), Object::Ref(reference))).unwrap(); + + saved.overlay_changes_from(¤t).unwrap(); + assert_eq!(*saved.get(reference), Object::Int(99), "generation {generation}"); + let back = Document::open(Arc::new(write_incremental(&saved, &SaveOptions::default()).unwrap())).unwrap(); + assert!(back.repair_log().is_empty()); + assert_eq!(*back.get(reference), Object::Int(99), "generation {generation}"); + let catalog = back.get(back.root().unwrap()); + assert_eq!(*back.resolve(catalog.as_dict().unwrap().get(b"Current").unwrap()), Object::Int(99)); + } +} + +#[test] +fn full_save_does_not_turn_stale_references_into_live_ones() { + let doc = Document::open(reused_object()).unwrap(); + for object_streams in [false, true] { + let saved = write_full(&doc, &SaveOptions { object_streams, ..SaveOptions::default() }).unwrap(); + let saved = Arc::new(saved); + let oracle = hayro_syntax::Pdf::new(saved.clone()).unwrap(); + let catalog = oracle.xref().get::>(ObjectIdentifier::new(1, 0)).unwrap(); + assert_eq!(catalog.get::(b"Current"), Some(42)); + assert_eq!(catalog.get::(b"Stale"), None); + let back = Document::open(saved).unwrap(); + assert!(back.repair_log().is_empty()); + let catalog = back.get(back.root().unwrap()); + let catalog = catalog.as_dict().unwrap(); + assert_eq!(catalog.get(b"Stale"), Some(&Object::Null)); + assert_eq!(*back.resolve(catalog.get(b"Current").unwrap()), Object::Int(42)); + } +} + +#[test] +fn compressed_objects_only_accept_generation_zero() { + let doc = Document::open(reused_object()).unwrap(); + let saved = write_full(&doc, &SaveOptions::default()).unwrap(); + let back = Document::open(Arc::new(saved)).unwrap(); + let catalog = back.get(back.root().unwrap()); + let current = catalog.as_dict().unwrap().reference(b"Current").unwrap(); + assert_eq!(current.generation, 0); + assert!(matches!(back.xref_entry(current.num), Some(XrefEntry::InStream { .. }))); + assert_eq!(*back.get(current), Object::Int(42)); + assert_eq!(*back.get(ObjRef::new(current.num, 1)), Object::Null); +} + +#[test] +fn indirect_stream_lengths_require_the_referenced_generation() { + for generation in [0, 1] { + // The embedded endstream is data when /Length resolves, but the existing missing- + // length recovery stops there when the reference is stale. + let stream = format!("<< /Length 4 {generation} R >>\nstream\nA\nendstream\nB\nendstream"); + let doc = Document::open(object_file(&[ + (1, 0, "<< /Type /Catalog /Pages 2 0 R >>"), + (2, 0, "<< /Type /Pages /Kids [] /Count 0 >>"), + (3, 0, &stream), + (4, 1, "13"), + ])) + .unwrap(); + let object = doc.get(ObjRef::new(3, 0)); + let Object::Stream(stream) = &*object else { panic!("expected stream") }; + assert_eq!(&stream.raw[..], if generation == 1 { b"A\nendstream\nB".as_slice() } else { b"A".as_slice() }); + } +} + +/// A newer valid row must not hide damage in a retained older revision. +fn shadowed_generation_disagreement(materialized: bool, shifted: bool) -> Arc> { + let original = contradictory_dictionary(); + let text = std::str::from_utf8(&original).unwrap(); + let previous: usize = text.rsplit("startxref\n").next().unwrap().lines().next().unwrap().parse().unwrap(); + let mut bytes = (*original).clone(); + let offset = bytes.len(); + bytes.extend_from_slice(b"3 1 obj\n<< /Value 77 >>\nendobj\n"); + let xref = bytes.len(); + let size = if materialized { 3 } else { 4 }; + bytes.extend_from_slice( + format!("xref\n3 1\n{offset:010} 00001 n \ntrailer\n<< /Size {size} /Root 1 0 R /Prev {previous} >>\nstartxref\n{xref}\n%%EOF\n").as_bytes(), + ); + if shifted { + bytes.splice(..0, b"junk before the header\n".iter().copied()); + } + Arc::new(bytes) +} + +#[test] +fn retained_generation_disagreements_are_reported_in_materialized_mode_too() { + for materialized in [false, true] { + for shifted in [false, true] { + let doc = Document::open(shadowed_generation_disagreement(materialized, shifted)).unwrap(); + assert_eq!(doc.revisions().len(), 2, "the xref chain is retained"); + assert!(!doc.reconstructed()); + assert_eq!(doc.get(ObjRef::new(3, 1)).as_dict().unwrap().int(b"Value"), Some(77)); + assert!( + doc.repair_log().iter().any(|line| line.contains("object 3: xref generation 0 disagrees with object header generation 1")), + "materialized={materialized}, shifted={shifted}: {:?}", + doc.repair_log() + ); + } + } +} + +#[test] +fn repair_aware_edits_keep_both_aliases_and_full_save_normalizes_them() { + let mut doc = Document::open(contradictory_entry(false)).unwrap(); + let snapshot = doc.clone(); + doc.set_existing(ObjRef::new(3, 0), Object::Int(77)).unwrap(); + for generation in [0, 1] { + assert_eq!(*doc.get(ObjRef::new(3, generation)), Object::Int(77)); + assert_eq!(*snapshot.get(ObjRef::new(3, generation)), Object::Int(42)); + } + let state = doc.read_state(); + assert!(doc.set_existing(ObjRef::new(3, 2), Object::Int(99)).is_err()); + assert!(state.matches(&doc)); + assert_eq!(*doc.get(ObjRef::new(3, 2)), Object::Null); + let mut restored = snapshot.clone(); + restored.overlay_changes_from(&doc).unwrap(); + for changed in [&doc, &restored] { + assert!(changed.full_save_required()); + for bytes in [ + write_incremental(changed, &SaveOptions::default()).unwrap(), + write_full(changed, &SaveOptions { object_streams: false, ..SaveOptions::default() }).unwrap(), + ] { + let bytes = Arc::new(bytes); + let oracle = hayro_syntax::Pdf::new(bytes.clone()).unwrap(); + let catalog = oracle.xref().get::>(ObjectIdentifier::new(1, 0)).unwrap(); + assert_eq!(catalog.get::(b"Current"), Some(77)); + assert_eq!(catalog.get::(b"Stale"), Some(77), "both source aliases stay live"); + let back = Document::open(bytes).unwrap(); + assert!(back.repair_log().is_empty(), "normalized output has no generation repairs"); + } + } + // Explicit replacement keeps its narrower, caller-supplied generation contract. + doc.set(ObjRef::new(3, 2), Object::Int(99)); + assert_eq!(*doc.get(ObjRef::new(3, 0)), Object::Null); + assert_eq!(*doc.get(ObjRef::new(3, 1)), Object::Null); + assert_eq!(*doc.get(ObjRef::new(3, 2)), Object::Int(99)); +} + +#[test] +fn hybrid_generation_diagnostics_observe_rows_before_the_table_hides_them() { + let mut bytes = b"%PDF-1.7\n".to_vec(); + let mut offsets = Vec::new(); + for (num, generation, body) in [(1, 0, "<< /Type /Catalog /Pages 2 0 R >>"), (2, 0, "<< /Type /Pages /Kids [] /Count 0 >>"), (3, 1, "42")] { + offsets.push(bytes.len()); + bytes.extend_from_slice(format!("{num} {generation} obj\n{body}\nendobj\n").as_bytes()); + } + let hybrid = bytes.len(); + offsets.push(hybrid); + bytes.extend_from_slice(b"4 0 obj\n<< /Type /XRef /Size 5 /W [1 4 2] /Index [3 1] /Length 7 >>\nstream\n"); + bytes.push(1); + bytes.extend_from_slice(&(offsets[2] as u32).to_be_bytes()); + bytes.extend_from_slice(&0u16.to_be_bytes()); + bytes.extend_from_slice(b"\nendstream\nendobj\n"); + let xref = bytes.len(); + bytes.extend_from_slice(b"xref\n0 5\n0000000000 65535 f \n"); + for (i, offset) in offsets.into_iter().enumerate() { + let generation = u16::from(i == 2); + bytes.extend_from_slice(format!("{offset:010} {generation:05} n \n").as_bytes()); + } + bytes.extend_from_slice(format!("trailer\n<< /Size 5 /Root 1 0 R /XRefStm {hybrid} >>\nstartxref\n{xref}\n%%EOF\n").as_bytes()); + let doc = Document::open(Arc::new(bytes)).unwrap(); + assert!(!doc.reconstructed()); + assert!(doc.xref_oddities().iter().any(|(_, num)| *num == 3), "the conflicting hybrid row is set aside"); + assert_eq!(*doc.get(ObjRef::new(3, 1)), Object::Int(42)); + assert_eq!(*doc.get(ObjRef::new(3, 0)), Object::Null, "the table still determines accepted references"); + assert!(doc.repair_log().iter().any(|line| line.contains("object 3: xref generation 0 disagrees with object header generation 1"))); +} + +#[test] +fn discarded_xref_repair_notes_do_not_survive_object_header_reconstruction() { + let bytes = + String::from_utf8((*contradictory_dictionary()).clone()).unwrap().replace("/Root 1 0 R", "/Root 99 0 R").replace("/Size 4", "/Size 3"); + let doc = Document::open(Arc::new(bytes.into_bytes())).unwrap(); + assert!(doc.reconstructed()); + assert!(doc.repair_log().iter().all(|line| !line.contains("xref generation")), "{:?}", doc.repair_log()); +} + +#[test] +fn restoring_equal_values_still_restores_the_accepted_generation_aliases() { + for edited in [false, true] { + for value in [42, 77] { + let mut current = Document::open(contradictory_entry(false)).unwrap(); + if edited { + current.set_existing(ObjRef::new(3, 0), Object::Int(42)).unwrap(); + } + let mut saved = Document::open(object_file(&[ + (1, 0, "<< /Type /Catalog /Pages 2 0 R /Stale 3 0 R /Current 3 1 R >>"), + (2, 0, "<< /Type /Pages /Kids [] /Count 0 >>"), + (3, 0, &value.to_string()), + ])) + .unwrap(); + assert_eq!(*saved.get(ObjRef::new(3, 1)), Object::Null); + saved.overlay_changes_from(¤t).unwrap(); + for generation in [0, 1] { + assert_eq!(*saved.get(ObjRef::new(3, generation)), Object::Int(42), "edited={edited}, value={value}"); + } + assert!(saved.full_save_required()); + } + } +} + +#[test] +fn repaired_page_tree_aliases_keep_one_identity_in_full_saves() { + let source = object_file_with_entry( + &[ + (1, 0, "<< /Type /Catalog /Pages 2 1 R /Stale 2 2 R >>"), + (2, 1, "<< /Type /Pages /Kids [3 0 R] /Count 1 >>"), + (3, 0, "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 10 10] >>"), + ], + Some((2, 0, false)), + ); + for edited in [false, true] { + let mut doc = Document::open(source.clone()).unwrap(); + if edited { + doc.update_dict(ObjRef::new(2, 0), |dict| dict.set(b"Edited".to_vec(), Object::Bool(true))).unwrap(); + } + let mut restored = Document::open(source.clone()).unwrap(); + restored.overlay_changes_from(&doc).unwrap(); + for current in [&doc, &restored] { + for object_streams in [false, true] { + let opts = SaveOptions { object_streams, ..SaveOptions::default() }; + let mut saves = vec![write_full(current, &opts).unwrap()]; + if edited { + saves.push(write_incremental(current, &opts).unwrap()); + } + for bytes in saves { + let back = Document::open(Arc::new(bytes)).unwrap(); + let catalog = back.get(back.root().unwrap()); + let catalog = catalog.as_dict().unwrap(); + let pages_ref = catalog.reference(b"Pages").unwrap(); + let pages = back.get(pages_ref); + let child_ref = pages.as_dict().unwrap().get(b"Kids").unwrap().as_array().unwrap()[0].as_ref().unwrap(); + let child = back.get(child_ref); + assert_eq!(child.as_dict().unwrap().reference(b"Parent"), Some(pages_ref), "edited={edited}, packed={object_streams}"); + assert!(matches!(catalog.get(b"Stale"), Some(Object::Null)), "unrelated generations stay null"); + assert!(back.repair_log().is_empty()); + } + } + } + } +} + +#[test] +fn restoring_explicit_sets_removes_old_generation_aliases() { + for value in [42, 77] { + let mut current = Document::open(contradictory_entry(false)).unwrap(); + current.set(ObjRef::new(3, 0), Object::Int(42)); + for repaired_overlay in [false, true] { + let mut saved = Document::open(contradictory_entry(false)).unwrap(); + if repaired_overlay { + saved.set_existing(ObjRef::new(3, 0), Object::Int(value)).unwrap(); + } + saved.overlay_changes_from(¤t).unwrap(); + assert_eq!(*saved.get(ObjRef::new(3, 0)), Object::Int(42)); + assert_eq!(*saved.get(ObjRef::new(3, 1)), Object::Null, "value={value}, overlay={repaired_overlay}"); + } + } +} + +/// `accepted_generations` lists exactly the generations `get` resolves, for every kind of slot: +/// a clean file slot, a damaged one (row and header disagree), a repair-aware edit, a plain +/// edit, a freed slot, a free row, an object stream member and an unknown number. +#[test] +fn accepted_generations_are_the_generations_get_resolves() { + let agrees = |doc: &Document, num: u32, expected: &[u16]| { + assert_eq!(doc.accepted_generations(num), expected, "object {num}"); + for generation in 0..8 { + let live = !matches!(*doc.get(ObjRef::new(num, generation)), Object::Null); + assert_eq!(live, expected.contains(&generation), "object {num} generation {generation}"); + } + }; + let doc = Document::open(contradictory_entry(false)).unwrap(); + agrees(&doc, 1, &[0]); + agrees(&doc, 3, &[0, 1]); + agrees(&doc, 9, &[]); + let mut repaired = doc.clone(); + repaired.set_existing(ObjRef::new(3, 0), Object::Int(77)).unwrap(); + agrees(&repaired, 3, &[0, 1]); + let mut narrowed = doc.clone(); + narrowed.set(ObjRef::new(3, 2), Object::Int(99)); + agrees(&narrowed, 3, &[2]); + let mut freed = doc.clone(); + freed.free(ObjRef::new(3, 0)); + agrees(&freed, 3, &[]); + agrees(&Document::open(contradictory_entry(true)).unwrap(), 3, &[]); + let mut doc = + Document::open(object_file(&[(1, 0, "<< /Type /Catalog /Pages 2 0 R >>"), (2, 0, "<< /Type /Pages /Kids [] /Count 0 >>")])).unwrap(); + let member = doc.add(Object::Dict(Default::default())); + doc.update_dict(ObjRef::new(1, 0), |c| c.set(b"Extra".to_vec(), Object::Ref(member))).unwrap(); + let packed = Document::open(Arc::new(write_full(&doc, &SaveOptions { object_streams: true, ..SaveOptions::default() }).unwrap())).unwrap(); + assert!(matches!(packed.xref_entry(member.num), Some(XrefEntry::InStream { .. })), "the member is in an object stream"); + agrees(&packed, member.num, &[0]); +} diff --git a/crates/cos/tests/it/main.rs b/crates/cos/tests/it/main.rs index 52cf05d6c2..1bd7ee1fd6 100644 --- a/crates/cos/tests/it/main.rs +++ b/crates/cos/tests/it/main.rs @@ -3,6 +3,7 @@ mod bounded_writer; mod corpus; mod encryption; +mod generations; mod proptests; mod shared_streams; mod writer; diff --git a/crates/cos/tests/it/writer.rs b/crates/cos/tests/it/writer.rs index eb03523189..2716b9242e 100644 --- a/crates/cos/tests/it/writer.rs +++ b/crates/cos/tests/it/writer.rs @@ -358,11 +358,32 @@ fn objects_referenced_by_the_encrypt_dictionary_stay_outside_object_streams() { let cf = doc.get(enc_ref).as_dict().unwrap().get(b"CF").cloned().unwrap(); let cf_ref = doc.add(cf); doc.update_dict(enc_ref, |d| d.set(b"CF".to_vec(), Object::Ref(cf_ref))).unwrap(); - let packed = write_full(&doc, &SaveOptions::default()).unwrap(); - let back = Document::open(Arc::new(packed.clone())).expect("reopens"); - assert!(back.repair_log().is_empty(), "{:?}", back.repair_log()); - assert_eq!(title(&back), b"Caf\xe9 (draft)"); - assert_eq!(hayro_pages(&packed), 4); + // Visit this alias from Root before the encryption walk reaches /CF's other alias. + doc.update_dict(doc.root().unwrap(), |d| d.set(b"CFProbe".to_vec(), Object::Ref(cf_ref))).unwrap(); + let classic = write_full(&doc, &SaveOptions { object_streams: false, ..SaveOptions::default() }).unwrap(); + let saved = Document::open(Arc::new(classic.clone())).unwrap(); + let cf_num = saved.get(saved.root().unwrap()).as_dict().unwrap().reference(b"CFProbe").unwrap().num; + for damaged in [false, true] { + let mut bytes = classic.clone(); + if damaged { + // Equal-length edits preserve the xref offsets and its generation-zero row. + for (from, to) in [(format!("\n{cf_num} 0 obj"), format!("\n{cf_num} 1 obj")), (format!("/CF {cf_num} 0 R"), format!("/CF {cf_num} 1 R"))] + { + let start = bytes.windows(from.len()).position(|w| w == from.as_bytes()).unwrap(); + bytes[start..start + from.len()].copy_from_slice(to.as_bytes()); + } + } + let doc = Document::open(Arc::new(bytes)).unwrap(); + let packed = write_full(&doc, &SaveOptions::default()).unwrap(); + let back = Document::open(Arc::new(packed.clone())).expect("reopens"); + assert!(back.repair_log().is_empty(), "{:?}", back.repair_log()); + assert_eq!(title(&back), b"Caf\xe9 (draft)"); + assert_eq!(hayro_pages(&packed), 4); + let probe = back.get(back.root().unwrap()).as_dict().unwrap().reference(b"CFProbe").unwrap(); + let encrypt = back.get(back.trailer().reference(b"Encrypt").unwrap()); + assert_eq!(encrypt.as_dict().unwrap().reference(b"CF"), Some(probe)); + assert!(matches!(back.xref_entry(probe.num), Some(pdfcraft_cos::XrefEntry::InFile { .. })), "crypt filters must stay outside object streams"); + } } #[test] diff --git a/crates/create/src/lib.rs b/crates/create/src/lib.rs index 6346e85120..53a587728f 100644 --- a/crates/create/src/lib.rs +++ b/crates/create/src/lib.rs @@ -92,6 +92,26 @@ pub fn blank(width: f64, height: f64, pages: usize) -> Result)]) -> Result { + if pages.is_empty() || pages.len() > 10_000 { + return Err(CreateError::Invalid("invalid page count".into())); + } + if pages.iter().any(|(w, h, _, _)| !(w.is_finite() && h.is_finite() && (3.0..=MAX_SIDE).contains(w) && (3.0..=MAX_SIDE).contains(h))) { + return Err(CreateError::Invalid("invalid page size".into())); + } + let mut doc = Document::new_empty(); + let mut tree = Pages::of(&doc)?; + for (w, h, resources, content) in pages { + let content = if content.is_empty() { None } else { Some(content.clone()) }; + tree.add(&mut doc, *w, *h, resources.clone(), content); + } + tree.finish(&mut doc)?; + Ok(doc) +} + // ── images ────────────────────────────────────────────────────────────────────────────────── /// An image ready to embed: its XObject dictionary, encoded data, optional soft mask, the ICC diff --git a/crates/crypt/src/lib.rs b/crates/crypt/src/lib.rs index ff92d9af66..e697bc8553 100644 --- a/crates/crypt/src/lib.rs +++ b/crates/crypt/src/lib.rs @@ -82,17 +82,24 @@ pub enum Auth { User, } -/// Permission bits from `/P` (§7.6.4.2, Table 22). Owners have every permission. +/// Permission flags from `/P`, read the way the handler's revision reads them (ISO 32000-1:2008, +/// Table 22). Owners have every permission. #[derive(Clone, Copy, Debug, PartialEq, Eq)] pub struct Permissions { pub bits: i32, pub owner: bool, + /// The `/R` value of the handler that read `bits`. + pub revision: i64, } impl Permissions { fn bit(&self, n: u32) -> bool { self.owner || (self.bits >> (n - 1)) & 1 == 1 } + /// Revision 3 and later read the full flag list of Table 22; revision 2 has its own. + fn is_revision_3_or_later(&self) -> bool { + self.revision >= 3 + } pub fn print(&self) -> bool { self.bit(3) } @@ -107,17 +114,22 @@ impl Permissions { self.bit(6) } pub fn fill_forms(&self) -> bool { - self.bit(9) || self.bit(6) + if self.is_revision_3_or_later() { self.bit(9) || self.bit(6) } else { self.bit(6) } } + /// Revision 2 covers accessibility with the copy flag (bit 5); revision 3 and later with bit 10. pub fn extract_for_accessibility(&self) -> bool { - self.bit(10) + if self.is_revision_3_or_later() { self.bit(10) } else { self.bit(5) } } - /// Insert, rotate or delete pages; create bookmarks and thumbnails. + /// Insert, rotate or delete pages; create bookmarks and thumbnails. Revision 3 and later grant + /// it by bit 11 alone. Table 22 defines bit 11 for those revisions only, so at revision 2 the + /// modify flag (bit 4) covers it. pub fn assemble(&self) -> bool { - self.bit(11) || self.bit(4) + if self.is_revision_3_or_later() { self.bit(11) } else { self.bit(4) } } + /// Revision 3 and later need bit 3 as well as bit 12. Table 22 defines bit 12 for those + /// revisions only: at revision 2 the print flag (bit 3) is all there is. pub fn print_high_quality(&self) -> bool { - self.bit(12) + if self.is_revision_3_or_later() { self.print() && self.bit(12) } else { self.print() } } } @@ -250,7 +262,7 @@ impl SecurityHandler { /// edited `/P` grants nothing extra. Without a readable `/Perms`, `/P` applies as before. pub fn permissions(&self) -> Permissions { let bits = self.perms_p().map_or(self.dict.p, |p| self.dict.p & p); - Permissions { bits, owner: self.auth == Auth::Owner } + Permissions { bits, owner: self.auth == Auth::Owner, revision: self.dict.r } } /// `/P` as `/Perms` holds it: AES-256 ECB of one block (= CBC with a zero IV), marked `adb`. diff --git a/crates/crypt/src/tests.rs b/crates/crypt/src/tests.rs index 6064176d78..09a88ed129 100644 --- a/crates/crypt/src/tests.rs +++ b/crates/crypt/src/tests.rs @@ -240,11 +240,11 @@ fn unsupported_handlers_are_reported() { #[test] fn permission_bits_follow_table_22() { - let p = Permissions { bits: -3904, owner: false }; // 0xFFFFF0C0: only reserved bits set + let p = Permissions { bits: -3904, owner: false, revision: 3 }; // 0xFFFFF0C0: only reserved bits set assert!(!p.print() && !p.modify() && !p.copy() && !p.annotate() && !p.assemble()); - let p = Permissions { bits: -1, owner: false }; + let p = Permissions { bits: -1, owner: false, revision: 3 }; assert!(p.print() && p.modify() && p.copy() && p.annotate() && p.assemble() && p.fill_forms()); - let p = Permissions { bits: 0, owner: true }; + let p = Permissions { bits: 0, owner: true, revision: 3 }; assert!(p.modify(), "owners ignore /P"); } diff --git a/crates/crypt/tests/it/main.rs b/crates/crypt/tests/it/main.rs index 8b39bf1956..e7742bbd75 100644 --- a/crates/crypt/tests/it/main.rs +++ b/crates/crypt/tests/it/main.rs @@ -1,4 +1,5 @@ //! The crate's integration tests, linked into one test binary. +mod permissions_revision; mod proptests; mod qpdf_literals; diff --git a/crates/crypt/tests/it/permissions_revision.rs b/crates/crypt/tests/it/permissions_revision.rs new file mode 100644 index 0000000000..bad709cd51 --- /dev/null +++ b/crates/crypt/tests/it/permissions_revision.rs @@ -0,0 +1,126 @@ +//! The permission accessors of `pdfcraft_crypt::Permissions`, checked against the flag table of +//! ISO 32000-1:2008 (Table 22). Expected values come from that table only. +//! +//! Bit n has the mask `1 << (n - 1)`. Readers use bits 3 to 6 and 9 to 12 and ignore the rest. + +use pdfcraft_crypt::{Algorithm, Auth, NewEncryption, Permissions, SecurityHandler, create}; + +const ID: &[u8] = b"0123456789abcdef"; +const T: bool = true; +const F: bool = false; + +/// The eight accessors in this order: print, high-quality print, modify, copy, accessibility, +/// annotate, fill forms, assemble. +fn cells(p: &Permissions) -> [bool; 8] { + [p.print(), p.print_high_quality(), p.modify(), p.copy(), p.extract_for_accessibility(), p.annotate(), p.fill_forms(), p.assemble()] +} + +/// Creates a file with `flags` at revision `revision`, opens it with the user password, and +/// returns the user's permissions. +fn user_view(alg: Algorithm, revision: i64, flags: i32) -> Permissions { + let made = create( + &NewEncryption { algorithm: alg, user_password: "user", owner_password: "owner", permissions: flags, encrypt_metadata: true, seed: [42; 32] }, + ID, + ) + .expect("the handler is created"); + assert_eq!(made.dict().r, revision, "the algorithm writes revision {revision}"); + let opened = SecurityHandler::open(made.dict().clone(), ID, Some("user")).expect("the user password opens the file"); + assert_eq!(opened.auth(), Auth::User, "the user password gives user access"); + opened.permissions() +} + +/// Revision 3 and later: each row is a flag word and the expected eight cells. +const REVISION_3_ROWS: &[(&str, i32, [bool; 8])] = &[ + ("every meaningful flag set, ignored flags clear", -196, [T, T, T, T, T, T, T, T]), + ("every flag set, ignored flags set too", -1, [T, T, T, T, T, T, T, T]), + ("reserved flags only", -4096, [F, F, F, F, F, F, F, F]), + ("ignored flags set, meaningful flags clear", -3901, [F, F, F, F, F, F, F, F]), + ("modify flag clear", -9, [T, T, F, T, T, T, T, T]), + ("copy flag clear", -17, [T, T, T, F, T, T, T, T]), + ("annotate flag clear", -33, [T, T, T, T, T, F, T, T]), + ("annotate and fill-existing flags clear", -289, [T, T, T, T, T, F, F, T]), + ("annotate flag set, fill-existing flag clear", -257, [T, T, T, T, T, T, T, T]), + ("accessibility flag clear", -513, [T, T, T, T, F, T, T, T]), + ("high-quality flag clear, print flag set", -2049, [T, F, T, T, T, T, T, T]), + ("bits 3 to 12 set, reserved bits clear", 4092, [T, T, T, T, T, T, T, T]), + ("positive flag word, bit 32 clear", 2_147_483_647, [T, T, T, T, T, T, T, T]), + ("revision 2 example read at revision 3", -44, [T, T, F, T, T, F, T, T]), +]; + +#[test] +fn revision_three_rows_hold_through_a_created_and_reopened_file() { + for (name, bits, want) in REVISION_3_ROWS { + assert_eq!(cells(&user_view(Algorithm::Rc4_128, 3, *bits)), *want, "{name}: flag word {bits:#010x}"); + } +} + +#[test] +fn owner_access_gives_every_flag() { + let made = create( + &NewEncryption { + algorithm: Algorithm::Rc4_128, + user_password: "user", + owner_password: "owner", + permissions: 0, + encrypt_metadata: true, + seed: [42; 32], + }, + ID, + ) + .expect("the handler is created"); + assert_eq!(made.auth(), Auth::Owner, "the owner password gives owner access"); + assert_eq!(cells(&made.permissions()), [T; 8]); +} + +#[test] +fn revision_two_print_modify_copy_and_annotate_follow_the_table() { + for (bits, want) in [(-4, [T, T, T, T]), (-44, [T, F, T, F]), (-17, [T, T, F, T])] { + let p = user_view(Algorithm::Rc4_40, 2, bits); + assert_eq!([p.print(), p.modify(), p.copy(), p.annotate()], want, "revision 2, flag word {bits}"); + } +} + +#[test] +fn revision_two_with_every_flag_set_grants_accessibility_and_form_filling() { + let p = user_view(Algorithm::Rc4_40, 2, -4); + assert!(p.extract_for_accessibility(), "bit 5 is set, and at revision 2 it covers accessibility"); + assert!(p.fill_forms(), "bit 6 is set"); +} + +#[test] +fn revision_three_assembly_needs_its_own_flag() { + let p = user_view(Algorithm::Rc4_128, 3, -1025); + assert!(p.modify(), "bit 4 is set"); + assert!(!p.assemble(), "assembly needs bit 11; bit 4 alone does not grant it"); +} + +#[test] +fn revision_three_high_quality_print_needs_the_print_flag() { + let p = user_view(Algorithm::Rc4_128, 3, -5); + assert!(!p.print(), "bit 3 is clear"); + assert!(!p.print_high_quality(), "high-quality print needs bit 3 and bit 12"); +} + +#[test] +fn revision_two_accessibility_follows_the_copy_flag() { + let p = user_view(Algorithm::Rc4_40, 2, -17); + assert!(!p.copy(), "bit 5 is clear"); + assert!(!p.extract_for_accessibility(), "at revision 2 accessibility is bit 5"); +} + +#[test] +fn revision_two_form_filling_ignores_bit_nine() { + let p = user_view(Algorithm::Rc4_40, 2, -44); + assert!(!p.annotate(), "bit 6 is clear"); + assert!(!p.fill_forms(), "at revision 2 form filling is bit 6 alone"); +} + +#[test] +fn revision_two_ignores_the_flags_only_later_revisions_define() { + // Bit 4 is clear and bit 11 set: bit 11 means nothing at revision 2, so assembly stays refused. + let p = user_view(Algorithm::Rc4_40, 2, -44); + assert!(!p.assemble(), "at revision 2 assembly is bit 4 alone"); + // Bit 3 is set and bit 12 clear: printing at revision 2 has one quality. + let p = user_view(Algorithm::Rc4_40, 2, -2049); + assert!(p.print_high_quality(), "at revision 2 high-quality print is bit 3 alone"); +} diff --git a/crates/edit/src/images.rs b/crates/edit/src/images.rs index f4680bcc5c..72656fa1bb 100644 --- a/crates/edit/src/images.rs +++ b/crates/edit/src/images.rs @@ -1691,6 +1691,12 @@ pub(crate) fn selectable_images( .collect()) } +/// Whether an XObject drawn at or after operator `from` of the page's joined operators shows +/// inside `rect` (user space). +pub(crate) fn painted_over(doc: &Document, page: usize, rect: [f64; 4], from: usize) -> Result { + Ok(page_images(doc, page)?.iter().any(|i| i.op >= from && i.visible_rect.and_then(|v| intersection(v, rect)).is_some())) +} + fn transform_image( doc: &mut Document, page: &pdfcraft_model::Page, diff --git a/crates/edit/src/lib.rs b/crates/edit/src/lib.rs index 44a6730acc..2a0add8d12 100644 --- a/crates/edit/src/lib.rs +++ b/crates/edit/src/lib.rs @@ -77,8 +77,8 @@ use marks::{matrix_number, n}; pub(crate) use place::undrawable; #[cfg(test)] use place::{ISOLATION_MAX_DEPTH, OLD_WRAP_CLOSE, WRAP_CLOSE, WRAP_OPEN, is_wrap_close, page_fonts, stream_bytes, stream_tag, tagged, wrapper_role}; -use place::{check, check_on_top, contents, page_list, place_tagged}; -pub use place::{stamp, undrawable_message}; +use place::{Reach, check, check_on_top, contents, draw_on_top, lift_on_top, page_list, place_tagged, revealed_under_paint}; +pub use place::{add_type3_fonts, stamp, stamp_with_fonts, undrawable_message}; #[cfg(test)] mod marked_isolation_tests; diff --git a/crates/edit/src/place.rs b/crates/edit/src/place.rs index b535d60563..6460aab066 100644 --- a/crates/edit/src/place.rs +++ b/crates/edit/src/place.rs @@ -371,6 +371,76 @@ pub(crate) fn place_tagged(doc: &mut Document, page: &pdfcraft_model::Page, tag: Ok(mark_ref) } +/// Draw `ops`, written for the space whose CTM is `ctm`, on top of everything on `page`, tagged +/// `Edited`. A rewritten invisible paragraph goes here when [`revealed_under_paint`]: an OCR +/// layer written before the scan image it lies under would keep the revealed text hidden (#3973). +pub(crate) fn draw_on_top(doc: &mut Document, page: &pdfcraft_model::Page, ctm: [f64; 6], ops: &[pdfcraft_content::Op]) -> Result<(), EditError> { + if ops.is_empty() { + return Ok(()); + } + use pdfcraft_content::Op; + let mut all = vec![Op::new("q", vec![]), Op::new("cm", ctm.iter().map(|v| pdfcraft_content::num(*v)).collect())]; + all.extend_from_slice(ops); + all.push(Op::new("Q", vec![])); + place_tagged(doc, page, "Edited", pdfcraft_content::serialize_ops(&all), false).map(drop) +} + +/// Where a rewritten paragraph can be drawn, in user space ([`revealed_under_paint`]). +pub(crate) struct Reach { + /// The paragraph's old box, where its cover goes. + pub rect: [f64; 4], + /// The move asked for (`BlockStyle::offset`). + pub offset: [f64; 2], + /// The widest line the rewrite may draw, how many lines it has, and how far apart they are. + pub width: f64, + pub rows: usize, + pub pitch: f64, + pub upright: bool, +} + +impl Reach { + /// The old box and the moved one, grown to `width` either way (alignment may grow a line left + /// or right), down to `rows` lines and up one `pitch` above its top (a larger size on the first line, however many lines the old box had); turned text grows by the lines' diagonal every way. + /// Unbounded when a number is not finite. + pub(crate) fn bounds(&self) -> [f64; 4] { + const UNBOUNDED: [f64; 4] = [f64::NEG_INFINITY, f64::NEG_INFINITY, f64::INFINITY, f64::INFINITY]; + let ([dx, dy], r) = (self.offset, self.rect); + if !r.iter().chain(&self.offset).chain([&self.width, &self.pitch]).all(|v| v.is_finite()) { + return UNBOUNDED; + } + let moved = [r[0] + dx, r[1] + dy, r[2] + dx, r[3] + dy]; + let rows = f64::from(u32::try_from(self.rows).unwrap_or(u32::MAX)); + let (width, height) = (self.width.abs(), self.pitch.abs() * rows); + let grown = if self.upright { + [moved[0].min(moved[2] - width), moved[1].min(moved[3] - height), moved[2].max(moved[0] + width), moved[3] + self.pitch.abs()] + } else { + let d = width.hypot(height); + [moved[0] - d, moved[1] - d, moved[2] + d, moved[3] + d] + }; + let b = [r[0].min(grown[0]), r[1].min(grown[1]), r[2].max(grown[2]), r[3].max(grown[3])]; + if b.iter().all(|v| v.is_finite()) { b } else { UNBOUNDED } + } +} + +/// Whether a rewritten paragraph that was `invisible` goes on top of page `page`: something after +/// operator `at` of the page's joined `ops` paints over `reach` (user space, [`Reach::bounds`]), as the scan image +/// does over an OCR layer written before it (#3973). XObjects count where they are drawn; an inline +/// image or a shading after it counts anywhere, and so does any XObject when the page's images +/// cannot be listed. +pub(crate) fn revealed_under_paint(doc: &Document, page: usize, invisible: bool, reach: [f64; 4], ops: &[pdfcraft_content::Op], at: usize) -> bool { + let later = ops.get(at..).unwrap_or_default(); + if !invisible || later.iter().any(|op| op.is("BI") || op.is("sh")) { + return invisible; + } + later.iter().any(|op| op.is("Do")) && crate::images::painted_over(doc, page, reach, at).unwrap_or(true) +} + +/// The operators to draw on top of the page ([`draw_on_top`]): all of `paragraph` when `over`, +/// none otherwise. The page was checked to take them ([`check_on_top`]) before the edit began. +pub(crate) fn lift_on_top(over: bool, paragraph: &mut Vec) -> Vec { + if over { std::mem::take(paragraph) } else { Vec::new() } +} + fn too_deep_on_top() -> EditError { EditError::Invalid("the page's marked content is nested too deeply to add anything on top of it".into()) } @@ -408,14 +478,79 @@ pub(crate) fn place_inside_wrapper(doc: &mut Document, page: &pdfcraft_model::Pa /// it away); the original content is wrapped in q/Q first. Used for applied redaction boxes. /// Returns the new content stream. pub fn stamp(doc: &mut Document, page: usize, tag: &str, content: Vec) -> Result { + stamp_with_fonts(doc, page, tag, &mut pdfcraft_fonts::TextPlan::default(), |_| content) +} + +/// Draw the content `make` builds on top of a page, with standard Helvetica as `/PCHelv` and +/// the plan's Type3 fonts added to the page's resources first (so the content can reference +/// them by the names the plan carries). The original content is wrapped in q/Q first; the +/// stream is tagged `tag` (not a mark kind, so Remove never takes it away). This is how the +/// OCR text layers get their Unicode faces (see [`add_type3_fonts`]). Returns the new content +/// stream. +pub fn stamp_with_fonts( + doc: &mut Document, + page: usize, + tag: &str, + plan: &mut pdfcraft_fonts::TextPlan, + make: impl FnOnce(&pdfcraft_fonts::TextPlan) -> Vec, +) -> Result { let all = page_list(doc); check(&[page], all.len())?; check_on_top(doc, &all[page])?; + add_type3_fonts(doc, page, plan)?; + let content = make(plan); + let all = page_list(doc); add_resources(doc, &all[page], None, Some(&content))?; let p = page_list(doc).swap_remove(page); place_tagged(doc, &p, tag, content, false) } +/// Add the plan's Type3 fonts to a page's (0-based) resources, building each font once and +/// sharing the object with every page that uses it. Fills in each planned font's resource name +/// (`PCOcr`, `PCOcr1`, … — a name free on the page being registered: a font built for an +/// earlier page keeps its name only where that name is free or already that font) and object; +/// the content generators read the names from the plan, so this must run before they are +/// called for each page. +pub fn add_type3_fonts(doc: &mut Document, page: usize, plan: &mut pdfcraft_fonts::TextPlan) -> Result<(), EditError> { + if plan.fonts.is_empty() { + return Ok(()); + } + let all = page_list(doc); + check(&[page], all.len())?; + let p = &all[page]; + let mut res = p.dict.get(b"Resources").map(|r| doc.resolve(r)).and_then(|r| r.as_dict().cloned()).unwrap_or_default(); + let mut fonts = res.get(b"Font").map(|f| doc.resolve(f)).and_then(|f| f.as_dict().cloned()).unwrap_or_default(); + for i in 0..plan.fonts.len() { + let object = match plan.fonts[i].object { + Some(r) => { + // Built for an earlier page under a name that was free THERE. On this page the + // name may already be another font (the page's own, or an earlier OCR layer's): + // then this page gets a free name of its own, and the plan carries it for the + // content written next. + let taken = fonts.get(plan.fonts[i].name.as_bytes()).is_some_and(|o| o.as_ref() != Some(r)); + if taken { + plan.fonts[i].name = super::text::unique_font_name(&fonts, "PCOcr"); + } + r + } + None => { + let glyphs = plan.fonts[i].glyphs.iter().map(|(ch, g)| (*ch, super::text::glyph_path(g), g.width)).collect::>(); + let dict = super::text::type3_font_dict(doc, &glyphs)?; + let r = doc.add(Object::Dict(dict)); + plan.fonts[i].name = super::text::unique_font_name(&fonts, "PCOcr"); + plan.fonts[i].object = Some(r); + r + } + }; + fonts.set(plan.fonts[i].name.as_bytes().to_vec(), Object::Ref(object)); + } + res.set(b"Font".to_vec(), Object::Dict(fonts)); + // The page gets its own (possibly copied) resources; shared dictionaries stay untouched. + let obj = p.obj; + doc.update_dict(obj, |d| d.set(b"Resources".to_vec(), Object::Dict(res)))?; + Ok(()) +} + pub(crate) fn begin(kind: MarkKind, subtype: &str, matrix: [f64; 6]) -> String { format!( "q\n/Artifact <> BDC\n{} {} {} {} {} {} cm\n", diff --git a/crates/edit/src/tests.rs b/crates/edit/src/tests.rs index 53fb786bd1..61ef7c4bcc 100644 --- a/crates/edit/src/tests.rs +++ b/crates/edit/src/tests.rs @@ -3756,7 +3756,10 @@ fn page_content_bytes(doc: &Document, page: usize) -> Vec { } } -/// A page drawing image XObject /Im0 at 100,100 size 200 × 100 (pixels 4 × 2). +/// A page drawing image XObject /Im0 at 100,100 size 200 × 100 (pixels 4 × 2), on a real optional +/// content layer: the catalog's `/OCProperties` declares object 7 as the only OCG and the page +/// resources' `/Properties` maps `/L0` to it, so `/OC /L0 BDC` marks a valid optional-content +/// member rather than a dangling name. fn image_page() -> Document { image_page_with("q 200 0 0 100 100 100 cm /Im0 Do Q BT /F1 12 Tf 72 700 Td (Caption) Tj ET") } @@ -3776,12 +3779,13 @@ fn remove_crop(doc: &mut Document, page: usize, index: usize) -> Result<(), Edit fn image_page_with(content: &str) -> Document { let objs: Vec = vec![ - "<< /Type /Catalog /Pages 2 0 R >>".into(), + "<< /Type /Catalog /Pages 2 0 R /OCProperties << /OCGs [7 0 R] /D << /Order [7 0 R] >> >> >>".into(), "<< /Type /Pages /Kids [3 0 R] /Count 1 >>".into(), - "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 600 800] /Contents 4 0 R /Resources << /Font << /F1 5 0 R >> /XObject << /Im0 6 0 R >> /ExtGState << /GS0 << /Type /ExtGState /ca 0.5 /CA 0.5 >> /GSalpha << /Type /ExtGState /ca 0.6 >> /GSblend << /Type /ExtGState /BM /Multiply >> /GS_translucent << /Type /ExtGState /ca 0.5 >> >> >> >>".into(), + "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 600 800] /Contents 4 0 R /Resources << /Font << /F1 5 0 R >> /XObject << /Im0 6 0 R >> /ExtGState << /GS0 << /Type /ExtGState /ca 0.5 /CA 0.5 >> /GSalpha << /Type /ExtGState /ca 0.6 >> /GSblend << /Type /ExtGState /BM /Multiply >> /GS_translucent << /Type /ExtGState /ca 0.5 >> >> /Properties << /L0 7 0 R >> >> >>".into(), format!("<< /Length {} >>\nstream\n{content}\nendstream", content.len()), "<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica >>".into(), "<< /Type /XObject /Subtype /Image /Width 4 /Height 2 /ColorSpace /DeviceGray /BitsPerComponent 8 /Length 8 >>\nstream\n\u{0}\u{0}\u{0}\u{0}\u{0}\u{0}\u{0}\u{0}\nendstream".into(), + "<< /Type /OCG /Name (Layer 0) >>".into(), ]; let mut out = b"%PDF-1.7\n".to_vec(); let mut offs = Vec::new(); @@ -3995,6 +3999,30 @@ fn repeated_group_moves_of_a_cropped_image_change_one_clip_transform() { assert_eq!(caption, ["Caption"]); } +#[test] +fn moving_an_image_inside_marked_content_leaves_every_sequence_intact() { + // `translate_images` changes the image's own `cm`; its `Do` stays where it was, so a moved + // image stays in the sequence that tags it (an image on a layer sits in `/OC … BDC`). + for content in [ + "/P <> BDC q 100 0 0 100 50 50 cm /Im0 Do Q EMC /P <> BDC BT /F1 12 Tf 72 700 Td (Text) Tj ET EMC", + "/OC /L0 BDC q 100 0 0 100 50 50 cm /Im0 Do Q EMC /P <> BDC BT /F1 12 Tf 72 700 Td (Text) Tj ET EMC", + ] { + let mut doc = image_page_with(content); + let before = images::page_images(&doc, 0).unwrap()[0].rect; + images::translate_images(&mut doc, 0, &[0], [10.0, 0.0]).unwrap(); + let after = images::page_images(&doc, 0).unwrap()[0].rect; + assert!(close(after, [before[0] + 10.0, before[1], before[2] + 10.0, before[3]]), "{content}: {after:?}"); + let ops = pdfcraft_content::parse(&page_content_bytes(&doc, 0)).ops; + let count = |name: &str| ops.iter().filter(|op| op.is(name)).count(); + assert_eq!((count("BDC"), count("EMC")), (2, 2), "{content}: both sequences remain"); + let first = ops.iter().position(|op| op.is("BDC")).unwrap(); + let end = ops.iter().position(|op| op.is("EMC")).unwrap(); + let image = ops.iter().position(|op| op.is("Do")).unwrap(); + assert!(first < image && image < end, "{content}: the image is still inside its own sequence"); + assert!(!ops[end..].iter().any(|op| op.is("Do")), "{content}: the other sequence gained no image"); + } +} + #[test] fn page_images_move_turn_replace_and_delete() { let mut doc = image_page(); @@ -6455,6 +6483,126 @@ fn group_selection_refuses_incomplete_content_and_resource_budget_overruns() { assert!(editable_objects(&doc, 0).is_err(), "an undecodable stream is refused"); } +/// Invisible text moves on top only when something painted after it covers it: an XObject drawn +/// over its box (an image, or a form holding one), or any inline image or shading. An image +/// clipped away entirely, or drawn elsewhere, does not move it. +#[test] +fn revealed_text_moves_on_top_only_when_painted_over() { + let on_top_of = |doc: Document| { + let block = text::text_blocks(&doc, 0).unwrap().remove(0); + let ops = pdfcraft_content::parse(&page_content_bytes(&doc, 0)).ops; + let at = ops.iter().position(|o| o.is("BT")).unwrap(); + revealed_under_paint(&doc, 0, block.invisible, block.rect, &ops, at) + }; + let on_top = |content: &str| on_top_of(image_page_with(content)); + let scan = "q 200 0 0 100 100 100 cm /Im0 Do Q"; + assert!(on_top(&format!("BT /F1 12 Tf 3 Tr 110 150 Td (Under) Tj ET {scan}")), "the scan covers it"); + assert!(!on_top(&format!("BT /F1 12 Tf 3 Tr 400 600 Td (Away) Tj ET {scan}")), "an image elsewhere"); + assert!(!on_top(&format!("{scan} BT /F1 12 Tf 3 Tr 110 150 Td (Over) Tj ET")), "the scan is drawn first"); + assert!(!on_top(&format!("BT /F1 12 Tf 0 Tr 110 150 Td (Shown) Tj ET {scan}")), "visible text stays where it is"); + assert!(!on_top("BT /F1 12 Tf 3 Tr 110 150 Td (Path) Tj ET 0 0 600 800 re f"), "a filled path does not move it"); + assert!(on_top("BT /F1 12 Tf 3 Tr 400 600 Td (Shade) Tj ET /Sh sh"), "a shading after it"); + assert!(on_top("BT /F1 12 Tf 3 Tr 400 600 Td (Inline) Tj ET BI /W 1 /H 1 /CS /G /BPC 8 ID x EI"), "an inline image after it"); + let under = "BT /F1 12 Tf 3 Tr 110 150 Td (Under) Tj ET"; + let elsewhere = "q 50 0 0 50 400 600 cm /Im0 Do Q"; + assert!(!on_top(&format!("{under} q 0 0 10 10 re W n {scan} Q")), "the scan is clipped away entirely"); + assert!(on_top(&format!("{under} {elsewhere} {scan}")), "two images, the second over it"); + assert!(on_top(&format!("{under} {scan} {elsewhere}")), "two images, the first over it"); + assert!(!on_top(&format!("{under} {elsewhere} {elsewhere}")), "two images, neither over it"); + let in_form = |content: &str| { + let page = "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 600 800] /Contents 4 0 R /Resources << /Font << /F1 5 0 R >> /XObject << /Fm0 7 0 R >> >> >>"; + let image = stream("/Type /XObject /Subtype /Image /Width 1 /Height 1 /ColorSpace /DeviceGray /BitsPerComponent 8", "A"); + let form = stream("/Type /XObject /Subtype /Form /BBox [0 0 1 1] /Resources << /XObject << /Im0 6 0 R >> >>", "/Im0 Do"); + let objs = [ + "<< /Type /Catalog /Pages 2 0 R >>".to_string(), + "<< /Type /Pages /Kids [3 0 R] /Count 1 >>".into(), + page.into(), + stream("", content), + "<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica >>".into(), + image, + form, + ]; + on_top_of(build(&objs.iter().map(String::as_str).collect::>())) + }; + assert!(in_form(&format!("{under} q 200 0 0 100 100 100 cm /Fm0 Do Q")), "a form holding the scan, over it"); + assert!(!in_form(&format!("{under} q 50 0 0 50 400 600 cm /Fm0 Do Q")), "a form holding an image elsewhere"); +} + +/// Revealed OCR text is checked where the rewrite lands, not only where it was: an image drawn +/// after it over just the moved or grown text still lifts the text on top, on both rewrite paths. +#[test] +fn an_ocr_rewrite_goes_on_top_of_paint_over_where_it_lands() { + let scan = "q 200 0 0 100 100 100 cm /Im0 Do Q"; + let moved = text::BlockStyle { offset: Some([-290.0, -450.0]), ..Default::default() }; + let sized = text::BlockStyle { size: Some(24.0), ..Default::default() }; + let spaced = text::BlockStyle { line_spacing: Some(3.0), ..Default::default() }; + let range = text::RangeStyle { + range: 0..2, + family: None, + resource: None, + size: Some(120.0), + color: None, + underline: None, + char_spacing: None, + scale: None, + }; + let enlarged = text::BlockStyle { ranges: vec![range], ..Default::default() }; + let above = "q 200 0 0 60 100 200 cm /Im0 Do Q"; + let cases = [ + ("BT /F1 12 Tf 3 Tr 400 600 Td (Away) Tj ET", "Moved", &moved), + ("BT /F1 12 Tf 3 Tr 400 600 Td (Away) Tj ET", "Moved\nTwice", &moved), + ("BT /F1 12 Tf 3 Tr 40 150 Td (Hi) Tj ET", "Hello there, this line now runs well past the scan's left edge", &text::BlockStyle::default()), + ("BT /F1 12 Tf 3 Tr 40 150 Td (Hi) Tj ET", "Hello there,\nthis line now runs well past the scan's left edge", &text::BlockStyle::default()), + ("BT /F1 12 Tf 3 Tr 110 230 Td (Above) Tj ET", "One\nTwo\nThree\nFour\nFive\nSix", &text::BlockStyle::default()), + ]; + let low = "q 200 0 0 30 100 300 cm /Im0 Do Q"; + let cases = cases.into_iter().map(|(c, e, s)| (c, e, s, scan)).chain([ + ("BT /F1 12 Tf 3 Tr 110 500 Td (Above) Tj ET", "One\nTwo\nThree\nFour\nFive\nSix", &sized, low), + ("BT /F1 12 Tf 3 Tr 110 500 Td (Above) Tj ET", "One\nTwo\nThree\nFour\nFive\nSix", &spaced, low), + ("BT /F1 12 Tf 3 Tr 110 150 Td (Hi there) Tj ET", "Hi there", &enlarged, above), + ]); + for (content, edit, style, scan) in cases { + let mut doc = image_page_with(&format!("{content} {scan}")); + let before = text::text_blocks(&doc, 0).unwrap().remove(0); + let ops = pdfcraft_content::parse(&page_content_bytes(&doc, 0)).ops; + let at = ops.iter().position(|o| o.is("BT")).unwrap(); + assert!(!revealed_under_paint(&doc, 0, true, before.rect, &ops, at), "{edit:?}: the old box alone is clear of the scan"); + text::rewrite_block(&mut doc, 0, 0, Some(edit), style).unwrap(); + let ops = pdfcraft_content::parse(&page_content_bytes(&doc, 0)).ops; + let scan_at = ops.iter().position(|o| o.is("Do")).unwrap(); + let shows = ["Tj", "TJ", "'", "\""]; + assert!(ops.iter().skip(scan_at).any(|o| shows.iter().any(|s| o.is(s))), "{edit:?}: the rewrite is drawn over the scan"); + } +} + +/// [`crate::Reach`] grows the old box to the move, the width either way and the lines below. +#[test] +fn a_rewrites_reach_covers_its_move_width_and_lines() { + let reach = |offset, width, rows, upright| crate::Reach { rect: [10.0, 90.0, 30.0, 100.0], offset, width, rows, pitch: 10.0, upright }.bounds(); + assert_eq!(reach([0.0, 0.0], 20.0, 1, true), [10.0, 90.0, 30.0, 110.0]); + assert_eq!(reach([5.0, -50.0], 20.0, 1, true), [10.0, 40.0, 35.0, 100.0]); + assert_eq!(reach([0.0, 0.0], 50.0, 3, true), [-20.0, 70.0, 60.0, 110.0]); + assert_eq!(reach([0.0, 0.0], 30.0, 4, false), [-40.0, 40.0, 80.0, 150.0]); + assert_eq!(reach([f64::NAN, 0.0], 20.0, 1, true)[0], f64::NEG_INFINITY); + let larger = crate::Reach { rect: [10.0, 90.0, 30.0, 100.0], offset: [0.0, 0.0], width: 20.0, rows: 1, pitch: 24.0, upright: true }; + assert_eq!(larger.bounds(), [10.0, 76.0, 30.0, 124.0], "a larger size grows the line up as well as down"); +} + +/// Revealed OCR text may have to go on top of the page; a page whose marked content nests too deep +/// for that is refused before either rewrite path changes anything. +#[test] +fn an_ocr_rewrite_refused_on_top_leaves_the_document_as_it_was() { + let content = format!("{} BT /F1 20 Tf 3 Tr 42 44 Td (John) Tj ET /Im Do {}", "/X BMC ".repeat(10_001), "EMC ".repeat(10_001)); + let mut doc = text_page(&content); + assert!(text::text_blocks(&doc, 0).unwrap()[0].invisible); + let before = write_full(&doc, &SaveOptions::default()).unwrap(); + for edit in ["Jane", "Jane\nDoe"] { + let error = text::rewrite_block(&mut doc, 0, 0, Some(edit), &text::BlockStyle::default()).unwrap_err(); + assert!(error.to_string().contains("nested too deeply"), "{edit:?}: {error}"); + assert!(write_full(&doc, &SaveOptions::default()).unwrap() == before, "{edit:?}: a refused rewrite changes nothing"); + } +} + #[test] fn group_move_budget_refusals_keep_the_original_page_selectable() { let text = text_page(&format!("{} BT /F1 12 Tf 40 400 Td (Target) Tj ET", "n ".repeat(99_995))); @@ -10351,6 +10499,86 @@ fn styled_existing_justification_is_refused_instead_of_silently_left_aligning() assert_eq!(page_content_bytes(&reopen(&doc), 0), before); } +/// The OCR stamp path registers the plan's Type3 faces (a free `PCOcr…` name per stamp, so a +/// page that is OCR'd twice keeps both layers' fonts), and the stamped codes extract back +/// through the font's ToUnicode CMap. +#[test] +fn stamp_with_fonts_registers_the_planned_type3_faces() { + if without_craft_fonts("stamp_with_fonts_registers_the_planned_type3_faces") { + return; + } + let mut doc = text_page("BT /F2 12 Tf 72 700 Td (ab) Tj ET"); + let mut plan = pdfcraft_fonts::TextPlan::for_words(["日本"]); + assert_eq!(plan.missing(), 0); + stamp_with_fonts(&mut doc, 0, "OCR", &mut plan, |p| { + let codes: Vec = p.encoding(0).iter().map(|&(_, c)| c).collect(); + let mut c = format!("BT 3 Tr /PCHelv 1 Tf\n/{} 1 Tf 100 700 Td ", p.fonts[0].name).into_bytes(); + c.extend_from_slice(&pdfcraft_fonts::literal(&codes)); + c.extend_from_slice(b" Tj ET"); + c + }) + .unwrap(); + assert!(plan.fonts[0].name.starts_with("PCOcr"), "{:?}", plan.fonts[0].name); + // The registered font carries a ToUnicode CMap; the engine's own reader decodes the run. + let reopened = Document::open(Arc::new(write_incremental(&doc, &SaveOptions::default()).unwrap())).unwrap(); + let lines = text::text_lines(&reopened, 0).unwrap(); + assert!(lines.iter().any(|l| l.text.contains("日本")), "{lines:?}"); + // A second stamp on the same page takes the next free name; the first layer's font stands. + let mut second = pdfcraft_fonts::TextPlan::for_words(["語"]); + stamp_with_fonts(&mut doc, 0, "OCR", &mut second, |p| format!("BT 3 Tr /{} 1 Tf 100 600 Td (\u{1}) Tj ET", p.fonts[0].name).into_bytes()) + .unwrap(); + assert_eq!(second.fonts[0].name, format!("{}1", plan.fonts[0].name)); +} + +/// A Type3 face built for page 1 is reused on page 2, but page 2 already has a font under the +/// name page 1 gave it: page 2's own font keeps its name, and the shared face is registered on +/// page 2 under a free one that the plan carries for page 2's content. (The plan is written out +/// by hand, so this runs without the craft-fonts build input.) +#[test] +fn a_reused_type3_face_never_replaces_a_later_page_s_font_of_the_same_name() { + let mut doc = build(&[ + "<< /Type /Catalog /Pages 2 0 R >>", + "<< /Type /Pages /Kids [3 0 R 4 0 R] /Count 2 >>", + "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 600 800] /Contents 5 0 R >>", + "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 600 800] /Contents 6 0 R /Resources << /Font << /PCOcr 7 0 R >> >> >>", + &stream("", ""), + &stream("", "BT /PCOcr 12 Tf 72 700 Td (own) Tj ET"), + "<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica /Encoding /WinAnsiEncoding >>", + ]); + let glyph = pdfcraft_fonts::GlyphOutline { contours: vec![vec![[0.1, 0.0], [0.9, 0.0], [0.5, 0.8]]], width: 1.0, bbox: [0.1, 0.0, 0.9, 0.8] }; + let mut plan = pdfcraft_fonts::TextPlan::default(); + plan.fonts.push(pdfcraft_fonts::PlannedType3 { family: "Test", glyphs: vec![('日', glyph)], name: String::new(), object: None }); + let draw = |p: &pdfcraft_fonts::TextPlan| format!("BT 3 Tr /{} 1 Tf 100 600 Td (\u{1}) Tj ET", p.fonts[0].name).into_bytes(); + stamp_with_fonts(&mut doc, 0, "OCR", &mut plan, draw).unwrap(); + assert_eq!(plan.fonts[0].name, "PCOcr", "free on page 1"); + let face = plan.fonts[0].object.expect("built once"); + stamp_with_fonts(&mut doc, 1, "OCR", &mut plan, draw).unwrap(); + assert_ne!(plan.fonts[0].name, "PCOcr", "page 2 already has a PCOcr"); + let page2 = &pdfcraft_model::pages(&doc)[1]; + let res = doc.resolve(page2.dict.get(b"Resources").unwrap()); + let fonts = doc.resolve(res.as_dict().unwrap().get(b"Font").unwrap()); + let fonts = fonts.as_dict().unwrap(); + assert_eq!(fonts.get(b"PCOcr").and_then(|o| o.as_ref()), Some(pdfcraft_cos::ObjRef::new(7, 0)), "page 2's own font keeps its name"); + assert_eq!(fonts.get(plan.fonts[0].name.as_bytes()).and_then(|o| o.as_ref()), Some(face), "the shared face, under a free name"); + assert_eq!(plan.fonts[0].object, Some(face), "still built once"); +} + +#[test] +fn planned_type3_charproc_has_valid_metrics_and_encloses_its_ink() { + let glyph = pdfcraft_fonts::GlyphOutline { + contours: vec![vec![[-0.1001, -0.2001], [0.9001, 1.1001]]], + width: 0.8, + bbox: [-0.1001, -0.2001, 0.9001, 1.1001], + }; + let parsed = pdfcraft_content::parse(&text::glyph_path(&glyph)); + assert_eq!(parsed.skipped, 0); + let metrics = &parsed.ops[0]; + assert!(metrics.is("d1")); + assert_eq!(metrics.operands.len(), 6); + assert_eq!((metrics.num(0), metrics.num(1)), (Some(800.0), Some(0.0))); + assert_eq!((metrics.num(2), metrics.num(3), metrics.num(4), metrics.num(5)), (Some(-101.0), Some(-201.0), Some(901.0), Some(1101.0))); +} + #[test] fn image_arrange_refuses_crossed_scope_boundaries_without_changes() { for content in ["/P <> BDC q 10 0 0 10 0 0 cm /Im0 Do EMC Q /Im0 Do", "q /P <> BDC /Im0 Do Q EMC /Im0 Do"] { diff --git a/crates/edit/src/text.rs b/crates/edit/src/text.rs index 987931f5af..2394a3d01d 100644 --- a/crates/edit/src/text.rs +++ b/crates/edit/src/text.rs @@ -52,6 +52,7 @@ use fallback::{ pub(crate) use fallback::{SOURCE_SPACE_CALLS, take_source_character_measurements}; #[cfg(test)] use fallback::{add_type3_font, is_type3_fallback}; +pub(crate) use fallback::{glyph_path, type3_font_dict, unique_font_name}; use interp::{Origin, Paint, union}; use lines::page_dict; pub(crate) use lines::{FormBudget, content_streams, content_streams_within, dict_in_place, form_call, resource_dict, rewritten_contents, splice}; diff --git a/crates/edit/src/text/blocks.rs b/crates/edit/src/text/blocks.rs index 9ddf4db4b1..48fe956f4f 100644 --- a/crates/edit/src/text/blocks.rs +++ b/crates/edit/src/text/blocks.rs @@ -31,7 +31,8 @@ pub struct TextBlock { /// Byte ranges in `editor_text` and their source line indexes. pub spans: Vec<(std::ops::Range, usize)>, /// Every line is invisible text (rendering mode 3): an OCR layer over a scanned image, whose - /// pixels show the words. A rewrite draws the new text visibly (see [`BlockStyle::cover`]). + /// pixels show the words. A rewrite draws the new text visibly, on top of the page when + /// the scan is drawn after it (see [`BlockStyle::cover`]). pub invisible: bool, } diff --git a/crates/edit/src/text/fallback.rs b/crates/edit/src/text/fallback.rs index 4b93b75afe..26203f7586 100644 --- a/crates/edit/src/text/fallback.rs +++ b/crates/edit/src/text/fallback.rs @@ -494,6 +494,92 @@ pub(super) fn add_type3_font(doc: &mut Document, fonts_res: &mut Dict, face: &Cr name } +/// A glyph's CharProc body at 1000 units per em, with a six-operand `d1` and its actual ink box. +pub(crate) fn glyph_path(glyph: &pdfcraft_fonts::GlyphOutline) -> Vec { + let scale = 1000.0; + let b = if glyph.contours.is_empty() { [0.0; 4] } else { glyph.bbox }; + let mut out = format!( + "{} 0 {} {} {} {} d1\n", + pdf_num(glyph.width * scale), + pdf_num((b[0] * scale).floor()), + pdf_num((b[1] * scale).floor()), + pdf_num((b[2] * scale).ceil()), + pdf_num((b[3] * scale).ceil()) + ) + .into_bytes(); + for contour in &glyph.contours { + let Some(first) = contour.first() else { continue }; + out.extend_from_slice(format!("{} {} m\n", pdf_num(first[0] * scale), pdf_num(first[1] * scale)).as_bytes()); + for p in contour.iter().skip(1) { + out.extend_from_slice(format!("{} {} l\n", pdf_num(p[0] * scale), pdf_num(p[1] * scale)).as_bytes()); + } + out.extend_from_slice(b"h\n"); + } + out.extend_from_slice(b"f\n"); + out +} + +/// A resource name free in `fonts_res`: `base`, then `base1`, `base2`, … (a page that was +/// OCR'd twice keeps both layers' fonts distinct). +pub(crate) fn unique_font_name(fonts_res: &Dict, base: &str) -> String { + let mut name = String::from(base); + let mut suffix = 0usize; + while fonts_res.contains(name.as_bytes()) { + suffix = suffix.saturating_add(1); + name = format!("{base}{suffix}"); + } + name +} + +/// The Type3 font dictionary for `(character, CharProc body, advance in em)` glyphs, codes from +/// 1: one CharProc per code, `/Differences` glyph names, real `/Widths`, and a `/ToUnicode` CMap +/// so the text extracts back (this is what makes the output Unicode-correct, not just visible). +pub(crate) fn type3_font_dict(doc: &mut Document, glyphs: &[(char, Vec, f64)]) -> Result { + let mut charprocs = Dict::new(); + let mut differences = vec![Object::Int(1)]; + let mut widths = Vec::with_capacity(glyphs.len()); + let mut cmap = String::from( + "/CIDInit /ProcSet findresource begin\n12 dict begin\nbegincmap\n/CMapType 2 def\n1 begincodespacerange\n<01> \nendcodespacerange\n", + ); + cmap.push_str(&format!("{} beginbfchar\n", glyphs.len())); + let mut codes = Vec::with_capacity(glyphs.len()); + for (i, (ch, path, width)) in glyphs.iter().enumerate() { + let code = u8::try_from(i + 1).map_err(|_| EditError::Invalid("too many unique characters for one font".into()))?; + let glyph_name = format!("g{code:02X}"); + let mut pd = Dict::new(); + pd.set(b"Length".to_vec(), path.len() as i64); + let proc_ref = doc.add(Object::Stream(Stream::from_raw(pd, path.clone()))); + charprocs.set(glyph_name.as_bytes().to_vec(), Object::Ref(proc_ref)); + differences.push(Object::name(&glyph_name)); + widths.push(Object::Real((width * 1000.0).round())); + cmap.push_str(&format!("<{code:02X}> <{}>\n", unicode_hex(*ch))); + codes.push((*ch, code, *width)); + } + cmap.push_str("endbfchar\nendcmap\nCMapName currentdict /CMap defineresource pop\nend\nend\n"); + let mut cmap_dict = Dict::new(); + cmap_dict.set(b"Length".to_vec(), cmap.len() as i64); + let cmap_ref = doc.add(Object::Stream(Stream::from_raw(cmap_dict, cmap.into_bytes()))); + let mut encoding = Dict::new(); + encoding.set(b"Type".to_vec(), Object::name("Encoding")); + encoding.set(b"Differences".to_vec(), Object::Array(differences)); + let mut font = Dict::new(); + font.set(b"Type".to_vec(), Object::name("Font")); + font.set(b"Subtype".to_vec(), Object::name("Type3")); + font.set(b"Name".to_vec(), Object::name("PCFallback")); + font.set(b"FontBBox".to_vec(), Object::Array(vec![Object::Int(0), Object::Int(-300), Object::Int(1000), Object::Int(1000)])); + font.set( + b"FontMatrix".to_vec(), + Object::Array(vec![Object::Real(0.001), Object::Int(0), Object::Int(0), Object::Real(0.001), Object::Int(0), Object::Int(0)]), + ); + font.set(b"FirstChar".to_vec(), Object::Int(1)); + font.set(b"LastChar".to_vec(), Object::Int(codes.len() as i64)); + font.set(b"Widths".to_vec(), Object::Array(widths)); + font.set(b"Encoding".to_vec(), Object::Dict(encoding)); + font.set(b"CharProcs".to_vec(), Object::Dict(charprocs)); + font.set(b"ToUnicode".to_vec(), Object::Ref(cmap_ref)); + Ok(font) +} + /// The runs that show `text` in the fallback: consecutive characters in one font form one run, so /// a font is switched only where the text moves to another font. `None` if a character has no code. pub(super) fn type3_encode(fallback: &Type3Fallback, text: &str) -> Option)>> { diff --git a/crates/edit/src/text/interp.rs b/crates/edit/src/text/interp.rs index 2a9f6da61a..5114a76263 100644 --- a/crates/edit/src/text/interp.rs +++ b/crates/edit/src/text/interp.rs @@ -18,13 +18,33 @@ pub(super) struct TextState { /// graphics state the text was painted with that [`TextState::ops`] doesn't restate. pub(super) stroke: Vec, pub(super) line_width: Option, - ext_gstate: Vec>, + ext_gstate: Vec, /// The dash pattern and line cap in force (see [`Ts`]): compared, not restated. pub(super) dash: Dash, pub(super) line_cap: i64, } impl TextState { + /// The state a page's content starts in (ISO 32000-2 §8.4.1, §9.3.1): what is in force on + /// top of the page, after its content wrapped in `q`/`Q`. + pub(crate) fn initial() -> Self { + TextState { + font: None, + char_spacing: 0.0, + word_spacing: 0.0, + scale: 1.0, + leading: 0.0, + rise: 0.0, + fill: Vec::new(), + render_mode: 0, + stroke: Vec::new(), + line_width: None, + ext_gstate: Vec::new(), + dash: (Vec::new(), 0.0), + line_cap: 0, + } + } + pub(super) fn of(ts: &Ts) -> Self { TextState { font: ts.font.as_ref().map(|(n, _)| (String::from_utf8_lossy(n).into_owned(), ts.size)), @@ -54,13 +74,17 @@ impl TextState { /// /// A `gs` has no inverse, so a selection `current` has and this state doesn't stays in force: /// run these inside a `q`/`Q` that starts from `current`, one per state drawn. - pub(crate) fn paint_ops(&self, current: &TextState) -> Vec { + /// + /// `ctm` is the CTM where they run. A soft mask is placed by the CTM its `gs` ran at + /// (ISO 32000-2 §11.6.5.2), so one selected at another CTM is selected there and the CTM + /// set back. + pub(crate) fn paint_ops(&self, current: &TextState, ctm: [f64; 6]) -> Vec { let n = pdfcraft_content::num; let mut v = Vec::new(); // The selections both share stay in force; the ones after them are selected again in order. let shared = self.ext_gstate.iter().zip(¤t.ext_gstate).take_while(|(a, b)| a == b).count(); - for name in self.ext_gstate.iter().skip(shared) { - v.push(Op::new("gs", vec![Object::Name(name.clone())])); + for sel in self.ext_gstate.iter().skip(shared) { + sel.push_ops(&mut v, ctm); } // A reselected ExtGState may set `/LW`, so the width (already the effective one) follows it. let reselected = !v.is_empty(); @@ -222,7 +246,7 @@ pub(super) struct Ts { /// The line width a `w` or an ExtGState's `/LW` set (`None`: the initial 1.0). pub(super) line_width: Option, /// The ExtGState resources selected with `gs`, oldest first (see [`select_gs`]). - pub(super) ext_gstate: Vec>, + pub(super) ext_gstate: Vec, /// The dash pattern a `d` or an ExtGState's `/D` set (initially solid: `[] 0`). pub(super) dash: Dash, /// The line cap a `J` or an ExtGState's `/LC` set (initially 0, butt). @@ -255,12 +279,38 @@ const MAX_GS: usize = 32; /// Record `gs /name` in `list`. A `gs` sets every parameter its dictionary names to an absolute /// value, so a later selection of the same dictionary overrides the earlier one completely: the /// earlier entry goes, which keeps replaying the list equivalent to replaying the stream. -fn select_gs(list: &mut Vec>, name: &[u8]) { - list.retain(|n| n != name); +fn select_gs(list: &mut Vec, name: &[u8], anchor: Option<[f64; 6]>) { + list.retain(|s| s.name != name); if list.len() >= MAX_GS { list.remove(0); } - list.push(name.to_vec()); + list.push(GsSelection { name: name.to_vec(), anchor }); +} + +/// One `gs` selection: the ExtGState's resource name, and the CTM it ran at when the dictionary +/// has a soft mask, which that CTM places (ISO 32000-2 §11.6.5.2). +#[derive(Clone, Debug, PartialEq)] +pub(crate) struct GsSelection { + name: Vec, + anchor: Option<[f64; 6]>, +} + +impl GsSelection { + /// Select this ExtGState from where the CTM is `ctm`: at its anchor, then back to `ctm`. Both + /// moves need an invertible matrix; without one it is selected where it runs. + fn push_ops(&self, v: &mut Vec, ctm: [f64; 6]) { + let gs = Op::new("gs", vec![Object::Name(self.name.clone())]); + let moves = self.anchor.filter(|a| *a != ctm).and_then(|a| { + let (anchor, here) = (Matrix(a), Matrix(ctm)); + Some((anchor.then(&here.invert()?), here.then(&anchor.invert()?))) + }); + let Some((to, back)) = moves else { + v.push(gs); + return; + }; + let cm = |m: Matrix| Op::new("cm", m.0.iter().map(|x| pdfcraft_content::num(*x)).collect()); + v.extend([cm(to), gs, cm(back)]); + } } /// The text-showing operators of one stream with their text, font, box and baseline. @@ -879,14 +929,15 @@ pub(super) fn interpret( b"Ts" => ts.rise = op.num(0).unwrap_or(0.0), b"Tr" => ts.render_mode = op.operands.first().and_then(Object::as_int).unwrap_or(-1), b"gs" => { - if let Some(name) = op.name(0) { - select_gs(&mut ts.ext_gstate, name); - } let mut held = None; let params = match op.name(0).and_then(|name| states.get(name)) { Some(p) => super::dict_in_place(doc, p, &mut held), None => None, }; + if let Some(name) = op.name(0) { + let masked = params.and_then(|p| p.get(b"SMask")).is_some_and(|m| doc.resolve(m).as_dict().is_some()); + select_gs(&mut ts.ext_gstate, name, masked.then_some(ts.ctm.0)); + } if let Some(params) = params { ts.gs.apply(doc, params); // `/LW` sets the line width as `w` does (ISO 32000-2 §8.4.5). diff --git a/crates/edit/src/text/rewrite.rs b/crates/edit/src/text/rewrite.rs index a0ab3583b8..7ec08e59e5 100644 --- a/crates/edit/src/text/rewrite.rs +++ b/crates/edit/src/text/rewrite.rs @@ -720,11 +720,13 @@ fn block_ops(plan: &BlockPlan<'_>, writer: &LineWriter<'_>, wrapped: &[(String, let mut ops = vec![Op::new("q", vec![])]; ops.extend(reveal_ops(plan.block, plan.lines, plan.style, plan.emit.ctm)?); // Moved: a translation inside that state. + let mut ctm = Matrix(plan.emit.ctm); if let Some((ax, ay)) = plan.movement { ops.push(Op::new("cm", vec![n(1.0), n(0.0), n(0.0), n(1.0), n(ax), n(ay)])); + ctm = Matrix([1.0, 0.0, 0.0, 1.0, ax, ay]).then(&ctm); } // Painted as the paragraph was, including what its text object set before the first show. - ops.extend(plan.state.paint_ops(plan.after)); + ops.extend(plan.state.paint_ops(plan.after, ctm.0)); ops.push(Op::new("BT", vec![])); let mut state = plan.state.clone(); state.word_spacing = 0.0; @@ -834,6 +836,11 @@ pub(crate) fn rewrite_block_from( let first = members.first().copied().ok_or_else(|| EditError::Invalid("the paragraph has no source runs".into()))?; members.iter().try_for_each(|line| line.ensure_editable())?; let p = page_dict(doc, page)?; + // Revealed OCR text may go on top of the page (see `revealed_under_paint`): a page that can't + // take that is refused here, before either rewrite path changes anything. + if b.invisible { + crate::check_on_top(doc, &p)?; + } let mut res = p.dict.get(b"Resources").map(|r| doc.resolve(r)).and_then(|r| r.as_dict().cloned()).unwrap_or_default(); let mut fonts_res = res.get(b"Font").map(|f| doc.resolve(f)).and_then(|f| f.as_dict().cloned()).unwrap_or_default(); let dispatch = choose_rewrite_path(doc, &fonts_res, &b, &members, first, text, style)?; @@ -973,9 +980,28 @@ pub(crate) fn rewrite_block_from( char_spacing: o_state.char_spacing, lead, }; - let plan = - BlockPlan { block: &b, lines: &lines, style, emit, after, movement, state: o_state, tm, size, underline: style.underline == Some(true) }; + let at = if split { start } else { emit.bt_op }; + // Revealed OCR text that something is painted over goes on top of the page, where it starts + // from the initial state. + let widest = wrapped.iter().map(|(l, _, _)| advance.of(l)).fold(width, f64::max); + let reach = crate::Reach { rect: b.rect, offset: [dx, dy], width: widest, rows: wrapped.len(), pitch: lead * k, upright }; + let over = crate::revealed_under_paint(doc, page, b.invisible, reach.bounds(), &ops, at); + let initial = TextState::initial(); + let drawn_from = if over { &initial } else { after }; + let plan = BlockPlan { + block: &b, + lines: &lines, + style, + emit, + after: drawn_from, + movement, + state: o_state, + tm, + size, + underline: style.underline == Some(true), + }; let mut block_ops = block_ops(&plan, &writer, &wrapped, font.show_font.clone())?; + let on_top = crate::lift_on_top(over, &mut block_ops); let n = pdfcraft_content::num; block_ops.extend(after.ops().into_iter().filter(|op| !op.is("Tf") || after.font.is_some())); if split { @@ -983,7 +1009,6 @@ pub(crate) fn rewrite_block_from( block_ops.push(Op::new("BT", vec![])); block_ops.push(Op::new("Tm", emit.tlm.iter().map(|v| n(*v)).collect())); } - let at = if split { start } else { emit.bt_op }; let new_data = splice(&pieces, &ops, |i| { let insert = if i == at { std::mem::take(&mut block_ops) } else { Vec::new() }; (insert, !dropped.contains(&i)) @@ -1000,12 +1025,14 @@ pub(crate) fn rewrite_block_from( d.set(b"Resources".to_vec(), Object::Dict(res)); } })?; + crate::draw_on_top(doc, &p, emit.ctm, &on_top)?; Ok(LineEdit { substituted: font.substituted }) } /// The cover (when [`BlockStyle::cover`] asks for one) and visible rendering that start a /// rewritten invisible paragraph, drawn in the space whose CTM is `ctm`. Nothing for visible text. -/// The paragraph's `q`/`Q` restores the rendering mode for the invisible text after it. +/// The paragraph's `q`/`Q` restores the rendering mode for the invisible text after it; one +/// painted over where it was is drawn on top of the page instead ([`crate::draw_on_top`]). pub(crate) fn reveal_ops(block: &TextBlock, lines: &[TextLine], style: &BlockStyle, ctm: [f64; 6]) -> Result, EditError> { if !block.invisible { return Ok(Vec::new()); diff --git a/crates/edit/src/text_regions/styled.rs b/crates/edit/src/text_regions/styled.rs index 885a0a7158..7a9204a8c0 100644 --- a/crates/edit/src/text_regions/styled.rs +++ b/crates/edit/src/text_regions/styled.rs @@ -1,3 +1,4 @@ +use super::super::interp::TextState; use super::grouping::{across, is_centred, row_rect}; use super::{ BlockStyle, CraftFont, Dict, Document, EditError, HashMap, LineEdit, MAX_TYPE3_GLYPHS, Matrix, Metrics, Object, Op, PdfString, Pieces, @@ -401,6 +402,16 @@ fn paragraph_geometry(doc: &Document, page: &pdfcraft_model::Page, block: &TextB Ok((dx, dy, style.width.unwrap_or(natural).clamp(1.0, 14_400.0))) } +/// How far below a row of `size` the next one is written: the source's leading unless a size +/// or line spacing is chosen. +fn row_step(style: &BlockStyle, size: f64, source_leading: f64) -> f64 { + if style.line_spacing.is_none() && style.size.is_none() { + source_leading + } else { + style.line_spacing.unwrap_or(1.5).clamp(0.5, 5.0) * style.size.unwrap_or(size) + } +} + type WrappedStyledRows = (StyledRows, RowRanges, Vec, Vec, Vec); fn wrap_styled_rows( @@ -976,6 +987,7 @@ pub(crate) fn rewrite_styled_block( let o = &lead.origin; // What is in force where the rewrite is drawn, and what the text after it expects. let after = if split { &o.state } else { &o.bt_state }; + let at = if split { start } else { o.bt_op }; validate_contextual_rows(&rows, &kept)?; let measured: Vec<_> = rows.iter().map(|row| measure(row)).collect::>()?; let shaped = prepare_shaped_rows(doc, &fonts, &rows, &kept, advance)?; @@ -989,6 +1001,7 @@ pub(crate) fn rewrite_styled_block( .ok_or_else(|| EditError::Invalid("the paragraph is drawn in a space it cannot be edited in".into()))?; let (tx, ty) = back.apply(dx, dy); block_ops.push(Op::new("cm", vec![n(1.0), n(0.0), n(0.0), n(1.0), n(tx), n(ty)])); + let run_ctm = Matrix([1.0, 0.0, 0.0, 1.0, tx, ty]).then(&Matrix(o.ctm)).0; // A new row can land one leading above the next paragraph and look like the same // paragraph. When every row is in this rewrite, a paragraph marker keeps them in one // scope so the neighbour stays out (#2517). A row left in the source stream cannot share @@ -1007,6 +1020,19 @@ pub(crate) fn rewrite_styled_block( }) .rfind(|v| *v > 0.0) .unwrap_or(first.size * 1.5); + // Revealed OCR text that something is painted over goes on top of the page, where it starts + // from the initial state. + let largest = + block.lines.iter().filter_map(|i| lines.get(*i)).map(|l| l.size).chain(style.ranges.iter().filter_map(|r| r.size)).fold(first.size, f64::max); + // At least one line of the largest glyphs, so a larger range reaches up as far as it draws. + const LINE_EM: f64 = 1.2; + let pitch = source_leading.max(row_step(style, largest, source_leading)).max(largest * LINE_EM); + let widest = measured.iter().copied().fold(width, f64::max); + // This path lays rows out along user-space baselines, so the paragraph is upright here. + let reach = crate::Reach { rect: block.rect, offset: [dx, dy], width: widest, rows: rows.len(), pitch, upright: true }; + let over = crate::revealed_under_paint(doc, page, block.invisible, reach.bounds(), &ops, at); + let initial = TextState::initial(); + let drawn_from = if over { &initial } else { after }; let mut baseline = first.origin.baseline; for (row_index, row) in rows.iter().enumerate() { let anchor = block.rows.get(row_index).and_then(|r| r.first()).and_then(|i| lines.get(*i)); @@ -1073,7 +1099,7 @@ pub(crate) fn rewrite_styled_block( // object (`gs`, stroke, line width, `Tr`; #3508/#3509) from the state the rewrite is // drawn in: a `gs` has no inverse, so one run's ExtGState would stay for the next. block_ops.push(Op::new("q", vec![])); - block_ops.extend(state.paint_ops(after)); + block_ops.extend(state.paint_ops(drawn_from, run_ctm)); if let Some((_, shaped, _)) = shaped_runs.get(&(row_index, pos)) { let resource = glyph_resources.get(&(row_index, pos)).ok_or_else(|| EditError::Invalid("shaped font is missing".into()))?; let mut properties = Dict::new(); @@ -1164,21 +1190,17 @@ pub(crate) fn rewrite_styled_block( x += segment_width; pos = end; } - baseline -= if style.line_spacing.is_none() && style.size.is_none() { - source_leading - } else { - style.line_spacing.unwrap_or(1.5).clamp(0.5, 5.0) * style.size.unwrap_or(anchor.size) - }; + baseline -= row_step(style, anchor.size, source_leading); } push_whole_paragraph_marker(&mut block_ops, marked, "EMC"); block_ops.push(Op::new("Q", vec![])); + let on_top = crate::lift_on_top(over, &mut block_ops); block_ops.extend(after.ops()); if split { block_ops.insert(0, Op::new("ET", vec![])); block_ops.push(Op::new("BT", vec![])); block_ops.push(Op::new("Tm", o.tlm.iter().map(|v| n(*v)).collect())); } - let at = if split { start } else { o.bt_op }; // A kept source row sits outside `block_ops`. Bound the whole paragraph when a marker is // safe. The close is not part of `block_ops`: that vector can end inside a text object // when the rewrite splits one, and a kept row's operators come later in the stream. @@ -1190,5 +1212,6 @@ pub(crate) fn rewrite_styled_block( d.set(b"Contents".to_vec(), contents); d.set(b"Resources".to_vec(), Object::Dict(res)); })?; + crate::draw_on_top(doc, &p, o.ctm, &on_top)?; Ok(LineEdit { substituted }) } diff --git a/crates/engine/src/actions.rs b/crates/engine/src/actions.rs index 3e942f6a2f..249279e09d 100644 --- a/crates/engine/src/actions.rs +++ b/crates/engine/src/actions.rs @@ -197,14 +197,14 @@ fn apply_step(s: &mut Session, id: DocId, step: &Step, log: &mut Vec) -> return Ok(None); } }; - // Steps that find nothing to do (no comments to flatten, …) aren't failures. A removal step - // that didn't remove is: the file would be passed on as cleaned with the data still in it. + // A step that finds nothing to do is not a failure. Any other error fails the file, so nothing is + // published for it: the file would otherwise be passed on without the step it was asked to apply. let removes = matches!(step, Step::RemoveHiddenInformation | Step::Sanitize); match s.apply(id, edit) { Ok(()) => log.push(step.label().to_string()), Err(EditError::Redact(pdfcraft_redact::RedactError::NothingToApply)) if removes => log.push(format!("{}: nothing to remove", step.label())), - Err(e) if removes => return Err(e), - Err(e) => log.push(format!("{}: {e}", step.label())), + Err(EditError::Redact(pdfcraft_redact::RedactError::NothingToApply)) => log.push(format!("{}: nothing to do", step.label())), + Err(e) => return Err(e), } Ok(None) } diff --git a/crates/engine/src/camera_clean.rs b/crates/engine/src/camera_clean.rs index 9c4120be56..d1768a2a83 100644 --- a/crates/engine/src/camera_clean.rs +++ b/crates/engine/src/camera_clean.rs @@ -1,6 +1,8 @@ -//! Clean up a straightened camera photo (Enhance ▸ Camera image ▸ enhancement level): even out -//! the lighting so the paper turns white (room-light colour casts, a hand's shadow, darker -//! edges go with it), then raise the contrast and drop isolated specks. Colour ink stays colour. +//! Clean up a straightened camera photo (Enhance ▸ Camera image ▸ enhancement level) or a +//! scanned page's picture (Enhance ▸ Scanned document, [`crate::scan_clean`]): even out the +//! lighting so the paper turns white (room-light colour casts, a hand's shadow, darker edges and +//! book gutters go with it), then raise the contrast and drop isolated specks. Colour ink stays +//! colour. //! //! The paper's illumination is estimated per channel on a coarse grid of blocks, from each //! block's bright end (the paper, not the ink), smoothed into a continuous map and divided out. @@ -8,6 +10,8 @@ /// The strongest a level can be. pub const MAX_LEVEL: u8 = 100; +/// The level from which [`clean`] also drops isolated specks. +pub const DESPECKLE_FROM: u8 = 60; /// How an enhanced camera page looks. #[derive(Clone, Copy, Debug, Default, PartialEq, Eq)] @@ -66,34 +70,158 @@ pub fn apply(rgb: &[u8], width: u32, height: u32, look: Look, level: u8) -> Opti let even = clean(rgb, width, height, 50)?; let t = 0.45 + 0.4 * level.min(MAX_LEVEL) as f32 / MAX_LEVEL as f32; let mut grey: Vec = even.as_chunks::<3>().0.iter().map(|p| if (lum(p) as f32 / 255.0) < t { 0 } else { 255 }).collect(); - despeckle_grey(&mut grey, w, h); + despeckle(&mut grey, 1, w, h); Pixels::Grey(grey) } }) } -/// Clean `rgb` (RGB8, `width` × `height`) at `level` (0 = unchanged … 100 = strongest). -/// Returns `None` when the buffer doesn't match the size. +/// Clean `rgb` (RGB8, `width` × `height`) at `level` (0 = unchanged … 100 = strongest), from +/// [`DESPECKLE_FROM`] without isolated specks. Returns `None` when the buffer doesn't match the +/// size. pub fn clean(rgb: &[u8], width: u32, height: u32, level: u8) -> Option> { + clean_samples(rgb, 3, width, height, CleanUp { level, edge_shadows: false, despeckle: level >= DESPECKLE_FROM }) +} + +/// What [`clean_samples`] does. +#[derive(Clone, Copy, Debug, PartialEq, Eq)] +pub struct CleanUp { + /// 0 (unchanged) … 100: even out the paper's illumination and raise the contrast, as + /// [`clean`]'s level. + pub level: u8, + /// Even out shadows along the picture's edges first (a book's gutter, a copier's dark edge): + /// lightened line by line to the paper beside them, within the outer fifth of each side. + pub edge_shadows: bool, + /// Drop isolated dark specks. + pub despeckle: bool, +} + +/// [`clean`] for samples of `n` components (1: grey, 3: RGB), with each step chosen apart (each +/// works at level 0 too). Returns `None` when `n` is neither or the buffer doesn't match the size. +pub fn clean_samples(px: &[u8], n: usize, width: u32, height: u32, how: CleanUp) -> Option> { let (w, h) = (width as usize, height as usize); - if w == 0 || h == 0 || w.checked_mul(h)?.checked_mul(3)? != rgb.len() { + if !(n == 1 || n == 3) || w == 0 || h == 0 || w.checked_mul(h)?.checked_mul(n)? != px.len() { return None; } - let level = level.min(MAX_LEVEL); - if level == 0 { - return Some(rgb.to_vec()); + let level = how.level.min(MAX_LEVEL); + let mut out = px.to_vec(); + if how.edge_shadows { + even_edges(&mut out, n, w, h); } + if level > 0 { + out = even_out(&out, n, w, h, level); + } + if how.despeckle { + despeckle(&mut out, n, w, h); + } + Some(out) +} + +/// The share of the width (and of the height) at each side where [`even_edges`] works. +const EDGE_BAND: f32 = 0.2; +/// The most [`even_edges`] lightens a line by; a darker line (a scanner's black border, a +/// picture) is not taken for shadowed paper and stays as it is. +const MAX_EDGE_GAIN: f32 = 3.0; + +/// Even out shadows that run along the picture's edges, in place: a book's gutter, a copier's +/// dark border, a page that curved away from the glass. Within the outer [`EDGE_BAND`] of each +/// side, each column (then each row) gets the paper's level along it (per channel, its 90th +/// percentile, so the text across it doesn't count), and is lightened to the brightest paper of +/// that band. The block grid of [`background`] is too coarse for such a narrow shadow: its +/// blocks borrow the brighter paper next to them. +fn even_edges(px: &mut [u8], n: usize, w: usize, h: usize) { + for vertical in [true, false] { + // Columns: lines across the width, each running down the height; rows the other way. + let (lines, along) = if vertical { (w, h) } else { (h, w) }; + let band = ((lines as f32 * EDGE_BAND) as usize).min(lines / 2); + if band == 0 { + continue; + } + let at = |line: usize, i: usize| if vertical { i * w + line } else { line * w + i }; + // The paper's level of every line in the bands and a few beyond them. + let reach = band.saturating_add(3).min(lines); + let measured = |line: usize| line < reach || line >= lines.saturating_sub(reach); + let mut level: Vec> = vec![None; lines]; + let mut hist = [[0u32; 256]; 3]; + for (line, slot) in level.iter_mut().enumerate().filter(|(line, _)| measured(*line)) { + hist.iter_mut().for_each(|hh| hh.fill(0)); + for i in 0..along { + let p = at(line, i) * n; + for (c, hh) in hist.iter_mut().enumerate().take(n) { + if let Some(v) = px.get(p + c) + && let Some(count) = hh.get_mut(*v as usize) + { + *count += 1; + } + } + } + let want = (along as f32 * 0.9) as u32; + *slot = Some(std::array::from_fn(|c| { + let mut seen = 0; + let v = hist.get(c).and_then(|hh| { + hh.iter().position(|&count| { + seen += count; + seen > want + }) + }); + v.unwrap_or(255) as f32 / 255.0 + })); + } + // Smoothed: the median of five neighbouring lines (fewer at the very edge, centred all + // the same, so a gradient isn't shifted), so one odd line (a rule, a fold) doesn't stand + // out. + let smooth: Vec> = (0..lines) + .map(|line| { + level.get(line).copied().flatten()?; + let r = line.min(lines - 1 - line).min(2); + Some(std::array::from_fn(|c| { + let mut near: Vec = (line - r..=line + r).filter_map(|l| level.get(l).copied().flatten()).map(|v| v[c]).collect(); + near.sort_by(f32::total_cmp); + near.get(near.len() / 2).copied().unwrap_or(1.0) + })) + }) + .collect(); + // Each side's band, and the lines its paper is measured on (the band and a few beyond). + for (shadowed, measured) in [(0..band, 0..reach), (lines - band..lines, lines.saturating_sub(reach)..lines)] { + // The brightest paper there: what the shadow is lightened to. + let paper: [f32; 3] = + std::array::from_fn(|c| measured.clone().filter_map(|l| smooth.get(l).copied().flatten()).map(|v| v[c]).fold(0.0, f32::max)); + for line in shadowed { + let Some(own) = smooth.get(line).copied().flatten() else { continue }; + let gain: [f32; 3] = std::array::from_fn(|c| { + let g = paper[c] / own[c].max(1e-3); + if (1.0..=MAX_EDGE_GAIN).contains(&g) { g } else { 1.0 } + }); + if gain.iter().take(n).all(|&g| g == 1.0) { + continue; + } + for i in 0..along { + let p = at(line, i) * n; + for (c, g) in gain.iter().enumerate().take(n) { + if let Some(v) = px.get_mut(p + c) { + *v = (*v as f32 * g).round().clamp(0.0, 255.0) as u8; + } + } + } + } + } + } +} + +/// The lighting evened out and the contrast raised at `level` (1–100); `px` holds `w` × `h` +/// pixels of `n` components. +fn even_out(px: &[u8], n: usize, w: usize, h: usize, level: u8) -> Vec { let k = level as f32 / MAX_LEVEL as f32; - let background = background(rgb, w, h); + let background = background(px, n, w, h); // Lighting is evened out fully from half strength on; contrast keeps growing to the top. let flatten = (2.0 * k).min(1.0); let white = 1.0 - 0.12 * k; // what reaches white let black = 0.30 * k; // what reaches black - let mut out = Vec::with_capacity(rgb.len()); - for (i, px) in rgb.as_chunks::<3>().0.iter().enumerate() { + let mut out = Vec::with_capacity(px.len()); + for (i, p) in px.chunks_exact(n).enumerate() { let (x, y) = (i % w, i / w); let bg = background.at(x, y); - for (c, &v) in px.iter().enumerate() { + for (c, &v) in p.iter().enumerate() { let v = v as f32 / 255.0; let b = bg.get(c).copied().unwrap_or(1.0).max(0.05); let even = (v / b).min(1.0); @@ -102,10 +230,7 @@ pub fn clean(rgb: &[u8], width: u32, height: u32, level: u8) -> Option> out.push((stretched * 255.0).round() as u8); } } - if level >= 60 { - despeckle(&mut out, w, h); - } - Some(out) + out } /// The paper's illumination: a grid of per-channel values, read back with bilinear @@ -140,9 +265,10 @@ impl Background { } } -/// Estimate the paper's illumination: per block, the 90th percentile of each channel (bright -/// paper, not dark ink); blocks that are mostly ink borrow from their neighbours; then smoothed. -fn background(rgb: &[u8], w: usize, h: usize) -> Background { +/// Estimate the paper's illumination: per block, the 90th percentile of each of the `n` channels +/// (bright paper, not dark ink); blocks that are mostly ink borrow from their neighbours; then +/// smoothed. Channels past `n` stay white. +fn background(px: &[u8], n: usize, w: usize, h: usize) -> Background { let block = (w.min(h) / 16).clamp(8, 128); let (cols, rows) = (w.div_ceil(block), h.div_ceil(block)); let mut cells = Vec::with_capacity(cols * rows); @@ -150,20 +276,20 @@ fn background(rgb: &[u8], w: usize, h: usize) -> Background { for by in 0..rows { for bx in 0..cols { hist.iter_mut().for_each(|hh| hh.fill(0)); - let mut n = 0u32; + let mut count = 0u32; for y in by * block..((by + 1) * block).min(h) { for x in bx * block..((bx + 1) * block).min(w) { - if let Some(p) = rgb.get((y * w + x) * 3..(y * w + x) * 3 + 3) { + if let Some(p) = px.get((y * w + x) * n..(y * w + x) * n + n) { for (c, &v) in p.iter().enumerate() { if let Some(slot) = hist.get_mut(c).and_then(|hh| hh.get_mut(v as usize)) { *slot += 1; } } - n += 1; + count += 1; } } } - let want = (n as f32 * 0.9) as u32; + let want = (count as f32 * 0.9) as u32; cells.push(std::array::from_fn(|c| { let mut seen = 0; let v = hist.get(c).and_then(|hh| { @@ -219,19 +345,15 @@ fn background(rgb: &[u8], w: usize, h: usize) -> Background { Background { cols, rows, block: block as f32, cells } } -/// [`despeckle`] for one grey byte per pixel. -fn despeckle_grey(grey: &mut [u8], w: usize, h: usize) { - let mut rgb: Vec = grey.iter().flat_map(|&v| [v, v, v]).collect(); - despeckle(&mut rgb, w, h); - for (g, p) in grey.iter_mut().zip(rgb.as_chunks::<3>().0.iter()) { - *g = p[0]; - } -} - -/// Lighten dark pixels with at most one dark neighbour (specks), in place. -fn despeckle(rgb: &mut [u8], w: usize, h: usize) { - let lum = |p: &[u8; 3]| (p[0] as u32 * 299 + p[1] as u32 * 587 + p[2] as u32 * 114) / 1000; - let dark: Vec = rgb.as_chunks::<3>().0.iter().map(|p| lum(p) < 128).collect(); +/// Lighten dark pixels with at most one dark neighbour (specks), in place; `px` holds `w` × `h` +/// pixels of `n` components (1 or 3). +fn despeckle(px: &mut [u8], n: usize, w: usize, h: usize) { + let lum = |p: &[u8]| match p { + [r, g, b] => (*r as u32 * 299 + *g as u32 * 587 + *b as u32 * 114) / 1000, + [v, ..] => *v as u32, + [] => 255, + }; + let dark: Vec = px.chunks_exact(n.max(1)).map(|p| lum(p) < 128).collect(); let is_dark = |x: isize, y: isize| { x >= 0 && y >= 0 && (x as usize) < w && (y as usize) < h && dark.get(y as usize * w + x as usize).copied().unwrap_or(false) }; @@ -243,7 +365,7 @@ fn despeckle(rgb: &mut [u8], w: usize, h: usize) { let neighbours = (-1..=1).flat_map(|dy| (-1..=1).map(move |dx| (dx, dy))).filter(|&(dx, dy)| (dx, dy) != (0, 0) && is_dark(x + dx, y + dy)).count(); if neighbours <= 1 - && let Some(p) = rgb.get_mut((y as usize * w + x as usize) * 3..(y as usize * w + x as usize) * 3 + 3) + && let Some(p) = px.get_mut((y as usize * w + x as usize) * n..(y as usize * w + x as usize) * n + n) { p.fill(255); } @@ -388,5 +510,40 @@ mod tests { assert_eq!(clean(&[], 0, 0, 50), None); assert_eq!(clean(&[0; 12], 2, 3, 50), None); assert_eq!(clean(&[9; 3], 1, 1, 255).map(|v| v.len()), Some(3)); + let every = CleanUp { level: 100, edge_shadows: true, despeckle: true }; + assert_eq!(clean_samples(&[0; 12], 2, 2, 3, every), None, "two components"); + assert_eq!(clean_samples(&[0; 5], 1, 2, 3, every), None); + for (w, h) in [(1, 1), (1, 9), (9, 1), (2, 2), (5, 3)] { + assert_eq!(clean_samples(&vec![100; w * h], 1, w as u32, h as u32, every).map(|v| v.len()), Some(w * h)); + } + } + + #[test] + fn edge_shadows_lighten_shadowed_paper_but_not_a_black_border_or_the_middle() { + // Grey paper (200) with a shadow darkening the left 20 columns to 120, a black scanner + // border in the right 8 columns, and a dark block in the middle. + let (w, h) = (100usize, 60usize); + let mut grey = vec![200u8; w * h]; + for y in 0..h { + for x in 0..w { + let v = if x >= w - 8 { + 5 + } else if (40..60).contains(&x) && (20..40).contains(&y) { + 30 + } else if x < 20 { + 120 + (80 * x / 20) as u8 + } else { + 200 + }; + grey[y * w + x] = v; + } + } + let out = clean_samples(&grey, 1, w as u32, h as u32, CleanUp { level: 0, edge_shadows: true, despeckle: false }).unwrap(); + for x in [0, 5, 19] { + assert!(out[30 * w + x].abs_diff(200) <= 3, "column {x}: {} → {}", grey[30 * w + x], out[30 * w + x]); + } + assert_eq!(out[30 * w + w - 3], 5, "a black border is no shadow"); + assert_eq!(out[30 * w + 50], 30, "the middle is untouched"); + assert_eq!(out[30 * w + 30], 200); } } diff --git a/crates/engine/src/catalog.rs b/crates/engine/src/catalog.rs index e479232bbb..d6be8fd4eb 100644 --- a/crates/engine/src/catalog.rs +++ b/crates/engine/src/catalog.rs @@ -265,9 +265,9 @@ pub static TOOL_GROUPS: &[ToolGroup] = &[ items: &[ item("In this file", "scan-text", "ocr.recognize", Ready), item("In multiple files", "files", "ocr.recognize_batch", Ready), - item("Enhance scanned file", "sparkles", "ocr.enhance", Planned("M10")), + item("Enhance scanned file", "sparkles", "ocr.enhance", Ready), item("Enhance camera image", "camera", "ocr.enhance_camera", Ready), - item("Correct recognized text", "text-select", "ocr.correct", Planned("M10")), + item("Correct recognized text", "text-select", "ocr.correct", Ready), ], }], }, diff --git a/crates/engine/src/commands.rs b/crates/engine/src/commands.rs index 0647726452..7cee14657e 100644 --- a/crates/engine/src/commands.rs +++ b/crates/engine/src/commands.rs @@ -220,6 +220,10 @@ pub const COMMANDS: &[CommandSpec] = &[ // Alt-drag for moving windows (several Linux window managers). // Copy events stay with the frontend so focused text inputs and page clipboard routing keep ownership. c("edit.copy", "Copy", EDIT, None, Copying, "copy"), + // Comments travel between documents as XFDF (M14): ⌘C with comments selected copies them, + // ⌘V pastes the copies onto the current page; the strip's Copy/Paste in the menus too. + c("comment.copy", "Copy comments", EDIT, None, Document, "copy"), + c("comment.paste", "Paste comments", EDIT, None, Annotate, "clipboard-paste"), c("edit.column_select", "Column select", EDIT, None, Document, "text-select"), c("edit.snapshot", "Take a snapshot", EDIT, None, Copying, "camera"), c("view.full_screen", "Full screen mode", VIEW, Some(Shortcut::cmd("L")), Document, "maximize"), @@ -355,7 +359,9 @@ pub const COMMANDS: &[CommandSpec] = &[ c("tools.document_js", "Document JavaScripts…", None, None, Modification, "file-code"), c("ocr.recognize", "Recognize text…", None, None, Modification, "scan-text"), c("ocr.enhance_camera", "Enhance camera image", None, None, Modification, "camera"), + c("ocr.enhance", "Enhance scanned file…", None, None, Modification, "sparkles"), c("ocr.recognize_batch", "Recognize text in multiple files…", None, None, Nothing, "files"), + c("ocr.correct", "Correct recognized text…", None, None, Document, "text-select"), c("a11y.check", "Check for accessibility…", None, None, Document, "accessibility"), c("a11y.find_untagged", "Find untagged content", None, None, Document, "search"), c("a11y.autotag", "Automatically tag PDF", None, None, Document, "tag"), diff --git a/crates/engine/src/copy_image.rs b/crates/engine/src/copy_image.rs index 53b890f27c..9d5668ccf5 100644 --- a/crates/engine/src/copy_image.rs +++ b/crates/engine/src/copy_image.rs @@ -112,7 +112,7 @@ pub(crate) fn added_picture(cos: &Cos, file: &[u8]) -> Option { } /// The largest raster image Form `form` draws through its resources (nested Forms included). -fn picture_in_form(cos: &Cos, form: ObjRef) -> Option { +pub(crate) fn picture_in_form(cos: &Cos, form: ObjRef) -> Option { let mut seen = HashSet::from([form]); let mut best: Option<(u64, ObjRef)> = None; let res = resources(cos, form)?; diff --git a/crates/engine/src/dates.rs b/crates/engine/src/dates.rs index c9382c1b02..5ede67c811 100644 --- a/crates/engine/src/dates.rs +++ b/crates/engine/src/dates.rs @@ -582,6 +582,18 @@ pub fn date_language(code: &str) -> Option<&'static DateLanguage> { DATE_LANGUAGES.iter().find(|l| l.code.eq_ignore_ascii_case(code.trim())) } +/// Form date fields (`AFDate_FormatEx`, `util.printd`, `util.scand`) write month and weekday +/// names in date language `code` from now on, as a localized Acrobat writes its own language's: +/// `dd-mmm-yy` is `18-ott-26` in Italian (#5303). An unknown code is English. +pub fn use_form_date_language(code: &str) { + let l = date_language(code).unwrap_or(&DATE_LANGUAGES[0]); + let names = pdfcraft_forms::af::DateNames { months: l.months, months_short: l.months_short, days: l.days, days_short: l.days_short }; + if pdfcraft_forms::af::date_names() != names { + pdfcraft_forms::af::set_date_names(names); + pdfcraft_js::set_date_names(l.months, l.months_short, l.days, l.days_short); + } +} + /// The characters of `text` that Fill & Sign can't write into a PDF yet, each once. Its text is /// drawn in Helvetica with WinAnsi encoding (Western European letters), so anything else (`ř`, /// Japanese, Chinese) would become `?` in the file. diff --git a/crates/engine/src/document.rs b/crates/engine/src/document.rs index fa98a18417..279dc8d563 100644 --- a/crates/engine/src/document.rs +++ b/crates/engine/src/document.rs @@ -539,7 +539,7 @@ impl Document { _ => "Attachments only", }; let pending = editor.cos.encryption_changed(); - let permissions = if pending { pdfcraft_cos::Permissions { bits: h.permissions().bits, owner: false } } else { h.permissions() }; + let permissions = if pending { pdfcraft_cos::Permissions { owner: false, ..h.permissions() } } else { h.permissions() }; Some(SecuritySummary { method: method.into(), owner: h.auth() == pdfcraft_cos::Auth::Owner && !pending, permissions, pending }) } diff --git a/crates/engine/src/edit.rs b/crates/engine/src/edit.rs index 8f43689aa9..1bcf24108f 100644 --- a/crates/engine/src/edit.rs +++ b/crates/engine/src/edit.rs @@ -3,10 +3,11 @@ use std::sync::Arc; use crate::{ - AddedContent, AddedText, Background, BoxSpec, EditError, FieldProps, FieldValue, FillMark, HeaderFooter, Hidden, ImageEdit, InitialView, Label, - LineEnding, LinkAction, LinkStyle, MarkFile, MarkKind, Markup, NewAnnotation, NewField, NoteIcon, ObjectTarget, PageBox, Protection, ReviewState, - Rgb, Shape, Standard, Style, TabOrder, TransitionSpec, Watermark, a11y, copy_image, decode_data_file, embed_image, field_table, fit_picture, - image_content_type, is_xfa_data_file, js, mark_source, measure, open_source, strip_xfa, unrestricted, xfa, xfa_after_data_change, + AddedContent, AddedText, Background, BorderShape, BoxSpec, EditError, FieldProps, FieldValue, FillMark, HeaderFooter, Hidden, ImageEdit, + InitialView, Label, LineEnding, LinkAction, LinkStyle, MarkFile, MarkKind, Markup, NewAnnotation, NewField, NoteIcon, ObjectTarget, PageBox, + Protection, ReviewState, Rgb, Shape, Standard, Style, TabOrder, TransitionSpec, Watermark, a11y, copy_image, decode_data_file, embed_image, + field_table, fit_picture, image_content_type, is_xfa_data_file, js, mark_source, measure, model_poster, open_source, strip_xfa, unrestricted, + xfa, xfa_after_data_change, }; mod labels; @@ -119,6 +120,15 @@ pub enum Edit { bold: Option, italic: Option, }, + /// Write `bookmarks` as children of `parent` from child `index` on, or with `replace` in + /// place of the whole outline (`parent` must then be `[]`). A bookmark whose page this + /// document lacks is kept without a destination (see [`pdfcraft_organize::import_outline`]). + ImportBookmarks { + parent: Vec, + index: usize, + bookmarks: Vec, + replace: bool, + }, /// Bookmarks for the tagged headings (H, H1–H6), nested by level under a new first /// "Untitled" bookmark (New Bookmarks from Structure). BookmarksFromStructure, @@ -157,6 +167,17 @@ pub enum Edit { file: MarkFile, author: String, }, + /// Rich Media ▸ Add 3D: a U3D or PRC `model` as a 3D annotation in `rect` on `page`, named + /// `name` (its alternate description). Its poster is the picture in `poster` (a PDF page or an + /// image) fitted inside `rect`, keeping its shape and centred, or a labelled frame without one. + AddModel3d { + page: usize, + rect: [f64; 4], + model: Arc>, + poster: Option, + activation: pdfcraft_annot::Activation3d, + name: String, + }, /// Delete the comment at `index` in the page's `/Annots`, with its pop-up and replies. DeleteAnnotation { page: usize, @@ -243,12 +264,35 @@ pub enum Edit { index: usize, fill: Option, }, + /// Draw a rectangle's, oval's, polygon's or text box's fill, or a text box's text, at an + /// opacity of its own (0–1), the line or border keeping the comment's; `None` leaves that + /// part as it is. `StyleAnnotation`'s opacity sets the whole comment's. + PartOpacity { + page: usize, + index: usize, + fill: Option, + text: Option, + }, /// Change the font size (pt) of a text box, typewriter text or callout. FontSizeAnnotation { page: usize, index: usize, size: f64, }, + /// Give a text box, typewriter text or callout a border colour of its own (`None`: the + /// text's again). + BorderColorAnnotation { + page: usize, + index: usize, + color: Option, + }, + /// Give a text box or typewriter text an oval, a circle or a rectangle for its border: the + /// text box keeps its place, the oval or circle goes around it. + BorderShapeAnnotation { + page: usize, + index: usize, + shape: BorderShape, + }, /// Comment properties ▸ General / note icon. SetAnnotationInfo { page: usize, @@ -386,6 +430,13 @@ pub enum Edit { page: usize, words: Vec, }, + /// Scan & OCR ▸ Enhance ▸ Scanned document: put the scan images + /// [`ScanCleanJob::run`](crate::scan_clean::ScanCleanJob::run) cleaned in place of the old + /// ones (each object must still be an image). Built by + /// [`Session::apply_scan_clean`](crate::Session::apply_scan_clean). + EnhanceScan { + images: Vec<(pdfcraft_cos::ObjRef, pdfcraft_cos::Stream)>, + }, /// Edit a PDF ▸ Edit text: replace the text of line `line` (from `Document::text_lines`) on /// `page`, in its own font when it can show it, else in Helvetica. EditTextLine { @@ -534,6 +585,15 @@ pub enum Edit { name: String, bytes: Arc>, }, + /// Paste copied comments (XFDF from [`Session::export_comments`]) onto `page` as new + /// comments: copies under names of their own, so nothing is replaced and pasting twice + /// gives two sets; replies keep their copied parent. + PasteComments { + xfdf: String, + page: usize, + /// Where the copies go: moved by (dx, dy) in user space, or at their own positions. + offset: Option<[f64; 2]>, + }, /// Remove Hidden Information: the chosen categories. RemoveHidden { which: Vec, diff --git a/crates/engine/src/edit/labels.rs b/crates/engine/src/edit/labels.rs index 5343804ef3..615ac0fea5 100644 --- a/crates/engine/src/edit/labels.rs +++ b/crates/engine/src/edit/labels.rs @@ -38,6 +38,7 @@ pub const EDIT_LABELS: &[&str] = &[ "Set bookmark destination", "Colour", "Style", + "Import bookmarks", "New bookmarks from structure", "New bookmarks from text styles", "Number pages", @@ -68,7 +69,9 @@ pub const EDIT_LABELS: &[&str] = &[ "Add inserted text", "Add file attachment", "Add stamp", + "Add 3D model", "Delete comment", + "Paste comments", "Edit comment", "Reply", "Set status None", @@ -99,6 +102,7 @@ pub const EDIT_LABELS: &[&str] = &[ "Set alternate text", "Mark figure as decorative", "Recognize text", + "Enhance scanned file", "Run JavaScript", "Run form script", "Save as {level}", @@ -193,6 +197,7 @@ impl Edit { // The bookmark Style and Colour menus' own names, already said in every language. Edit::SetBookmarkColor { .. } => Label::of("Colour"), Edit::SetBookmarkStyle { .. } => Label::of("Style"), + Edit::ImportBookmarks { .. } => Label::of("Import bookmarks"), Edit::BookmarksFromStructure => Label::of("New bookmarks from structure"), Edit::BookmarksFromTextStyles { .. } => Label::of("New bookmarks from text styles"), Edit::NumberPages { .. } => Label::of("Number pages"), @@ -205,7 +210,9 @@ impl Edit { Edit::SetMeasurementScale { .. } => Label::of("Set measurement scale"), Edit::AddAnnotation(a) => Label::of(add_annotation(&a.shape)), Edit::AddCustomStamp { .. } => Label::of("Add stamp"), + Edit::AddModel3d { .. } => Label::of("Add 3D model"), Edit::DeleteAnnotation { .. } => Label::of("Delete comment"), + Edit::PasteComments { .. } => Label::of("Paste comments"), Edit::SetAnnotationContents { .. } => Label::of("Edit comment"), Edit::ReplyToAnnotation { .. } => Label::of("Reply"), // The state's name is part of the wording: a language may need to agree it with the rest. @@ -225,9 +232,13 @@ impl Edit { Edit::MoveAnnotation { .. } => Label::of("Move comment"), Edit::ReshapeAnnotation { .. } => Label::of("Reshape comment"), Edit::ResizeAnnotation { .. } => Label::of("Resize comment"), - Edit::StyleAnnotation { .. } | Edit::FillAnnotation { .. } | Edit::FontSizeAnnotation { .. } | Edit::SetAnnotationInfo { .. } => { - Label::of("Change comment properties") - } + Edit::StyleAnnotation { .. } + | Edit::FillAnnotation { .. } + | Edit::PartOpacity { .. } + | Edit::FontSizeAnnotation { .. } + | Edit::BorderColorAnnotation { .. } + | Edit::BorderShapeAnnotation { .. } + | Edit::SetAnnotationInfo { .. } => Label::of("Change comment properties"), Edit::SetFieldValue { name, .. } => Label::with("Fill in {name}", name.as_str()), Edit::SetFieldImage { name, .. } => Label::with("Set the image of {name}", name.as_str()), Edit::ResetForm { .. } => Label::of("Clear form"), @@ -248,6 +259,7 @@ impl Edit { Edit::MarkDecorative { .. } => Label::of("Mark figure as decorative"), Edit::MarkTagArtifact { .. } => Label::of("Mark content as artifact"), Edit::AddOcrText { .. } => Label::of("Recognize text"), + Edit::EnhanceScan { .. } => Label::of("Enhance scanned file"), Edit::ApplyScriptChanges { .. } => Label::of("Run JavaScript"), Edit::XfaEvent { .. } => Label::of("Run form script"), // The level is the standard's own name ("PDF/A-2b"), so it goes in as it stands. diff --git a/crates/engine/src/edit/run.rs b/crates/engine/src/edit/run.rs index 5ed70aed7a..4719de302f 100644 --- a/crates/engine/src/edit/run.rs +++ b/crates/engine/src/edit/run.rs @@ -1,6 +1,6 @@ use super::{ AddedContent, Arc, Edit, EditError, FieldValue, ImageEdit, NewAnnotation, Shape, a11y, copy_image, decode_data_file, embed_image, fit_picture, - image_content_type, is_xfa_data_file, js, mark_source, measure, open_source, strip_xfa, unrestricted, xfa, xfa_after_data_change, + image_content_type, is_xfa_data_file, js, mark_source, measure, model_poster, open_source, strip_xfa, unrestricted, xfa, xfa_after_data_change, }; use crate::SourceProblem; use crate::sources::{comment_source, import_pdf_comments}; @@ -27,6 +27,7 @@ pub(crate) fn check_permission(edit: &Edit, p: &pdfcraft_cos::Permissions) -> Re | Edit::SetBookmarkPage { .. } | Edit::SetBookmarkColor { .. } | Edit::SetBookmarkStyle { .. } + | Edit::ImportBookmarks { .. } | Edit::BookmarksFromStructure | Edit::NumberPages { .. } => { if p.assemble() { @@ -48,7 +49,9 @@ pub(crate) fn check_permission(edit: &Edit, p: &pdfcraft_cos::Permissions) -> Re } Edit::AddMeasurement(_) | Edit::AddAnnotation(_) + | Edit::PasteComments { .. } | Edit::AddCustomStamp { .. } + | Edit::AddModel3d { .. } | Edit::DeleteAnnotation { .. } | Edit::SetAnnotationContents { .. } | Edit::ReplyToAnnotation { .. } @@ -62,7 +65,10 @@ pub(crate) fn check_permission(edit: &Edit, p: &pdfcraft_cos::Permissions) -> Re | Edit::ResizeAnnotation { .. } | Edit::StyleAnnotation { .. } | Edit::FillAnnotation { .. } + | Edit::PartOpacity { .. } | Edit::FontSizeAnnotation { .. } + | Edit::BorderColorAnnotation { .. } + | Edit::BorderShapeAnnotation { .. } | Edit::SetAnnotationInfo { .. } | Edit::SetMeasurementScale { .. } => { if p.annotate() { @@ -136,6 +142,7 @@ pub(crate) fn check_permission(edit: &Edit, p: &pdfcraft_cos::Permissions) -> Re | Edit::MarkDecorative { .. } | Edit::MarkTagArtifact { .. } | Edit::AddOcrText { .. } + | Edit::EnhanceScan { .. } | Edit::SetDocumentScript { .. } | Edit::SetFieldScript { .. } | Edit::RemoveStandardId { .. } @@ -454,6 +461,9 @@ pub(crate) fn run_edit(doc: &mut pdfcraft_cos::Document, edit: &Edit, cx: &mut E Edit::SetBookmarkPage { path, page } => pdfcraft_organize::set_bookmark_page(doc, path, *page)?, Edit::SetBookmarkColor { path, color } => pdfcraft_organize::set_bookmark_color(doc, path, *color)?, Edit::SetBookmarkStyle { path, bold, italic } => pdfcraft_organize::set_bookmark_style(doc, path, *bold, *italic)?, + Edit::ImportBookmarks { parent, index, bookmarks, replace } => { + pdfcraft_organize::import_outline(doc, parent, *index, bookmarks, *replace)?; + } Edit::BookmarksFromStructure => { let entries: Vec<_> = pdfcraft_a11y::headings(doc) .into_iter() @@ -489,6 +499,14 @@ pub(crate) fn run_edit(doc: &mut pdfcraft_cos::Document, edit: &Edit, cx: &mut E Edit::AddAnnotation(a) => { pdfcraft_annot::add_annotation(doc, a, &cx.meta())?; } + Edit::PasteComments { xfdf, page, offset } => { + // Names from the edit's own id generator: unique to this paste within the document. + let prefix = cx.meta().id; + let report = pdfcraft_xfdf::import_comments(doc, xfdf, Some(*page), Some(&prefix), *offset)?; + if report.comments == 0 { + return Err(EditError::Invalid("the clipboard holds no comments".into())); + } + } Edit::AddCustomStamp { page, rect, name, file, author } => { let src = mark_source(doc, file)?; let (sw, sh) = (src.size.0.max(1.0), src.size.1.max(1.0)); @@ -509,6 +527,16 @@ pub(crate) fn run_edit(doc: &mut pdfcraft_cos::Document, edit: &Edit, cx: &mut E let index = pdfcraft_annot::add_annotation(doc, &new, &cx.meta())?; pdfcraft_annot::orient_image_stamp(doc, *page, index, rotation)?; } + Edit::AddModel3d { page, rect, model, poster, activation, name } => { + // A file that is not a model is refused before its poster is brought in. + pdfcraft_annot::check_model(model)?; + let poster = match poster { + Some(file) => Some(model_poster(doc, *rect, file)?), + None => None, + }; + let new = pdfcraft_annot::NewModel3d { page: *page, rect: *rect, model, activation: *activation, poster, contents: name }; + pdfcraft_annot::add_3d_model(doc, &new, &cx.meta())?; + } Edit::DeleteAnnotation { page, index } => pdfcraft_annot::delete_annotation(doc, *page, *index)?, Edit::SetAnnotationContents { page, index, text } => pdfcraft_annot::set_contents(doc, *page, *index, text, &cx.meta())?, Edit::ReplyToAnnotation { page, index, text, author } => { @@ -559,7 +587,10 @@ pub(crate) fn run_edit(doc: &mut pdfcraft_cos::Document, edit: &Edit, cx: &mut E )?; } Edit::FillAnnotation { page, index, fill } => pdfcraft_annot::set_fill(doc, *page, *index, *fill, &cx.meta())?, + Edit::PartOpacity { page, index, fill, text } => pdfcraft_annot::set_part_opacity(doc, *page, *index, *fill, *text, &cx.meta())?, Edit::FontSizeAnnotation { page, index, size } => pdfcraft_annot::set_font_size(doc, *page, *index, *size, &cx.meta())?, + Edit::BorderColorAnnotation { page, index, color } => pdfcraft_annot::set_border_color(doc, *page, *index, *color, &cx.meta())?, + Edit::BorderShapeAnnotation { page, index, shape } => pdfcraft_annot::set_border_shape(doc, *page, *index, *shape, &cx.meta())?, Edit::SetAnnotationInfo { page, index, author, subject, icon } => { pdfcraft_annot::set_info(doc, *page, *index, author.as_deref(), subject.as_deref(), *icon, &cx.meta())?; } @@ -687,8 +718,9 @@ pub(crate) fn run_edit(doc: &mut pdfcraft_cos::Document, edit: &Edit, cx: &mut E Edit::RemoveStandardId { standard } => pdfcraft_organize::remove_standard_id(doc, *standard)?, Edit::SetDocumentScript { name, script } => pdfcraft_forms::set_document_script(doc, name, script.as_deref())?, Edit::AddOcrText { page, words } => { - pdfcraft_edit::stamp(doc, *page, "OCR", pdfcraft_ocr::text_layer(words))?; + crate::ocr::layer::add_text(doc, *page, words)?; } + Edit::EnhanceScan { images } => crate::scan_clean::write_back(doc, images)?, Edit::EditPageImage { page, index, change } => run_edit_page_image(doc, *page, *index, change)?, Edit::MoveObjects { page, objects, offset } => pdfcraft_edit::move_objects(doc, *page, objects, *offset)?, Edit::EditTextBlock { page, block, text, style } => { diff --git a/crates/engine/src/history.rs b/crates/engine/src/history.rs index fafa65b4c2..ce51df5e61 100644 --- a/crates/engine/src/history.rs +++ b/crates/engine/src/history.rs @@ -53,7 +53,9 @@ pub(crate) fn scope_of(edit: &Edit) -> Scope { Edit::AddAnnotation(a) if matches!(a.shape, Shape::Attachment { .. }) => Scope::Full, Edit::AddMeasurement(_) | Edit::AddAnnotation(_) + | Edit::PasteComments { .. } | Edit::AddCustomStamp { .. } + | Edit::AddModel3d { .. } | Edit::DeleteAnnotation { .. } | Edit::SetAnnotationContents { .. } | Edit::ReplyToAnnotation { .. } @@ -67,7 +69,10 @@ pub(crate) fn scope_of(edit: &Edit) -> Scope { | Edit::ResizeAnnotation { .. } | Edit::StyleAnnotation { .. } | Edit::FillAnnotation { .. } + | Edit::PartOpacity { .. } | Edit::FontSizeAnnotation { .. } + | Edit::BorderColorAnnotation { .. } + | Edit::BorderShapeAnnotation { .. } | Edit::SetAnnotationInfo { .. } => Scope::Comments, Edit::AddLink { .. } | Edit::SetLink { .. } | Edit::DeleteLink { .. } | Edit::RemoveLinks { .. } | Edit::AddLinks { .. } => Scope::Links, Edit::SetFieldValue { .. } | Edit::ResetForm { .. } | Edit::SetFieldImage { .. } | Edit::ApplyScriptChanges { .. } => Scope::Form, @@ -77,6 +82,7 @@ pub(crate) fn scope_of(edit: &Edit) -> Scope { | Edit::EditTextLine { .. } | Edit::EditTextBlock { .. } | Edit::EditPageImage { .. } + | Edit::EnhanceScan { .. } | Edit::AddText { .. } | Edit::AddImage { .. } | Edit::UpdateContent { .. } diff --git a/crates/engine/src/js.rs b/crates/engine/src/js.rs index d1ee5c15b0..cc78785815 100644 --- a/crates/engine/src/js.rs +++ b/crates/engine/src/js.rs @@ -199,6 +199,24 @@ impl Scripts for JsRunner { /// Run field `target`'s script for `event` over the form's fields, keeping what it showed and /// asked for; a refusal carries its last message. fn run(&mut self, event: FieldEvent, script: &str, target: &Field, value: &str, fields: &[Field]) -> ScriptResult { + // As in Acrobat, a failing script leaves the value alone (the error goes to the console). + self.run_logged(event, script, target, value, fields, true).unwrap_or_else(|| ScriptResult { + rc: true, + value: value.to_string(), + changes: Vec::new(), + message: None, + }) + } + + fn rerun_with_english_dates(&mut self, event: FieldEvent, script: &str, target: &Field, value: &str, fields: &[Field]) -> Option { + pdfcraft_js::with_english_printd(true, || self.run_logged(event, script, target, value, fields, false)) + } +} + +impl JsRunner { + /// [`Scripts::run`], keeping what the script showed and asked for only when `log`; `None` + /// when the script failed. + fn run_logged(&mut self, event: FieldEvent, script: &str, target: &Field, value: &str, fields: &[Field], log: bool) -> Option { let states: Vec<_> = fields .iter() .map(|f| { @@ -212,17 +230,18 @@ impl Scripts for JsRunner { let states = std::sync::Arc::new(states); let ev = pdfcraft_js::Event::field(event.name(), &target.name, value); let o = pdfcraft_js::run_shared(script, &ev, &self.doc, std::sync::Arc::clone(&states), &self.doc_scripts, Limits::default()); - self.output.absorb(&o); - if o.error.is_some() { - // As in Acrobat, a failing script leaves the value alone (the error goes to the console). - return ScriptResult { rc: true, value: value.to_string(), changes: Vec::new(), message: None }; + if log { + self.output.absorb(&o); } - ScriptResult { + if o.error.is_some() { + return None; + } + Some(ScriptResult { rc: o.rc, value: o.value.clone(), changes: changes(&states, &o.changed), message: (!o.rc).then(|| o.alerts.last().map(|m| m.text.clone())).flatten(), - } + }) } } diff --git a/crates/engine/src/lib.rs b/crates/engine/src/lib.rs index fe139164be..1c113d0d82 100644 --- a/crates/engine/src/lib.rs +++ b/crates/engine/src/lib.rs @@ -35,6 +35,7 @@ mod history; pub mod js; mod layers; pub mod links; +pub mod model3d; pub mod ocr; mod open; pub mod optimizer; @@ -43,6 +44,7 @@ mod pictures; pub mod print_raster; pub mod printcheck; mod protection; +pub mod scan_clean; mod scan_colours; mod scan_fields; mod session; @@ -71,7 +73,10 @@ pub use signature_scan::{PhotoRect, ScanError, SignaturePhoto, Turn}; pub use stamp_preview::{CommentLayer, StampPreview}; pub use text_preview::{TextEditPreview, UnmarkedPages}; -pub use pdfcraft_organize::{BoxSpec, FieldConflicts, PageBox, SplitBy, Transition, TransitionSpec, mime_for_name, split_ranges}; +pub use pdfcraft_organize::{ + BoxSpec, FieldConflicts, MAX_IMPORT_BOOKMARKS, OutlineNode, OutlineTree, PageBox, SplitBy, Transition, TransitionSpec, mime_for_name, + split_ranges, unmapped as outline_unmapped, view_arity as outline_view_arity, +}; pub use pdfcraft_organize::LabelStyle; pub use pdfcraft_organize::Standard; @@ -108,9 +113,10 @@ pub use pdfcraft_fonts::{MAX_SIGNATURE_CHARS, ScriptOutline, script_missing, scr pub use pdfcraft_annot::appearance as annot_text; pub use pdfcraft_annot::links::{Highlight as LinkHighlight, LinkAction, LinkItem, LinkStyle}; pub use pdfcraft_annot::{ - AppearanceCache, AttachIcon, FillMark, LineEnding, Markup, NOTE_SIZE, NewAnnotation, NoteIcon, OverlayFont, OverlayLook, Props as CommentProps, - ReviewState, Rgb, Shape, StampGroup, StampKind, Style, rect_quad, + Activation3d, AppearanceCache, AttachIcon, BorderShape, FillMark, LineEnding, Markup, NOTE_SIZE, NewAnnotation, NoteIcon, OverlayFont, + OverlayLook, Props as CommentProps, ReviewState, Rgb, Shape, StampGroup, StampKind, Style, rect_quad, }; +pub use pdfcraft_annot::{MAX_MODEL_BYTES as MAX_3D_MODEL_BYTES, ModelFormat as Model3dFormat}; pub use pdfcraft_forms::detect; pub use pdfcraft_optimize as optimize; pub use pdfcraft_preflight as pdfa; @@ -123,6 +129,7 @@ pub use pdfcraft_redact::sanitize::{HIDDEN, Hidden}; pub use pdfcraft_sign as sign; pub use pdfcraft_sign::{SignOptions, SignatureInfo, Status as SignatureStatus, TrustStore}; pub use pdfcraft_xfdf::Format as DataFormat; +pub use pdfcraft_xfdf::{Outline as CommentOutline, comment_count, comment_outlines, comments_bounds, is_xfdf_comments}; use bookmarks::outline_rows; pub use bookmarks::{BookmarkPage, BookmarkRow}; @@ -140,7 +147,7 @@ use layers::{apply_layer_state, set_default_layer_state, use_layer_choices}; pub use open::{OpenContext, PreparedDocument}; pub use page_filter::{PageOrientation, PageParity, filter_pages}; pub use pictures::{MarkFile, image_dimensions}; -use pictures::{embed_image, fit_picture, mark_source}; +use pictures::{embed_image, fit_picture, mark_source, model_poster}; use protection::unrestricted; pub use protection::{Changes, Printing, Protection, SecuritySummary}; pub use session::Session; diff --git a/crates/engine/src/model3d.rs b/crates/engine/src/model3d.rs new file mode 100644 index 0000000000..b95a332a22 --- /dev/null +++ b/crates/engine/src/model3d.rs @@ -0,0 +1,19 @@ +//! The 3D models on a page (ISO 32000-2 §13.6), read so a renderer can draw them. + +pub use pdfcraft_annot::{MAX_MODEL_BYTES, Model3d, Model3dKind}; + +use crate::Document; + +impl Document { + /// The `/Annots` positions of the 3D annotations on 0-based `page`, in order. + pub fn model_3d_annotations(&self, page: usize) -> Result, String> { + let editor = self.editor.as_ref().ok_or("the document can't be read")?; + pdfcraft_annot::model_3d_annotations(&editor.cos, page).map_err(|e| e.to_string()) + } + + /// The decoded model of the 3D annotation at `/Annots` position `index` on 0-based `page`. + pub fn model_3d(&self, page: usize, index: usize) -> Result { + let editor = self.editor.as_ref().ok_or("the document can't be read")?; + pdfcraft_annot::model_3d(&editor.cos, page, index).map_err(|e| e.to_string()) + } +} diff --git a/crates/engine/src/ocr.rs b/crates/engine/src/ocr.rs index 6b9869cead..18ea331d41 100644 --- a/crates/engine/src/ocr.rs +++ b/crates/engine/src/ocr.rs @@ -3,16 +3,104 @@ //! //! Recognition is slow (about a second a page), so it is split from the edit: [`Session::ocr_job`] //! captures what it needs, [`OcrJob::run`] works anywhere (the UI runs it on a worker thread) and -//! [`Session::apply_ocr`] applies the result as one undoable step. +//! [`Session::apply_ocr`] applies the result as one undoable step. The result is only applied +//! while the document is still the one that was read — [`Session::apply_ocr_checked`] refuses a +//! stale result with a named reason instead of writing into a changed document. use std::sync::{Arc, Mutex}; +#[cfg(not(target_arch = "wasm32"))] +use std::time::{Duration, Instant}; use pdfcraft_render::{PageInfo, PageRenderer, RenderConfig, RenderRequest, RequestKind}; -pub use pdfcraft_ocr::{LANGUAGES, Models, Ocr, OcrError, PlacedWord}; +pub use pdfcraft_ocr::{ + ConfidenceBand, LANGUAGES, MergeStrategy, Models, Ocr, OcrError, PlacedWord, Recognizer, band_for, is_suspect, low_confidence, +}; use crate::{DocId, Edit, EditError, Session}; +pub mod dictionary; +pub(crate) mod layer; + +/// Bound recognition/review inputs before cloning or planning glyphs. +pub fn validate_ocr_words<'a>(words: impl IntoIterator) -> Result { + layer::validate_words(words).map_err(|e| e.to_string()) +} + +/// Which recognition engine to use. The default is ocrs, the pure-Rust engine; "Automatic" is +/// an ensemble — ocrs reading first, Tesseract as a second opinion, merged per the strategy — +/// when both are available. +#[derive(Clone, Copy, Debug, Default, PartialEq, Eq, Hash)] +pub enum EngineChoice { + /// The pure-Rust ocrs engine (the default). + #[default] + Ocrs, + /// The installed `tesseract` binary, as an external process. + Tesseract, + /// Ensemble: the default engine first, the other one as a secondary when available. + Auto, +} + +impl EngineChoice { + /// Parse a choice; unknown names fall back to the default (ocrs), like Acrobat's dialogs + /// fall back for unknown persisted values. + pub fn parse(s: &str) -> EngineChoice { + match s.trim().to_ascii_lowercase().as_str() { + "tesseract" => EngineChoice::Tesseract, + "auto" | "automatic" | "ensemble" => EngineChoice::Auto, + _ => EngineChoice::Ocrs, + } + } + + /// The strategy this choice reads with when none was chosen. Automatic exists to bring in + /// the second engine, so it asks it when the first reading is weak or of unknown + /// confidence ([`MergeStrategy::ConfidenceWeighted`]); a single named engine reads alone. + pub fn default_strategy(self) -> MergeStrategy { + match self { + EngineChoice::Auto => MergeStrategy::ConfidenceWeighted, + EngineChoice::Ocrs | EngineChoice::Tesseract => MergeStrategy::PrimaryOnly, + } + } + + /// The name settings and tools use for this choice. + pub fn name(self) -> &'static str { + match self { + EngineChoice::Ocrs => "ocrs", + EngineChoice::Tesseract => "tesseract", + EngineChoice::Auto => "auto", + } + } +} + +/// What to produce from a recognition. **Searchable** (the default) adds the existing +/// invisible-text layer on the original page; **EditableText** writes a *new* document with the +/// reviewed words as visible text and no scan image (the writer is the Verify screen's; the mode +/// is plumbed through here). Unknown mode strings mean Searchable. +#[derive(Clone, Copy, Debug, Default, PartialEq, Eq, Hash)] +pub enum OutputMode { + #[default] + Searchable, + EditableText, +} + +impl OutputMode { + /// Parse a mode name (trimmed, case-insensitive); unknown strings mean Searchable. + pub fn parse(s: &str) -> OutputMode { + match s.trim().to_ascii_lowercase().as_str() { + "editable" | "editable-text" | "editable_text" | "editabletext" => OutputMode::EditableText, + _ => OutputMode::Searchable, + } + } + + /// The name settings and tools use for this mode. + pub fn name(self) -> &'static str { + match self { + OutputMode::Searchable => "searchable", + OutputMode::EditableText => "editable", + } + } +} + /// Recognize Text settings. #[derive(Clone, Debug, PartialEq)] pub struct OcrSettings { @@ -29,6 +117,40 @@ pub struct OcrSettings { /// that is read, so a scan with a few comments on it is still a scan. `false`: everything /// the page shows. pub native_text_only: bool, + /// Leave whole files that already have text alone (the batch flow's "Skip files that + /// already contain text"; a skipped file is its own result bucket, never an OCR success). + pub skip_text_files: bool, + /// Read pages even when the skip options would leave them alone ("Force OCR (override + /// skip options)"). + pub force_ocr: bool, + /// Which engine reads the pages. + pub engine: EngineChoice, + /// How the engines' readings combine (see [`MergeStrategy`]). + pub strategy: MergeStrategy, + /// Preprocessing, all off by default (a scan is passed through as rendered): + /// straighten pages scanned at 90°/180°/270°. + pub auto_rotate: bool, + /// Preprocessing: straighten skewed text lines. + pub deskew: bool, + /// Preprocessing: 3×3 median filter. + pub denoise: bool, + /// Preprocessing: Sauvola binarization. + pub binarize: bool, + /// What to produce (searchable text layer, or a new editable document). + pub output_mode: OutputMode, + /// Restrict recognition to this rectangle of the rendered page image, in pixels + /// `[x0, y0, x1, y1]` (the Verify screen's "Re-OCR this region"). + pub region: Option<[f32; 4]>, + /// The batch note's separate per-word threshold: words with `0 <= conf <= threshold` are + /// counted for "N low-confidence word(s) on page(s) …" (see [`low_confidence`]; unrelated + /// to the review overlay's bands, which use [`band_for`]). + pub low_confidence_threshold: f32, + /// Where the Tesseract program lives, when the user configured it (a preference — the + /// `PDFCRAFT_TESSERACT` environment variable is the fallback in the engine; never taken + /// from document data). A configured path must be absolute to an existing, executable + /// file; an invalid configuration makes the engine unavailable rather than silently + /// searching the PATH. `None` means the PATH search decides. + pub tesseract_path: Option, } /// A recognised word whose box is at least this much covered by the page's own text is that @@ -56,7 +178,24 @@ fn covered(word: [f32; 4], text: &[[f32; 4]]) -> f32 { impl Default for OcrSettings { fn default() -> Self { - OcrSettings { dpi: 300.0, language: "en".into(), skip_text_pages: true, native_text_only: false } + OcrSettings { + dpi: 300.0, + language: "en".into(), + skip_text_pages: true, + native_text_only: false, + skip_text_files: false, + force_ocr: false, + engine: EngineChoice::default(), + strategy: MergeStrategy::default(), + auto_rotate: false, + deskew: false, + denoise: false, + binarize: false, + output_mode: OutputMode::default(), + region: None, + low_confidence_threshold: 60.0, + tesseract_path: None, + } } } @@ -69,31 +208,406 @@ pub struct OcrPage { pub already_on_page: usize, /// Why the page was not read, if it was skipped. pub skipped: Option, + /// Notes about a partial reading: the words stand, but something on the way did not (an + /// ensemble partner failed and the page keeps the primary engine's reading alone). + pub notes: Vec, } impl OcrPage { pub fn text(&self) -> String { self.words.iter().map(|w| w.text.as_str()).collect::>().join(" ") } -} -/// The recogniser, loaded once (it takes a moment) and shared. -pub fn engine() -> Result, String> { - static OCR: Mutex>> = Mutex::new(None); - let mut slot = OCR.lock().unwrap_or_else(std::sync::PoisonError::into_inner); - if let Some(o) = slot.as_ref() { - return Ok(o.clone()); + /// Mean of the confidences the page's words report (`None` when none do). + pub fn mean_confidence(&self) -> Option { + let (sum, n) = self.words.iter().filter_map(|w| w.confidence).fold((0.0f64, 0usize), |(s, n), c| (s + c as f64, n + 1)); + (n > 0).then_some((sum / n as f64) as f32) + } + + /// The words to flag for review: their confidence is in the Low band ([`band_for`]). + /// Excluded are words the caller wants gone — removed words, words in the user dictionary, + /// skipped words — which is review state, so this reports the indices and the reviewer + /// filters them. + pub fn suspects(&self) -> Vec { + self.words.iter().enumerate().filter(|(_, w)| pdfcraft_ocr::is_suspect(w.confidence)).map(|(i, _)| i).collect() } - let o = Arc::new(Ocr::find().map_err(|e| e.to_string())?); - *slot = Some(o.clone()); - Ok(o) } -/// Whether the recognition models are installed. -pub fn available() -> bool { +/// The engines a job reads with: one primary and, for ensemble strategies, an optional +/// secondary. +pub struct Recognizers { + pub primary: Arc, + pub secondary: Option>, +} + +struct EngineCache { + ocrs: Option>, +} + +/// How long a negative engine probe (no program there) is remembered. A short TTL, not +/// forever: a tesseract installed while the app runs is found by the next job, without a +/// restart. An engine that was found is remembered for the session. +#[cfg(not(target_arch = "wasm32"))] +const NEGATIVE_TTL: Duration = Duration::from_secs(30); + +/// The remembered Tesseract probe: the configured location it answers for, the engine it +/// found (`None` for a probe that failed), and when that probe happened. +#[cfg(not(target_arch = "wasm32"))] +struct CachedTesseract { + key: Option, + found: Option>, + at: Instant, +} + +/// Whether a probe result still counts: a found engine for the session, a failed probe until +/// [`NEGATIVE_TTL`] passes (split out so tests can pass the clock). +#[cfg(not(target_arch = "wasm32"))] +fn probe_fresh(found: bool, at: Instant, now: Instant) -> bool { + found || now.duration_since(at) < NEGATIVE_TTL +} + +/// The remembered Tesseract probe (one slot: the last location asked about). +#[cfg(not(target_arch = "wasm32"))] +static TESSERACT_PROBE: Mutex> = Mutex::new(None); + +/// Whether a background probe, started for the UI by [`tesseract_engine_now`], is running now. +/// At most one runs at a time. +#[cfg(not(target_arch = "wasm32"))] +static PROBING: std::sync::atomic::AtomicBool = std::sync::atomic::AtomicBool::new(false); + +/// The Tesseract engine: found once (it answers `--list-langs` through a process) and then +/// remembered, per configured path; a failed probe is retried after [`NEGATIVE_TTL`]. No +/// static mutex is held while the discovery process runs: the cache is read under the lock, +/// released for the probe, and the result is written back when it arrives (concurrent probes +/// of the same location may both run; the last write wins and both answers are valid). A +/// user-configured path (`settings.tesseract_path`, else the `PDFCRAFT_TESSERACT` +/// environment variable) is validated before any process starts; an invalid configuration is +/// unavailable rather than a PATH fallback. This waits for the probe: the UI thread uses +/// [`tesseract_engine_now`]. +#[cfg(not(target_arch = "wasm32"))] +fn tesseract_engine(configured: Option<&str>) -> Option> { + if let Cached::Fresh(found) = cached_tesseract(configured) { + return found; + } + let key = configured.map(str::to_string); + // The probe (a `--list-langs` process, up to the timeout) runs outside every static mutex. + let found = match configured { + Some(path) => pdfcraft_ocr::tesseract::TesseractCli::at(std::path::PathBuf::from(path)).map(|t| Arc::new(t) as Arc), + None => pdfcraft_ocr::tesseract::TesseractCli::find().map(|t| Arc::new(t) as Arc), + }; + *TESSERACT_PROBE.lock().unwrap_or_else(|e| e.into_inner()) = Some(CachedTesseract { key, found: found.clone(), at: Instant::now() }); + found +} + +/// What the UI may know about the Tesseract probe without waiting for it. +#[cfg(not(target_arch = "wasm32"))] +enum Probe { + /// An answer: the engine, or `None` when there is none. It may be an old one that a + /// background probe is refreshing. + Known(Option>), + /// No answer yet: the first background probe is running (or could not be started). + Pending, +} + +/// What the cache holds for a location. +#[cfg(not(target_arch = "wasm32"))] +enum Cached { + /// A usable answer (see [`probe_fresh`]). + Fresh(Option>), + /// A failed probe older than [`NEGATIVE_TTL`]: still the best known answer, but due again. + Stale(Option>), + /// Nothing for this location. + Missing, +} + +/// [`tesseract_engine`] for the UI thread, which must never wait on a process: the cached +/// answer is returned at once, and when it is missing or due for a retry one background thread +/// runs the probe (it fills the cache). The answer is [`Probe::Pending`] only until the first +/// probe of a location lands, so a retry never makes the choice flicker. +#[cfg(not(target_arch = "wasm32"))] +fn tesseract_engine_now(configured: Option<&str>) -> Probe { + let known = match cached_tesseract(configured) { + Cached::Fresh(found) => return Probe::Known(found), + Cached::Stale(found) => Some(found), + Cached::Missing => None, + }; + if !PROBING.swap(true, std::sync::atomic::Ordering::AcqRel) { + let configured = configured.map(str::to_string); + let started = std::thread::Builder::new().name("pdfcraft-tesseract-probe".into()).spawn(move || { + let _ = tesseract_engine(configured.as_deref()); + PROBING.store(false, std::sync::atomic::Ordering::Release); + }); + if started.is_err() { + PROBING.store(false, std::sync::atomic::Ordering::Release); + } + } + known.map_or(Probe::Pending, Probe::Known) +} + +/// Whether the Tesseract probe is running in the background right now. The interface asks it +/// while it shows an engine choice that read as "still checking", to look again soon. +pub fn engine_probe_running() -> bool { + #[cfg(not(target_arch = "wasm32"))] + { + PROBING.load(std::sync::atomic::Ordering::Acquire) + } + #[cfg(target_arch = "wasm32")] + { + false + } +} + +/// What the cache holds for `configured`. Never starts a process. +#[cfg(not(target_arch = "wasm32"))] +fn cached_tesseract(configured: Option<&str>) -> Cached { + let slot = TESSERACT_PROBE.lock().unwrap_or_else(|e| e.into_inner()); + cache_lookup(slot.as_ref(), configured, Instant::now()) +} + +/// [`cached_tesseract`] over an explicit slot and clock (so tests can pass both). +#[cfg(not(target_arch = "wasm32"))] +fn cache_lookup(slot: Option<&CachedTesseract>, configured: Option<&str>, now: Instant) -> Cached { + match slot { + Some(c) if c.key.as_deref() == configured => { + if probe_fresh(c.found.is_some(), c.at, now) { + Cached::Fresh(c.found.clone()) + } else { + Cached::Stale(c.found.clone()) + } + } + _ => Cached::Missing, + } +} + +/// Whether the ocrs models are installed, remembered like the Tesseract probe (found: for the +/// session; missing: for [`NEGATIVE_TTL`]), so per-frame UI checks do not search the disk. +#[cfg(not(target_arch = "wasm32"))] +fn models_installed_cached() -> bool { + static MODELS: Mutex> = Mutex::new(None); + let mut slot = MODELS.lock().unwrap_or_else(|e| e.into_inner()); + if let Some((found, at)) = *slot + && probe_fresh(found, at, Instant::now()) + { + return found; + } + let found = Models::find().is_some(); + *slot = Some((found, Instant::now())); + found +} + +/// The web build has no disk search worth caching. +#[cfg(target_arch = "wasm32")] +fn models_installed_cached() -> bool { Models::find().is_some() } +/// The program's `--version` line, remembered per program path for the session (`ocr_status` +/// asks every time it is called; the version of an installed binary does not change under a +/// running app often enough to start a process each time). +#[cfg(not(target_arch = "wasm32"))] +fn tesseract_version(program: &str) -> Option { + static VERSIONS: Mutex)>> = Mutex::new(Vec::new()); + if let Some((_, v)) = VERSIONS.lock().unwrap_or_else(|e| e.into_inner()).iter().find(|(p, _)| p == program) { + return v.clone(); + } + // Validated (absolute, existing, executable) before the process starts; run outside the lock. + let path = std::path::PathBuf::from(program); + let version = pdfcraft_ocr::tesseract::program_is_usable(&path).then(|| pdfcraft_ocr::tesseract::TesseractCli::new(path).version()).flatten(); + let mut cache = VERSIONS.lock().unwrap_or_else(|e| e.into_inner()); + if cache.len() < 16 && !cache.iter().any(|(p, _)| p == program) { + cache.push((program.to_string(), version.clone())); + } + version +} + +/// The Tesseract program the engine would run, named for `ocr_status`: the user-configured +/// location when one is set, else what the PATH search resolves to. This function itself +/// starts no process (the PATH search only looks at files), so a path here says where the +/// engine looks, not that a program answers there; [`engines_status`], which reports it, does +/// run the cached `--list-langs` probe and a cached `--version`. `None` when nothing is +/// configured and the PATH holds no `tesseract` (or, on the web, when there is no process to +/// run). +#[cfg(not(target_arch = "wasm32"))] +pub fn tesseract_program_path(configured: Option<&str>) -> Option { + match configured { + Some(p) => Some(p.to_string()), + None => pdfcraft_ocr::tesseract::path_search().map(|p| p.to_string_lossy().into_owned()), + } +} + +/// One engine's status, for `ocr_status`: whether it can run, why not when it cannot, and for +/// Tesseract the program it would run and that program's version (both `None` on the web). +#[derive(Clone, Debug, PartialEq)] +pub struct EngineStatus { + pub id: &'static str, + pub available: bool, + pub why_not: Option, + pub program: Option, + pub version: Option, +} + +/// The engines' status, in the order the Engine dropdown lists them: ocrs (built in), then +/// tesseract (the external program). Tesseract's availability is judged the same way +/// [`recognizers`] judges it under the default settings: the `PDFCRAFT_TESSERACT` environment +/// variable when set, else the PATH. This waits for the (cached) Tesseract probe; the tools +/// and tests call it, never the UI thread. +pub fn engines_status() -> Vec { + engines_status_for(&OcrSettings::default()) +} + +/// [`engines_status`] for the location a job's `settings` name: `settings.tesseract_path` when +/// set, else the environment variable, else the PATH, as [`recognizers`] decides it. (No tool +/// takes a location, because an agent must not choose which program runs; the tools report the +/// default settings.) +pub fn engines_status_for(settings: &OcrSettings) -> Vec { + let ocrs = Models::find().is_some(); + let mut out = vec![EngineStatus { id: "ocrs", available: ocrs, why_not: (!ocrs).then(models_missing), program: None, version: None }]; + #[cfg(not(target_arch = "wasm32"))] + { + let configured = configured_tesseract(settings); + let configured = configured.as_deref(); + let available = tesseract_engine(configured).is_some(); + let program = tesseract_program_path(configured); + // The version is a `--version` run against the validated program path, remembered per + // path; a program that answers `--list-langs` but not `--version` still counts as + // available. + let version = if available { program.as_deref().and_then(tesseract_version) } else { None }; + let why_not = (!available).then(|| match configured { + Some(p) => format!("the configured tesseract location is not an absolute path to an existing, executable file: {p}"), + None => tesseract_unavailable(), + }); + out.push(EngineStatus { id: "tesseract", available, why_not, program, version }); + } + #[cfg(target_arch = "wasm32")] + { + let _ = settings; + out.push(EngineStatus { id: "tesseract", available: false, why_not: Some(tesseract_unavailable()), program: None, version: None }); + } + out +} + +/// The web build never spawns processes, so it never has Tesseract. +#[cfg(target_arch = "wasm32")] +fn tesseract_engine(configured: Option<&str>) -> Option> { + let _ = configured; + None +} + +/// The configured Tesseract location: the user preference wins; the environment variable is +/// the fallback (a preference file rarely survives onto another machine). Both are user +/// choices — this is never taken from document data. +fn configured_tesseract(settings: &OcrSettings) -> Option { + settings.tesseract_path.clone().or_else(|| std::env::var_os("PDFCRAFT_TESSERACT").map(|v| v.to_string_lossy().into_owned())) +} + +/// The engine selector: builds each engine once and hands out the pair `settings.engine` +/// asks for. "Automatic" is an ensemble when both engines are available; a named engine that +/// cannot start fails with its cause named. +pub fn recognizers(settings: &OcrSettings) -> Result { + // The Tesseract probe (a `--list-langs` process) runs before any static mutex is taken, so + // recognition on other threads is never blocked behind someone's discovery. A job that + // names its engines never starts it: only Tesseract and Automatic read anything from it, + // so an ocrs-only job pays no process and no wait. + let configured = configured_tesseract(settings); + let tesseract = + if matches!(settings.engine, EngineChoice::Tesseract | EngineChoice::Auto) { tesseract_engine(configured.as_deref()) } else { None }; + static ENGINES: Mutex> = Mutex::new(None); + let mut slot = ENGINES.lock().unwrap_or_else(|e| e.into_inner()); + let cache = slot.get_or_insert_with(|| EngineCache { ocrs: None }); + if cache.ocrs.is_none() { + // Cheap to re-attempt while the models are missing; expensive work is cached forever. + cache.ocrs = pdfcraft_ocr::OcrsRecognizer::find().ok().map(|r| Arc::new(r) as Arc); + } + let ocrs = cache.ocrs.clone(); + match settings.engine { + EngineChoice::Ocrs => Ok(Recognizers { primary: ocrs.ok_or_else(models_missing)?, secondary: None }), + EngineChoice::Tesseract => Ok(Recognizers { primary: tesseract.ok_or_else(tesseract_unavailable)?, secondary: None }), + EngineChoice::Auto => match (ocrs, tesseract) { + (Some(p), s) => Ok(Recognizers { primary: p, secondary: s }), + (None, Some(p)) => Ok(Recognizers { primary: p, secondary: None }), + (None, None) => Err(models_missing()), + }, + } +} + +/// Why Tesseract cannot run, named for the user. +#[cfg(not(target_arch = "wasm32"))] +fn tesseract_unavailable() -> String { + "the tesseract program was not found (install it, or add it to the PATH)".into() +} + +/// Why Tesseract cannot run, named for the user (the web build has no processes). +#[cfg(target_arch = "wasm32")] +fn tesseract_unavailable() -> String { + "the tesseract engine is not available in the web build".into() +} + +/// Why the ocrs engine cannot run, in the end-user wording the UI catalogs translate (the +/// same text as [`pdfcraft_ocr::OcrError::NoModels`]: no developer commands in front of users). +fn models_missing() -> String { + pdfcraft_ocr::OcrError::NoModels.to_string() +} + +/// Whether recognition can run at all: the ocrs models are installed, or (off the web) a +/// Tesseract program is found. This may run the Tesseract probe (a process, cached) and so +/// waits for it: tools and tests call it; the UI asks [`engine_unavailable_reason`], which +/// never waits. +pub fn available() -> bool { + let configured = std::env::var_os("PDFCRAFT_TESSERACT").map(|v| v.to_string_lossy().into_owned()); + Models::find().is_some() || tesseract_engine(configured.as_deref()).is_some() +} + +/// Why `engine` cannot run right now, named for the UI's tooltips (the Engine dropdown +/// disables the choice with this text); `None` when it can. The app UI has no Tesseract +/// location field of its own, so this judges the environment variable, exactly like +/// [`available`] (no tool takes a per-job location either; `OcrSettings::tesseract_path` is +/// for library callers, and [`recognizers`] validates it on its own). +/// +/// Called on the UI thread every frame, so it never waits on a process: the answers are +/// cached ([`NEGATIVE_TTL`] for a missing engine, refreshed in the background), and until the +/// first Tesseract probe lands the choice reads as "still checking" (see +/// [`engine_probe_running`], to ask for another frame meanwhile). +pub fn engine_unavailable_reason(engine: EngineChoice) -> Option { + let ocrs = models_installed_cached(); + #[cfg(not(target_arch = "wasm32"))] + let tesseract = { + let configured = std::env::var_os("PDFCRAFT_TESSERACT").map(|v| v.to_string_lossy().into_owned()); + match tesseract_engine_now(configured.as_deref()) { + Probe::Known(Some(_)) => Ok(()), + Probe::Known(None) => Err(tesseract_unavailable()), + Probe::Pending => Err(tesseract_checking()), + } + }; + #[cfg(target_arch = "wasm32")] + let tesseract: Result<(), String> = Err(tesseract_unavailable()); + match engine { + EngineChoice::Ocrs => (!ocrs).then(models_missing), + EngineChoice::Tesseract => tesseract.err(), + EngineChoice::Auto => match (ocrs, tesseract) { + (true, _) | (false, Ok(())) => None, + (false, Err(why)) => Some(format!("{}; {why}", models_missing())), + }, + } +} + +/// Why Tesseract cannot be chosen yet: its probe is still running in the background. +#[cfg(not(target_arch = "wasm32"))] +fn tesseract_checking() -> String { + "still checking whether the tesseract program can run; try again in a moment".into() +} + +/// What a started job expects of the document when its result is applied: the same open +/// document, unchanged. [`Session::stale_reason`] names any difference it finds. +#[derive(Clone)] +pub struct OcrGuard { + pub(crate) doc: DocId, + pub(crate) pages: usize, + pub(crate) revisions: usize, + /// The document's working bytes when the job started; any edit rewrites them, so a shared + /// `Arc` pointer means "untouched". + pub(crate) bytes: Arc>, +} + /// Everything recognition needs from a document, detached from the session. pub struct OcrJob { bytes: Arc>, @@ -104,40 +618,162 @@ pub struct OcrJob { native: Vec>, pub pages: Vec, pub settings: OcrSettings, + /// What to check before applying this job's result. + pub guard: OcrGuard, +} + +/// A page's "has text" probe for skip-text: at least one non-blank text run. A failed probe +/// (the text layer could not be read) never enables a skip. +fn page_has_text(text: Option<&Arc>) -> bool { + text.is_some_and(|t| t.glyphs.iter().any(|g| !g.text.trim().is_empty())) +} + +/// Clamp a region to the image, snapping outward to whole pixels. `None` when nothing of it +/// is inside the image (empty, inverted, NaN, or wholly outside). +pub fn clamp_region(region: [f32; 4], width: u32, height: u32) -> Option<[u32; 4]> { + let (w, h) = (f64::from(width), f64::from(height)); + if region.iter().any(|v| !v.is_finite()) { + return None; + } + let (x0, y0, x1, y1) = (region[0] as f64, region[1] as f64, region[2] as f64, region[3] as f64); + let (x0, y0) = ((x0.floor().max(0.0).min(w)) as u32, (y0.floor().max(0.0).min(h)) as u32); + let (x1, y1) = ((x1.ceil().max(0.0).min(w)) as u32, (y1.ceil().max(0.0).min(h)) as u32); + (x1 > x0 && y1 > y0).then_some([x0, y0, x1, y1]) +} + +/// Map a recognized word from its source page's user space into the editable output's page +/// user space (the new page is the source's displayed size, unrotated): user → view → new +/// user. The box never moves relative to what the reviewer saw. +fn editable_word(w: &PlacedWord, info: &PageInfo) -> PlacedWord { + let to_new = |p: [f64; 2]| { + let v = info.user_to_view(p[0] as f32, p[1] as f32); + [f64::from(v[0]), f64::from(info.height) - f64::from(v[1])] + }; + let origin = to_new(w.origin); + let br = to_new([w.origin[0] + w.across[0], w.origin[1] + w.across[1]]); + let tl = to_new([w.origin[0] + w.up[0], w.origin[1] + w.up[1]]); + PlacedWord { + text: w.text.clone(), + origin, + across: [br[0] - origin[0], br[1] - origin[1]], + up: [tl[0] - origin[0], tl[1] - origin[1]], + ..w.clone() + } } impl OcrJob { + /// The pixels-per-point this job renders `page` at (the real render dpi is this times 72). + /// The Verify screen uses it to express a region picked on its own preview in the pixels + /// the job will actually read. A scan is read at its own resolution (enlarging it only + /// blurs the letters), at most the chosen one. + pub fn render_scale(&self, page: usize) -> f32 { + let wanted = self.settings.dpi.clamp(72.0, 600.0) / 72.0; + let Some(native) = self.native.get(page).copied().flatten() else { return wanted }; + // A scan whose page is larger than a sheet (often a box given in pixels, one point per + // pixel, #1848) is measured as if shrunk to one: otherwise 1 px/pt is already its full + // image, and no "Downsample to" choice could ask for less. + let fit = self.infos.get(page).map_or(1.0, |i| (SHEET / i.width.max(i.height).max(1e-3)).min(1.0)); + native.clamp(fit, wanted * fit) + } + + /// Whether any of the job's pages already has text (at least one non-blank run): the + /// batch flow's "Skip files that already contain text" probe. A failed probe never + /// enables a skip, so a page whose text cannot be read counts as no text there. + pub fn has_text(&self) -> bool { + let mut r = PageRenderer::new(self.bytes.clone(), RenderConfig { password: self.password.as_deref().map(Arc::from), ..Default::default() }); + self.pages.iter().any(|&page| { + let t = r.render(RenderRequest { page, kind: RequestKind::Text, scale: 1.0, ..Default::default() }); + page_has_text(t.text.as_ref()) + }) + } + /// Read the pages. `progress(done, total)` is called before each page; returning `false` /// stops (the pages read so far are returned). - pub fn run(self, ocr: &Ocr, mut progress: impl FnMut(usize, usize) -> bool) -> Vec { + pub fn run(self, recognizers: &Recognizers, mut progress: impl FnMut(usize, usize) -> bool) -> Vec { let config = RenderConfig { password: self.password.as_deref().map(Arc::from), hide_comments: self.settings.native_text_only, ..Default::default() }; let mut r = PageRenderer::new(self.bytes.clone(), config); let mut out = Vec::new(); let total = self.pages.len(); + let options = pdfcraft_ocr::RecognizeOptions { language: self.settings.language.clone() }; for (done, &page) in self.pages.iter().enumerate() { if !progress(done, total) { break; } let Some(info) = self.infos.get(page) else { continue }; - let skip = |why: &str| OcrPage { page, words: Vec::new(), already_on_page: 0, skipped: Some(why.into()) }; + let skip = |why: &str| OcrPage { page, words: Vec::new(), already_on_page: 0, skipped: Some(why.into()), notes: Vec::new() }; // The boxes of the page's own text (an earlier OCR layer's included), in view space. let existing: Vec<[f32; 4]> = r .render(RenderRequest { page, kind: RequestKind::Text, scale: 1.0, ..Default::default() }) .text .map(|t| t.glyphs.iter().filter(|g| !g.text.trim().is_empty()).map(|g| g.rect).collect()) .unwrap_or_default(); - if self.settings.skip_text_pages && !existing.is_empty() { + if self.settings.skip_text_pages && !self.settings.force_ocr && !existing.is_empty() { out.push(skip("the page already contains text")); continue; } - let shot = r.render(RenderRequest { page, scale: self.scale(page, info), ..Default::default() }); + // A scan is read at its own resolution (enlarging it only blurs the letters), at + // most the chosen one. + let scale = self.render_scale(page); + let shot = r.render(RenderRequest { page, scale, ..Default::default() }); if let Some(e) = shot.error { out.push(skip(&e)); continue; } - let lines = match ocr.recognize(&shot.rgba, shot.width, shot.height) { - Ok(l) => l, + let full = match pdfcraft_ocr::OcrImage::new(shot.width, shot.height, shot.rgba.to_vec()) { + Ok(img) => img, + Err(e) => { + out.push(skip(&e.to_string())); + continue; + } + }; + // The optional region: crop the rendered page, clamped; an empty region skips the + // page with a named reason (nothing was asked of it). + let (image, offset) = match self.settings.region { + None => (full, [0.0f32; 2]), + Some(region) => match clamp_region(region, shot.width, shot.height) { + Some([x0, y0, x1, y1]) => match crop(&full, x0, y0, x1, y1) { + Some(c) => (c, [x0 as f32, y0 as f32]), + None => { + out.push(skip("the selected region is empty")); + continue; + } + }, + None => { + out.push(skip("the selected region is outside the page")); + continue; + } + }, + }; + // Preprocessing sits between the page raster and recognition; its word boxes come + // back in processed coordinates and are mapped to the ORIGINAL raster's before + // placement. The render dpi is the real one (pixels per point × 72), never a + // fixed value. + let pre = pdfcraft_ocr::preprocess::preprocess( + image, + scale * 72.0, + &pdfcraft_ocr::preprocess::PreprocessOptions { + auto_rotate: self.settings.auto_rotate, + deskew: self.settings.deskew, + denoise: self.settings.denoise, + binarize: self.settings.binarize, + }, + ); + let (image, inverse) = match pre { + Ok(p) => (p.image, p.inverse_transform), + Err(e) => { + out.push(skip(&e.to_string())); + continue; + } + }; + let outcome = match pdfcraft_ocr::recognize_with_strategy( + recognizers.primary.as_ref(), + recognizers.secondary.as_deref(), + self.settings.strategy, + &image, + &options, + ) { + Ok(o) => o, Err(e) => { out.push(skip(&e.to_string())); continue; @@ -145,30 +781,69 @@ impl OcrJob { }; let scale = shot.width as f32 / info.width.max(1e-3); let to_user = |x: f32, y: f32| { - let [u, v] = info.view_to_user(x / scale, y / scale); + // processed pixels → original raster pixels (the exact inverse) → past the + // region crop → page view space. + let (ox, oy) = inverse.apply(f64::from(x), f64::from(y)); + let [u, v] = info.view_to_user((ox as f32 + offset[0]) / scale, (oy as f32 + offset[1]) / scale); [u as f64, v as f64] }; - // Words the page's own text already has are not added a second time. - let (fresh, already): (Vec<_>, Vec<_>) = - lines.iter().flat_map(|l| &l.words).partition(|w| covered(w.rect.map(|v| v / scale), &existing) < ALREADY_ON_PAGE); - let words = fresh.into_iter().map(|w| PlacedWord::place(w, to_user)).collect(); - out.push(OcrPage { page, words, already_on_page: already.len(), skipped: None }); + // Validate recognition geometry and compare against existing text in view space, + // undoing preprocessing and region offsets before testing overlap. + let (fresh, already): (Vec<_>, Vec<_>) = outcome.lines.iter().flat_map(|l| &l.words).filter(|w| box_writable(w)).partition(|w| { + let [l, t, r, b] = w.rect; + let bounds = [(l, t), (r, t), (r, b), (l, b)].into_iter().fold( + [f32::INFINITY, f32::INFINITY, f32::NEG_INFINITY, f32::NEG_INFINITY], + |mut bounds, (x, y)| { + let (x, y) = inverse.apply(f64::from(x), f64::from(y)); + let x = (x as f32 + offset[0]) / scale; + let y = (y as f32 + offset[1]) / scale; + bounds[0] = bounds[0].min(x); + bounds[1] = bounds[1].min(y); + bounds[2] = bounds[2].max(x); + bounds[3] = bounds[3].max(y); + bounds + }, + ); + covered(bounds, &existing) < ALREADY_ON_PAGE + }); + let words = fresh.into_iter().map(|w| PlacedWord::place(w, to_user)).collect::>(); + let mut notes = outcome.notes; + // Only visible editable output needs display glyphs. Searchable output preserves + // every Unicode character with empty Type3 programs and ToUnicode maps. + if self.settings.output_mode == OutputMode::EditableText && !words.is_empty() { + let missing = pdfcraft_fonts::TextPlan::for_words(words.iter().map(|w| w.text.as_str())).missing(); + if missing > 0 { + notes.push(format!("{missing} characters had no font that could show them and appear as ? in the output")); + } + } + out.push(OcrPage { page, words, already_on_page: already.len(), skipped: None, notes }); } progress(total, total); out } +} - /// The render scale (pixels per point) `page` is read at. A scan is read at its own - /// resolution (enlarging it only blurs the letters), at most the chosen one. - fn scale(&self, page: usize, info: &PageInfo) -> f32 { - let wanted = self.settings.dpi.clamp(72.0, 600.0) / 72.0; - let Some(native) = self.native.get(page).copied().flatten() else { return wanted }; - // A scan whose page is larger than a sheet (often a box given in pixels, one point per - // pixel, #1848) is measured as if shrunk to one: otherwise 1 px/pt is already its full - // image, and no "Downsample to" choice could ask for less. - let fit = (SHEET / info.width.max(info.height).max(1e-3)).min(1.0); - native.clamp(fit, wanted * fit) +/// Whether a word's pixel box may reach the content stream: width and height (right − left, +/// bottom − top) must be positive and finite. Anything else — a zero, negative, NaN or +/// infinite size — is dropped before placement rather than written into the page. +fn box_writable(word: &pdfcraft_ocr::Word) -> bool { + let [l, t, r, b] = word.rect; + let (width, height) = (r - l, b - t); + width > 0.0 && height > 0.0 && width.is_finite() && height.is_finite() +} + +/// Crop `[x0, y0, x1, y1]` out of `image` as a new [`OcrImage`] (the sides are already clamped +/// by [`clamp_region`]). +fn crop(image: &pdfcraft_ocr::OcrImage, x0: u32, y0: u32, x1: u32, y1: u32) -> Option { + let (w, h) = (x1.checked_sub(x0)?, y1.checked_sub(y0)?); + let stride = image.width() as usize * 4; + let mut out = Vec::with_capacity((w as usize).checked_mul(h as usize)?.checked_mul(4)?); + for y in y0..y1 { + let start = y as usize * stride + x0 as usize * 4; + let row = image.rgba().get(start..start + (x1 - x0) as usize * 4)?; + out.extend_from_slice(row); } + pdfcraft_ocr::OcrImage::new(w, h, out).ok() } impl Session { @@ -193,7 +868,8 @@ impl Session { .reduce(f32::max) }) .collect(); - Some(OcrJob { bytes: doc.bytes.clone(), password: doc.password.clone(), infos: doc.info.pages.clone(), native, pages, settings }) + let guard = OcrGuard { doc: id, pages: all, revisions: doc.revision_ends().len(), bytes: doc.bytes.clone() }; + Some(OcrJob { bytes: doc.bytes.clone(), password: doc.password.clone(), infos: doc.info.pages.clone(), native, pages, settings, guard }) } /// Add the words found by [`OcrJob::run`] as one undoable step. Returns the number of words. @@ -207,14 +883,129 @@ impl Session { Ok(words) } + /// Why a job's result may not be applied to this session any more: the document was closed, + /// edited, re-saved, or its pages changed since the job started. `None` means it is safe. + pub fn stale_reason(&self, guard: &OcrGuard) -> Option { + let Some(doc) = self.get(guard.doc) else { + return Some("the document is no longer open".into()); + }; + if !Arc::ptr_eq(&doc.bytes, &guard.bytes) { + return Some("the document changed since recognition started".into()); + } + let now = doc.info.pages.len(); + if now != guard.pages { + return Some(format!("the page count changed from {} to {} since recognition started", guard.pages, now)); + } + let revisions = doc.revision_ends().len(); + if revisions != guard.revisions { + let had = guard.revisions; + return Some(format!("the document gained revisions ({had} → {revisions}) since recognition started")); + } + None + } + + /// What applying a result to document `id` expects of it right now. After a checked apply + /// succeeds the document has changed (that apply is its newest edit), so a caller that + /// applies more of the same review in further steps (page by page) takes a fresh guard + /// here; any other edit in between still makes the next apply stale. + pub fn ocr_guard(&self, id: DocId) -> Option { + let doc = self.get(id)?; + Some(OcrGuard { doc: id, pages: doc.info.pages.len(), revisions: doc.revision_ends().len(), bytes: doc.bytes.clone() }) + } + + /// [`Self::apply_ocr`], but only while the document is still the one that was read; a stale + /// result is refused with a named reason and nothing is written. + pub fn apply_ocr_checked(&mut self, id: DocId, found: &[OcrPage], guard: &OcrGuard) -> Result { + if guard.doc != id { + return Err("the result belongs to another document".into()); + } + if let Some(why) = self.stale_reason(guard) { + return Err(why); + } + self.apply_ocr(id, found).map_err(|e| e.to_string()) + } + + /// The editable-text output: a NEW document with the recognized words as VISIBLE text at + /// their original boxes and no scan image — one page per source page, blank where nothing + /// was recognized, each page the displayed size of its source. The open document is never + /// touched; the caller chooses where the bytes are written. Refuses when there are no + /// words at all (an empty document would silently lose the scan's content). + pub fn editable_ocr_bytes(&self, id: DocId, found: &[OcrPage]) -> Result>, String> { + let doc = self.get(id).ok_or("no such document")?; + let words = layer::validate_words(found.iter().flat_map(|p| &p.words)).map_err(|e| e.to_string())?; + if words == 0 { + return Err("the recognition found no words, so there is no editable text to write".into()); + } + // The words per page, in user space; and one plan over all of them, so a face's Type3 + // font is built once and shared by every page that uses it. + let page_words: Vec> = (0..doc.info.pages.len()) + .map(|p| { + let Some(info) = doc.info.pages.get(p) else { return Vec::new() }; + found.iter().find(|f| f.page == p).map(|f| f.words.iter().map(|w| editable_word(w, info)).collect()).unwrap_or_default() + }) + .collect(); + layer::validate_words(page_words.iter().flatten()).map_err(|e| e.to_string())?; + let mut plan = pdfcraft_fonts::TextPlan::for_words(page_words.iter().flat_map(|ws| ws.iter().map(|w| w.text.as_str()))); + let pages: Vec<(f64, f64, pdfcraft_cos::Dict, Vec)> = (0..doc.info.pages.len()) + .map(|p| { + let info = doc.info.pages.get(p); + (info.map_or(612.0, |i| f64::from(i.width)), info.map_or(792.0, |i| f64::from(i.height)), pdfcraft_cos::Dict::new(), Vec::new()) + }) + .collect(); + let mut created = pdfcraft_create::from_contents(&pages).map_err(|e| e.to_string())?; + // One shared Helvetica for every page: added once into the new document, then linked + // from each page's resources as /PCHelv (the name the text layers draw with). The + // plan's fonts are linked per page too; their objects are shared. + let mut font = pdfcraft_cos::Dict::new(); + font.set(b"Type".to_vec(), pdfcraft_cos::Object::name("Font")); + font.set(b"Subtype".to_vec(), pdfcraft_cos::Object::name("Type1")); + font.set(b"BaseFont".to_vec(), pdfcraft_cos::Object::name("Helvetica")); + font.set(b"Encoding".to_vec(), pdfcraft_cos::Object::name("WinAnsiEncoding")); + let fr = created.add(font); + // The page tree's kids, direct array or indirect reference (a fresh create document + // keeps /Kids inline) — the output's pages are linked through them. + let kids = created + .root() + .and_then(|r| created.get(r).as_dict().cloned()) + .and_then(|d| d.reference(b"Pages")) + .and_then(|p| created.get(p).as_dict().cloned()) + .and_then(|d| match d.get(b"Kids").map(|k| created.resolve(k)).as_deref() { + Some(pdfcraft_cos::Object::Array(a)) => Some(a.clone()), + Some(pdfcraft_cos::Object::Ref(r)) => created.get(*r).as_array().cloned(), + _ => None, + }); + for (pi, kid) in kids.into_iter().flatten().enumerate() { + let pdfcraft_cos::Object::Ref(page) = kid else { continue }; + let _ = created.update_dict(page, |d| { + let mut res = d.get(b"Resources").and_then(|o| o.as_dict().cloned()).unwrap_or_default(); + let mut fonts = res.get(b"Font").and_then(|o| o.as_dict().cloned()).unwrap_or_default(); + fonts.set(b"PCHelv".to_vec(), pdfcraft_cos::Object::Ref(fr)); + res.set(b"Font".to_vec(), pdfcraft_cos::Object::Dict(fonts)); + d.set(b"Resources".to_vec(), pdfcraft_cos::Object::Dict(res)); + }); + // The page's content is written after the fonts carry their resource names, so + // the runs can reference them. + if page_words.get(pi).is_some_and(|ws| !ws.is_empty()) { + pdfcraft_edit::add_type3_fonts(&mut created, pi, &mut plan).map_err(|e| e.to_string())?; + let word_offset = page_words.iter().take(pi).map(Vec::len).sum(); + let content = pdfcraft_ocr::visible_text_layer_at(&page_words[pi], &plan, word_offset); + let mut dict = pdfcraft_cos::Dict::new(); + dict.set(b"PCMark".to_vec(), pdfcraft_cos::Object::name("OCR")); + let stream = created.add(pdfcraft_cos::Object::Stream(pdfcraft_cos::Stream::flate(dict, &content))); + let _ = created.update_dict(page, |d| d.set(b"Contents".to_vec(), pdfcraft_cos::Object::Ref(stream))); + } + } + self.write_new(&created).map_err(|e| e.to_string()) + } + /// Recognise text on `pages` and add it, in one call (the CLI and agents use this). pub fn recognize_text(&mut self, id: DocId, pages: &[usize], settings: OcrSettings) -> Result, String> { let job = self.ocr_job(id, pages, settings).ok_or("no such document")?; if let Some(why) = self.get(id).and_then(|d| d.read_only_reason.clone()) { return Err(why); } - let ocr = engine()?; - let found = job.run(&ocr, |_, _| true); + let recognizers = recognizers(&job.settings)?; + let found = job.run(&recognizers, |_, _| true); self.apply_ocr(id, &found).map_err(|e| e.to_string())?; Ok(found) } @@ -226,22 +1017,68 @@ pub struct FileResult { /// The new file (incrementally saved), or the original when nothing was recognised. pub bytes: Arc>, pub pages: Vec, + /// Why the whole file was left alone ("Skip files that already contain text"): a skipped + /// file is its own result bucket, never counted as an OCR success. + pub file_skipped: Option, } impl FileResult { pub fn words(&self) -> usize { self.pages.iter().map(|p| p.words.len()).sum() } + + /// Whether the file was passed over entirely (never an OCR success). + pub fn skipped(&self) -> bool { + self.file_skipped.is_some() + } +} + +/// The output names a multi-file run would write more than once: the batch writes every +/// result into one folder under its input's file name, so two inputs of the same name from +/// different folders would overwrite each other. Compared case-insensitively (the Windows and +/// macOS file systems are), in first-seen order, each listed once. +pub fn duplicate_output_names<'a>(names: impl IntoIterator) -> Vec { + let mut seen = std::collections::HashSet::new(); + let mut listed = std::collections::HashSet::new(); + let mut out = Vec::new(); + for name in names { + let key = name.to_lowercase(); + if !seen.insert(key.clone()) && listed.insert(key) { + out.push(name.to_string()); + } + } + out +} + +/// The batch note's data: how many words sit at or under `threshold` ([`low_confidence`], the +/// separate batch rule), and the 1-based, sorted pages they are on. `None` when there are +/// none, so a note is only written when there is something to review. +pub fn low_confidence_pages(pages: &[OcrPage], threshold: f32) -> Option<(usize, Vec)> { + let (mut count, mut page_numbers) = (0usize, Vec::new()); + for p in pages { + if p.skipped.is_some() { + continue; + } + let on_page = p.words.iter().filter(|w| w.confidence.is_some_and(|c| pdfcraft_ocr::low_confidence(c, threshold))).count(); + if on_page > 0 { + count += on_page; + page_numbers.push(p.page + 1); + } + } + page_numbers.sort_unstable(); + (count > 0).then_some((count, page_numbers)) } /// Recognize text in multiple files: one PDF's bytes in, the searchable PDF out. `progress` -/// works as for [`OcrJob::run`]. +/// works as for [`OcrJob::run`]. A file that the skip-files option passes over comes back with +/// [`FileResult::file_skipped`] set and its original bytes; the caller decides what to do with +/// it, but it is never an OCR success. pub fn recognize_file( name: &str, bytes: Arc>, password: Option<&str>, settings: OcrSettings, - ocr: &Ocr, + recognizers: &Recognizers, progress: impl FnMut(usize, usize) -> bool, ) -> Result { let mut s = Session::new(); @@ -249,15 +1086,31 @@ pub fn recognize_file( if let Some(why) = s.get(id).and_then(|d| d.read_only_reason.clone()) { return Err(why); } - let job = s.ocr_job(id, &[], settings).ok_or("the document could not be read")?; - let pages = job.run(ocr, progress); + let job = s.ocr_job(id, &[], settings.clone()).ok_or("the document could not be read")?; + // The skip-files probe runs before any engine work: a file with text needs no engine at + // all, and the original bytes come back unchanged. + if settings.skip_text_files && !settings.force_ocr && job.has_text() { + let pages = job + .pages + .iter() + .map(|&page| OcrPage { + already_on_page: 0, + page, + words: Vec::new(), + skipped: Some("the file already contains text".into()), + notes: Vec::new(), + }) + .collect(); + return Ok(FileResult { bytes, pages, file_skipped: Some("the file already contains text".into()) }); + } + let pages = job.run(recognizers, progress); if s.apply_ocr(id, &pages).map_err(|e| e.to_string())? == 0 { - return Ok(FileResult { bytes, pages }); + return Ok(FileResult { bytes, pages, file_skipped: None }); } // Checked like a save before the caller writes it (#2166). let bytes = s.save_bytes(id).map_err(|e| e.to_string())?; s.check_save(id, &bytes).map_err(|e| e.to_string())?; - Ok(FileResult { bytes, pages }) + Ok(FileResult { bytes, pages, file_skipped: None }) } #[cfg(test)] @@ -282,28 +1135,6 @@ mod tests { assert_eq!(covered([100.5, 52.0, 134.0, 61.0], &[]), 0.0); } - /// #1848: an A4 scan at 300 dpi whose page box is its pixel size (2480 x 3508 pt, one pixel - /// per point) gets a smaller bitmap for each lower "Downsample to" choice, and a low-resolution - /// letter page is still never shrunk below one pixel per point. - #[test] - fn downsample_shrinks_a_page_box_given_in_pixels() { - let mut s = Session::new(); - let page = |w: u32, h: u32| export::encode_png(w, h, &vec![255; (w * h * 4) as usize]).unwrap(); - let bytes = s.create_from_images(&[("a4.png".into(), page(2480, 3508)), ("letter.png".into(), page(612, 792))]).unwrap(); - let id = s.open("scan.pdf", None, bytes, None).unwrap(); - let mut r = PageRenderer::new(s.get(id).unwrap().bytes.clone(), RenderConfig::default()); - for (dpi, a4) in [(72.0, [595, 842]), (150.0, [1240, 1754]), (300.0, [2480, 3508]), (600.0, [2480, 3508])] { - let job = s.ocr_job(id, &[], OcrSettings { dpi, ..Default::default() }).unwrap(); - for (page, want) in [(0, a4), (1, [612, 792])] { - let scale = job.scale(page, &job.infos[page]); - let shot = r.render(RenderRequest { page, scale, ..Default::default() }); - assert!(shot.error.is_none(), "{:?}", shot.error); - let near = |got: u32, want: u32| got.abs_diff(want) <= 1; - assert!(near(shot.width, want[0]) && near(shot.height, want[1]), "{dpi} dpi, page {page}: {}x{}", shot.width, shot.height); - } - } - } - #[test] fn coverage_of_overlapping_or_broken_boxes_stays_between_zero_and_one() { // A glyph drawn twice (fake bold) does not cover more than the whole word. @@ -317,4 +1148,846 @@ mod tests { let c = covered([0.0, 0.0, 10.0, 10.0], &broken); assert!((0.0..=1.0).contains(&c), "{c}"); } + use crate::tests::fixture; + use pdfcraft_render::TextGlyph; + + fn glyph(text: &str) -> TextGlyph { + TextGlyph { text: text.into(), rect: [0.0, 0.0, 1.0, 1.0], quad: [[0.0, 0.0], [1.0, 0.0], [1.0, 1.0], [0.0, 1.0]], direction: [1.0, 0.0] } + } + + /// The skip-text probe: at least one non-blank run; blank-only or missing text never + /// enables a skip. + #[test] + fn has_text_probe_needs_one_non_blank_run() { + let none: Option> = None; + assert!(!page_has_text(none.as_ref()), "a failed probe never enables a skip"); + assert!(!page_has_text(Some(Arc::new(pdfcraft_render::PageText::default())).as_ref())); + let blank = pdfcraft_render::PageText { glyphs: vec![glyph(" "), glyph("")], ..Default::default() }; + assert!(!page_has_text(Some(Arc::new(blank)).as_ref())); + let has = pdfcraft_render::PageText { glyphs: vec![glyph(" "), glyph("a")], ..Default::default() }; + assert!(page_has_text(Some(Arc::new(has)).as_ref())); + } + + #[test] + fn output_mode_parsing_falls_back_to_searchable() { + assert_eq!(OutputMode::parse("editable"), OutputMode::EditableText); + assert_eq!(OutputMode::parse(" Editable-Text "), OutputMode::EditableText); + assert_eq!(OutputMode::parse("searchable"), OutputMode::Searchable); + assert_eq!(OutputMode::parse("nonsense"), OutputMode::Searchable, "unknown strings mean Searchable"); + assert_eq!(OutputMode::parse(""), OutputMode::Searchable); + } + + #[test] + fn engine_choice_parsing() { + assert_eq!(EngineChoice::parse("tesseract"), EngineChoice::Tesseract); + assert_eq!(EngineChoice::parse(" AUTO "), EngineChoice::Auto); + assert_eq!(EngineChoice::parse("ocrs"), EngineChoice::Ocrs); + assert_eq!(EngineChoice::parse("wat"), EngineChoice::Ocrs, "unknown names fall back to the default"); + assert_eq!(EngineChoice::default().name(), "ocrs"); + } + + /// Automatic reads with the second engine when the first is unsure unless a strategy was + /// chosen: "Primary only" would never run it. A named engine reads alone. + #[test] + fn automatic_defaults_to_the_confidence_weighted_strategy() { + assert_eq!(EngineChoice::Auto.default_strategy(), MergeStrategy::ConfidenceWeighted); + assert_eq!(EngineChoice::Ocrs.default_strategy(), MergeStrategy::PrimaryOnly); + assert_eq!(EngineChoice::Tesseract.default_strategy(), MergeStrategy::PrimaryOnly); + } + + /// Settings default to today's behavior: ocrs, PrimaryOnly, no preprocessing, searchable. + #[test] + fn settings_default_to_the_ocrs_status_quo() { + let s = OcrSettings::default(); + assert_eq!(s.engine, EngineChoice::Ocrs); + assert_eq!(s.strategy, MergeStrategy::PrimaryOnly); + assert_eq!(s.output_mode, OutputMode::Searchable); + assert!(!s.auto_rotate && !s.deskew && !s.denoise && !s.binarize); + assert_eq!(s.region, None); + assert_eq!(s.low_confidence_threshold, 60.0); + assert_eq!(s.tesseract_path, None, "no configured tesseract path by default: the PATH decides"); + } + + /// A configured tesseract path that is not an absolute, existing, executable file makes + /// the Tesseract choice unavailable — the explicit setting is never silently replaced by + /// a PATH search (and this holds whatever this machine has installed). + #[test] + fn an_invalid_configured_tesseract_path_is_unavailable_not_a_fallback() { + let bogus = OcrSettings { engine: EngineChoice::Tesseract, tesseract_path: Some("not/a/program".into()), ..Default::default() }; + match recognizers(&bogus) { + Ok(r) => panic!("a bogus configured path must not resolve: {:?}", r.primary.id()), + Err(e) => assert!(e.contains("tesseract"), "{e}"), + } + } + + /// A failed engine probe is remembered only for the negative TTL — a program installed + /// while the app runs is found by a later job, without a restart — while a found engine is + /// remembered for the session. + #[test] + fn a_negative_probe_expires_a_found_engine_does_not() { + let now = Instant::now(); + let old = now.checked_sub(Duration::from_secs(31)).unwrap(); + assert!(!probe_fresh(false, old, now), "past the TTL the probe is retried"); + assert!(probe_fresh(false, now - Duration::from_secs(5), now), "a fresh failure is not re-probed"); + assert!(probe_fresh(true, old, now), "a found engine is kept"); + } + + /// `ocr_status` and a job judge the same location: one named by the job's settings is the + /// one reported (here an invalid one, so no process starts and the answer is the same on + /// every machine). + #[test] + fn the_status_follows_the_location_a_job_names() { + let bogus = OcrSettings { tesseract_path: Some("not/a/program".into()), ..Default::default() }; + let tesseract = engines_status_for(&bogus).into_iter().find(|e| e.id == "tesseract").expect("tesseract is always listed"); + assert!(!tesseract.available, "{tesseract:?}"); + assert!(tesseract.why_not.as_deref().is_some_and(|why| why.contains("not/a/program")), "{tesseract:?}"); + assert_eq!(tesseract.program.as_deref(), Some("not/a/program")); + } + + /// What the interface sees of the cache: a fresh answer as is, an expired failed probe as + /// the best known answer (shown while it is retried, so the choice does not flicker), and + /// nothing for a location nobody asked about. + #[test] + fn the_cache_lookup_keeps_a_stale_answer_and_knows_its_location() { + let now = Instant::now(); + let old = now.checked_sub(Duration::from_secs(31)).unwrap(); + let failed = |key: Option<&str>, at| CachedTesseract { key: key.map(str::to_string), found: None, at }; + assert!(matches!(cache_lookup(None, None, now), Cached::Missing)); + assert!(matches!(cache_lookup(Some(&failed(None, now)), None, now), Cached::Fresh(None))); + assert!(matches!(cache_lookup(Some(&failed(None, old)), None, now), Cached::Stale(None))); + assert!(matches!(cache_lookup(Some(&failed(Some("/x"), now)), None, now), Cached::Missing), "another location is another question"); + let found = CachedTesseract { key: None, found: Some(Arc::new(Fake::new())), at: old }; + assert!(matches!(cache_lookup(Some(&found), None, now), Cached::Fresh(Some(_))), "a found engine stays fresh"); + } + + /// Engine discovery never deadlocks or corrupts the cache: concurrent lookups with + /// different configured locations (all invalid, so they are refused before any process + /// starts) all finish. + #[test] + fn concurrent_engine_lookups_all_finish() { + let handles: Vec<_> = (0..8) + .map(|i| { + std::thread::spawn(move || { + let configured = if i % 2 == 0 { Some(format!("not/a/program-{i}")) } else { None }; + let _ = tesseract_engine(configured.as_deref()); + }) + }) + .collect(); + for h in handles { + h.join().expect("no panic, no deadlock"); + } + } + + /// The preprocessing flags are not just plumbing: a page read with all four on still + /// yields the same words, in the same places (needs the models). + #[test] + fn preprocessing_flags_flow_through_the_job() { + if !available() { + eprintln!("skipped: no OCR engine available"); + return; + } + let mut s = Session::new().with_clock(|| 1_700_000_000); + let text = s.create_from_text("t", "The quick brown fox jumps over the lazy dog.").unwrap(); + let id = s.open("text.pdf", None, text, None).unwrap(); + let png = crate::export::Exporter::new(s.get(id).unwrap()).png(0, 150.0).unwrap(); + let scan = s.create_from_images(&[("scan.png".into(), png)]).unwrap(); + let id = s.open("scan.pdf", None, scan, None).unwrap(); + let settings = OcrSettings { auto_rotate: true, deskew: true, denoise: true, binarize: true, ..Default::default() }; + let found = s.recognize_text(id, &[], settings).unwrap(); + let text = crate::tests::page_texts(&s, id)[0].to_lowercase(); + // Binarizing an anti-aliased render thins the thin strokes, so edge letters drop + // ("lazy" reads as "azy", "brown" as "rown"); the match rests on what stays stable — + // the middle of "quick" (some CPUs' kernels read the model's "q" as an "a", the same + // "uick" the OCR test in crate::tests pins) and whole short words. + for w in ["uick", "fox", "over"] { + assert!(text.contains(w), "{text}"); + } + // The words were placed, meaning boxes survived the trip back to raster coordinates. + assert!(!found[0].words.is_empty()); + } + + /// suspects(): only Low-band words, by index. + #[test] + fn page_suspects_are_the_low_band_words() { + let word = |t: &str, c: Option| PlacedWord { + text: t.into(), + origin: [0.0; 2], + across: [1.0, 0.0], + up: [0.0, 1.0], + confidence: c, + source: "ocrs".into(), + }; + let page = OcrPage { + already_on_page: 0, + page: 0, + words: vec![word("a", Some(95.0)), word("b", Some(50.0)), word("c", None), word("d", Some(69.0))], + skipped: None, + notes: Vec::new(), + }; + assert_eq!(page.suspects(), vec![1, 3]); + let mean = page.mean_confidence().unwrap_or_default(); + assert!((mean - 71.333).abs() < 0.01, "{mean}"); + } + + /// Region clamping: interior kept, negative extents snapped outward, garbage refused. + #[test] + fn regions_clamp_and_refuse_garbage() { + assert_eq!(clamp_region([10.0, 10.0, 20.0, 20.0], 100, 100), Some([10, 10, 20, 20])); + assert_eq!(clamp_region([-5.0, -5.5, 20.4, 20.0], 100, 100), Some([0, 0, 21, 20]), "snapped outward, clamped to the image"); + assert_eq!(clamp_region([50.0, 50.0, 200.0, 200.0], 100, 100), Some([50, 50, 100, 100])); + assert_eq!(clamp_region([10.0, 10.0, 10.0, 20.0], 100, 100), None, "empty"); + assert_eq!(clamp_region([20.0, 10.0, 10.0, 20.0], 100, 100), None, "inverted"); + assert_eq!(clamp_region([f32::NAN, 0.0, 1.0, 1.0], 100, 100), None, "NaN"); + assert_eq!(clamp_region([-20.0, -20.0, -5.0, -5.0], 100, 100), None, "wholly outside"); + } + + /// A word box only reaches the content stream with positive, finite geometry: NaN, + /// infinite, negative and zero-sized boxes are dropped before placement, so nothing + /// non-finite is ever stamped onto the page. + #[test] + fn non_finite_and_negative_word_boxes_are_dropped_before_placement() { + let word = |rect: [f32; 4]| pdfcraft_ocr::Word::new("w", rect, "ocrs"); + for (why, rect) in [ + ("NaN width", [f32::NAN, 0.0, 10.0, 10.0]), + ("NaN height", [0.0, 0.0, 10.0, f32::NAN]), + ("infinite right", [0.0, 0.0, f32::INFINITY, 10.0]), + ("1e39 height overflows to inf", [0.0, 0.0, 10.0, 1e39f64 as f32]), + ("negative width", [10.0, 0.0, 0.0, 10.0]), + ("negative height", [0.0, 10.0, 10.0, 0.0]), + ("zero size", [5.0, 5.0, 5.0, 5.0]), + ] { + assert!(!box_writable(&word(rect)), "{why}: {rect:?}"); + } + assert!(box_writable(&word([0.0, 0.0, 10.0, 10.0])), "a sane box is placed"); + } + + /// A crop takes exactly the asked pixels (this feeds words back to page space, so edges + /// matter). + #[test] + fn crops_take_the_asked_pixels() { + let img = pdfcraft_ocr::OcrImage::new(3, 2, (0..6).flat_map(|p| vec![p as u8; 4]).collect()).unwrap(); + let c = crop(&img, 1, 0, 3, 2).unwrap(); + assert_eq!((c.width(), c.height()), (2, 2)); + assert_eq!(c.rgba()[0], 1, "the crop starts at x=1"); + assert_eq!(c.rgba()[4], 2, "the first row's second column is x=2"); + assert_eq!(c.rgba()[8], 4, "the second row starts at y=1"); + } + + /// The stale guard: edits and closing are named and the result is refused; a page-count or + /// revision difference is named too (any edit rewrites the working bytes, so those branches + /// are reached with a hand-built guard). + #[test] + fn stale_results_are_named_and_refused() { + let mut s = Session::new(); + let id = s.open("t.pdf", None, Arc::new(fixture(2)), None).unwrap(); + let job = s.ocr_job(id, &[], OcrSettings::default()).unwrap(); + assert_eq!(s.stale_reason(&job.guard), None, "freshly captured"); + // Any edit rewrites the working bytes. + s.apply(id, Edit::AddOcrText { page: 0, words: vec![] }).unwrap(); + assert_eq!(s.stale_reason(&job.guard).as_deref(), Some("the document changed since recognition started")); + assert!(s.apply_ocr_checked(id, &[], &job.guard).is_err(), "a stale result is never applied"); + // A fresh guard passes. + let fresh = s.ocr_job(id, &[], OcrSettings::default()).unwrap(); + assert_eq!(s.stale_reason(&fresh.guard), None); + // A page count and a revision count from another life are named too. + let doc = s.get(id).unwrap(); + let (bytes, revisions) = (doc.bytes.clone(), doc.revision_ends().len()); + let guard = OcrGuard { doc: id, pages: 5, revisions, bytes }; + assert!(s.stale_reason(&guard).is_some_and(|r| r.contains("page count changed")), "{}", s.stale_reason(&guard).unwrap()); + let guard = OcrGuard { doc: id, pages: 2, revisions: revisions + 3, bytes: s.get(id).unwrap().bytes.clone() }; + assert!(s.stale_reason(&guard).is_some_and(|r| r.contains("revisions")), "{}", s.stale_reason(&guard).unwrap()); + // A guard for a different document refuses up front. + let other = s.open("other.pdf", None, Arc::new(fixture(1)), None).unwrap(); + let cross = s.apply_ocr_checked(other, &[], &fresh.guard); + assert!(cross.as_ref().is_err_and(|e| e.contains("another document")), "{cross:?}"); + // And so is a closed document. + s.close(id); + assert_eq!(s.stale_reason(&fresh.guard).as_deref(), Some("the document is no longer open")); + } + + /// The stale guard, exercised end to end through the recognition flow itself (the flow the + /// UI's Verify screen and the automation's two-phase path drive): capture a job on a + /// synthetic scanned page, run the recognition, edit the document, and the captured result + /// is refused with its named reason and nothing written; a fresh recognition over the + /// edited document applies. + #[test] + fn stale_results_are_refused_end_to_end_through_the_recognition_flow() { + if !available() { + eprintln!("skipped: no OCR engine available"); + return; + } + let mut s = Session::new().with_clock(|| 1_700_000_000); + let text = s.create_from_text("t", "The quick brown fox jumps over the lazy dog.").unwrap(); + let id = s.open("text.pdf", None, text, None).unwrap(); + let png = crate::export::Exporter::new(s.get(id).unwrap()).png(0, 150.0).unwrap(); + let scan = s.create_from_images(&[("scan.png".into(), png)]).unwrap(); + let id = s.open("scan.pdf", None, scan, None).unwrap(); + + // Recognize: the job (carrying its guard) is captured first, then the reading runs. + let job = s.ocr_job(id, &[], OcrSettings::default()).unwrap(); + let guard = job.guard.clone(); + let r = recognizers(&job.settings).unwrap(); + let found = job.run(&r, |_, _| true); + assert!(!found[0].words.is_empty(), "the scan was read: {:?}", found[0].skipped); + + // The document is edited while the result sits around; any edit rewrites the bytes. + s.apply(id, Edit::AddOcrText { page: 0, words: vec![] }).unwrap(); + + // Applying the now-stale result is refused with its named reason and writes nothing. + let after_edit = s.get(id).unwrap().bytes.clone(); + assert!(!Arc::ptr_eq(&after_edit, &guard.bytes), "the edit rewrote the working bytes"); + let revisions = s.get(id).unwrap().revision_ends().len(); + let why = s.apply_ocr_checked(id, &found, &guard).unwrap_err(); + assert_eq!(why, "the document changed since recognition started", "{why}"); + assert_eq!(s.get(id).unwrap().revision_ends().len(), revisions, "the refusal wrote no revision"); + assert!(Arc::ptr_eq(&s.get(id).unwrap().bytes, &after_edit), "the refused apply wrote nothing"); + assert!(crate::tests::page_texts(&s, id)[0].trim().is_empty(), "no text layer landed"); + + // A fresh recognition over the current document applies: same flow, valid now. + let job = s.ocr_job(id, &[], OcrSettings::default()).unwrap(); + let guard = job.guard.clone(); + assert_eq!(s.stale_reason(&guard), None, "a fresh job is not stale"); + let found = job.run(&r, |_, _| true); + let words = s.apply_ocr_checked(id, &found, &guard).unwrap(); + assert!(words > 0, "the re-run found the words again"); + let text = crate::tests::page_texts(&s, id)[0].to_lowercase(); + assert!(text.contains("fox"), "{text}"); + } + + /// The selector refuses an unusable engine with its cause named, and resolves the ensemble + /// when both engines exist (here: whatever this machine has for ocrs and tesseract). + #[test] + fn selector_names_the_missing_engine() { + let tesseract = OcrSettings { engine: EngineChoice::Tesseract, ..Default::default() }; + match recognizers(&tesseract) { + Ok(r) => assert_eq!(r.primary.id(), "tesseract"), + Err(e) => assert!(e.contains("tesseract"), "{e}"), + } + let ocrs = OcrSettings { engine: EngineChoice::Ocrs, ..Default::default() }; + match recognizers(&ocrs) { + Ok(r) => assert_eq!(r.primary.id(), "ocrs"), + Err(e) => assert!(e.contains("models"), "{e}"), + } + } + + /// An ocrs-only choice never starts the Tesseract probe: the probe is a `--list-langs` + /// process, and a job that names its engine needs no reading from Tesseract. The test's + /// fake program touches a marker file when it runs, so the probe's start is observable; + /// it needs a Unix shell, like the other process fixtures. + #[cfg(unix)] + #[test] + fn an_ocrs_only_choice_never_probes_tesseract() { + use std::os::unix::fs::PermissionsExt; + let probe = |name: &str| { + let dir = std::env::temp_dir().join(format!("pdfcraft-ocr-probe-{}-{name}", std::process::id())); + std::fs::create_dir_all(&dir).unwrap(); + let program = dir.join("fake-tesseract"); + std::fs::write(&program, "#!/bin/sh\ntouch \"$0.probed\"\nexit 1\n").unwrap(); + let mut perms = std::fs::metadata(&program).unwrap().permissions(); + perms.set_mode(0o755); + std::fs::set_permissions(&program, perms).unwrap(); + (dir.join("fake-tesseract.probed"), program.to_string_lossy().into_owned()) + }; + // An ocrs-only job: the configured location stays untouched. + let (marker, program) = probe("ocrs-only"); + let ocrs_only = OcrSettings { engine: EngineChoice::Ocrs, tesseract_path: Some(program), ..Default::default() }; + let _ = recognizers(&ocrs_only); + assert!(!marker.exists(), "an ocrs-only job must not start the tesseract probe"); + std::fs::remove_file(marker.parent().unwrap().join("fake-tesseract")).ok(); + + // A Tesseract job does validate its program (that is how it is found usable). + let (marker, program) = probe("tesseract"); + let wants = OcrSettings { engine: EngineChoice::Tesseract, tesseract_path: Some(program), ..Default::default() }; + let _ = recognizers(&wants); + assert!(marker.exists(), "a tesseract job probes its program"); + } + + /// A fake engine for the batch skip logic: reads one confident word from anything, counts + /// its calls, and never touches the filesystem. + struct Fake { + calls: std::sync::atomic::AtomicUsize, + } + + impl Fake { + fn new() -> Fake { + Fake { calls: std::sync::atomic::AtomicUsize::new(0) } + } + fn calls(&self) -> usize { + self.calls.load(std::sync::atomic::Ordering::SeqCst) + } + } + + impl Recognizer for Fake { + fn id(&self) -> &'static str { + "fake" + } + fn available(&self) -> bool { + true + } + fn languages(&self) -> Vec<(String, String)> { + vec![("en".into(), "English".into())] + } + fn recognize(&self, _image: &pdfcraft_ocr::OcrImage, _options: &pdfcraft_ocr::RecognizeOptions) -> Result, OcrError> { + self.calls.fetch_add(1, std::sync::atomic::Ordering::SeqCst); + Ok(vec![pdfcraft_ocr::Line { words: vec![pdfcraft_ocr::Word::new("word", [1.0, 1.0, 9.0, 9.0], "fake").with_confidence(95.0)] }]) + } + } + + fn fake_recognizers() -> (Recognizers, Arc) { + let fake = Arc::new(Fake::new()); + (Recognizers { primary: fake.clone(), secondary: None }, fake) + } + + /// Skip-files: a file that already has text is passed over entirely — the engine never + /// runs, the original bytes come back, and the result is its own skipped bucket. + #[test] + fn skip_files_passes_text_files_over_without_running_the_engine() { + let s = Session::new(); + let bytes = s.create_from_text("t", "Already typed here").unwrap(); + let (r, fake) = fake_recognizers(); + let settings = OcrSettings { skip_text_files: true, ..Default::default() }; + let out = recognize_file("t.pdf", bytes.clone(), None, settings, &r, |_, _| true).unwrap(); + assert_eq!(fake.calls(), 0, "the engine never ran"); + assert_eq!(out.bytes.as_ptr(), bytes.as_ptr(), "the original bytes came back"); + assert_eq!(out.file_skipped.as_deref(), Some("the file already contains text")); + assert!(out.skipped() && out.words() == 0); + } + + /// Force OCR overrides the file skip: the file is read even though it has text, and the + /// result is an OCR success (new bytes, words placed), never a skipped bucket. + #[test] + fn force_ocr_overrides_the_file_skip() { + let s = Session::new(); + let bytes = s.create_from_text("t", "Already typed here").unwrap(); + let (r, fake) = fake_recognizers(); + let settings = OcrSettings { skip_text_files: true, force_ocr: true, ..Default::default() }; + let out = recognize_file("t.pdf", bytes, None, settings, &r, |_, _| true).unwrap(); + assert_eq!(fake.calls(), 1); + assert_eq!(out.file_skipped, None); + assert_eq!(out.words(), 1); + assert!(!out.bytes.is_empty()); + } + + /// Force OCR also overrides the per-page skip: a page with text is read anyway. + #[test] + fn force_ocr_overrides_the_page_skip() { + let s = Session::new(); + let bytes = s.create_from_text("t", "A page with text").unwrap(); + let (r, fake) = fake_recognizers(); + let settings = OcrSettings { force_ocr: true, ..Default::default() }; + let out = recognize_file("t.pdf", bytes, None, settings, &r, |_, _| true).unwrap(); + assert_eq!(fake.calls(), 1, "the page's own text did not stop the run"); + assert_eq!(out.pages[0].skipped, None); + } + + /// The per-page skip stays the default: with force off, a text page is passed over (and a + /// second run over a searchable output is a no-op — nothing left to read). + #[test] + fn page_skip_leaves_text_pages_alone_and_a_rerun_is_a_noop() { + let s = Session::new(); + let bytes = s.create_from_text("t", "A page with text").unwrap(); + let (r, fake) = fake_recognizers(); + let out = recognize_file("t.pdf", bytes, None, OcrSettings::default(), &r, |_, _| true).unwrap(); + assert_eq!(fake.calls(), 0); + assert_eq!(out.pages[0].skipped.as_deref(), Some("the page already contains text")); + assert_eq!(out.words(), 0); + // Second run over the same (already text-bearing) file: skipped pages again, no calls. + let again = recognize_file("t.pdf", out.bytes, None, OcrSettings::default(), &r, |_, _| true).unwrap(); + assert_eq!(fake.calls(), 0); + assert_eq!(again.pages[0].skipped.as_deref(), Some("the page already contains text")); + } + + /// Two inputs of one name (from different folders, or differing only in case) would + /// write the same output: both are found, each once. + #[test] + fn duplicate_output_names_are_found_case_insensitively() { + assert!(duplicate_output_names(["a.pdf", "b.pdf"]).is_empty()); + assert_eq!(duplicate_output_names(["a.pdf", "b.pdf", "A.PDF", "b.pdf", "a.pdf"]), ["A.PDF", "b.pdf"]); + } + + /// The batch note's data: counts words at or under the threshold per page (1-based, sorted; + /// skipped pages never contribute), and says nothing when there is nothing to review. + #[test] + fn low_confidence_pages_counts_the_batch_rule() { + let word = |c: f32| PlacedWord { + text: "w".into(), + origin: [0.0; 2], + across: [1.0, 0.0], + up: [0.0, 1.0], + confidence: Some(c), + source: "ocrs".into(), + }; + let page = |n: usize, confs: &[f32], skipped: Option<&str>| OcrPage { + already_on_page: 0, + page: n, + words: confs.iter().map(|c| word(*c)).collect(), + skipped: skipped.map(str::to_string), + notes: Vec::new(), + }; + let pages = vec![ + page(0, &[95.0, 40.0], None), + page(1, &[60.0, 60.5, 10.0], None), + page(2, &[59.0], Some("skipped pages never count")), + page(3, &[0.0], None), + ]; + let (count, pages_on) = low_confidence_pages(&pages, 60.0).unwrap(); + assert_eq!((count, pages_on), (4, vec![1, 2, 4]), "60 counts, 60.5 does not, the skipped page is left out"); + let all_high = vec![page(0, &[95.0, 90.5], None)]; + assert_eq!(low_confidence_pages(&all_high, 60.0), None, "no low-confidence words, no note"); + } + + /// The editable writer: refuses an empty payload, writes the words as visible text at + /// their boxes on pages of the source's displayed size, adds no scan image, and never + /// touches the open document. + #[test] + fn editable_writer_refuses_empty_and_writes_visible_words() { + let mut s = Session::new(); + let id = s.open("src.pdf", None, Arc::new(fixture(2)), None).unwrap(); + assert!( + s.editable_ocr_bytes(id, &[OcrPage { already_on_page: 0, page: 0, words: vec![], skipped: None, notes: Vec::new() }]).is_err(), + "zero words are refused" + ); + let info = s.get(id).unwrap().info.pages[0].clone(); + // A word in the source page's user space, high on the page (the fixture's pages are + // 200 x 300 pt, so the box stays inside them). + let w = PlacedWord { + text: "Editable".into(), + origin: [20.0, 200.0], + across: [120.0, 0.0], + up: [0.0, 12.0], + confidence: Some(88.0), + source: "ocrs".into(), + }; + let found = vec![ + OcrPage { already_on_page: 0, page: 0, words: vec![w], skipped: None, notes: Vec::new() }, + OcrPage { already_on_page: 0, page: 1, words: vec![], skipped: None, notes: Vec::new() }, + ]; + let before = s.get(id).unwrap().bytes.clone(); + let bytes = s.editable_ocr_bytes(id, &found).unwrap(); + // The open document was never touched. + assert!(Arc::ptr_eq(&s.get(id).unwrap().bytes, &before), "the source bytes are the same allocation"); + let mut out = Session::new(); + let oid = out.open("ocr.pdf", None, bytes, None).unwrap(); + let otext = crate::tests::page_texts(&out, oid)[0].clone().to_lowercase(); + assert!(otext.contains("editable"), "{otext}"); + let odoc = out.get(oid).unwrap(); + assert_eq!(odoc.info.pages.len(), 2, "the blank source page stays a page"); + assert!(crate::tests::page_texts(&out, oid)[1].trim().is_empty(), "the page with no words stays blank"); + assert!(odoc.page_images(0).is_empty(), "no scan image is carried over"); + let (w, h) = (odoc.info.pages[0].width, odoc.info.pages[0].height); + assert_eq!((w, h), (info.width, info.height), "pages keep the source's displayed size"); + } + + /// The editable writer maps through a rotated source page: a word drawn in user space + /// reads upright on the new page, and the new page has the source's displayed size. + #[test] + fn editable_writer_maps_rotated_pages() { + let mut s = Session::new().with_clock(|| 1_700_000_000); + let text = s.create_from_text("t", "rotated source").unwrap(); + let id = s.open("r.pdf", None, text, None).unwrap(); + s.apply(id, Edit::RotatePages { pages: vec![0], degrees: 90 }).unwrap(); + let info = s.get(id).unwrap().info.pages[0].clone(); + assert_eq!(info.rotation, 90); + // The word sits at view (20, 48)..(120, 60) on the displayed page; its box comes in + // user space, exactly as recognition left it. + let (o, br, tl) = (info.view_to_user(20.0, 60.0), info.view_to_user(120.0, 60.0), info.view_to_user(20.0, 48.0)); + let w = PlacedWord { + text: "Word".into(), + origin: [f64::from(o[0]), f64::from(o[1])], + across: [f64::from(br[0] - o[0]), f64::from(br[1] - o[1])], + up: [f64::from(tl[0] - o[0]), f64::from(tl[1] - o[1])], + confidence: None, + source: "ocrs".into(), + }; + let bytes = s.editable_ocr_bytes(id, &[OcrPage { already_on_page: 0, page: 0, words: vec![w], skipped: None, notes: Vec::new() }]).unwrap(); + let mut out = Session::new(); + let oid = out.open("ocr.pdf", None, bytes, None).unwrap(); + let otext = crate::tests::page_texts(&out, oid)[0].clone(); + assert!(otext.contains("Word"), "{otext}"); + // The new page is the displayed size (portrait, not the rotated-away landscape). + let oinfo = out.get(oid).unwrap().info.pages[0].clone(); + assert_eq!((oinfo.width, oinfo.height), (info.width, info.height)); + } + + /// The searchable layer round-trips the recognized words through the engine's own + /// extraction, characters beyond WinAnsi included, without needing any display fonts. + #[test] + fn searchable_layer_round_trips_non_win_ansi_words() { + let mut s = Session::new().with_clock(|| 1_700_000_000); + let id = s.open("src.pdf", None, Arc::new(fixture(1)), None).unwrap(); + let word = PlacedWord { + text: "日本語".into(), + origin: [20.0, 250.0], + across: [80.0, 0.0], + up: [0.0, 10.0], + confidence: None, + source: "ocrs".into(), + }; + s.apply(id, Edit::AddOcrText { page: 0, words: vec![word] }).unwrap(); + let text = crate::tests::page_texts(&s, id)[0].clone(); + assert!(text.contains("日本語"), "{text}"); + } + + /// The fontless searchable layer retains its original pixels, handles multiple code + /// fonts and supplementary Unicode, and does not replace existing resource names. + #[test] + fn searchable_unicode_preserves_pixels_save_undo_and_resources() { + let mut s = Session::new().with_clock(|| 1_700_000_000); + let id = s.open("src.pdf", None, Arc::new(fixture(1)), None).unwrap(); + let before = crate::export::Exporter::new(s.get(id).unwrap()).png(0, 72.0).unwrap(); + let mut lines = vec!["中文 English 2026".to_string()]; + let chars: Vec = (0x4e00..0x4f10).filter_map(char::from_u32).chain(['𠮷']).collect(); + lines.extend(chars.chunks(90).map(|c| c.iter().collect())); + let text = lines.join("\n"); + let words: Vec<_> = lines + .iter() + .enumerate() + .map(|(i, line)| PlacedWord { + text: line.clone(), + origin: [20.0, 80.0 - 10.0 * i as f64], + across: [160.0, 0.0], + up: [0.0, 5.0], + confidence: None, + source: "test".into(), + }) + .collect(); + s.apply(id, Edit::AddOcrText { page: 0, words: words.clone() }).unwrap(); + assert!(crate::tests::page_texts(&s, id)[0].contains(&text)); + assert_eq!(crate::export::Exporter::new(s.get(id).unwrap()).png(0, 72.0).unwrap(), before); + let saved = s.save_bytes(id).unwrap(); + let reopened = s.open("searchable.pdf", None, saved, None).unwrap(); + assert!(crate::tests::page_texts(&s, reopened)[0].contains(&text)); + assert_eq!(crate::export::Exporter::new(s.get(reopened).unwrap()).png(0, 72.0).unwrap(), before); + s.undo(id).unwrap(); + assert!(!crate::tests::page_texts(&s, id)[0].contains(&text)); + s.redo(id).unwrap(); + let mut second = words[0].clone(); + second.text = "另一行".into(); + second.origin[1] = 100.0; + s.apply(id, Edit::AddOcrText { page: 0, words: vec![second] }).unwrap(); + let extracted = crate::tests::page_texts(&s, id)[0].clone(); + assert!(extracted.contains(&text) && extracted.contains("另一行"), "{extracted}"); + } + + #[test] + fn searchable_unicode_rejects_excessive_or_non_finite_input_atomically() { + let mut s = Session::new(); + let id = s.open("src.pdf", None, Arc::new(fixture(1)), None).unwrap(); + let word = + PlacedWord { text: "日".into(), origin: [20.0, 80.0], across: [30.0, 0.0], up: [0.0, 5.0], confidence: None, source: "test".into() }; + let before = crate::tests::page_texts(&s, id); + let mut non_finite = word.clone(); + non_finite.origin[0] = f64::NAN; + let mut overflow = word.clone(); + overflow.origin[0] = f64::MAX; + overflow.across[0] = f64::MAX; + let mut too_long = word.clone(); + too_long.text = "a".repeat(4 * 1024 * 1024 + 1); + for words in [vec![non_finite], vec![overflow], vec![too_long], vec![word; 65537]] { + assert!(s.apply(id, Edit::AddOcrText { page: 0, words }).is_err()); + assert_eq!(crate::tests::page_texts(&s, id), before); + } + } + + #[test] + fn editable_export_refuses_invalid_or_excessive_aggregate_input_without_changing_source() { + let mut s = Session::new(); + let id = s.open("src.pdf", None, Arc::new(fixture(2)), None).unwrap(); + let before = s.save_bytes(id).unwrap(); + let word = + PlacedWord { text: "word".into(), origin: [20.0, 80.0], across: [30.0, 0.0], up: [0.0, 5.0], confidence: None, source: "test".into() }; + let page = |page, words| OcrPage { already_on_page: 0, page, words, skipped: None, notes: Vec::new() }; + let mut invalid = word.clone(); + invalid.origin[0] = f64::INFINITY; + let mut long = word.clone(); + long.text = "a".repeat(2 * 1024 * 1024 + 1); + for found in [ + vec![page(0, vec![invalid])], + vec![page(0, vec![long.clone()]), page(1, vec![long])], + vec![page(0, vec![word.clone(); 32768]), page(1, vec![word; 32769])], + ] { + assert!(s.editable_ocr_bytes(id, &found).is_err()); + assert_eq!(s.save_bytes(id).unwrap(), before); + } + } + + /// The editable output carries non-WinAnsi words as visible text that extraction reads + /// back, in a face beyond Helvetica when the build has one. + #[test] + fn editable_output_round_trips_non_win_ansi_words() { + let mut s = Session::new().with_clock(|| 1_700_000_000); + let id = s.open("src.pdf", None, Arc::new(fixture(1)), None).unwrap(); + let w = PlacedWord { + text: "Привет".into(), + origin: [20.0, 250.0], + across: [80.0, 0.0], + up: [0.0, 10.0], + confidence: None, + source: "ocrs".into(), + }; + let bytes = s.editable_ocr_bytes(id, &[OcrPage { already_on_page: 0, page: 0, words: vec![w], skipped: None, notes: Vec::new() }]).unwrap(); + let mut out = Session::new(); + let oid = out.open("ocr.pdf", None, bytes, None).unwrap(); + let text = crate::tests::page_texts(&out, oid)[0].clone(); + if pdfcraft_fonts::CRAFT_FONTS.is_empty() { + assert!(text.contains("??????"), "{text}"); + } else { + assert!(text.contains("Привет"), "{text}"); + } + } + + /// The editable output's VISIBLE layer — a Type3 run for the Cyrillic word, Helvetica for + /// the Latin one — is redacted end to end: a mark over the Cyrillic word removes exactly + /// it (the renderer's extractor no longer reads it, before or after a save and reopen), and + /// the Latin word elsewhere on the page stays. + #[test] + fn the_editable_type3_layer_is_redacted_end_to_end() { + if pdfcraft_fonts::CRAFT_FONTS.is_empty() { + eprintln!("skipping: built without craft-fonts (set CRAFT_FONTS_DIR to run it)"); + return; + } + let mut s = Session::new().with_clock(|| 1_700_000_000); + let id = s.open("src.pdf", None, Arc::new(fixture(1)), None).unwrap(); + let word = |text: &str, y: f64, w: f64| PlacedWord { + text: text.into(), + origin: [20.0, y], + across: [w, 0.0], + up: [0.0, 10.0], + confidence: None, + source: "ocrs".into(), + }; + let words = vec![word("Привет", 250.0, 80.0), word("Hello", 100.0, 50.0)]; + let bytes = s.editable_ocr_bytes(id, &[OcrPage { already_on_page: 0, page: 0, words, skipped: None, notes: Vec::new() }]).unwrap(); + let mut out = Session::new().with_clock(|| 1_700_000_000); + let oid = out.open("ocr.pdf", None, bytes, None).unwrap(); + let before = crate::tests::page_texts(&out, oid)[0].clone(); + assert!(before.contains("Привет") && before.contains("Hello"), "{before}"); + let shape = crate::Shape::Redact { + quads: vec![pdfcraft_annot::rect_quad([10.0, 240.0, 110.0, 270.0])], + overlay: String::new(), + look: Default::default(), + }; + let mark = Edit::AddAnnotation(crate::NewAnnotation { + page: 0, + style: crate::Style::default_for(&shape), + shape, + contents: String::new(), + author: "Ada".into(), + }); + out.apply(oid, mark).unwrap(); + out.apply(oid, Edit::ApplyRedactions { pages: None }).unwrap(); + let after = crate::tests::page_texts(&out, oid)[0].clone(); + assert!(!after.contains("Привет") && !after.contains("Пр"), "the Type3 word is gone: {after}"); + assert!(after.contains("Hello"), "the word outside the mark stays: {after}"); + // Reopened from the saved bytes, an independent read finds the same. + let saved = out.save_bytes(oid).unwrap(); + let mut again = Session::new(); + let aid = again.open("r.pdf", None, saved, None).unwrap(); + let reread = crate::tests::page_texts(&again, aid)[0].clone(); + assert!(!reread.contains("Привет") && reread.contains("Hello"), "{reread}"); + } + + /// A shared document plan serves a multi-page editable export: each page's words must be + /// written with their OWN encodings (the plan's offset for that page), so page 2 never + /// shows page 1's text. + #[test] + fn editable_output_keeps_each_pages_own_words() { + if pdfcraft_fonts::CRAFT_FONTS.is_empty() { + eprintln!("skipping: built without craft-fonts (set CRAFT_FONTS_DIR to run it)"); + return; + } + let mut s = Session::new().with_clock(|| 1_700_000_000); + let id = s.open("src.pdf", None, Arc::new(fixture(2)), None).unwrap(); + let word = |text: &str| PlacedWord { + text: text.into(), + origin: [20.0, 250.0], + across: [80.0, 0.0], + up: [0.0, 10.0], + confidence: None, + source: "ocrs".into(), + }; + // Different scripts, so each page's words live in different planned fonts (and page 2 + // would inherit page 1's codes under page-local indexing into the shared plan). + let found = vec![ + OcrPage { already_on_page: 0, page: 0, words: vec![word("日本語")], skipped: None, notes: Vec::new() }, + OcrPage { already_on_page: 0, page: 1, words: vec![word("Привет")], skipped: None, notes: Vec::new() }, + ]; + let bytes = s.editable_ocr_bytes(id, &found).unwrap(); + let mut out = Session::new(); + let oid = out.open("ocr.pdf", None, bytes, None).unwrap(); + let texts = crate::tests::page_texts(&out, oid); + assert!(texts[0].contains("日本語") && !texts[0].contains("Привет"), "{texts:?}"); + assert!(texts[1].contains("Привет") && !texts[1].contains("日本語"), "{texts:?}"); + } + + /// Redacting a recognized page takes the Unicode searchable layer away for good: the + /// apply's own verification passes over the Type3 font and an independent extractor finds + /// nothing, including in builds with no display fonts. + #[test] + fn redaction_removes_the_unicode_searchable_layer() { + let mut s = Session::new().with_clock(|| 1_700_000_000); + let id = s.open("src.pdf", None, Arc::new(fixture(1)), None).unwrap(); + let word = PlacedWord { + text: "日本語のテキスト".into(), + origin: [20.0, 250.0], + across: [100.0, 0.0], + up: [0.0, 10.0], + confidence: None, + source: "ocrs".into(), + }; + s.apply(id, Edit::AddOcrText { page: 0, words: vec![word] }).unwrap(); + assert!(crate::tests::page_texts(&s, id)[0].contains("日本語")); + let shape = pdfcraft_annot::Shape::Redact { + quads: vec![pdfcraft_annot::rect_quad([18.0, 248.0, 122.0, 262.0])], + overlay: String::new(), + look: Default::default(), + }; + let mark = Edit::AddAnnotation(pdfcraft_annot::NewAnnotation { + page: 0, + style: pdfcraft_annot::Style::default_for(&shape), + shape, + contents: String::new(), + author: "Ada".into(), + }); + s.apply(id, mark).unwrap(); + s.apply(id, Edit::ApplyRedactions { pages: None }).unwrap(); + assert!(!crate::tests::page_texts(&s, id)[0].contains("日本語"), "the layer is gone, not just hidden"); + } + + /// The job's render scale is what the run actually renders at: the chosen dpi, or the + /// page image's own resolution when that is lower. + #[test] + fn render_scale_matches_the_job_resolution() { + let mut s = Session::new(); + let text = s.create_from_text("t", "scale probe").unwrap(); + let id = s.open("t.pdf", None, text, None).unwrap(); + let job = s.ocr_job(id, &[0], OcrSettings { dpi: 144.0, ..Default::default() }).unwrap(); + assert!((job.render_scale(0) - 2.0).abs() < 1e-4, "144 dpi = 2 px/pt, got {}", job.render_scale(0)); + let job = s.ocr_job(id, &[0], OcrSettings { dpi: 600.0, ..Default::default() }).unwrap(); + assert!((job.render_scale(0) - 600.0 / 72.0).abs() < 1e-4, "a text page has no image, so the wanted dpi wins"); + let page = job.pages[0]; + assert_eq!(page, 0); + } + + /// #1848: an A4 scan at 300 dpi whose page box is its pixel size (2480 x 3508 pt, one pixel + /// per point) gets a smaller bitmap for each lower "Downsample to" choice, and a low-resolution + /// letter page is still never shrunk below one pixel per point. + #[test] + fn downsample_shrinks_a_page_box_given_in_pixels() { + let mut s = Session::new(); + let page = |w: u32, h: u32| export::encode_png(w, h, &vec![255; (w * h * 4) as usize]).unwrap(); + let bytes = s.create_from_images(&[("a4.png".into(), page(2480, 3508)), ("letter.png".into(), page(612, 792))]).unwrap(); + let id = s.open("scan.pdf", None, bytes, None).unwrap(); + let mut r = PageRenderer::new(s.get(id).unwrap().bytes.clone(), RenderConfig::default()); + for (dpi, a4) in [(72.0, [595, 842]), (150.0, [1240, 1754]), (300.0, [2480, 3508]), (600.0, [2480, 3508])] { + let job = s.ocr_job(id, &[], OcrSettings { dpi, ..Default::default() }).unwrap(); + for (page, want) in [(0, a4), (1, [612, 792])] { + let scale = job.render_scale(page); + let shot = r.render(RenderRequest { page, scale, ..Default::default() }); + assert!(shot.error.is_none(), "{:?}", shot.error); + let near = |got: u32, want: u32| got.abs_diff(want) <= 1; + assert!(near(shot.width, want[0]) && near(shot.height, want[1]), "{dpi} dpi, page {page}: {}x{}", shot.width, shot.height); + } + } + } } diff --git a/crates/engine/src/ocr/dictionary.rs b/crates/engine/src/ocr/dictionary.rs new file mode 100644 index 0000000000..cbefa9f7b5 --- /dev/null +++ b/crates/engine/src/ocr/dictionary.rs @@ -0,0 +1,80 @@ +//! The UI dictionary format: one case-insensitive word per UTF-8 line. +use std::collections::BTreeSet; + +pub const MAX_BYTES: usize = 4 * 1024 * 1024; +const MAX_WORDS: usize = 65536; + +fn word(value: &str) -> Result { + let value = value.trim(); + if value.is_empty() || value.len() > 1024 || value.chars().any(char::is_control) { + return Err("dictionary words must be nonempty single lines of at most 1024 bytes".into()); + } + let normalized = value.to_lowercase(); + if normalized.len() > 1024 { + return Err("normalized dictionary words must be at most 1024 bytes".into()); + } + Ok(normalized) +} + +pub fn words(text: &str) -> Result, String> { + if text.len() > MAX_BYTES { + return Err("OCR dictionary exceeds 4 MiB".into()); + } + let mut words = BTreeSet::new(); + for line in text.lines().filter(|line| !line.trim().is_empty()) { + words.insert(word(line)?); + if words.len() > MAX_WORDS { + return Err("OCR dictionary exceeds 65536 words".into()); + } + } + Ok(words) +} + +/// Normalize, update and bound the complete dictionary before the caller writes it. +pub fn update(text: &str, add: &[&str], remove: &[&str]) -> Result { + if add.len() > MAX_WORDS || remove.len() > MAX_WORDS { + return Err("OCR dictionary exceeds 65536 words".into()); + } + let mut entries = words(text)?; + for value in remove { + entries.remove(&word(value)?); + } + for value in add { + entries.insert(word(value)?); + if entries.len() > MAX_WORDS { + return Err("OCR dictionary exceeds 65536 words".into()); + } + } + let mut out = String::new(); + for entry in entries { + if out.len().checked_add(entry.len()).and_then(|n| n.checked_add(1)).is_none_or(|n| n > MAX_BYTES) { + return Err("OCR dictionary exceeds 4 MiB".into()); + } + out.push_str(&entry); + out.push('\n'); + } + Ok(out) +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn dictionary_updates_round_trip_and_refuse_expanding_normalization() { + let text = update("Quartz\n日本語\n", &["QUARTZ", "extra"], &["日本語"]).unwrap(); + assert_eq!(text, "extra\nquartz\n"); + assert_eq!(words(&text).unwrap(), BTreeSet::from(["extra".into(), "quartz".into()])); + assert!(update("original\n", &[&"İ".repeat(512)], &[]).unwrap_err().contains("normalized")); + assert!(update("original\n", &["two\nlines"], &[]).is_err()); + } + + #[test] + fn dictionary_limits_cover_existing_and_added_entries() { + let full: String = (0..65536).map(|i| format!("word{i}\n")).collect(); + assert_eq!(words(&full).unwrap().len(), 65536); + assert!(update(&full, &["extra"], &[]).is_err()); + assert_eq!(update(&full, &["extra"], &["word0"]).unwrap().lines().count(), 65536); + assert!(words(&"a".repeat(MAX_BYTES + 1)).is_err()); + } +} diff --git a/crates/engine/src/ocr/layer.rs b/crates/engine/src/ocr/layer.rs new file mode 100644 index 0000000000..ec2bac5fa8 --- /dev/null +++ b/crates/engine/src/ocr/layer.rs @@ -0,0 +1,145 @@ +//! Unicode search text has no visible glyphs. A generated Type 3 font supplies advance widths +//! and a ToUnicode map, with empty character programs; no display font is needed or embedded. +//! Adapted from the searchable-layer implementation in #433 (32e09fbe). + +use std::collections::BTreeMap; + +use pdfcraft_cos::{Dict, Document, Object, Stream}; +use pdfcraft_edit::EditError; +use pdfcraft_ocr::PlacedWord; + +/// Refuse oversized text or invalid geometry before cloning words or allocating fonts. +/// These limits apply to the complete input, including all editable export pages. +pub(crate) fn validate_words<'a>(words: impl IntoIterator) -> Result { + let mut count = 0usize; + let mut length = 0usize; + for w in words { + count += 1; + length = length.checked_add(w.text.len()).ok_or_else(|| EditError::Invalid("OCR text layer is too large".into()))?; + if count > 65536 || length > 4 * 1024 * 1024 { + return Err(EditError::Invalid("OCR text layer is too large".into())); + } + if !w.origin.iter().chain(&w.across).chain(&w.up).all(|v| v.is_finite()) + || !w.across[0].hypot(w.across[1]).is_finite() + || !w.up[0].hypot(w.up[1]).is_finite() + || (0..2).any(|i| !(w.origin[i] + 0.2 * w.up[i]).is_finite() || !(w.origin[i] + w.across[i] + 0.2 * w.up[i]).is_finite()) + { + return Err(EditError::Invalid("OCR text coordinates must be finite".into())); + } + } + Ok(count) +} + +pub(crate) fn add_text(doc: &mut Document, page: usize, words: &[PlacedWord]) -> Result<(), EditError> { + validate_words(words)?; + // Keep the existing Latin layout, including Helvetica's proportional advance widths. + if words.iter().all(|w| { + w.text + .chars() + .all(|c| matches!(c, '\u{20}'..='\u{7e}' | '\u{a0}'..='\u{ff}' | '€' | '‚' | '„' | '…' | '‘' | '’' | '“' | '”' | '•' | '–' | '—' | '™')) + }) { + pdfcraft_edit::stamp(doc, page, "OCR", pdfcraft_ocr::text_layer(words, &pdfcraft_fonts::TextPlan::default()))?; + return Ok(()); + } + let p = pdfcraft_model::pages(doc).get(page).cloned().ok_or_else(|| EditError::Invalid("OCR page does not exist".into()))?; + let mut res = p.dict.get(b"Resources").map(|r| doc.resolve(r)).and_then(|r| r.as_dict().cloned()).unwrap_or_default(); + let mut fonts = res.get(b"Font").map(|f| doc.resolve(f)).and_then(|f| f.as_dict().cloned()).unwrap_or_default(); + let chars: Vec = words.iter().flat_map(|w| w.text.chars()).collect::>().into_iter().collect(); + // Match the extractor's em box (-200..800). Baselines below are offset by 0.2 em, + // so search/selection rectangles span the detected image box. + let blank = doc.add(Object::Stream(Stream::from_raw(Dict::new(), b"1000 0 0 -200 1000 800 d1\n".to_vec()))); + let mut codes = BTreeMap::new(); + let mut names = Vec::new(); + for (index, chunk) in chars.chunks(255).enumerate() { + let mut name = format!("PCOcr{index}"); + while fonts.contains(name.as_bytes()) { + name.push('x'); + } + let mut differences = vec![Object::Int(1)]; + let mut procs = Dict::new(); + let mut cmap = String::from( + "/CIDInit /ProcSet findresource begin\n12 dict begin\nbegincmap\n/CIDSystemInfo << /Registry (PdfCraft) /Ordering (Unicode) /Supplement 0 >> def\n/CMapName /PdfCraftOCR def\n/CMapType 2 def\n1 begincodespacerange\n<01> \nendcodespacerange\n", + ); + for (batch, part) in chunk.chunks(100).enumerate() { + cmap.push_str(&format!("{} beginbfchar\n", part.len())); + for (i, ch) in part.iter().enumerate() { + let code = (batch * 100 + i + 1) as u8; + let glyph = format!("g{code:02X}"); + differences.push(Object::name(&glyph)); + procs.set(glyph.into_bytes(), Object::Ref(blank)); + let mut units = [0u16; 2]; + let hex: String = ch.encode_utf16(&mut units).iter().map(|u| format!("{u:04X}")).collect(); + cmap.push_str(&format!("<{code:02X}> <{hex}>\n")); + codes.insert(*ch, (index, code)); + } + cmap.push_str("endbfchar\n"); + } + cmap.push_str("endcmap\nCMapName currentdict /CMap defineresource pop\nend\nend\n"); + let map = doc.add(Object::Stream(Stream::flate(Dict::new(), cmap.as_bytes()))); + let mut encoding = Dict::new(); + encoding.set(b"Type".to_vec(), Object::name("Encoding")); + encoding.set(b"Differences".to_vec(), Object::Array(differences)); + let mut font = Dict::new(); + font.set(b"Type".to_vec(), Object::name("Font")); + font.set(b"Subtype".to_vec(), Object::name("Type3")); + font.set(b"FontBBox".to_vec(), Object::Array(vec![0.into(), (-200).into(), 1000.into(), 800.into()])); + font.set(b"FontMatrix".to_vec(), Object::Array(vec![Object::Real(0.001), 0.into(), 0.into(), Object::Real(0.001), 0.into(), 0.into()])); + font.set(b"FirstChar".to_vec(), Object::Int(1)); + font.set(b"LastChar".to_vec(), Object::Int(chunk.len() as i64)); + font.set(b"Widths".to_vec(), Object::Array(vec![Object::Int(1000); chunk.len()])); + font.set(b"Encoding".to_vec(), Object::Dict(encoding)); + font.set(b"CharProcs".to_vec(), Object::Dict(procs)); + font.set(b"Resources".to_vec(), Object::Dict(Dict::new())); + font.set(b"ToUnicode".to_vec(), Object::Ref(map)); + fonts.set(name.as_bytes().to_vec(), Object::Ref(doc.add(Object::Dict(font)))); + names.push(name); + } + res.set(b"Font".to_vec(), Object::Dict(fonts)); + doc.update_dict(p.obj, |d| d.set(b"Resources".to_vec(), Object::Dict(res)))?; + // `3 Tr` and the font are graphics state, which `ET` doesn't reset: like + // `pdfcraft_ocr::text_layer` (#3029), the text is drawn inside `q … Q` so the invisible + // rendering mode cannot leak into later content streams on the page. + let mut content = String::from("/OCR BMC\nq\nBT\n3 Tr\n"); + for w in words { + let count = w.text.chars().count(); + if count == 0 || w.up[0].hypot(w.up[1]) <= 0.0 || w.across[0].hypot(w.across[1]) <= 0.0 { + continue; + } + let mut start = 0; + let mut run = Vec::new(); + let mut font_index = None; + let emit = |content: &mut String, font: usize, start: usize, run: &[u8]| { + let n = count as f64; + let at = start as f64 / n; + let hex: String = run.iter().map(|b| format!("{b:02X}")).collect(); + content.push_str(&format!( + "/{} 1 Tf\n{} {} {} {} {} {} Tm <{hex}> Tj\n", + names[font], + w.across[0] / n, + w.across[1] / n, + w.up[0], + w.up[1], + w.origin[0] + at * w.across[0] + 0.2 * w.up[0], + w.origin[1] + at * w.across[1] + 0.2 * w.up[1] + )); + }; + for (i, ch) in w.text.chars().enumerate() { + let Some(&(font, code)) = codes.get(&ch) else { continue }; + if let Some(previous) = font_index + && previous != font + { + emit(&mut content, previous, start, &run); + run.clear(); + start = i; + } + font_index = Some(font); + run.push(code); + } + if let Some(font) = font_index { + emit(&mut content, font, start, &run); + } + } + content.push_str("ET\nQ\nEMC\n"); + pdfcraft_edit::stamp(doc, page, "OCR", content.into_bytes())?; + Ok(()) +} diff --git a/crates/engine/src/pictures.rs b/crates/engine/src/pictures.rs index 49d76f07c0..d8b8835dc7 100644 --- a/crates/engine/src/pictures.rs +++ b/crates/engine/src/pictures.rs @@ -60,6 +60,18 @@ pub(crate) fn mark_source(doc: &mut pdfcraft_cos::Document, f: &MarkFile) -> Res } } +/// A 3D model's poster: the picture in `file` brought into `doc` and fitted inside `rect` +/// (user space), keeping its shape, and centred. It is drawn in user space and not counter-rotated +/// (unlike a stamp), so on a page with `/Rotate` it turns with the page, as the model does. +pub(crate) fn model_poster(doc: &mut pdfcraft_cos::Document, rect: [f64; 4], file: &MarkFile) -> Result { + let src = mark_source(doc, file)?; + let (sw, sh) = src.size; + if !(sw.is_finite() && sh.is_finite() && sw > 0.0 && sh > 0.0) { + return Err(EditError::Source(format!("{}: the poster has no area", file.name))); + } + Ok(pdfcraft_annot::PosterPicture { xobject: src.xobject, image: src.image, size: (sw, sh), at: fit_picture(rect, sw, sh) }) +} + /// The dimensions of an image from its bytes (JPEG, PNG, TIFF), without decoding the full raster. pub fn image_dimensions(bytes: &[u8]) -> Option<(u32, u32)> { let format = image::guess_format(bytes).ok()?; diff --git a/crates/engine/src/scan_clean.rs b/crates/engine/src/scan_clean.rs new file mode 100644 index 0000000000..cef7538309 --- /dev/null +++ b/crates/engine/src/scan_clean.rs @@ -0,0 +1,308 @@ +//! Scan & OCR ▸ Enhance ▸ Scanned document: clean up the pictures scanned pages are made of. +//! +//! Unlike Enhance ▸ Camera image, the page is not rendered and replaced: each scan image is +//! decoded, cleaned ([`crate::camera_clean::clean_samples`]) and written back **into the same +//! image object**, at the same size and in the same colour space. So everything else on the +//! page stays as it was: an invisible OCR text layer, comments, links, form fields, other +//! content, and the image's own soft mask, `/Interpolate` and other entries. A JPEG stays a JPEG +//! (at quality 90), any other grey or RGB image is stored losslessly (Flate), and a black and +//! white (1-bit) image stays 1-bit. An image several pages draw is cleaned once. +//! +//! A page is a scan when its pictures cover at least half of it (as for Recognize text); the +//! pictures each covering at least a tenth of the page are its scan. What is not cleaned is +//! reported, never skipped silently: pages that are no scan, and scan images in a form PdfCraft +//! can't rewrite yet (CMYK, indexed, JPEG 2000, JBIG2, CCITT fax, a stencil mask, a remapping +//! decode array, a scan drawn inside grouped Form artwork) or whose data is broken. +//! +//! Like Optimize, the work is a job: [`Session::scan_clean_job`] captures the document, +//! [`ScanCleanJob::run`] cleans anywhere (a worker thread), and [`Session::apply_scan_clean`] +//! makes the result one undoable step, "Enhance scanned file" (the name the Scan & OCR tool has). + +use pdfcraft_cos::{ObjRef, Object, Stream}; + +use crate::camera_clean::{CleanUp, MAX_LEVEL, clean_samples}; +use crate::edit::check_permission; +use crate::{DocId, Edit, EditError, Session, guard, optimize}; + +pub use crate::camera_clean::DESPECKLE_FROM; + +/// The JPEG quality a cleaned JPEG scan is written back at (as Enhance camera image). +const QUALITY: u8 = 90; +/// On a scanned page, an image covering at least this share of the page is part of the scan; +/// smaller ones (a logo or a signature pasted on) are left alone. +const PART_OF_SCAN: f64 = 0.1; +/// Why a scan drawn through grouped Form artwork stays as it is. +const IN_FORM: &str = "it is drawn inside grouped artwork (a form XObject), which can't be cleaned up yet"; + +/// How scanned pages are cleaned up. +#[derive(Clone, Copy, Debug, PartialEq, Eq)] +pub struct ScanCleanSettings { + /// Background removal, 0 (off) … 100 (strongest), as Enhance camera image's clean-up level: + /// the paper's illumination is evened out (fully from 50), so a grey, yellowed or unevenly lit + /// background turns white; then the contrast is raised so faint strokes darken. Colour ink + /// stays colour. + pub level: u8, + /// Edge shadow removal: shadows along the scan's edges (a book's gutter, a copier's dark + /// border) are lightened to the paper next to them, line by line, before the level applies. + pub edge_shadows: bool, + /// Despeckle: isolated dark specks (dust, scanner noise), dark pixels with at most one dark + /// neighbour, turn white. The only change a black and white (1-bit) scan takes. + pub despeckle: bool, +} + +impl ScanCleanSettings { + /// The level `level` (capped at 100) with edge shadow removal (unless the level is 0), and + /// despeckling from [`DESPECKLE_FROM`] as the camera tool does. + pub fn at(level: u8) -> Self { + let level = level.min(MAX_LEVEL); + Self { level, edge_shadows: level > 0, despeckle: level >= DESPECKLE_FROM } + } + + /// Whether these settings change anything at all. + pub fn changes_something(self) -> bool { + self.level > 0 || self.edge_shadows || self.despeckle + } +} + +impl Default for ScanCleanSettings { + fn default() -> Self { + Self::at(50) + } +} + +/// What a clean-up did. +#[derive(Clone, Debug, Default, PartialEq)] +pub struct ScanCleanReport { + /// Pages (0-based) whose scan was cleaned, in order. + pub pages: Vec, + /// Images rewritten. One drawn on several pages counts once. + pub images: usize, + /// Pages left alone because no picture covers most of them. + pub not_scanned: Vec, + /// Scan images left as they were, and why. + pub skipped: Vec, +} + +/// A scan image that was not cleaned. +#[derive(Clone, Debug, PartialEq)] +pub struct SkippedImage { + /// The pages (0-based) that draw it, among those asked for. + pub pages: Vec, + /// Its resource name on the first of them. + pub name: String, + /// Why, in a sentence that follows "skipped because". + pub reason: String, +} + +/// One image to clean: its object, its name and the pages that draw it. +struct Target { + object: ObjRef, + name: String, + pages: Vec, +} + +/// What cleaning scanned pages needs, independent of the session. +pub struct ScanCleanJob { + cos: pdfcraft_cos::Document, + targets: Vec, + settings: ScanCleanSettings, + /// The document's edit generation it was taken at: a result is for that state only. + generation: u64, + report: ScanCleanReport, +} + +/// The cleaned images, ready for [`Session::apply_scan_clean`], and the report. +pub struct ScanCleaned { + images: Vec<(ObjRef, Stream)>, + generation: u64, + pub report: ScanCleanReport, +} + +impl Session { + /// Capture what cleaning up the scans on `pages` (0-based; empty = all) of document `id` + /// needs. Errors when the document can't be changed (read-only, or its permissions forbid + /// changes) or a page doesn't exist. + pub fn scan_clean_job(&self, id: DocId, pages: &[usize], settings: ScanCleanSettings) -> Result { + let doc = self.get(id).ok_or(EditError::NoDocument)?; + if let Some(why) = &doc.read_only_reason { + return Err(EditError::ReadOnly(why.clone())); + } + let editor = doc.editor.as_ref().ok_or_else(|| EditError::ReadOnly(String::new()))?; + if let Some(p) = editor.cos.permissions() { + check_permission(&Edit::EnhanceScan { images: Vec::new() }, &p)?; + } + let all = doc.info.pages.len(); + let mut wanted: Vec = if pages.is_empty() { (0..all).collect() } else { pages.to_vec() }; + if let Some(p) = wanted.iter().find(|p| **p >= all) { + return Err(EditError::Invalid(format!("there is no page {}: the document has {all}", p.saturating_add(1)))); + } + wanted.sort_unstable(); + wanted.dedup(); + let mut report = ScanCleanReport::default(); + let mut targets: Vec = Vec::new(); + // Form artwork already reported, and where in `report.skipped`. + let mut forms_reported: Vec<(ObjRef, usize)> = Vec::new(); + for page in wanted { + let Some(info) = doc.info.pages.get(page) else { continue }; + // Ordinary images and scan-holding Forms are classified together, so a page that + // mixes them gets its images cleaned and its Forms reported. + let Some((images, forms, own_scan)) = crate::scan_fields::scan_parts(doc, page, info) else { + report.not_scanned.push(page); + continue; + }; + let forms: Vec<_> = forms.into_iter().filter(|(_, share)| *share >= PART_OF_SCAN).collect(); + if !own_scan && forms.is_empty() { + report.not_scanned.push(page); + continue; + } + // A scan drawn through a Form XObject is a scan all the same: say why it stays. + for (form, _) in forms { + let known = form.object.and_then(|o| forms_reported.iter().find(|(r, _)| *r == o)).and_then(|(_, k)| report.skipped.get_mut(*k)); + match known { + Some(s) if !s.pages.contains(&page) => s.pages.push(page), + Some(_) => {} + None => { + if let Some(o) = form.object { + forms_reported.push((o, report.skipped.len())); + } + report.skipped.push(SkippedImage { pages: vec![page], name: form.name, reason: IN_FORM.into() }); + } + } + } + for (image, _) in images.into_iter().filter(|(_, share)| *share >= PART_OF_SCAN) { + let Some(object) = image.object else { + report.skipped.push(SkippedImage { + pages: vec![page], + name: image.name, + reason: "it is stored inside the page's resources rather than as an object of its own".into(), + }); + continue; + }; + match targets.iter_mut().find(|t| t.object == object) { + Some(t) if !t.pages.contains(&page) => t.pages.push(page), + Some(_) => {} + None => targets.push(Target { object, name: image.name, pages: vec![page] }), + } + } + } + Ok(ScanCleanJob { cos: editor.cos.clone(), targets, settings, generation: doc.edit_generation(), report }) + } + + /// Write the images [`ScanCleanJob::run`] cleaned into document `id` as one undoable step, + /// "Enhance scanned file" (none when nothing was cleaned), and return the report. Refused + /// when the document changed since the job was taken. + pub fn apply_scan_clean(&mut self, id: DocId, cleaned: ScanCleaned) -> Result { + let doc = self.get(id).ok_or(EditError::NoDocument)?; + if doc.edit_generation() != cleaned.generation { + return Err(EditError::Invalid("the document changed while its scans were being cleaned up; nothing was changed, try again".into())); + } + if !cleaned.images.is_empty() { + self.apply(id, Edit::EnhanceScan { images: cleaned.images })?; + } + Ok(cleaned.report) + } + + /// Enhance ▸ Scanned document in one call: [`Self::scan_clean_job`], run, then + /// [`Self::apply_scan_clean`] (the CLI and agents use this). + pub fn enhance_scanned_pages(&mut self, id: DocId, pages: &[usize], settings: ScanCleanSettings) -> Result { + let job = self.scan_clean_job(id, pages, settings)?; + let cleaned = job.run(|_, _| true)?; + self.apply_scan_clean(id, cleaned) + } +} + +impl ScanCleanJob { + /// The images to clean (one drawn on several pages counts once). + pub fn total(&self) -> usize { + self.targets.len() + } + + /// Clean every scan image. `progress(done, total)` is called before each image; returning + /// `false` stops with [`EditError::Cancelled`]. A panic is returned as an error, so a worker + /// thread always reports back. + pub fn run(self, mut progress: impl FnMut(usize, usize) -> bool) -> Result { + guard(move || self.work(&mut progress)).map_err(|m| EditError::Invalid(format!("cleaning up the scans failed unexpectedly ({m})")))? + } + + fn work(self, progress: &mut dyn FnMut(usize, usize) -> bool) -> Result { + let ScanCleanJob { cos, targets, settings, generation, mut report } = self; + let total = targets.len(); + let mut images = Vec::new(); + let mut changed_pages: Vec = Vec::new(); + // One decoding allowance for the whole run, sized by the file. + let mut budget = pdfcraft_cos::ImageBudget::new(&cos); + for (done, target) in targets.into_iter().enumerate() { + if !progress(done, total) { + return Err(EditError::Cancelled); + } + let mut skip = |reason: String| report.skipped.push(SkippedImage { pages: target.pages.clone(), name: target.name.clone(), reason }); + match clean_image(&cos, target.object, settings, &mut budget) { + Ok(Some(stream)) => { + images.push((target.object, stream)); + changed_pages.extend(target.pages.iter().copied()); + } + Ok(None) => skip("it is already clean: nothing changes at these settings".into()), + Err(reason) => skip(reason), + } + } + changed_pages.sort_unstable(); + changed_pages.dedup(); + report.images = images.len(); + report.pages = changed_pages; + report.skipped.sort_by_key(|s| s.pages.first().copied()); + Ok(ScanCleaned { images, generation, report }) + } +} + +/// Image `object` cleaned with `settings`: the new stream, `None` when nothing changes, or why it +/// can't be cleaned. +fn clean_image( + cos: &pdfcraft_cos::Document, + object: ObjRef, + settings: ScanCleanSettings, + budget: &mut pdfcraft_cos::ImageBudget, +) -> Result, String> { + let obj = cos.get(object); + let Object::Stream(s) = &*obj else { return Err("it is not an image".into()) }; + if s.dict.name(b"Subtype") != Some(b"Image") { + return Err("it is not an image".into()); + } + // A colour-key mask names sample values to leave unpainted: cleaned samples would match others. + if matches!(s.dict.get(b"Mask").map(|m| cos.resolve(m)).as_deref(), Some(Object::Array(_))) { + return Err("it has a colour-key mask, which cleaning would change".into()); + } + let samples = optimize::decode_samples(cos, s, budget).map_err(|e| e.to_string())?; + let cleaned = if samples.bilevel { + if !settings.despeckle { + return Err("it is black and white, which only Despeckle changes, and Despeckle is off".into()); + } + clean_samples(&samples.data, 1, samples.width, samples.height, CleanUp { level: 0, edge_shadows: false, despeckle: true }) + } else { + if !settings.changes_something() { + return Ok(None); + } + let how = CleanUp { level: settings.level, edge_shadows: settings.edge_shadows, despeckle: settings.despeckle }; + clean_samples(&samples.data, samples.components, samples.width, samples.height, how) + } + .ok_or("its samples don't match its size")?; + if cleaned == samples.data { + return Ok(None); + } + let cleaned = optimize::Samples { data: cleaned, ..samples }; + optimize::encode_samples(s, &cleaned, QUALITY).map(Some).ok_or_else(|| "it could not be written back".into()) +} + +/// [`Edit::EnhanceScan`]: put the cleaned images in place. Each object must still be an image. +pub(crate) fn write_back(doc: &mut pdfcraft_cos::Document, images: &[(ObjRef, Stream)]) -> Result<(), EditError> { + for (object, stream) in images { + let is_image = matches!(&*doc.get(*object), Object::Stream(s) if s.dict.name(b"Subtype") == Some(b"Image")); + if !is_image { + return Err(EditError::Invalid(format!("object {} is no longer an image; the scan was not changed", object.num))); + } + doc.set(*object, Object::Stream(stream.clone())); + } + Ok(()) +} + +#[cfg(test)] +mod tests; diff --git a/crates/engine/src/scan_clean/tests.rs b/crates/engine/src/scan_clean/tests.rs new file mode 100644 index 0000000000..e58ab3e5ef --- /dev/null +++ b/crates/engine/src/scan_clean/tests.rs @@ -0,0 +1,463 @@ +use std::sync::Arc; + +use pdfcraft_testkit::Pdf; + +use super::*; + +const W: usize = 300; +const H: usize = 400; +/// Isolated specks on the paper, away from the text rows. +const SPECKS: [(usize, usize); 3] = [(150, 25), (200, 85), (120, 125)]; + +/// The light falling on the paper at `x`: full in the middle, darker towards a shadowed left +/// edge (as a book's gutter) and a little at the right edge. +fn light(x: usize) -> f32 { + let left = (x as f32 / 60.0).min(1.0); + let right = ((W - 1 - x) as f32 / 30.0).min(1.0); + (0.55 + 0.45 * left) * (0.8 + 0.2 * right) +} + +fn is_stroke(x: usize, y: usize) -> bool { + (60..360).contains(&y) && (y - 60) % 40 < 6 && (60..240).contains(&x) +} + +/// A synthetic scan, `n` components (1 grey, 3 RGB): greyish yellowed paper, a shadowed gutter, +/// dark text strokes and a few specks. +fn scan_pixels(n: usize) -> Vec { + let mut out = Vec::with_capacity(W * H * n); + for y in 0..H { + for x in 0..W { + let ink = if is_stroke(x, y) || SPECKS.contains(&(x, y)) { 0.12 } else { 0.8 }; + let v = ink * light(x); + let px = [v, v, v * 0.85]; + for c in px.iter().take(n) { + out.push((c * 255.0).round().clamp(0.0, 255.0) as u8); + } + } + } + out +} + +fn jpeg(px: &[u8], n: usize) -> Vec { + let mut out = Vec::new(); + let ty = if n == 1 { image::ExtendedColorType::L8 } else { image::ExtendedColorType::Rgb8 }; + image::ImageEncoder::write_image(image::codecs::jpeg::JpegEncoder::new_with_quality(&mut out, 95), px, W as u32, H as u32, ty).unwrap(); + out +} + +/// Three pages: 1 a scan (the image `image_dict`/`data` as object 4) with an invisible text +/// layer and a comment, 2 a page of real text, 3 the same scan image drawn again. +fn scan_pdf(image_dict: &str, data: &[u8]) -> Vec { + let resources = "/Resources << /XObject << /Im0 4 0 R >> /Font << /F1 5 0 R >> >>"; + Pdf::new() + .obj("<< /Type /Catalog /Pages 2 0 R >>") + .obj("<< /Type /Pages /Kids [3 0 R 8 0 R 9 0 R] /Count 3 >>") + .obj(format!("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {W} {H}] {resources} /Contents 6 0 R /Annots [7 0 R] >>")) + .stream(&format!("/Type /XObject /Subtype /Image {image_dict}"), data) + .obj("<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica >>") + .stream("", format!("q {W} 0 0 {H} 0 0 cm /Im0 Do Q BT 3 Tr /F1 12 Tf 20 380 Td (Invisible OCR words) Tj ET")) + .obj("<< /Type /Annot /Subtype /Square /Rect [10 10 50 50] /C [1 0 0] /Contents (A comment) >>") + .obj(format!("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {W} {H}] {resources} /Contents 10 0 R >>")) + .obj(format!("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {W} {H}] {resources} /Contents 11 0 R >>")) + .stream("", "BT /F1 12 Tf 20 380 Td (Real text) Tj ET") + .stream("", format!("q {W} 0 0 {H} 0 0 cm /Im0 Do Q")) + .build() +} + +fn grey_scan() -> Vec { + scan_pdf(&format!("/Width {W} /Height {H} /ColorSpace /DeviceGray /BitsPerComponent 8 /Interpolate true"), &scan_pixels(1)) +} + +fn open(bytes: Vec) -> (Session, DocId) { + let mut s = Session::new().with_clock(|| 1_700_000_000); + let id = s.open("scan.pdf", None, Arc::new(bytes), None).unwrap(); + (s, id) +} + +fn object(s: &Session, id: DocId, num: u32) -> Arc { + s.get(id).unwrap().editor.as_ref().unwrap().cos.get(ObjRef::new(num, 0)) +} + +/// The samples of the image page 1 draws, now. +fn samples(s: &Session, id: DocId) -> optimize::Samples { + let doc = s.get(id).unwrap(); + let object = doc.page_images(0)[0].object.unwrap(); + let cos = &doc.editor.as_ref().unwrap().cos; + let Object::Stream(img) = &*cos.get(object) else { panic!("the image") }; + optimize::decode_samples(cos, img, &mut pdfcraft_cos::ImageBudget::with_work(u64::MAX)).unwrap() +} + +fn grey_at(px: &optimize::Samples, x: usize, y: usize) -> u8 { + px.data[(y * W + x) * px.components] +} + +#[test] +fn a_scan_gets_white_paper_even_edges_dark_strokes_and_no_specks_in_place() { + let (mut s, id) = open(grey_scan()); + let before = samples(&s, id); + let kept: Vec> = [3, 5, 6, 7, 8, 10, 11].iter().map(|&n| object(&s, id, n)).collect(); + let text_before = s.get(id).unwrap().text_lines(0); + let original = s.get(id).unwrap().bytes.clone(); + + let report = s.enhance_scanned_pages(id, &[], ScanCleanSettings::at(100)).unwrap(); + assert_eq!(report.pages, vec![0, 2], "the shared image counts for both its pages"); + assert_eq!(report.images, 1); + assert_eq!(report.not_scanned, vec![1], "the text page is no scan"); + assert!(report.skipped.is_empty(), "{:?}", report.skipped); + assert_eq!(s.get(id).unwrap().can_undo(), Some("Enhance scanned file")); + + let after = samples(&s, id); + assert_eq!((after.width, after.height, after.components, after.jpeg), (W as u32, H as u32, 1, false)); + // Paper, also in the shadowed gutter and at the right edge, turns (nearly) white. + for (x, y) in [(150, 40), (150, 380), (3, 40), (3, 380), (30, 200), (W - 2, 40)] { + assert!(grey_at(&before, x, y) < 210, "the fixture's paper is grey at ({x}, {y})"); + assert!(grey_at(&after, x, y) >= 235, "paper at ({x}, {y}): {} → {}", grey_at(&before, x, y), grey_at(&after, x, y)); + } + // The edge is evened out: the gutter is as white as the middle. + assert!(grey_at(&after, 150, 40).abs_diff(grey_at(&after, 3, 40)) <= 12); + // Strokes stay dark (darker, with the contrast), also in the gutter's light. + for (x, y) in [(150, 62), (61, 102), (238, 342)] { + assert!(grey_at(&after, x, y) <= 40, "stroke at ({x}, {y}): {}", grey_at(&after, x, y)); + } + for (x, y) in SPECKS { + assert!(grey_at(&before, x, y) < 60); + assert!(grey_at(&after, x, y) >= 200, "speck at ({x}, {y}): {}", grey_at(&after, x, y)); + } + // Everything but the image is as it was: the page, its invisible text layer, the comment, the + // font, the other pages. The image keeps its entries. + let now: Vec> = [3, 5, 6, 7, 8, 10, 11].iter().map(|&n| object(&s, id, n)).collect(); + assert_eq!(kept, now); + assert_eq!(s.get(id).unwrap().info.pages.len(), 3); + let text_after = s.get(id).unwrap().text_lines(0); + assert!(text_after.iter().any(|l| l.text.contains("Invisible OCR words")), "{text_after:?}"); + assert_eq!(text_before.len(), text_after.len()); + let Object::Stream(img) = &*object(&s, id, 4) else { panic!("the image") }; + assert_eq!(img.dict.get(b"Interpolate"), Some(&Object::Bool(true))); + assert_eq!(img.dict.name(b"ColorSpace"), Some(&b"DeviceGray"[..])); + + // Undo brings the original image, and the original file, back. + s.undo(id).unwrap(); + assert_eq!(samples(&s, id), before); + assert_eq!(s.get(id).unwrap().bytes, original); + assert_eq!(s.get(id).unwrap().can_undo(), None); +} + +#[test] +fn a_jpeg_scan_stays_a_jpeg_in_colour() { + let px = scan_pixels(3); + let (mut s, id) = open(scan_pdf(&format!("/Width {W} /Height {H} /ColorSpace /DeviceRGB /BitsPerComponent 8 /Filter /DCTDecode"), &jpeg(&px, 3))); + let report = s.enhance_scanned_pages(id, &[0], ScanCleanSettings::at(80)).unwrap(); + // Only page 1 was asked for; the image it shares with page 3 is changed for both. + assert_eq!((report.pages.as_slice(), report.images), (&[0][..], 1)); + let after = samples(&s, id); + assert!(after.jpeg && after.components == 3); + let Object::Stream(img) = &*object(&s, id, 4) else { panic!("the image") }; + assert_eq!(img.dict.name(b"Filter"), Some(&b"DCTDecode"[..])); + // Yellowed paper becomes white; the strokes stay dark. + let at = |x: usize, y: usize| after.data[(y * W + x) * 3..][..3].to_vec(); + for (x, y) in [(150, 40), (5, 300)] { + assert!(at(x, y).iter().all(|&v| v >= 230), "paper at ({x}, {y}): {:?}", at(x, y)); + } + assert!(at(150, 63).iter().all(|&v| v <= 60), "stroke: {:?}", at(150, 63)); + // The file still renders. + let doc = s.get(id).unwrap(); + let shot = pdfcraft_render::PageRenderer::new(doc.bytes.clone(), doc.config.clone()).render(pdfcraft_render::RenderRequest { + page: 0, + scale: 1.0, + ..Default::default() + }); + assert!(shot.error.is_none(), "{:?}", shot.error); +} + +#[test] +fn despeckle_is_its_own_switch() { + let speck = |settings: ScanCleanSettings| { + let (mut s, id) = open(grey_scan()); + s.enhance_scanned_pages(id, &[0], settings).unwrap(); + let after = samples(&s, id); + SPECKS.iter().map(|&(x, y)| grey_at(&after, x, y)).max().unwrap_or(0) + }; + assert!(speck(ScanCleanSettings { despeckle: false, ..ScanCleanSettings::at(80) }) < 100, "kept without despeckle"); + assert!(speck(ScanCleanSettings::at(80)) >= 200); + // Despeckle alone, at level 0, changes nothing but the specks. + let (mut s, id) = open(grey_scan()); + let before = samples(&s, id); + s.enhance_scanned_pages(id, &[], ScanCleanSettings { level: 0, edge_shadows: false, despeckle: true }).unwrap(); + let after = samples(&s, id); + let changed: Vec = before.data.iter().zip(&after.data).enumerate().filter(|(_, (a, b))| a != b).map(|(i, _)| i).collect(); + let mut specks: Vec = SPECKS.iter().map(|&(x, y)| y * W + x).collect(); + specks.sort_unstable(); + assert_eq!(changed, specks); + // Nothing to do: no undo step, and the report says why. + let (mut s, id) = open(grey_scan()); + let report = s.enhance_scanned_pages(id, &[], ScanCleanSettings { level: 0, edge_shadows: false, despeckle: false }).unwrap(); + assert_eq!(report.images, 0); + assert!(report.pages.is_empty()); + assert_eq!(report.skipped.len(), 1); + assert!(report.skipped[0].reason.contains("already clean"), "{:?}", report.skipped); + assert_eq!(s.get(id).unwrap().can_undo(), None); + assert!(!s.get(id).unwrap().dirty); +} + +#[test] +fn a_black_and_white_scan_stays_one_bit_and_loses_its_specks() { + // 64 × 64, one bit per pixel, 1 = white: a black bar and two specks. + let (w, h) = (64usize, 64usize); + let mut bits = vec![0xFFu8; w / 8 * h]; + let mut black = |x: usize, y: usize| bits[y * w / 8 + x / 8] &= !(0x80 >> (x % 8)); + for y in 20..30 { + for x in 10..50 { + black(x, y); + } + } + black(5, 5); + black(58, 58); + let bytes = Pdf::new() + .obj("<< /Type /Catalog /Pages 2 0 R >>") + .obj("<< /Type /Pages /Kids [3 0 R] /Count 1 >>") + .obj("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 64 64] /Resources << /XObject << /Im0 4 0 R >> >> /Contents 5 0 R >>") + .stream("/Type /XObject /Subtype /Image /Width 64 /Height 64 /ColorSpace /DeviceGray /BitsPerComponent 1", &bits) + .stream("", "q 64 0 0 64 0 0 cm /Im0 Do Q") + .build(); + let (mut s, id) = open(bytes); + // Only Despeckle changes a black and white scan: without it the image is reported. + let report = s.enhance_scanned_pages(id, &[], ScanCleanSettings { despeckle: false, ..ScanCleanSettings::at(100) }).unwrap(); + assert_eq!(report.images, 0); + assert!(report.skipped.first().is_some_and(|k| k.reason.contains("black and white")), "{:?}", report.skipped); + let report = s.enhance_scanned_pages(id, &[], ScanCleanSettings::at(100)).unwrap(); + assert_eq!(report.images, 1); + let Object::Stream(img) = &*object(&s, id, 4) else { panic!("the image") }; + assert_eq!(img.dict.int(b"BitsPerComponent"), Some(1)); + let px = samples(&s, id); + assert!(px.bilevel); + let at = |x: usize, y: usize| px.data[y * w + x]; + assert_eq!((at(5, 5), at(58, 58)), (255, 255), "specks gone"); + assert_eq!((at(10, 20), at(30, 25), at(49, 29)), (0, 0, 0), "the bar stays"); + assert_eq!(px.data.iter().filter(|&&v| v == 0).count(), 40 * 10); +} + +#[test] +fn hostile_images_are_reported_not_a_panic() { + // Each page draws one image over the whole page. + let images: [(&str, Vec); 6] = [ + ("/Width 100000 /Height 100000 /ColorSpace /DeviceGray /BitsPerComponent 8", vec![0; 64]), + ("/Width 300 /Height 400 /ColorSpace /DeviceGray /BitsPerComponent 8 /Filter /DCTDecode", vec![0xFF, 0xD8, 0xFF, 0xE0, 0, 3, 9]), + ("/Width 300 /Height 400 /ColorSpace /DeviceGray /BitsPerComponent 8", vec![200; 1000]), + ("/Width 1 /Height 1 /ColorSpace /DeviceGray /BitsPerComponent 8", vec![128]), + ("/Width 8 /Height 8 /ColorSpace /DeviceCMYK /BitsPerComponent 8", vec![0; 256]), + ("/Width 300 /Height 400 /ColorSpace /DeviceGray /BitsPerComponent 8 /Mask [250 255]", vec![200; 300 * 400]), + ]; + let n = images.len(); + let kids: String = (0..n).map(|i| format!("{} 0 R ", 3 + 3 * i)).collect(); + let mut pdf = Pdf::new().obj("<< /Type /Catalog /Pages 2 0 R >>").obj(format!("<< /Type /Pages /Kids [{kids}] /Count {n} >>")); + for (i, (dict, data)) in images.iter().enumerate() { + let (img, contents) = (4 + 3 * i, 5 + 3 * i); + pdf = pdf + .obj(format!( + "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 300 400] /Resources << /XObject << /Im0 {img} 0 R >> >> /Contents {contents} 0 R >>" + )) + .stream(&format!("/Type /XObject /Subtype /Image {dict}"), data) + .stream("", "q 300 0 0 400 0 0 cm /Im0 Do Q"); + } + let (mut s, id) = open(pdf.build()); + let report = s.enhance_scanned_pages(id, &[], ScanCleanSettings::at(100)).unwrap(); + assert_eq!(report.images, 0); + assert_eq!(report.not_scanned, vec![3], "a 1 × 1 image is no scan"); + let why: Vec<(usize, &str)> = report.skipped.iter().map(|k| (k.pages[0], k.reason.as_str())).collect(); + assert_eq!(why.len(), 5, "{why:?}"); + assert!(why[0].0 == 0 && why[0].1.contains("80 million"), "{why:?}"); + assert!(why[1].0 == 1 && why[1].1.contains("could not be decoded"), "{why:?}"); + assert!(why[2].0 == 2 && why[2].1.contains("shorter"), "{why:?}"); + assert!(why[3].0 == 4 && why[3].1.contains("colour space"), "{why:?}"); + assert!(why[4].0 == 5 && why[4].1.contains("colour-key mask"), "{why:?}"); + assert_eq!(s.get(id).unwrap().can_undo(), None); + // Pages that don't exist are an error, not a panic. + assert!(matches!(s.enhance_scanned_pages(id, &[99], ScanCleanSettings::default()), Err(EditError::Invalid(_)))); +} + +#[test] +fn a_scan_drawn_inside_a_form_xobject_is_reported_not_called_no_scan() { + // Pages 1 and 3 draw the scan through Form Fm0 (as some scanners write it), page 2 is text. + let page = |contents: usize| { + format!( + "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {W} {H}] /Resources << /XObject << /Fm0 4 0 R >> /Font << /F1 9 0 R >> >> /Contents {contents} 0 R >>" + ) + }; + let bytes = Pdf::new() + .obj("<< /Type /Catalog /Pages 2 0 R >>") + .obj("<< /Type /Pages /Kids [3 0 R 7 0 R 10 0 R] /Count 3 >>") + .obj(page(6)) + .stream( + &format!("/Type /XObject /Subtype /Form /BBox [0 0 {W} {H}] /Resources << /XObject << /Im0 5 0 R >> >>"), + format!("q {W} 0 0 {H} 0 0 cm /Im0 Do Q"), + ) + .stream(&format!("/Type /XObject /Subtype /Image /Width {W} /Height {H} /ColorSpace /DeviceGray /BitsPerComponent 8"), scan_pixels(1)) + .stream("", "/Fm0 Do") + .obj(page(8)) + .stream("", "BT /F1 12 Tf 20 380 Td (Real text) Tj ET") + .obj("<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica >>") + .obj(page(11)) + .stream("", "q /Fm0 Do Q") + .build(); + let (mut s, id) = open(bytes); + let before = object(&s, id, 5); + let report = s.enhance_scanned_pages(id, &[], ScanCleanSettings::at(100)).unwrap(); + assert!(report.pages.is_empty() && report.images == 0, "{report:?}"); + assert_eq!(report.not_scanned, vec![1], "only the text page is no scan"); + assert_eq!(report.skipped.len(), 1, "the form is reported once: {report:?}"); + let k = &report.skipped[0]; + assert_eq!((k.pages.as_slice(), k.name.as_str()), (&[0, 2][..], "Fm0")); + assert!(k.reason.contains("form XObject"), "{k:?}"); + assert_eq!(object(&s, id, 5), before, "the picture is left as it was"); + assert_eq!(s.get(id).unwrap().can_undo(), None); + + // A page of vector artwork in a Form, holding only a small logo, is no scan. + let logo = Pdf::new() + .obj("<< /Type /Catalog /Pages 2 0 R >>") + .obj("<< /Type /Pages /Kids [3 0 R] /Count 1 >>") + .obj(format!("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {W} {H}] /Resources << /XObject << /Fm0 4 0 R >> >> /Contents 6 0 R >>")) + .stream( + &format!("/Type /XObject /Subtype /Form /BBox [0 0 {W} {H}] /Resources << /XObject << /Im0 5 0 R >> >>"), + format!("0 0 {W} {H} re f q 40 0 0 20 10 370 cm /Im0 Do Q"), + ) + .stream("/Type /XObject /Subtype /Image /Width 40 /Height 20 /ColorSpace /DeviceGray /BitsPerComponent 8", [128; 800]) + .stream("", "/Fm0 Do") + .build(); + let (mut s, id) = open(logo); + let report = s.enhance_scanned_pages(id, &[], ScanCleanSettings::at(100)).unwrap(); + assert_eq!((report.not_scanned.as_slice(), report.skipped.len()), (&[0][..], 0), "{report:?}"); +} + +#[test] +fn a_page_mixing_its_own_image_and_a_scan_in_a_form_cleans_the_one_and_reports_the_other() { + // Page 1: its own image covers 40% and a Form holding a scan 20%: a scan only together. + // Page 2: its own image covers the page and the Form 20% more. Both pages clean Im1 and + // report Fm0. + let page = |contents: usize| { + format!( + "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {W} {H}] /Resources << /XObject << /Fm0 4 0 R /Im1 6 0 R >> >> /Contents {contents} 0 R >>" + ) + }; + let gray = format!("/Type /XObject /Subtype /Image /Width {W} /Height {H} /ColorSpace /DeviceGray /BitsPerComponent 8"); + let bytes = Pdf::new() + .obj("<< /Type /Catalog /Pages 2 0 R >>") + .obj("<< /Type /Pages /Kids [3 0 R 8 0 R] /Count 2 >>") + .obj(page(7)) + .stream( + &format!("/Type /XObject /Subtype /Form /BBox [0 0 {W} {H}] /Resources << /XObject << /Im0 5 0 R >> >>"), + format!("q {W} 0 0 {H} 0 0 cm /Im0 Do Q"), + ) + .stream(&gray, scan_pixels(1)) + .stream(&gray, scan_pixels(1)) + .stream("", format!("q {W} 0 0 160 0 240 cm /Im1 Do Q q 1 0 0 0.2 0 0 cm /Fm0 Do Q")) + .obj(page(9)) + .stream("", format!("q {W} 0 0 {H} 0 0 cm /Im1 Do Q q 1 0 0 0.2 0 0 cm /Fm0 Do Q")) + .build(); + let (mut s, id) = open(bytes); + let (form_picture, own) = (object(&s, id, 5), object(&s, id, 6)); + let report = s.enhance_scanned_pages(id, &[], ScanCleanSettings::at(100)).unwrap(); + assert_eq!((report.pages.as_slice(), report.images), (&[0, 1][..], 1), "Im1 is cleaned on both pages: {report:?}"); + assert!(report.not_scanned.is_empty(), "{report:?}"); + assert_eq!(report.skipped.len(), 1, "the form is reported once: {report:?}"); + let k = &report.skipped[0]; + assert_eq!((k.pages.as_slice(), k.name.as_str()), (&[0, 1][..], "Fm0")); + assert!(k.reason.contains("form XObject"), "{k:?}"); + assert_ne!(object(&s, id, 6), own, "the page's own image is cleaned"); + assert_eq!(object(&s, id, 5), form_picture, "the picture in the form is left as it was"); +} + +#[test] +fn protected_documents_keep_their_encryption_and_restrictions() { + let protect = |user: &str, permissions: i32| { + let mut doc = pdfcraft_cos::Document::open(Arc::new(grey_scan())).unwrap(); + doc.set_encryption(&pdfcraft_cos::NewEncryption { + algorithm: pdfcraft_cos::Algorithm::Aes256, + user_password: user, + owner_password: "owner", + permissions, + encrypt_metadata: true, + seed: [7; 32], + }) + .unwrap(); + Arc::new(pdfcraft_cos::write_full(&doc, &Default::default()).unwrap()) + }; + // Opened with the password: cleaned, saved still encrypted, and the clean scan reads back. + let mut s = Session::new(); + let id = s.open("x.pdf", None, protect("pw", -1), Some("pw")).unwrap(); + assert_eq!(s.enhance_scanned_pages(id, &[], ScanCleanSettings::at(100)).unwrap().images, 1); + let saved = s.save_bytes(id).unwrap(); + assert!(pdfcraft_cos::Document::open(saved.clone()).is_err(), "still protected"); + let id2 = s.open("y.pdf", None, saved, Some("pw")).unwrap(); + assert!(grey_at(&samples(&s, id2), 3, 40) >= 235); + // Changes withheld: refused before any work. + let id = s.open("r.pdf", None, protect("", 0b0100), None).unwrap(); + assert_eq!(s.enhance_scanned_pages(id, &[], ScanCleanSettings::default()).unwrap_err(), EditError::NotPermitted("changes to the document")); + assert!(!s.get(id).unwrap().dirty); +} + +#[test] +fn a_result_for_an_older_state_is_refused() { + let (mut s, id) = open(grey_scan()); + let job = s.scan_clean_job(id, &[], ScanCleanSettings::at(70)).unwrap(); + assert_eq!(job.total(), 1); + let cleaned = job.run(|_, _| true).unwrap(); + s.apply(id, Edit::RotatePages { pages: vec![1], degrees: 90 }).unwrap(); + assert!(matches!(s.apply_scan_clean(id, cleaned), Err(EditError::Invalid(_)))); + assert_eq!(s.get(id).unwrap().can_undo(), Some("Rotate page")); + // Cancelled from the progress callback. + let job = s.scan_clean_job(id, &[], ScanCleanSettings::at(70)).unwrap(); + assert!(matches!(job.run(|_, _| false), Err(EditError::Cancelled))); +} + +#[test] +fn edge_shadow_removal_whitens_the_gutter_background_removal_alone_leaves() { + let gutter = |edge_shadows: bool| { + let (mut s, id) = open(grey_scan()); + s.enhance_scanned_pages(id, &[0], ScanCleanSettings { edge_shadows, ..ScanCleanSettings::at(100) }).unwrap(); + let after = samples(&s, id); + [(3, 40), (3, 380), (10, 200)].map(|(x, y)| grey_at(&after, x, y)) + }; + assert!(gutter(false).iter().any(|&v| v < 200), "{:?}", gutter(false)); + assert!(gutter(true).iter().all(|&v| v >= 235), "{:?}", gutter(true)); + // On its own, at level 0, it lightens the gutter's paper to the paper beside it. + let (mut s, id) = open(grey_scan()); + let before = samples(&s, id); + s.enhance_scanned_pages(id, &[0], ScanCleanSettings { level: 0, edge_shadows: true, despeckle: false }).unwrap(); + let after = samples(&s, id); + assert!(grey_at(&after, 3, 40).abs_diff(grey_at(&before, 150, 40)) <= 4, "{} vs {}", grey_at(&after, 3, 40), grey_at(&before, 150, 40)); + assert_eq!(grey_at(&after, 150, 40), grey_at(&before, 150, 40), "the middle is left alone"); +} + +/// Two pages, each with its own scan image (objects 4 and 7), so cleaning one page can be told +/// apart from cleaning the other. +fn two_scans_pdf() -> Vec { + let dict = format!("/Type /XObject /Subtype /Image /Width {W} /Height {H} /ColorSpace /DeviceGray /BitsPerComponent 8"); + let draw = format!("q {W} 0 0 {H} 0 0 cm /Im0 Do Q"); + Pdf::new() + .obj("<< /Type /Catalog /Pages 2 0 R >>") + .obj("<< /Type /Pages /Kids [3 0 R 6 0 R] /Count 2 >>") + .obj(format!("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {W} {H}] /Resources << /XObject << /Im0 4 0 R >> >> /Contents 5 0 R >>")) + .stream(&dict, scan_pixels(1)) + .stream("", &draw) + .obj(format!("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {W} {H}] /Resources << /XObject << /Im0 7 0 R >> >> /Contents 8 0 R >>")) + .stream(&dict, scan_pixels(1)) + .stream("", &draw) + .build() +} + +#[test] +fn only_the_chosen_pages_scan_is_cleaned() { + // Clean page 1, then page 2 of a fresh copy: the other page's image stays byte-identical. + for (chosen, image, other) in [(0usize, 4u32, 7u32), (1, 7, 4)] { + let (mut s, id) = open(two_scans_pdf()); + let (image_before, other_before) = (object(&s, id, image), object(&s, id, other)); + assert_eq!(s.get(id).unwrap().page_images(chosen)[0].object, Some(ObjRef::new(image, 0)), "the fixture"); + let report = s.enhance_scanned_pages(id, &[chosen], ScanCleanSettings::at(100)).unwrap(); + assert_eq!((report.pages.as_slice(), report.images), (&[chosen][..], 1), "page {}", chosen + 1); + assert_ne!(object(&s, id, image), image_before, "page {}'s image was not cleaned", chosen + 1); + assert_eq!(object(&s, id, other), other_before, "cleaning page {} changed the other page's image", chosen + 1); + let (Object::Stream(was), Object::Stream(now)) = (&*other_before, &*object(&s, id, other)) else { panic!("the images") }; + assert_eq!(was.raw, now.raw, "the other image's stream bytes"); + assert_eq!(s.get(id).unwrap().can_undo(), Some("Enhance scanned file")); + } +} diff --git a/crates/engine/src/scan_fields.rs b/crates/engine/src/scan_fields.rs index 8287743fc5..05906d75b2 100644 --- a/crates/engine/src/scan_fields.rs +++ b/crates/engine/src/scan_fields.rs @@ -19,30 +19,94 @@ const SCALE: f32 = 2.0; const MAX_SIDE: f32 = 3000.0; /// A page is a scan when its images cover at least this share of it. const COVERED: f64 = 0.5; +/// A picture inside Form artwork is taken for a scan when it has at least this resolution over +/// the area the artwork covers (a quarter of the pixels a 72 dpi scan would have). +const MIN_FORM_DPI: f64 = 36.0; /// Recognised "words" less tall than this (points) are specks read as letters, and would make /// the box they sit on look filled in. const MIN_WORD_HEIGHT: f64 = 4.0; /// Whether `page` is a picture of a page: images cover most of it. pub(crate) fn is_scan(doc: &Document, page: usize, info: &PageInfo) -> bool { - let [cx0, cy0, cx1, cy1] = info.crop.map(f64::from); - let page_area = (cx1 - cx0) * (cy1 - cy0); - if !page_area.is_finite() || page_area <= 0.0 { - return false; - } - let covered: f64 = doc + scan_images(doc, page, info).is_some() +} + +/// The raster images (larger than one pixel) `page` draws, each with the share of the page (its +/// crop box) it covers, when together they cover most of it ([`is_scan`]); `None` otherwise. +/// Enhance ▸ Scanned document cleans these up ([`crate::scan_clean`]). +pub(crate) fn scan_images(doc: &Document, page: usize, info: &PageInfo) -> Option> { + let area = page_area(info)?; + let images: Vec<(pdfcraft_edit::PageImage, f64)> = doc .page_images(page) - .iter() + .into_iter() .filter(|i| !i.is_form && i.width > 1 && i.height > 1) - .map(|i| { - let r = i.visible_rect.unwrap_or(i.rect); - let w = (r[2].min(cx1) - r[0].max(cx0)).max(0.0); - let h = (r[3].min(cy1) - r[1].max(cy0)).max(0.0); - w * h - }) - .filter(|a| a.is_finite()) - .sum(); - covered >= COVERED * page_area + .filter_map(|i| Some((share(info, area, &i)?, i))) + .map(|(share, i)| (i, share)) + .collect(); + let covered: f64 = images.iter().map(|(_, share)| share).sum(); + (covered >= COVERED).then_some(images) +} + +/// A page that is a scan, classified in one pass: its raster images (larger than one pixel) and +/// the grouped Form artwork that holds a raster picture, each with the share of the page it +/// covers, when together they cover most of the page; `None` otherwise. The flag says whether the +/// raster images alone do ([`scan_images`]). Enhance ▸ Scanned document cleans the images and +/// reports the Forms, which it can't clean yet, so a page mixing both gets both. +#[allow(clippy::type_complexity)] +pub(crate) fn scan_parts( + doc: &Document, + page: usize, + info: &PageInfo, +) -> Option<(Vec<(pdfcraft_edit::PageImage, f64)>, Vec<(pdfcraft_edit::PageImage, f64)>, bool)> { + let area = page_area(info)?; + let editor = doc.editor.as_ref(); + let (mut images, mut forms) = (Vec::new(), Vec::new()); + let (mut own, mut covered) = (0.0, 0.0); + for i in doc.page_images(page) { + let Some(share) = share(info, area, &i) else { continue }; + if !i.is_form { + if i.width > 1 && i.height > 1 { + own += share; + covered += share; + images.push((i, share)); + } + } else if let (Some(form), Some(editor)) = (i.object, editor) + && holds_scan(&editor.cos, form, share * area) + { + covered += share; + forms.push((i, share)); + } + } + (covered >= COVERED).then_some((images, forms, own >= COVERED)) +} + +/// Whether Form `form`, covering `points` square points of the page, draws a picture that could +/// be a scan of it: one with at least [`MIN_FORM_DPI`] over that area, not a logo inside a page +/// of vector artwork. +fn holds_scan(cos: &pdfcraft_cos::Document, form: pdfcraft_cos::ObjRef, points: f64) -> bool { + let Some(picture) = crate::copy_image::picture_in_form(cos, form) else { return false }; + let obj = cos.get(picture); + let pdfcraft_cos::Object::Stream(s) = &*obj else { return false }; + let px = |k: &[u8]| s.dict.int(k).unwrap_or(0).max(0) as f64; + let per_point = MIN_FORM_DPI / 72.0; + px(b"Width") * px(b"Height") >= points * per_point * per_point +} + +/// The area of `info`'s crop box, when it is a usable one. +fn page_area(info: &PageInfo) -> Option { + let [cx0, cy0, cx1, cy1] = info.crop.map(f64::from); + let area = (cx1 - cx0) * (cy1 - cy0); + (area.is_finite() && area > 0.0).then_some(area) +} + +/// The share of the crop box (of area `area`) that `image` covers, when it is a number. +fn share(info: &PageInfo, area: f64, image: &pdfcraft_edit::PageImage) -> Option { + let [cx0, cy0, cx1, cy1] = info.crop.map(f64::from); + let r = image.visible_rect.unwrap_or(image.rect); + let w = (r[2].min(cx1) - r[0].max(cx0)).max(0.0); + let h = (r[3].min(cy1) - r[1].max(cy0)).max(0.0); + let share = w * h / area; + share.is_finite().then_some(share) } /// The page as displayed, in points with y up: where detection on a scan works. @@ -122,10 +186,14 @@ impl Session { if !ocr::available() { return Vec::new(); } - let (Ok(engine), Some(job)) = (ocr::engine(), self.ocr_job(id, &[page], ocr::OcrSettings::default())) else { + let settings = ocr::OcrSettings::default(); + let Some(job) = self.ocr_job(id, &[page], settings.clone()) else { return Vec::new(); }; - job.run(&engine, |_, _| true) + let Ok(recognizers) = ocr::recognizers(&settings) else { + return Vec::new(); + }; + job.run(&recognizers, |_, _| true) .into_iter() .flat_map(|p| p.words) .filter_map(|w| { diff --git a/crates/engine/src/session/files.rs b/crates/engine/src/session/files.rs index 553ddd6aa3..44bed85cde 100644 --- a/crates/engine/src/session/files.rs +++ b/crates/engine/src/session/files.rs @@ -114,6 +114,19 @@ impl Session { .map_err(EditError::Write) } + /// XFDF with these comments (0-based page, position in the page's `/Annots`) and their + /// replies: what Copy puts on the clipboard; [`Edit::PasteComments`] takes it back in. + pub fn export_comments(&self, id: DocId, which: &[(usize, usize)]) -> Result { + let doc = self.get(id).ok_or(EditError::NoDocument)?; + let cos = match doc.editor.as_ref() { + Some(e) => e.cos.clone(), + None => { + pdfcraft_cos::Document::open_with_password(doc.bytes.clone(), doc.password.as_deref()).map_err(|e| EditError::Write(e.to_string()))? + } + }; + guard(|| pdfcraft_xfdf::export_xfdf_comments(&cos, which, &doc.name)).map_err(EditError::Write) + } + /// The print-ready PDF for `settings` (sheets laid out for the paper; see `pdfcraft-print`). pub fn print_pdf(&self, id: DocId, settings: &print::Settings) -> Result, EditError> { let doc = self.get(id).ok_or(EditError::NoDocument)?; @@ -384,7 +397,7 @@ impl Session { /// A new file (Combine, Extract, Split, Create) as bytes, checked like a save before anyone /// writes it (#1698, #2166): it reopens, and shows the pages it was built from. - fn write_new(&self, doc: &pdfcraft_cos::Document) -> Result>, EditError> { + pub(crate) fn write_new(&self, doc: &pdfcraft_cos::Document) -> Result>, EditError> { let bytes = self.write_unchecked(doc)?; let expected = pdfcraft_organize::page_count(doc)?; pdfcraft_cos::Document::open(bytes.clone()).map_err(|e| EditError::Reopen(e.to_string()))?; diff --git a/crates/engine/src/session/view.rs b/crates/engine/src/session/view.rs index c0e7417c8e..5ef135d6c8 100644 --- a/crates/engine/src/session/view.rs +++ b/crates/engine/src/session/view.rs @@ -168,6 +168,14 @@ impl Session { Some(BookmarkPage { rows, next: (end < all.len()).then_some(end), truncated: doc.info.outline_more }) } + /// The bookmark at `path` with everything under it, or the whole outline for `[]`, as a tree + /// another document can take ([`crate::Edit::ImportBookmarks`]). `None` when no document has + /// `id`, or its structure could not be read for editing. + pub fn outline_tree(&self, id: DocId, path: &[usize]) -> Option> { + let editor = self.get(id)?.editor.as_ref()?; + Some(pdfcraft_organize::outline_tree(&editor.cos, path)) + } + /// Top-level bookmarks as split points: (first page of each part, its bookmark's title). pub fn bookmark_splits(&self, id: DocId) -> Vec<(usize, String)> { let Some(doc) = self.get(id) else { return Vec::new() }; diff --git a/crates/engine/src/tests.rs b/crates/engine/src/tests.rs index 5d95486c32..688dcec049 100644 --- a/crates/engine/src/tests.rs +++ b/crates/engine/src/tests.rs @@ -549,7 +549,86 @@ fn pdf_page_stamps_fit_inside_an_explicit_box() { assert!(content.contains("/Pic Do") && !content.contains("BI"), "{content}"); } -fn page_texts(s: &Session, id: DocId) -> Vec { +/// #1297: a 3D model on a page keeps the whole box as its rectangle and draws a PDF poster as a +/// form XObject fitted inside it, keeping its shape and centred. +#[test] +fn a_3d_model_gets_a_fitted_vector_poster_and_a_file_that_is_not_one_is_refused() { + let mut session = Session::new().with_clock(|| 1_700_000_000); + let blank = session.create_blank(595.0, 842.0, 1).unwrap(); + let id = session.open_new("model.pdf", blank).unwrap(); + let mut prc = b"PRC".to_vec(); + prc.extend_from_slice(&[1; 64]); + let add = |model: Vec| Edit::AddModel3d { + page: 0, + rect: [56.0, 400.0, 539.0, 700.0], + model: Arc::new(model), + poster: Some(MarkFile { name: "poster.pdf".into(), bytes: Arc::new(half_red_page()), page: 0 }), + activation: Activation3d::default(), + name: "Bracket".into(), + }; + assert!(session.apply(id, add(b"%PDF-1.7 not a model".to_vec())).is_err()); + session.apply(id, add(prc)).unwrap(); + + let doc = pdfcraft_cos::Document::open(session.save_bytes(id).unwrap()).unwrap(); + let page = &pdfcraft_model::pages(&doc)[0]; + let annots = doc.resolve(page.dict.get(b"Annots").unwrap()); + let [only] = annots.as_array().unwrap().as_slice() else { panic!("the refused model added nothing: {annots:?}") }; + let annot = doc.get(only.as_ref().unwrap()); + let annot = annot.as_dict().unwrap(); + assert_eq!(annot.name(b"Subtype"), Some(&b"3D"[..])); + let rect: Vec = annot.get(b"Rect").unwrap().as_array().unwrap().iter().filter_map(pdfcraft_cos::Object::as_f64).collect(); + assert_eq!(rect, [56.0, 400.0, 539.0, 700.0], "the whole box, not the fitted picture"); + assert_eq!(doc.resolve(annot.get(b"3DA").unwrap()).as_dict().unwrap().name(b"A"), Some(&b"XA"[..])); + let ap = doc.resolve(annot.get(b"AP").unwrap()); + let pdfcraft_cos::Object::Stream(normal) = doc.get(ap.as_dict().unwrap().reference(b"N").unwrap()).as_ref().clone() else { + panic!("an appearance stream") + }; + let pic = doc.resolve(normal.dict.get(b"Resources").unwrap()).as_dict().unwrap().clone(); + let pic = doc.resolve(pic.get(b"XObject").unwrap()).as_dict().unwrap().reference(b"Pic").unwrap(); + let pdfcraft_cos::Object::Stream(picture) = doc.get(pic).as_ref().clone() else { panic!("the picture is a stream") }; + assert_eq!(picture.dict.name(b"Subtype"), Some(&b"Form"[..]), "the page is kept as a form XObject"); + // The 80 × 40 page in the 483 × 300 box: 483 × 241.5 from y 429.25. + let content = String::from_utf8(normal.decoded().unwrap().to_vec()).unwrap(); + assert!(content.contains(" 56 429.25 cm /Pic Do"), "{content}"); +} + +/// A PDF poster whose page has no area is refused rather than drawn as nothing. +#[test] +fn a_3d_poster_page_without_area_is_refused() { + let mut session = Session::new().with_clock(|| 1_700_000_000); + let blank = session.create_blank(595.0, 842.0, 1).unwrap(); + let id = session.open_new("model.pdf", blank).unwrap(); + let flat = String::from_utf8(half_red_page()).unwrap().replace("[0 0 80 40]", "[0 0 80 0 ]"); + let add = Edit::AddModel3d { + page: 0, + rect: [56.0, 400.0, 539.0, 700.0], + model: Arc::new(b"PRC rest".to_vec()), + poster: Some(MarkFile { name: "flat.pdf".into(), bytes: Arc::new(flat.into_bytes()), page: 0 }), + activation: Activation3d::default(), + name: "Bracket".into(), + }; + assert!(matches!(session.apply(id, add), Err(EditError::Source(_)))); + assert!(session.get(id).unwrap().info.annotations.is_empty(), "nothing was added"); +} + +/// Adding a 3D model is a comment change: refused without the comments permission, and only the +/// comments are rebuilt after it. +#[test] +fn a_3d_model_needs_the_comments_permission_and_changes_the_comments() { + let edit = Edit::AddModel3d { + page: 0, + rect: [0.0, 0.0, 100.0, 100.0], + model: Arc::new(b"PRC rest".to_vec()), + poster: None, + activation: Activation3d::default(), + name: String::new(), + }; + let p = pdfcraft_cos::Permissions { bits: 0, owner: false, revision: 3 }; + assert_eq!(check_permission(&edit, &p), Err(EditError::NotPermitted("comments"))); + assert_eq!(crate::history::scope_of(&edit), crate::history::Scope::Comments); +} + +pub(crate) fn page_texts(s: &Session, id: DocId) -> Vec { let doc = s.get(id).unwrap(); let config = pdfcraft_render::RenderConfig { password: doc.password.as_deref().map(Arc::from), ..Default::default() }; let mut r = pdfcraft_render::PageRenderer::new(doc.bytes.clone(), config); @@ -2132,7 +2211,7 @@ fn rect_comment(page: usize, rect: [f64; 4]) -> Edit { Edit::AddAnnotation(NewAnnotation { page, shape: Shape::Rectangle { rect }, - style: Style { color: [1.0, 0.0, 0.0], opacity: 1.0, width: 2.0, fill: Some([1.0, 0.0, 0.0]) }, + style: Style { color: [1.0, 0.0, 0.0], opacity: 1.0, width: 2.0, fill: Some([1.0, 0.0, 0.0]), ..Style::default() }, contents: "Look here".into(), author: "Reviewer".into(), }) @@ -5969,7 +6048,7 @@ fn measurement_edits_are_transactional_and_honour_annotation_permissions() { assert!(session.get(id).unwrap().measurements().unwrap().measurements.is_empty()); session.redo(id).unwrap(); assert_eq!(session.get(id).unwrap().bytes, before); - let p = pdfcraft_cos::Permissions { bits: 0, owner: false }; + let p = pdfcraft_cos::Permissions { bits: 0, owner: false, revision: 3 }; assert_eq!(check_permission(&Edit::AddMeasurement(m), &p), Err(EditError::NotPermitted("comments"))); assert_eq!( check_permission( @@ -7243,12 +7322,13 @@ fn datasets_streams(bytes: &Arc>) -> usize { /// and noted. #[test] fn hostile_xfa_open_scripts_are_merged_capped_kept_from_printing_and_noted() { - // On open: a calculate setting its own value 50 000 times, an initialize script showing a - // thousand messages and asking to print, open a link and save. + // On open: a calculate setting its own value 20 000 times (50 000 sat right at the 1 s + // unasked-script timeout and abandoned whenever the test machine was loaded), an + // initialize script showing a thousand messages and asking to print, open a link and save. let tpl = pdfcraft_xfa::fixtures::scripted_template() .replace( r#""#, - r#""#, + r#""#, ) .replace( r#"if (qty.rawValue === null) qty.rawValue = 2;"#, @@ -7258,9 +7338,11 @@ fn hostile_xfa_open_scripts_are_merged_capped_kept_from_printing_and_noted() { let mut s = Session::new().with_clock(|| 1_700_000_000); let started = std::time::Instant::now(); let id = s.open("hostile.pdf", None, bytes, None).expect("opens"); - assert!(started.elapsed() < std::time::Duration::from_secs(5), "{:?}", started.elapsed()); + // 30 s and not 5 s: the run is wall-clock timed and a loaded test machine (all suites in + // parallel) can take several seconds for the 20 000 iterations; a real hang still fails. + assert!(started.elapsed() < std::time::Duration::from_secs(30), "{:?}", started.elapsed()); let doc = s.get(id).unwrap(); - assert_eq!(doc.form.iter().find(|f| f.name == "total").unwrap().value, vec!["49999".to_string()]); + assert_eq!(doc.form.iter().find(|f| f.name == "total").unwrap().value, vec!["19999".to_string()]); // One datasets stream for the scripts' revision, not one per value set. assert!(datasets_streams(&doc.bytes) <= 3, "{} datasets streams", datasets_streams(&doc.bytes)); // What opening changed is noted with the form's other XFA notes. @@ -10106,6 +10188,42 @@ fn need_appearances_buttons_are_displayed_from_their_captions() { } } +/// Copy and paste: the copies of a comment and its reply land on the asked-for page of another +/// document under names of their own (nothing replaced; a second paste gives a second set), +/// undo takes a paste away as one step, and a clipboard without comments is refused. +#[test] +fn pasting_copied_comments_adds_copies_that_undo_as_one_step() { + let (mut s, id) = session_with(2); + s.apply(id, rect_comment(1, [80.0, 80.0, 120.0, 120.0])).unwrap(); + s.apply(id, Edit::ReplyToAnnotation { page: 1, index: 0, text: "Done".into(), author: "Ada".into() }).unwrap(); + let xfdf = s.export_comments(id, &[(1, 0)]).unwrap(); + assert!(xfdf.contains(" MutexGuard<'static, ()> { + static LOCK: Mutex<()> = Mutex::new(()); + let lock = LOCK.lock().unwrap_or_else(PoisonError::into_inner); + use_form_date_language("en"); + lock +} + +/// `util.printd(fmt)` of 18 October 2026, then of what `util.scand(fmt, typed)` reads. +fn printd(fmt: &str, typed: &str) -> String { + let doc = DocInfo { file_name: "form.pdf".into(), num_pages: 1, page: 0, info: Vec::new(), xfa: false }; + let script = format!("event.value = util.printd('{fmt}', new Date(2026, 9, 18)) + '|' + util.printd('{fmt}', util.scand('{fmt}', '{typed}'));"); + let o = run(&script, &Event::field("Format", "date", ""), &doc, &[], &[], Limits::default()); + assert_eq!(o.error, None); + o.value +} + +#[test] +fn form_dates_follow_the_date_language() { + let _names = names_lock(); + let oct18 = DateTime { y: 2026, m: 10, d: 18, hh: 0, mm: 0, ss: 0 }; + let dd_mmm_yy = Format::Date("dd-mmm-yy".into()); + // English until the app says otherwise. + assert_eq!(format_date(oct18, "dd-mmm-yy dddd ddd mmmm"), "18-Oct-26 Sunday Sun October"); + use_form_date_language("it"); + // AFDate_FormatEx("dd-mmm-yy"), what the issue's form used. + assert_eq!(format_value(&dd_mmm_yy, "18-10-26"), "18-ott-26"); + assert_eq!(format_date(oct18, "dddd d mmmm yyyy, ddd"), "domenica 18 ottobre 2026, dom"); + // What it writes reads back, and so does English typing. + assert_eq!(parse_date("18-ott-26", "dd-mmm-yy"), Some(oct18)); + assert_eq!(parse_date("18 ottobre 2026", "dd mmmm yyyy"), Some(oct18)); + assert_eq!(parse_date("18-Oct-26", "dd-mmm-yy"), Some(oct18)); + assert!(keystroke(&dd_mmm_yy, "date", "18-ott-26").is_ok()); + // util.printd writes Italian names; util.scand reads them. + assert_eq!(printd("dd-mmm-yy", "18-ott-26"), "18-ott-26|18-ott-26"); + assert_eq!(printd("dd-mmm-yy", "18-Oct-26"), "18-ott-26|18-ott-26", "English typing reads too"); + assert_eq!(printd("dddd d mmmm yyyy", "domenica 18 ottobre 2026"), "domenica 18 ottobre 2026|domenica 18 ottobre 2026"); + // German short names end in a dot; they are written and read whole. + use_form_date_language("de"); + assert_eq!(format_value(&dd_mmm_yy, "18-10-26"), "18-Okt.-26"); + assert_eq!(parse_date("18-Okt.-26", "dd-mmm-yy"), Some(oct18)); + // An unknown language is English again. + use_form_date_language("xx"); + assert_eq!(format_value(&dd_mmm_yy, "18-10-26"), "18-Oct-26"); + assert_eq!(printd("dd-mmm-yy", "18-Oct-26"), "18-Oct-26|18-Oct-26"); + every_date_language_reads_back_what_it_writes(); + // Hostile typing never panics and stays quick: long, odd and mixed-script text. + use_form_date_language("th"); + let junk = ["", " ", "ม.", "ม.ค.ม.ค.ม.ค.", "\u{0}", "1月1月1月", "tháng tháng", "Ottobre", "🙂ott🙂", "İstanbul ott 26"]; + for t in junk.iter().map(|s| s.to_string()).chain(["ม.ค.1".repeat(5000), "9".repeat(100_000)]) { + let started = std::time::Instant::now(); + let _ = parse_date(&t, "dd-mmm-yy"); + let _ = keystroke(&dd_mmm_yy, "date", &t); + assert!(started.elapsed() < std::time::Duration::from_secs(2), "{} chars took {:?}", t.len(), started.elapsed()); + } + use_form_date_language("en"); +} + +/// An English weekday typed before the month fills the format's weekday, so a month name that +/// the date language shares with a weekday (Italian, Spanish, Romanian and Albanian "mar") is +/// still read as the month: "Tue 17 Mar 2026" read as "Tue 17 Tuesday 2026" (#5303). +#[test] +fn an_english_weekday_before_a_shared_month_name_reads_as_english() { + let _names = names_lock(); + let doc = DocInfo { file_name: "form.pdf".into(), num_pages: 1, page: 0, info: Vec::new(), xfa: false }; + let mar17 = DateTime { y: 2026, m: 3, d: 17, hh: 0, mm: 0, ss: 0 }; + let mut bad = Vec::new(); + for language in ["it", "es", "ro", "sq"] { + use_form_date_language(language); + let ddd_d_mmm = "ddd d mmm yyyy"; + let typed = [ + // What the language writes, then English, then English and localized mixed. + (ddd_d_mmm, format_date(mar17, ddd_d_mmm)), + (ddd_d_mmm, "Tue 17 Mar 2026".into()), + (ddd_d_mmm, "Tuesday 17 Mar 2026".into()), + (ddd_d_mmm, "tue 17 mar 2026".into()), + (ddd_d_mmm, format!("Tue 17 {}", format_date(mar17, "mmm yyyy"))), + (ddd_d_mmm, format!("{} 17 Mar 2026", format_date(mar17, "ddd"))), + ("mmm d yyyy ddd", "Mar 17 2026 Tue".into()), + ("dddd d mmmm yyyy", "Tuesday 17 March 2026".into()), + ]; + for (fmt, text) in typed { + if parse_date(&text, fmt) != Some(mar17) { + bad.push(format!("{language} forms {fmt}: {text:?} -> {:?}", parse_date(&text, fmt))); + } + let script = format!("var d = util.scand('{fmt}', {text:?}); event.value = d ? util.printd('yyyy-mm-dd', d) : 'null';"); + let o = run(&script, &Event::field("Format", "date", ""), &doc, &[], &[], Limits::default()); + if o.value != "2026-03-17" { + bad.push(format!("{language} scand {fmt}: {text:?} -> {:?} {:?}", o.value, o.error)); + } + } + } + use_form_date_language("en"); + assert!(bad.is_empty(), "{}", bad.join("; ")); +} + +/// Every date language: what a date field writes reads back as the same date, for each date +/// letter that writes a name, through the form path and through `util.scand`. Called from the +/// one test, which owns the process-wide names. +fn every_date_language_reads_back_what_it_writes() { + let doc = DocInfo { file_name: "form.pdf".into(), num_pages: 1, page: 0, info: Vec::new(), xfa: false }; + let mut bad = Vec::new(); + for l in pdfcraft_engine::dates::DATE_LANGUAGES.iter() { + use_form_date_language(l.code); + for m in 1..=12u32 { + let dt = DateTime { y: 2026, m, d: 18, hh: 0, mm: 0, ss: 0 }; + for fmt in ["dd-mmm-yy", "d mmmm yyyy", "dddd, mmmm d, yyyy", "ddd d mmm yyyy", "mmm d, yyyy ddd", "yyyymmmdd"] { + let text = format_date(dt, fmt); + // A short month ending in a number ("Thg 1") glued to the day is ambiguous + // ("Thg 118": January 18 or November 8): nothing can read it back. + if fmt == "yyyymmmdd" && l.months_short.iter().any(|n| n.ends_with(|c: char| c.is_ascii_digit())) { + continue; + } + if parse_date(&text, fmt).map(|p| (p.y, p.m, p.d)) != Some((2026, m, 18)) { + bad.push(format!("{} forms {fmt}: {text:?}", l.code)); + } + let script = format!("var d = util.scand('{fmt}', {text:?}); event.value = d ? util.printd('yyyy-mm-dd', d) : 'null';"); + let o = run(&script, &Event::field("Format", "date", ""), &doc, &[], &[], Limits::default()); + if o.value != format!("2026-{m:02}-18") { + bad.push(format!("{} scand {fmt}: {text:?} -> {:?} {:?}", l.code, o.value, o.error)); + } + } + } + } + use_form_date_language("en"); + assert!(bad.is_empty(), "{} failures:\n{}", bad.len(), bad.iter().take(40).cloned().collect::>().join("\n")); +} + +/// A form whose text fields are each `(name, font, format script)`: `/Helv` is Helvetica with +/// WinAnsi, `/Song` a Chinese font with a Unicode CMap, which also holds Cyrillic. +fn form(fields: &[(&str, &str, &str)]) -> Document { + let n = fields.len(); + let first = 6; + let refs = |from: usize| (from..from + n).map(|i| format!("{i} 0 R")).collect::>().join(" "); + let mut objs: Vec = vec![ + "<< /Type /Catalog /Pages 2 0 R /AcroForm 4 0 R >>".into(), + "<< /Type /Pages /Kids [3 0 R] /Count 1 /MediaBox [0 0 600 800] >>".into(), + format!("<< /Type /Page /Parent 2 0 R /Annots [{}] >>", refs(first)), + format!("<< /Fields [{}] /DA (/Helv 0 Tf 0 g) /DR << /Font << /Helv 5 0 R /Song {} 0 R >> >> >>", refs(first), first + n), + "<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica /Encoding /WinAnsiEncoding >>".into(), + ]; + for (i, (name, font, js)) in fields.iter().enumerate() { + let js = js.replace('\\', "\\\\").replace('(', "\\(").replace(')', "\\)"); + let y = 700 - 30 * i; + objs.push(format!( + "<< /Type /Annot /Subtype /Widget /FT /Tx /T ({name}) /DA (/{font} 10 Tf 0 g) /Rect [50 {y} 300 {}] /P 3 0 R /AA << /F << /S /JavaScript /JS ({js}) >> >> >>", + y + 20 + )); + } + objs.push("<< /Type /Font /Subtype /Type0 /BaseFont /STSong-Light /Encoding /UniGB-UCS2-H /DescendantFonts [<< /Type /Font /Subtype /CIDFontType0 /BaseFont /STSong-Light /CIDSystemInfo << /Registry (Adobe) /Ordering (GB1) /Supplement 4 >> >>] >>".into()); + let mut out = b"%PDF-1.7 +" + .to_vec(); + let mut offsets = Vec::new(); + for (i, o) in objs.iter().enumerate() { + offsets.push(out.len()); + out.extend_from_slice( + format!( + "{} 0 obj +{o} +endobj +", + i + 1 + ) + .as_bytes(), + ); + } + let xref = out.len(); + out.extend_from_slice( + format!( + "xref +0 {} +0000000000 65535 f +", + objs.len() + 1 + ) + .as_bytes(), + ); + for o in offsets { + out.extend_from_slice( + format!( + "{o:010} 00000 n +" + ) + .as_bytes(), + ); + } + out.extend_from_slice( + format!( + "trailer +<< /Size {} /Root 1 0 R >> +startxref +{xref} +%%EOF +", + objs.len() + 1 + ) + .as_bytes(), + ); + Document::open(Arc::new(out)).expect("opens") +} + +/// The content of field `name`'s normal appearance. +fn appearance(doc: &Document, name: &str) -> Vec { + let all = pdfcraft_forms::fields(doc); + let f = all.iter().find(|f| f.name == name).expect("field"); + let widget = doc.get(f.widgets[0].obj); + let normal = widget.as_dict().and_then(|d| d.get(b"AP")).and_then(|a| doc.resolve(a).as_dict().and_then(|a| a.reference(b"N"))).expect("/AP /N"); + match &*doc.get(normal) { + Object::Stream(s) => s.decoded().expect("decodes"), + other => panic!("not a stream: {other:?}"), + } +} + +/// What a Unicode-CMap (UCS-2) font is given for `text`. +fn ucs2(text: &str) -> Vec { + text.encode_utf16().flat_map(u16::to_be_bytes).collect() +} + +fn contains(hay: &[u8], needle: &[u8]) -> bool { + hay.windows(needle.len()).any(|w| w == needle) +} + +/// Fills `form`'s field `name` with 18 October 2026 through the JavaScript runner. +fn fill(doc: &mut Document, name: &str) { + fill_with(doc, name, "18-10-26"); +} + +/// Fills `form`'s field `name` with `value` through the JavaScript runner. +fn fill_with(doc: &mut Document, name: &str, value: &str) { + let mut js = JsRunner::new(doc, "form.pdf"); + pdfcraft_forms::set_value_with(doc, name, &FieldValue::Text(value.into()), &mut js).expect("fills"); +} + +/// Month names in the date language that no font of the field can draw are drawn in English, +/// never as `?` ("18-????.-26" in Russian); a font that holds them keeps them (#5303). +#[test] +fn date_names_no_field_font_can_draw_are_drawn_in_english() { + let _names = names_lock(); + let date = "AFDate_FormatEx(\"dd-mmm-yy\");"; + let fields = [("helv", "Helv", date), ("song", "Song", date)]; + for (language, local) in [("ru", "окт."), ("cs", "říj"), ("ja", "10月"), ("el", "Οκτ")] { + use_form_date_language(language); + let mut doc = form(&fields); + fill(&mut doc, "helv"); + fill(&mut doc, "song"); + let helv = appearance(&doc, "helv"); + assert!(contains(&helv, b"(18-Oct-26) Tj"), "{language}: {}", String::from_utf8_lossy(&helv)); + assert!(!helv.contains(&b'?'), "{language}: {}", String::from_utf8_lossy(&helv)); + assert!(pdfcraft_forms::undrawn_fields(&doc).is_empty(), "{language}"); + let song = appearance(&doc, "song"); + assert!(contains(&song, &ucs2(&format!("18-{local}-26"))), "{language} keeps its name: {song:?}"); + } + // Names Helvetica holds are kept: Italian "ott". + use_form_date_language("it"); + let mut doc = form(&fields); + fill(&mut doc, "helv"); + assert!(contains(&appearance(&doc, "helv"), b"(18-ott-26) Tj")); + use_form_date_language("en"); +} + +/// A Format script's `util.printd` is run again with English names for a field whose fonts +/// can't draw the date language's, and kept where they can (#5303). +#[test] +fn format_scripts_print_dates_in_english_where_the_field_cannot_draw_them() { + let _names = names_lock(); + let script = "event.value = event.value ? util.printd(\"dd-mmm-yy\", util.scand(\"dd-mm-yy\", event.value)) + \" !\" : \"\";"; + let fields = [("helv", "Helv", script), ("song", "Song", script)]; + use_form_date_language("ru"); + let mut doc = form(&fields); + fill(&mut doc, "helv"); + fill(&mut doc, "song"); + let helv = appearance(&doc, "helv"); + assert!(contains(&helv, b"(18-Oct-26 !) Tj"), "{}", String::from_utf8_lossy(&helv)); + assert!(!helv.contains(&b'?'), "{}", String::from_utf8_lossy(&helv)); + assert!(pdfcraft_forms::undrawn_fields(&doc).is_empty()); + assert!(contains(&appearance(&doc, "song"), &ucs2("18-окт.-26 !"))); + // A script whose text has no English either keeps it, reported as undrawn. + let greeting = [("hi", "Helv", "event.value = \"Привет \" + util.printd(\"mmm\", new Date(2026, 9, 18));")]; + let mut doc = form(&greeting); + fill(&mut doc, "hi"); + assert_eq!(pdfcraft_forms::undrawn_fields(&doc), [("hi".to_string(), 'П')]); + // The English run is the appearance's alone: scripts go on printing the date language. + assert_eq!(printd("dd-mmm-yy", "18-окт.-26"), "18-окт.-26|18-окт.-26"); + use_form_date_language("en"); +} + +/// A Format script's English run is kept when it ran, even where it prints the stored value +/// again: a Helvetica field storing "18-Oct-26", printed again as `dd-mmm-yy`, showed +/// "18-????.-26" in Russian. A script that fails keeps what it showed, reported undrawn (#5303). +#[test] +fn an_english_run_that_prints_the_stored_value_is_drawn() { + let _names = names_lock(); + let script = "event.value = util.printd(\"dd-mmm-yy\", util.scand(\"dd-mmm-yy\", event.value));"; + use_form_date_language("ru"); + let mut doc = form(&[("helv", "Helv", script), ("fails", "Helv", "throw new Error(\"no\");")]); + fill_with(&mut doc, "helv", "18-Oct-26"); + let helv = appearance(&doc, "helv"); + assert!(contains(&helv, b"(18-Oct-26) Tj"), "{}", String::from_utf8_lossy(&helv)); + assert!(!helv.contains(&b'?'), "{}", String::from_utf8_lossy(&helv)); + assert!(pdfcraft_forms::undrawn_fields(&doc).is_empty()); + fill_with(&mut doc, "fails", "18-окт.-26"); + assert_eq!(pdfcraft_forms::undrawn_fields(&doc), [("fails".to_string(), 'о')]); + use_form_date_language("en"); +} + +/// What `util.scand(fmt, text)` reads, as `yyyy-mm-dd`, or `null`. +fn scanned(fmt: &str, text: &str) -> String { + let doc = DocInfo { file_name: "form.pdf".into(), num_pages: 1, page: 0, info: Vec::new(), xfa: false }; + let script = format!("var d = util.scand('{fmt}', {text:?}); event.value = d ? util.printd('yyyy-mm-dd', d) : 'null';"); + let o = run(&script, &Event::field("Format", "date", ""), &doc, &[], &[], Limits::default()); + assert_eq!(o.error, None); + o.value +} + +/// A short weekday typed without its dot ("mar" for French "mar.", Tuesday) is the weekday where +/// the format has one, not the English "Mar" (March) that is the only other name of its length: +/// "mar 6 janv 2026" read as no date in French and Romanian (#5303). +#[test] +fn a_short_weekday_typed_without_its_dot_is_the_weekday() { + let _names = names_lock(); + let mut bad = Vec::new(); + let cases = [ + ("fr", "ddd d mmm yyyy", "Mar 6 janv 2026", "2026-01-06"), + ("fr", "ddd d mmm yyyy", "mar 17 mars 2026", "2026-03-17"), + ("fr", "mmm d yyyy ddd", "janv 6 2026 mar", "2026-01-06"), + ("fr", "ddd dd/mm/yyyy", "mar 06/01/2026", "2026-01-06"), + ("ro", "ddd d mmm yyyy", "mar 6 ian 2026", "2026-01-06"), + ("ro", "ddd dd/mm/yyyy", "mar 06/01/2026", "2026-01-06"), + ]; + for (language, fmt, text, want) in cases { + use_form_date_language(language); + let got = scanned(fmt, text); + if got != want { + bad.push(format!("{language} scand {fmt}: {text:?} -> {got}")); + } + let read = parse_date(text, fmt).map(|d| format!("{:04}-{:02}-{:02}", d.y, d.m, d.d)); + if read.as_deref() != Some(want) { + bad.push(format!("{language} forms {fmt}: {text:?} -> {read:?}")); + } + } + use_form_date_language("en"); + assert!(bad.is_empty(), "{}", bad.join("; ")); +} + +/// A weekday left out of a format with one ("17 mar 2026" for "ddd d mmm yyyy") lets a name the +/// date language shares between a weekday and a month be the month: Italian, Spanish and +/// Albanian "mar" (Tuesday, March) and Albanian "sht" (Saturday, September) were read as the +/// weekday, leaving no month (#5303). Typed weekdays still go first. +#[test] +fn a_shared_name_is_the_month_when_the_weekday_is_left_out() { + let _names = names_lock(); + let mut bad = Vec::new(); + let cases = [ + ("it", "ddd d mmm yyyy", "17 Mar 2026", (2026, 3, 17)), + ("es", "ddd d mmm yyyy", "17 Mar 2026", (2026, 3, 17)), + ("sq", "ddd d mmm yyyy", "17 Mar 2026", (2026, 3, 17)), + ("it", "ddd d mmm yyyy", "17 mar 2026", (2026, 3, 17)), + ("it", "dddd, mmmm d, yyyy", "mar 17, 2026", (2026, 3, 17)), + ("it", "dddd, mmm d, yyyy", "mar 17 2026", (2026, 3, 17)), + ("sq", "ddd d mmm yyyy", "5 sht 2026", (2026, 9, 5)), + ("it", "ddd d mmm yyyy", "mar 17 Mar 2026", (2026, 3, 17)), + ("it", "ddd d mmm yyyy", "mar 17 mar 2026", (2026, 3, 17)), + ]; + for (language, fmt, text, want) in cases { + use_form_date_language(language); + let read = parse_date(text, fmt).map(|d| (d.y, d.m, d.d)); + if read != Some(want) { + bad.push(format!("{language} forms {fmt}: {text:?} -> {read:?}")); + } + if keystroke(&Format::Date(fmt.into()), "date", text).is_err() { + bad.push(format!("{language} keystroke {fmt}: {text:?} refused")); + } + } + use_form_date_language("en"); + assert!(bad.is_empty(), "{}", bad.join("; ")); +} + +/// Turkish names typed in capitals are read: lower-casing letter by letter made "MAYIS" "mayis" +/// (not "mayıs"), "SALI" "sali" and "EKİM" "eki̇m" (an i with a dot above), so no name matched. +/// Dotted, dotless and capital i are one letter when names are read (#5303). +#[test] +fn turkish_names_typed_in_capitals_are_read() { + let _names = names_lock(); + use_form_date_language("tr"); + let mut bad = Vec::new(); + for (fmt, text, want) in [ + ("d mmmm yyyy", "1 MAYIS 2026", (2026, 5, 1)), + ("d mmmm yyyy", "1 EKİM 2026", (2026, 10, 1)), + ("d mmmm yyyy", "1 Ekim 2026", (2026, 10, 1)), + ("d mmm yyyy", "1 EKİ 2026", (2026, 10, 1)), + ("dddd d mmmm yyyy", "SALI 6 OCAK 2026", (2026, 1, 6)), + ("dddd d mmmm yyyy", "Salı 6 Ocak 2026", (2026, 1, 6)), + ] { + let read = parse_date(text, fmt).map(|d| (d.y, d.m, d.d)); + if read != Some(want) { + bad.push(format!("forms {fmt}: {text:?} -> {read:?}")); + } + let (y, m, d) = want; + let got = scanned(fmt, text); + if got != format!("{y:04}-{m:02}-{d:02}") { + bad.push(format!("scand {fmt}: {text:?} -> {got}")); + } + } + // English is read as before, capitals and all. + use_form_date_language("en"); + for (fmt, text) in [("d mmmm yyyy", "1 APRIL 2026"), ("dddd d mmmm yyyy", "FRIDAY 1 MAY 2026"), ("d mmm yyyy", "1 MAY 2026")] { + if parse_date(text, fmt).is_none() || scanned(fmt, text) == "null" { + bad.push(format!("en {fmt}: {text:?}")); + } + } + assert!(bad.is_empty(), "{}", bad.join("; ")); +} + +/// #5464 review: Greek names typed in capitals end in a capital Σ, which is the final ς of the +/// name ("ΙΑΝΟΥΆΡΙΟΣ" for "Ιανουάριος"); they read as they did before names were folded. +#[test] +fn greek_names_typed_in_capitals_are_read() { + let _names = names_lock(); + use_form_date_language("el"); + let mut bad = Vec::new(); + for (fmt, text, want) in [ + ("d mmmm yyyy", "1 ΙΑΝΟΥΆΡΙΟΣ 2026", (2026, 1, 1)), + ("d mmmm yyyy", "1 Ιανουάριος 2026", (2026, 1, 1)), + ("d mmmm yyyy", "1 ιανουάριοσ 2026", (2026, 1, 1)), + ("mmmm d, yyyy", "ΜΆΙΟΣ 5, 2026", (2026, 5, 5)), + ("dddd, mmmm d, yyyy", "FRIDAY, ΙΑΝΟΥΆΡΙΟΣ 2, 2026", (2026, 1, 2)), + ] { + let read = parse_date(text, fmt).map(|d| (d.y, d.m, d.d)); + if read != Some(want) { + bad.push(format!("forms {fmt}: {text:?} -> {read:?}")); + } + let (y, m, d) = want; + let got = scanned(fmt, text); + if got != format!("{y:04}-{m:02}-{d:02}") { + bad.push(format!("scand {fmt}: {text:?} -> {got}")); + } + } + assert!(bad.is_empty(), "{}", bad.join("; ")); +} diff --git a/crates/engine/tests/it/ocr_edit.rs b/crates/engine/tests/it/ocr_edit.rs index 8009ef1bbf..2d358e327e 100644 --- a/crates/engine/tests/it/ocr_edit.rs +++ b/crates/engine/tests/it/ocr_edit.rs @@ -12,7 +12,14 @@ const WORD: [u32; 4] = [40, 40, 100, 60]; /// A 200 × 100 pt "scan" on tinted paper, with OCR text over its word. fn scanned() -> (Session, pdfcraft_engine::DocId) { - scanned_with(vec![PlacedWord { text: "John".into(), origin: [40.0, 40.0], across: [60.0, 0.0], up: [0.0, 20.0] }]) + scanned_with(vec![PlacedWord { + text: "John".into(), + origin: [40.0, 40.0], + across: [60.0, 0.0], + up: [0.0, 20.0], + confidence: None, + source: "ocrs".into(), + }]) } fn scanned_with(words: Vec) -> (Session, pdfcraft_engine::DocId) { @@ -86,11 +93,18 @@ fn an_explicit_colour_is_kept_for_the_new_text() { /// descender line when it has descenders, up to the cap height. fn scanned_word(text: &str, x0: f64, x1: f64, baseline: f64, size: f64) -> PlacedWord { let low = if text.chars().any(|c| "gjpqy".contains(c)) { 0.21 * size } else { 0.0 }; - PlacedWord { text: text.into(), origin: [x0, baseline - low], across: [x1 - x0, 0.0], up: [0.0, 0.718 * size + low] } + PlacedWord { + text: text.into(), + origin: [x0, baseline - low], + across: [x1 - x0, 0.0], + up: [0.0, 0.718 * size + low], + confidence: None, + source: "ocrs".into(), + } } fn word(text: &str, x0: f64, x1: f64, baseline: f64) -> PlacedWord { - PlacedWord { text: text.into(), origin: [x0, baseline], across: [x1 - x0, 0.0], up: [0.0, 10.0] } + PlacedWord { text: text.into(), origin: [x0, baseline], across: [x1 - x0, 0.0], up: [0.0, 10.0], confidence: None, source: "ocrs".into() } } /// OCR words keep the scan's layout: a phrase is one paragraph, a value in the next table cell and @@ -207,3 +221,192 @@ fn visible_stretched_text_rewrites_unchanged() { let i = ((y * out.width + x) * 4) as usize; assert_eq!(&out.rgba[i..i + 3], &[255, 255, 255]); } + +/// A scan whose OCR layer is drawn first and the opaque page image over it, as some OCR tools +/// write it (#3973): invisible Helvetica `text` on the scanned word, then the image. +fn ocr_under_scan(text: &str) -> std::sync::Arc> { + under_scan(&format!("BT /F1 20 Tf 3 Tr 42 44 Td ({text}) Tj ET")) +} + +/// A page whose `text` content is followed by the opaque 200 × 100 scan image. +fn under_scan(text: &str) -> std::sync::Arc> { + under_scan_with(text, |_| pdfcraft_cos::Dict::new()) +} + +/// As [`under_scan`], with the page's ExtGState resources made by `ext_gstates`. +fn under_scan_with(text: &str, ext_gstates: impl FnOnce(&mut pdfcraft_cos::Document) -> pdfcraft_cos::Dict) -> std::sync::Arc> { + use pdfcraft_cos::{Dict, Document, Object, SaveOptions, Stream}; + let (w, h) = (200u32, 100u32); + let mut rgb = Vec::new(); + for y in 0..h { + for x in 0..w { + let ink = (WORD[0]..WORD[2]).contains(&x) && (WORD[1]..WORD[3]).contains(&y); + rgb.extend(if ink { INK } else { PAPER }); + } + } + let mut d = Document::new_empty(); + let pages = d.add(Object::Dict(Dict::new())); + let content = format!("{text} q 200 0 0 100 0 0 cm /Im Do Q"); + let contents = d.add(Object::Stream(Stream::flate(Dict::new(), content.as_bytes()))); + let mut image = Dict::new(); + image.set(b"Type".to_vec(), Object::name("XObject")); + image.set(b"Subtype".to_vec(), Object::name("Image")); + image.set(b"Width".to_vec(), Object::Int(i64::from(w))); + image.set(b"Height".to_vec(), Object::Int(i64::from(h))); + image.set(b"ColorSpace".to_vec(), Object::name("DeviceRGB")); + image.set(b"BitsPerComponent".to_vec(), Object::Int(8)); + let image = d.add(Object::Stream(Stream::flate(image, &rgb))); + let mut font = Dict::new(); + font.set(b"Type".to_vec(), Object::name("Font")); + font.set(b"Subtype".to_vec(), Object::name("Type1")); + font.set(b"BaseFont".to_vec(), Object::name("Helvetica")); + let mut fonts = Dict::new(); + fonts.set(b"F1".to_vec(), Object::Dict(font)); + let mut xobjects = Dict::new(); + xobjects.set(b"Im".to_vec(), Object::Ref(image)); + let mut res = Dict::new(); + res.set(b"Font".to_vec(), Object::Dict(fonts)); + res.set(b"XObject".to_vec(), Object::Dict(xobjects)); + res.set(b"ExtGState".to_vec(), Object::Dict(ext_gstates(&mut d))); + let mut p = Dict::new(); + p.set(b"Type".to_vec(), Object::name("Page")); + p.set(b"Parent".to_vec(), Object::Ref(pages)); + p.set(b"Contents".to_vec(), Object::Ref(contents)); + p.set(b"Resources".to_vec(), Object::Dict(res)); + p.set(b"MediaBox".to_vec(), Object::Array([0, 0, 200, 100].map(Object::Int).to_vec())); + let page = d.add(Object::Dict(p)); + d.update_dict(pages, |p| { + p.set(b"Type".to_vec(), Object::name("Pages")); + p.set(b"Kids".to_vec(), Object::Array(vec![Object::Ref(page)])); + p.set(b"Count".to_vec(), Object::Int(1)); + }) + .unwrap(); + let mut root = Dict::new(); + root.set(b"Type".to_vec(), Object::name("Catalog")); + root.set(b"Pages".to_vec(), Object::Ref(pages)); + let root = d.add(Object::Dict(root)); + d.trailer_mut().set(b"Root".to_vec(), Object::Ref(root)); + std::sync::Arc::new(pdfcraft_cos::write_full(&d, &SaveOptions::default()).unwrap()) +} + +/// The scanned word's box shows the cover and new ink, not the scan. +fn assert_edit_shows(s: &Session, id: pdfcraft_engine::DocId, label: &str) { + let px = word_pixels(s, id); + let paper = px.iter().filter(|p| near(**p, PAPER)).count(); + let ink = px.iter().filter(|p| near(**p, INK)).count(); + assert!(paper * 10 > px.len() * 4, "{label}: the cover is drawn over the scan: {paper} of {} paper", px.len()); + assert!(ink * 100 > px.len() * 3, "{label}: the new text shows: {ink} of {} ink", px.len()); +} + +fn texts(s: &Session, id: pdfcraft_engine::DocId) -> Vec { + s.get(id).unwrap().text_blocks(0).iter().map(|b| b.text.replace('\n', " ")).collect() +} + +fn edit_block(s: &mut Session, id: pdfcraft_engine::DocId, block: usize, text: &str) { + s.apply(id, Edit::EditTextBlock { page: 0, block, text: text.into(), style: Default::default() }).unwrap(); +} + +/// The OCR layer is under the scan image: the rewritten paragraph is drawn on top of the image, +/// not in the OCR text's place where the image would hide it (#3973). One line takes the +/// single-style rewrite, a typed line break the styled one. Undo restores the hidden layer, and +/// the edit survives redo, a save and reopen, and being edited again. +#[test] +fn an_ocr_layer_under_the_scan_image_shows_its_edit() { + for edit in ["Jane", "Jane\nDoe"] { + let mut s = Session::new(); + let id = s.open("ocr-under-scan.pdf", None, ocr_under_scan("John"), None).unwrap(); + let blocks = s.get(id).unwrap().text_blocks(0); + assert_eq!(blocks.len(), 1); + assert!(blocks[0].invisible, "the OCR layer is invisible text"); + assert!(word_pixels(&s, id).iter().all(|p| near(*p, INK)), "the scanned word fills its box"); + + edit_block(&mut s, id, 0, edit); + assert_eq!(texts(&s, id), [edit.replace('\n', " ")], "{edit:?}"); + assert_edit_shows(&s, id, edit); + + s.undo(id).unwrap(); + assert_eq!(texts(&s, id), ["John"], "{edit:?}: undo"); + assert!(s.get(id).unwrap().text_blocks(0)[0].invisible, "{edit:?}: undo hides the layer again"); + assert!(word_pixels(&s, id).iter().all(|p| near(*p, INK)), "{edit:?}: undo shows the scan again"); + s.redo(id).unwrap(); + assert_edit_shows(&s, id, &format!("{edit:?} redo")); + + let saved = s.save_bytes(id).unwrap(); + let mut reopened = Session::new(); + let id = reopened.open("saved.pdf", None, saved, None).unwrap(); + assert_eq!(texts(&reopened, id), [edit.replace('\n', " ")], "{edit:?}: reopened"); + assert_edit_shows(&reopened, id, &format!("{edit:?} reopened")); + + edit_block(&mut reopened, id, 0, "Ann"); + assert_eq!(texts(&reopened, id), ["Ann"], "{edit:?}: edited again"); + assert_edit_shows(&reopened, id, &format!("{edit:?} edited again")); + } +} + +/// Visible text earlier in the same text object makes the rewrite split it; the OCR phrase still +/// goes on top of the scan, and the visible text stays. +#[test] +fn an_ocr_phrase_split_from_visible_text_under_the_scan_shows_its_edit() { + for edit in ["Jane", "Jane\nDoe"] { + let mut s = Session::new(); + let page = under_scan("BT /F1 20 Tf 10 80 Td (Head) Tj 3 Tr 32 -36 Td (John) Tj ET"); + let id = s.open("split.pdf", None, page, None).unwrap(); + let blocks = s.get(id).unwrap().text_blocks(0); + let john = blocks.iter().position(|b| b.text == "John").unwrap(); + assert!(blocks[john].invisible && blocks.iter().any(|b| b.text == "Head" && !b.invisible), "{blocks:?}"); + edit_block(&mut s, id, john, edit); + let mut after = texts(&s, id); + after.sort(); + let mut want = vec!["Head".to_string(), edit.replace('\n', " ")]; + want.sort(); + assert_eq!(after, want, "{edit:?}"); + assert_edit_shows(&s, id, &format!("{edit:?} split")); + } +} + +/// `/SM`: a luminosity soft mask that is white over 30..110 × 30..70 in the space it is selected +/// in, black elsewhere. +fn soft_mask_state(d: &mut pdfcraft_cos::Document) -> pdfcraft_cos::Dict { + use pdfcraft_cos::{Dict, Object, Stream}; + let mut group = Dict::new(); + group.set(b"S".to_vec(), Object::name("Transparency")); + group.set(b"CS".to_vec(), Object::name("DeviceRGB")); + let mut form = Dict::new(); + form.set(b"Type".to_vec(), Object::name("XObject")); + form.set(b"Subtype".to_vec(), Object::name("Form")); + form.set(b"BBox".to_vec(), Object::Array([0, 0, 200, 100].map(Object::Int).to_vec())); + form.set(b"Group".to_vec(), Object::Dict(group)); + let mut mask = Dict::new(); + mask.set(b"Type".to_vec(), Object::name("Mask")); + mask.set(b"S".to_vec(), Object::name("Luminosity")); + mask.set(b"G".to_vec(), Object::Ref(d.add(Object::Stream(Stream::flate(form, b"1 g 30 30 80 40 re f"))))); + let mut state = Dict::new(); + state.set(b"Type".to_vec(), Object::name("ExtGState")); + state.set(b"SMask".to_vec(), Object::Dict(mask)); + let mut states = Dict::new(); + states.set(b"SM".to_vec(), Object::Dict(state)); + states +} + +/// The soft mask is selected before a translation, so it is placed by the CTM before it. The +/// paragraph lifted over the scan selects it again after its own `cm`: there it must be placed +/// as it was, not 100 pt to the right where it would hide the new text, so the edit renders +/// exactly as the same edit on the page without the mask. Both rewrite paths. +#[test] +fn a_soft_mask_selected_before_a_cm_stays_put_when_the_edit_is_lifted() { + for edit in ["Jane", "Jane\nDoe"] { + let mut plain = Session::new(); + let page = under_scan("q 1 0 0 1 100 0 cm BT /F1 20 Tf 3 Tr -58 44 Td (John) Tj ET Q"); + let plain_id = plain.open("plain.pdf", None, page, None).unwrap(); + edit_block(&mut plain, plain_id, 0, edit); + let mut s = Session::new(); + let page = under_scan_with("q /SM gs 1 0 0 1 100 0 cm BT /F1 20 Tf 3 Tr -58 44 Td (John) Tj ET Q", soft_mask_state); + let id = s.open("masked.pdf", None, page, None).unwrap(); + assert_eq!(texts(&s, id), ["John"]); + edit_block(&mut s, id, 0, edit); + assert_eq!(texts(&s, id), [edit.replace('\n', " ")], "{edit:?}"); + assert_edit_shows(&s, id, &format!("{edit:?} masked")); + let ink = |s: &Session, id| word_pixels(s, id).iter().filter(|p| near(**p, INK)).count(); + assert_eq!(ink(&s, id), ink(&plain, plain_id), "{edit:?}: the mask does not hide the new text"); + } +} diff --git a/crates/fonts/README.md b/crates/fonts/README.md index b20b930a91..096f1a263e 100644 --- a/crates/fonts/README.md +++ b/crates/fonts/README.md @@ -59,7 +59,12 @@ set). Without that variable `CRAFT_FONTS` is empty and everything below copes: and shaped Thai Add Text content. Harfrust positions stacked vowel/tone marks and decomposes Sara Am; cluster ToUnicode entries preserve the original text. The shared Type 3 writer embeds outlines, so reopening does not depend on installed fonts. +- `craft_glyph(face, ch)` and `face_has(bytes, ch)`: bounded outlines and glyph coverage for any bundled face. - `SHIPPORI_MINCHO`: Shippori Mincho's bytes, or `None`. +- `TextPlan::for_words`: how the OCR output writes text beyond WinAnsi — per word the first + face (document face first) whose glyphs cover it, planned once per page into Type 3 fonts + (codes, advances, `?` stand-ins) that `pdfcraft-edit` registers and the `pdfcraft-ocr` layers + draw with; characters no face has are counted (`missing`) so the output reports them. Generated Japanese Type 3 fonts record the fallback face's family and style in an indirect `FontDescriptor`. `pdf::Metrics::base_font` uses its `FontName` when the Type 3 dictionary has diff --git a/crates/fonts/src/lib.rs b/crates/fonts/src/lib.rs index 61061f5509..45cb843ab0 100644 --- a/crates/fonts/src/lib.rs +++ b/crates/fonts/src/lib.rs @@ -14,6 +14,7 @@ pub mod embedding; mod encodings; pub mod fallback; pub mod pdf; +pub mod plan; mod script; mod shaping; pub mod source_shaping; @@ -24,8 +25,10 @@ pub use craft::{ document_japanese_font_for_style, document_japanese_fonts_for_style, document_latin_fonts, document_thai_font, ui_arabic_fonts, ui_chinese_fonts, ui_cjk_fonts, ui_devanagari_fonts, ui_hebrew_fonts, ui_japanese_fonts, ui_telugu_fonts, ui_thai_fonts, }; +pub use plan::{MAX_PLAN_GLYPHS, MAX_TYPE3_GLYPHS, PlannedType3, TextPlan}; pub use script::{ - GlyphError, GlyphOutline, INTER_REGULAR, MAX_SIGNATURE_CHARS, ScriptOutline, japanese_glyph, japanese_glyph_from, script_missing, script_outline, + GlyphError, GlyphOutline, INTER_REGULAR, MAX_SIGNATURE_CHARS, ScriptOutline, craft_glyph, face_has, japanese_glyph, japanese_glyph_from, + script_missing, script_outline, }; pub use shaping::{ ShapedCluster, ShapedScript, arabic_glyph, arabic_has, hebrew_has, shape_arabic, shape_hebrew, shape_thai, shaped_glyph, thai_has, @@ -273,6 +276,16 @@ pub fn set_undrawn_mark(d: &mut pdfcraft_cos::Dict, c: Option) { } } +/// The byte used by Unicode text plans, from the shared WinAnsi table (tabs become spaces). +pub fn win_ansi_char(c: char) -> Option { + if c == '\t' { Some(b' ') } else { win_ansi_byte(c) } +} + +/// Whether every character can be represented without substitution in WinAnsi. +pub fn fits_win_ansi(s: &str) -> bool { + s.chars().all(win_ansi_encodable) +} + /// Bytes as a PDF literal string, `(` … `)`, with delimiters escaped. pub fn literal(bytes: &[u8]) -> Vec { let mut out = Vec::with_capacity(bytes.len() + 2); diff --git a/crates/fonts/src/plan.rs b/crates/fonts/src/plan.rs new file mode 100644 index 0000000000..9c22a38fd3 --- /dev/null +++ b/crates/fonts/src/plan.rs @@ -0,0 +1,338 @@ +//! Planning visible Unicode text output for OCR: which characters the standard WinAnsi +//! fonts show as is, which need a Type3 font drawn from the craft-fonts faces, and which have +//! no glyph anywhere (written as `?` — and counted, never silently dropped). +//! +//! The plan is a pure function of the words ([`TextPlan::for_words`]). The content generators +//! (`pdfcraft_ocr::text_layer` / `visible_text_layer`) read the per-character codes and +//! advances from it, and whoever registers the fonts (`pdfcraft_edit::add_type3_fonts`) fills +//! in each face's resource name and object. The build must work without craft-fonts too: there +//! [`CRAFT_FONTS`] is empty, nothing can be planned, and every non-WinAnsi character is +//! reported as a substitution instead. + +use std::collections::HashMap; + +use pdfcraft_cos::ObjRef; + +use crate::{CraftFont, GlyphOutline, craft_glyph, document_japanese_font, face_has, fits_win_ansi, win_ansi_char}; + +/// Unique characters per Type3 font: the same bound the text-editing fallback uses, so one +/// hostile word cannot balloon into thousands of glyph streams. A face with more characters +/// simply gets another font. +pub const MAX_TYPE3_GLYPHS: usize = 240; + +/// Planned glyphs per plan in total (10 full fonts). A page's recognition cannot realistically +/// exceed it; hostile input beyond it degrades to `?` substitutions, which are counted. +pub const MAX_PLAN_GLYPHS: usize = 2400; + +/// One planned Type3 font: the glyphs of one craft-fonts face, one character per code +/// (`code = position + 1`, as in the text-editing fallback). `name` and `object` are filled in +/// when the font is added to a page's resources; until then the content generators fall back to +/// the WinAnsi path for the words that reference it. +pub struct PlannedType3 { + /// The face family the glyphs are drawn from ("Shippori Mincho"). + pub family: &'static str, + /// (character, outline with its advance) per code, in code order. + pub glyphs: Vec<(char, GlyphOutline)>, + /// The resource name the content references (`PCOcr`, `PCOcr1`, …); empty until registered. + pub name: String, + /// The built font object, shared by every page that uses the font; set at registration. + pub object: Option, +} + +impl PlannedType3 { + fn new(family: &'static str) -> Self { + PlannedType3 { family, glyphs: Vec::new(), name: String::new(), object: None } + } + + fn push(&mut self, ch: char, glyph: GlyphOutline) -> u8 { + self.glyphs.push((ch, glyph)); + self.glyphs.len() as u8 + } +} + +/// How the text layers write a plan's characters: each planned character is a +/// `(font index, code)` pair in its font's encoding; a character no face had a glyph for is +/// either the font's `?` code or absent from the run (both counted in [`TextPlan::missing`]). +#[derive(Default)] +pub struct TextPlan { + /// One planned font per face chunk, in first-use order. + pub fonts: Vec, + /// Per input word: `(font, code)` per character; empty for a word the standard WinAnsi + /// fonts write as is (or for which nothing could be planned). + words: Vec>, + /// character → where its glyph lives, across all fonts (codes are font-local). + index: HashMap, + /// Characters with no code in their word's font: they appear as `?` (or, when not even a + /// `?` glyph exists, are left out) and are reported, never silently lost. + missing: usize, +} + +impl TextPlan { + /// Plan the output for `words`, in order: characters the WinAnsi fonts can't show get codes + /// in Type3 fonts built from the craft-fonts faces (the document face preferred, then the + /// rest of the build input in manifest order); a word whose face can't show a character + /// substitutes its `?`. Faces are chosen per word, so a word of one script stays one face; + /// characters already planned for an earlier word keep their code. Built without + /// craft-fonts, the plan reports every non-WinAnsi character as a substitution. + pub fn for_words<'a>(words: impl IntoIterator) -> TextPlan { + let faces = document_faces(); + let mut coverage = Coverage::default(); + let mut plan = TextPlan::default(); + for word in words { + if fits_win_ansi(word) { + // The WinAnsi path substitutes '?' behind the caller's back; count it. + plan.missing += word.chars().filter(|&c| win_ansi_char(c).is_none()).count(); + plan.words.push(Vec::new()); + continue; + } + let Some(face) = choose_face(&faces, word, &mut coverage) else { + // No face for this word's script: the WinAnsi path writes '?' for what it + // can't encode. + plan.missing += word.chars().filter(|&c| win_ansi_char(c).is_none()).count(); + plan.words.push(Vec::new()); + continue; + }; + let mut enc: Vec<(usize, u8)> = Vec::with_capacity(word.len()); + let mut run: Option = None; + for ch in word.chars() { + if let Some(&where_) = plan.index.get(&ch) { + enc.push(where_); + run = Some(where_.0); + continue; + } + if let Ok(glyph) = craft_glyph(face, ch) + && plan.planned_glyphs() < MAX_PLAN_GLYPHS + && let Some((fi, code)) = plan.add_glyph(face, ch, glyph) + { + plan.index.insert(ch, (fi, code)); + enc.push((fi, code)); + run = Some(fi); + continue; + } + // The face has no glyph for the character (or the plan is full): substitute + // the font's '?', in the font the surrounding run is set in. + if let Some((fi, code)) = plan.question_mark(face, run.filter(|&f| plan.fonts[f].family == face.family)) { + enc.push((fi, code)); + run = Some(fi); + } + plan.missing += 1; + } + plan.words.push(enc); + } + plan + } + + /// The encoding of word `i` (parallel to the slice [`Self::for_words`] planned): a list of + /// `(font, code)` per character, or an empty slice for a word the WinAnsi fonts write. + /// Out-of-range indexes degrade to the empty slice rather than panicking — callers pass the + /// same word list they planned with. + pub fn encoding(&self, word: usize) -> &[(usize, u8)] { + self.words.get(word).map(Vec::as_slice).unwrap_or_default() + } + + /// How many characters no font could show (they appear as `?`, or are left out when the + /// word's face has no `?` glyph either). + pub fn missing(&self) -> usize { + self.missing + } + + /// The advance of one planned code, in em units (the Type3 glyphs' own widths). `0.0` for + /// anything out of range. + pub fn advance(&self, font: usize, code: u8) -> f64 { + self.fonts.get(font).and_then(|f| f.glyphs.get(usize::from(code).checked_sub(1)?)).map_or(0.0, |(_, g)| g.width) + } + + /// The sum of the advances of a word's planned codes, in em units. + pub fn advance_of(&self, encoding: &[(usize, u8)]) -> f64 { + encoding.iter().map(|&(f, c)| self.advance(f, c)).sum() + } + + /// Whether every font the encoding references has been registered (has a resource name). + /// Unregistered plans fall back to the WinAnsi path, the same output as before the + /// Unicode mapping existed. + pub fn registered(&self, encoding: &[(usize, u8)]) -> bool { + encoding.iter().all(|&(f, _)| self.fonts.get(f).is_some_and(|font| !font.name.is_empty())) + } + + /// Characters with a planned glyph so far. + fn planned_glyphs(&self) -> usize { + self.fonts.iter().map(|f| f.glyphs.len()).sum() + } + + /// Add a glyph to the face's current font (a fresh one when that font is full). + fn add_glyph(&mut self, face: &CraftFont, ch: char, glyph: GlyphOutline) -> Option<(usize, u8)> { + let fi = self.font_of_face(face.family)?; + let code = self.fonts[fi].push(ch, glyph); + Some((fi, code)) + } + + /// The index of the face's open font, or a fresh one. `None` without craft-fonts (the + /// face itself comes from there, so this cannot happen for a planned face). + fn font_of_face(&mut self, family: &'static str) -> Option { + if let Some(fi) = self.fonts.iter().rposition(|f| f.family == family && f.glyphs.len() < MAX_TYPE3_GLYPHS) { + return Some(fi); + } + if self.fonts.len() >= MAX_PLAN_GLYPHS / MAX_TYPE3_GLYPHS { + return None; + } + self.fonts.push(PlannedType3::new(family)); + Some(self.fonts.len() - 1) + } + + /// The code of the font's `?` stand-in, adding the glyph when the font lacks it. + /// `prefer` is the run's current font; `None` without craft-fonts. + fn question_mark(&mut self, face: &CraftFont, prefer: Option) -> Option<(usize, u8)> { + if let Some(fi) = prefer + && let Some(code) = self.fonts[fi].glyphs.iter().position(|(c, _)| *c == '?') + { + return Some((fi, code as u8 + 1)); + } + let fi = self.font_of_face(face.family)?; + if let Some(code) = self.fonts[fi].glyphs.iter().position(|(c, _)| *c == '?') { + return Some((fi, code as u8 + 1)); + } + let glyph = craft_glyph(face, '?').ok()?; + let code = self.fonts[fi].push('?', glyph); + Some((fi, code)) + } +} + +/// The faces that can draw document text, in preference order: the Japanese document face +/// (serif document text) first, then the rest of the build input in manifest order. +fn document_faces() -> Vec<&'static CraftFont> { + let mut out = Vec::new(); + if let Some(d) = document_japanese_font() { + out.push(d); + } + for f in crate::CRAFT_FONTS { + if !out.iter().any(|d| std::ptr::eq(*d, f)) { + out.push(f); + } + } + out +} + +/// The face for a word: the first (in preference order) whose glyphs cover every character, +/// else the one covering the most characters (ties go to the earlier face). `None` when no +/// face covers any character of the word. +fn choose_face<'f>(faces: &[&'f CraftFont], word: &str, coverage: &mut Coverage) -> Option<&'f CraftFont> { + if let Some(f) = faces.iter().enumerate().find(|(i, f)| word.chars().all(|c| coverage.has(*i, f, c))).map(|(_, f)| *f) { + return Some(f); + } + faces + .iter() + .enumerate() + .map(|(i, f)| (word.chars().filter(|&c| coverage.has(i, f, c)).count(), std::cmp::Reverse(i), *f)) + .max_by(|a, b| (a.0, a.1).cmp(&(b.0, b.1))) + .filter(|(covered, _, _)| *covered > 0) + .map(|(_, _, f)| f) +} + +/// Which faces have a glyph for which characters, remembered for one plan: a page of words +/// asks the same (face, character) pairs over and over, and each fresh answer parses the font. +#[derive(Default)] +struct Coverage { + known: std::collections::HashMap<(usize, char), bool>, + lookups: usize, +} + +impl Coverage { + /// Whether face number `i` (`face`) has `ch`; asks the font once per pair. + fn has(&mut self, i: usize, face: &CraftFont, ch: char) -> bool { + *self.known.entry((i, ch)).or_insert_with(|| { + self.lookups += 1; + face_has(face.bytes, ch) + }) + } +} + +#[cfg(test)] +mod tests { + use super::*; + + /// Face coverage is asked once per (face, character): repeating words cost no new font + /// lookups. + #[test] + fn face_coverage_is_remembered_per_face_and_character() { + let faces = document_faces(); + let Some(face) = faces.first() else { + eprintln!("skipping: built without craft-fonts (set CRAFT_FONTS_DIR to run it)"); + return; + }; + let mut coverage = Coverage::default(); + let first = (0..50).map(|_| choose_face(&faces, "日本語", &mut coverage)).last().flatten(); + assert!(first.is_some()); + assert!(coverage.lookups <= faces.len() * 3, "{} lookups for 3 characters", coverage.lookups); + assert_eq!(coverage.has(0, face, '日'), face_has(face.bytes, '日')); + } + + /// A page's vocabulary: German umlauts are WinAnsi, and a character no bundled face has + /// (here an emoji) is counted wherever it appears. + #[test] + fn win_ansi_words_get_no_encoding_and_count_only_real_losses() { + let plan = TextPlan::for_words(["Zürcher", "café", "size"]); + assert!(plan.words.iter().all(Vec::is_empty), "all WinAnsi: no Type3 fonts needed"); + assert!(plan.fonts.is_empty()); + assert_eq!(plan.missing(), 0); + let none = TextPlan::for_words(["plain text"]); + assert_eq!(none.missing(), 0); + // U+1F600 has no WinAnsi code and no glyph in any bundled face (with craft-fonts the + // JIS faces even cover the snowman): counted, shown as '?'. + let emoji = TextPlan::for_words(["\u{1f600}!"]); + assert_eq!(emoji.missing(), 1); + } + + #[test] + fn without_craft_fonts_every_non_win_ansi_character_is_reported() { + if !crate::CRAFT_FONTS.is_empty() { + eprintln!("built with craft-fonts: the planning tests with faces cover this instead"); + return; + } + let plan = TextPlan::for_words(["Привет", "日本", "abc"]); + assert!(plan.words.iter().all(Vec::is_empty)); + assert!(plan.fonts.is_empty()); + assert_eq!(plan.missing(), 8, "six Cyrillic + two CJK characters, each reported"); + } + + #[test] + fn encodings_are_aligned_and_total_the_advances() { + if crate::CRAFT_FONTS.is_empty() { + eprintln!("skipping: built without craft-fonts (set CRAFT_FONTS_DIR to run it)"); + assert!(TextPlan::for_words(["日本"]).fonts.is_empty()); + return; + } + let plan = TextPlan::for_words(["日本語", "語"]); + assert_eq!(plan.encoding(0).len(), 3); + assert_eq!(plan.encoding(1).len(), 1); + // 語 planned once keeps its code (and font) in both words. + let (f, c) = plan.encoding(1)[0]; + assert_eq!(plan.encoding(0)[2], (f, c), "the repeated character keeps its code"); + assert!(plan.advance(f, c) > 0.0); + assert_eq!(plan.advance(f, 0), 0.0, "codes start at 1"); + assert_eq!(plan.encoding(7), &[], "out of range degrades to the WinAnsi path"); + assert!(plan.advance_of(plan.encoding(0)) > plan.advance_of(plan.encoding(1))); + } + + #[test] + fn hostile_input_stays_bounded() { + // Thousands of DISTINCT characters: the covered ones fill fonts up to the caps, the + // rest are reported substitutions. + let word: String = (0..4000u32).filter_map(|i| char::from_u32(0x4e00 + i)).collect(); + let plan = TextPlan::for_words([word.as_str()]); + assert!(plan.planned_glyphs() <= MAX_PLAN_GLYPHS); + assert!(plan.fonts.iter().all(|f| f.glyphs.len() <= MAX_TYPE3_GLYPHS)); + assert!(plan.missing() > 0, "characters beyond the caps are reported, never silently dropped"); + } + + #[test] + fn substitution_and_unknown_scripts_come_out_counted() { + if crate::CRAFT_FONTS.is_empty() { + eprintln!("skipping: built without craft-fonts (set CRAFT_FONTS_DIR to run it)"); + return; + } + let plan = TextPlan::for_words(["日本\u{1f4a9}語"]); + // The emoji has no glyph in any bundled face; the kana/kanji do. + assert_eq!(plan.missing(), 1); + assert_eq!(plan.encoding(0).len(), 4, "the emoji still has an encoding: the font's ?"); + } +} diff --git a/crates/fonts/src/script.rs b/crates/fonts/src/script.rs index 70fc6ce092..94a8a93b8c 100644 --- a/crates/fonts/src/script.rs +++ b/crates/fonts/src/script.rs @@ -134,6 +134,17 @@ impl OutlinePen for Flatten { } } +/// Whether the face has a glyph for `ch` (a cheap character-map probe, no outline work). +pub fn face_has(bytes: &[u8], ch: char) -> bool { + FontRef::new(bytes).is_ok_and(|font| font.charmap().map(ch).is_some()) +} + +/// Return one glyph of `face` (a craft-fonts face), bounded so hostile replacement text cannot +/// allocate unbounded outline data. +pub fn craft_glyph(face: &crate::CraftFont, ch: char) -> Result { + glyph_in_face(face.bytes, ch) +} + /// Return one glyph of the Japanese document face ([`crate::document_japanese_font`], Shippori /// Mincho from craft-fonts), bounded so hostile replacement text cannot allocate unbounded /// outline data. [`GlyphError::NoFont`] when the build has no Japanese face. @@ -145,7 +156,11 @@ pub fn japanese_glyph(ch: char) -> Result { /// The bounded outline and advance from the explicitly selected craft-fonts face. /// Uses the same complexity and missing-glyph checks as [`japanese_glyph`]. pub fn japanese_glyph_from(face: &crate::CraftFont, ch: char) -> Result { - let Ok(font) = FontRef::new(face.bytes) else { return Err(GlyphError::Missing) }; + glyph_in_face(face.bytes, ch) +} + +fn glyph_in_face(bytes: &[u8], ch: char) -> Result { + let Ok(font) = FontRef::new(bytes) else { return Err(GlyphError::Missing) }; let loc = LocationRef::default(); let metrics = font.metrics(Size::unscaled(), loc); let scale = 1.0 / metrics.units_per_em.max(1) as f64; diff --git a/crates/forms/src/af.rs b/crates/forms/src/af.rs index 86c7971c31..f5167374d1 100644 --- a/crates/forms/src/af.rs +++ b/crates/forms/src/af.rs @@ -20,7 +20,10 @@ mod script; #[cfg(test)] use calc::{Notation, eval_notation, notation_name, notation_tokens, unique}; pub use calc::{calculate, calculation_reads, notation_js, reads_built_names, rename_in_calculation}; -pub use format::{DATE_PRESETS, DateTime, TIME_PRESETS, format_date, format_value, js_number, keystroke, make_number, parse_date, validate}; +pub use format::{ + DATE_PRESETS, DateNames, DateTime, ENGLISH_DATE_NAMES, TIME_PRESETS, date_names, format_date, format_date_in, format_value, format_value_in, + js_number, keystroke, make_number, parse_date, set_date_names, validate, +}; use script::{Arg, call}; pub use script::{NOTATION_NOT_STORABLE, calculate_js, format_js, notation_storable, parse_calculate, parse_format, parse_validate, validate_js}; diff --git a/crates/forms/src/af/format.rs b/crates/forms/src/af/format.rs index df4720bec7..37d2163cb0 100644 --- a/crates/forms/src/af/format.rs +++ b/crates/forms/src/af/format.rs @@ -89,6 +89,165 @@ const MONTHS: [&str; 12] = ["January", "February", "March", "April", "May", "Jun const DAYS: [&str; 7] = ["Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday"]; +/// The month and weekday names `mmmm`, `mmm`, `dddd` and `ddd` write (weekdays from Sunday), as +/// a localized Acrobat writes them in its own language (#5303). +#[derive(Clone, Copy, Debug, PartialEq, Eq)] +pub struct DateNames { + pub months: [&'static str; 12], + pub months_short: [&'static str; 12], + pub days: [&'static str; 7], + pub days_short: [&'static str; 7], +} + +/// English, the names until the app sets others. +pub const ENGLISH_DATE_NAMES: DateNames = DateNames { + months: MONTHS, + months_short: ["Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"], + days: DAYS, + days_short: ["Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"], +}; + +static DATE_NAMES: std::sync::RwLock = std::sync::RwLock::new(ENGLISH_DATE_NAMES); + +/// Sets the names date formats write from now on (the app's date language); typed dates are read +/// in them and in English. +pub fn set_date_names(names: DateNames) { + *DATE_NAMES.write().unwrap_or_else(std::sync::PoisonError::into_inner) = names; +} + +/// The names date formats write. +pub fn date_names() -> DateNames { + *DATE_NAMES.read().unwrap_or_else(std::sync::PoisonError::into_inner) +} + +/// `text` with the month and weekday names of `lists` (in the order months, short months, +/// weekdays, short weekdays) written as their English names, so a parser that reads English +/// reads them. The longest name wins at each place, case ignored, and only a whole name: one +/// that doesn't continue a word, nor end in a digit that a number continues. A name may hold +/// digits ("10月"), spaces ("tháng 10"), dots ("ม.ค.") or combining vowel signs, which a +/// word-by-word reader would split (#5303). Where a weekday and a month share a name (Italian +/// "mar"), the order of the format's name fields (`mmm`, `ddd`) decides. English names are +/// found too, so that one typed ("Tue" before "Mar") fills its field of the format as a +/// localized one does. +fn english_names(text: &str, fmt: &str, lists: [&[&str]; 4]) -> String { + const EN: [&[&str]; 4] = [&MONTHS, &MONTHS, &DAYS, &DAYS]; + // No date is this long: hostile text is left as it is rather than scanned name by name. + if text.len() > MAX_DATE_TEXT_BYTES { + return text.to_string(); + } + let chars: Vec = text.chars().collect(); + let typed_english: [&[&str]; 4] = [&MONTHS, &ENGLISH_DATE_NAMES.months_short, &DAYS, &ENGLISH_DATE_NAMES.days_short]; + // Each name as folded chars, whether it is a month, and its English name. + let mut names: Vec<(Vec, bool, &str)> = Vec::new(); + for (k, (list, en)) in lists.into_iter().zip(EN).chain(typed_english.into_iter().zip(EN)).enumerate() { + for (name, english) in list.iter().zip(en.iter()) { + // A short name ending in a dot ("mar.") is also read typed without it, as the same + // kind of name, so that where it equals another ("Mar", March) the format decides. + let bare = name.trim_end_matches('.'); + for name in std::iter::once(*name).chain((bare.len() < name.len()).then_some(bare)) { + let n: Vec = name.chars().map(fold).collect(); + if !n.is_empty() && n.len() <= MAX_NAME_CHARS { + names.push((n, k % 4 < 2, english)); + } + } + } + } + names.sort_by_key(|(n, ..)| std::cmp::Reverse(n.len())); + // The format's name fields in order: true for a month, false for a weekday. + let f: Vec = fmt.chars().collect(); + let mut expected = Vec::new(); + let mut j = 0; + while let Some(&c) = f.get(j) { + let run = f.get(j..).map_or(1, |r| r.iter().take_while(|x| **x == c).count()).max(1); + if run >= 3 && (c == 'm' || c == 'd') { + expected.push(c == 'm'); + } + j = j.saturating_add(run); + } + let mut slot = 0; + // A name's first letter may not continue a word; its last may not run on into a letter, nor + // (a digit) into a number. + let letter = |c: Option<&char>| c.is_some_and(|c| c.is_alphabetic()); + let mut out = String::with_capacity(text.len()); + let mut i = 0; + while let Some(&c) = chars.get(i) { + let fits = |n: &Vec| { + let end = i.saturating_add(n.len()); + let first = n.first().and_then(|e| e.chars().next()); + let last = n.last().and_then(|e| e.chars().next()); + chars.get(i..end).is_some_and(|t| t.iter().map(|c| fold(*c)).eq(n.iter().cloned())) + && !(first.is_some_and(char::is_alphabetic) && letter(i.checked_sub(1).and_then(|p| chars.get(p)))) + && !match last { + Some(l) if l.is_ascii_digit() => chars.get(end).is_some_and(char::is_ascii_digit), + Some(l) if l.is_alphabetic() => letter(chars.get(end)), + _ => false, + } + }; + let longest = names.iter().find(|(n, ..)| fits(n)).map(|(n, ..)| n.len()); + let hit = longest.and_then(|len| { + let mut same = names.iter().filter(|(n, ..)| n.len() == len && fits(n)); + let want = expected.get(slot).copied(); + let all: Vec<_> = same.by_ref().collect(); + all.iter().find(|(_, month, _)| Some(*month) == want).or_else(|| all.first()).copied() + }); + match hit { + Some((n, month, english)) => { + if expected.get(slot) == Some(month) { + slot += 1; + } + // Apart from what touches it ("10月18"), but no space added at either end. + if out.ends_with(char::is_alphanumeric) { + out.push(' '); + } + out.push_str(english); + i = i.saturating_add(n.len()); + if chars.get(i).is_some_and(|c| c.is_alphanumeric()) { + out.push(' '); + } + } + None => { + out.push(c); + i += 1; + } + } + } + out +} + +/// `c` as names are compared: lower case, with the capital I and the Turkish dotted and dotless +/// i all "i", and a combining dot above (U+0307) dropped, so that "EKİM", "SALI" and "MAYIS" are +/// Turkish "Ekim", "Salı" and "Mayıs", which lower-casing alone makes "ekim" with a dotted i, +/// "sali" and "mayis" (#5303). The Greek final sigma is a sigma, since a capital Σ, which +/// lower-cases to σ on its own, ends "ΙΑΝΟΥΆΡΙΟΣ" for "Ιανουάριος". One piece for each char, so +/// positions in the text stay aligned. +fn fold(c: char) -> String { + match c { + 'I' | 'ı' | 'İ' => "i".to_string(), + 'ς' => "σ".to_string(), + _ => c.to_lowercase().filter(|&l| l != '\u{307}').collect(), + } +} + +/// `s` [`fold`]ed char by char. +fn folded(s: &str) -> String { + s.chars().map(fold).collect() +} + +/// Typed dates longer than this are not searched for names. +const MAX_DATE_TEXT_BYTES: usize = 1024; + +/// Names longer than this are not looked for (settings are trusted, but a cap keeps the scan cheap). +const MAX_NAME_CHARS: usize = 32; + +/// The month (1 to 12) a typed word names: a start of at least three letters of a month's name, +/// or its whole short name, in `names` or in English. +fn month_named(word: &str, names: &DateNames) -> Option { + let w = folded(word); + let starts = |list: &[&str; 12]| list.iter().position(|m| w.chars().count() >= 3 && folded(m).starts_with(&w)); + let whole = |list: &[&str; 12]| list.iter().position(|m| folded(m).trim_end_matches('.') == w); + starts(&names.months).or_else(|| whole(&names.months_short)).or_else(|| starts(&MONTHS)).map(|i| i as u32 + 1) +} + /// A date and time read from text. #[derive(Clone, Copy, Debug, PartialEq, Eq)] pub struct DateTime { @@ -202,7 +361,20 @@ fn order(fmt: &str) -> Vec { /// Parse typed text against a date/time format (AFDate_KeystrokeEx's leniency: any /// separators, month names, two-digit years). `None` when it isn't a valid date. pub fn parse_date(text: &str, fmt: &str) -> Option { - let t = text.trim(); + let names = date_names(); + let lists: [&[&str]; 4] = [&names.months, &names.months_short, &names.days, &names.days_short]; + // Weekdays are shown, not typed (see `order`): where the format has one but none was typed, a + // name the language shares between a weekday and a month (Italian "mar") is the month, as + // the names are read again with no weekday expected (#5303). + read_date(&english_names(text, fmt, lists), fmt, &names).or_else(|| { + let no_weekday = fmt.replace("dddd", " ").replace("ddd", " "); + (no_weekday != fmt).then(|| read_date(&english_names(text, &no_weekday, lists), fmt, &names)).flatten() + }) +} + +/// [`parse_date`] of `english`, typed text whose month and weekday names are English. +fn read_date(english: &str, fmt: &str, names: &DateNames) -> Option { + let t = english.trim(); if t.is_empty() { return None; } @@ -223,10 +395,8 @@ pub fn parse_date(text: &str, fmt: &str) -> Option { } if cur.starts_with(|c: char| c.is_ascii_digit()) { nums.push_back(std::mem::take(cur)); - } else if cur.len() >= 3 - && let Some(i) = MONTHS.iter().position(|m| m.to_lowercase().starts_with(&cur.to_lowercase())) - { - *month = Some(i as u32 + 1); + } else if let Some(m) = month_named(cur, names) { + *month = Some(m); } cur.clear(); }; @@ -337,6 +507,11 @@ pub fn parse_date(text: &str, fmt: &str) -> Option { /// Format a date/time with Acrobat's date format letters. pub fn format_date(dt: DateTime, fmt: &str) -> String { + format_date_in(dt, fmt, &date_names()) +} + +/// [`format_date`] with month and weekday names from `names`. +pub fn format_date_in(dt: DateTime, fmt: &str, names: &DateNames) -> String { let chars: Vec = fmt.chars().collect(); let mut out = String::new(); let mut i = 0; @@ -354,9 +529,9 @@ pub fn format_date(dt: DateTime, fmt: &str) -> String { ('y', _) => { let _ = write!(out, "{:02}", dt.y.rem_euclid(100)); } - ('m', 3..) => match MONTHS.get((dt.m as usize).wrapping_sub(1)) { - Some(name) if n >= 4 => out.push_str(name), - Some(name) => out.push_str(name.get(..3).unwrap_or(name)), + ('m', 3..) => match (dt.m as usize).checked_sub(1).and_then(|i| names.months.get(i).zip(names.months_short.get(i))) { + Some((name, _)) if n >= 4 => out.push_str(name), + Some((_, short)) => out.push_str(short), None => { let _ = write!(out, "{}", dt.m); } @@ -368,8 +543,8 @@ pub fn format_date(dt: DateTime, fmt: &str) -> String { let _ = write!(out, "{}", dt.m); } ('d', 3..) => { - let name = DAYS.get(weekday(dt.y, dt.m, dt.d)).copied().unwrap_or_default(); - out.push_str(if n >= 4 { name } else { name.get(..3).unwrap_or(name) }); + let list = if n >= 4 { &names.days } else { &names.days_short }; + out.push_str(list.get(weekday(dt.y, dt.m, dt.d)).copied().unwrap_or_default()); } ('d', 2) => { let _ = write!(out, "{:02}", dt.d); @@ -451,6 +626,11 @@ fn apply_mask(mask: &str, text: &str) -> Option { /// The text shown for a value (the Format event). pub fn format_value(f: &Format, value: &str) -> String { + format_value_in(f, value, &date_names()) +} + +/// [`format_value`] with month and weekday names from `names`. +pub fn format_value_in(f: &Format, value: &str, names: &DateNames) -> String { if value.is_empty() { return String::new(); } @@ -479,7 +659,7 @@ pub fn format_value(f: &Format, value: &str) -> String { ), None => value.to_string(), }, - Format::Date(fmt) | Format::Time(fmt) => parse_date(value, fmt).map_or_else(|| value.to_string(), |dt| format_date(dt, fmt)), + Format::Date(fmt) | Format::Time(fmt) => parse_date(value, fmt).map_or_else(|| value.to_string(), |dt| format_date_in(dt, fmt, names)), Format::Special(n) => apply_mask(SPECIAL_MASKS[(*n).min(3) as usize], value).unwrap_or_else(|| value.to_string()), Format::Mask(m) => apply_mask(m, value).unwrap_or_else(|| value.to_string()), } diff --git a/crates/forms/src/appearance.rs b/crates/forms/src/appearance.rs index c9ff86d17a..8403b729d9 100644 --- a/crates/forms/src/appearance.rs +++ b/crates/forms/src/appearance.rs @@ -739,8 +739,18 @@ pub fn value_overflow(doc: &Document, f: &Field, w: &Widget) -> Option field_appearance_fit(doc, f, w, &f.value, true).2 } +/// Whether `text` draws in widget `wd` of `f` with no `?`: in the field's `/DR` font when that +/// holds it, as [`field_appearance_fit`] picks, or else in the Helvetica (WinAnsi) stand-in. +fn draws(doc: &Document, f: &Field, wd: &Dict, text: &str) -> bool { + let da = parse_da(wd.get(b"DA").and_then(|o| doc.resolve(o).as_string().map(|s| s.to_text())).as_deref().unwrap_or(&f.da)); + let enc = dr_font(doc, &da.font).map(|(_, enc)| enc).filter(|enc| enc.can_encode(text)).unwrap_or(FontText::WinAnsi); + enc.first_unencodable(text).is_none() +} + /// [`field_appearance_drawn`], and whether the text drawn is cut off by the box. fn field_appearance_fit(doc: &Document, f: &Field, w: &Widget, values: &[String], format: bool) -> (Stream, Option, Option) { + let wobj = doc.get(w.obj); + let wd = wobj.as_dict().cloned().unwrap_or_default(); // The Format event: what is shown, not what is stored. let formatted: Vec; let values = if format @@ -748,13 +758,16 @@ fn field_appearance_fit(doc: &Document, f: &Field, w: &Widget, values: &[String] && values.len() == 1 && f.actions.format != crate::af::Format::None { - formatted = vec![crate::af::format_value(&f.actions.format, &values[0])]; + let shown = crate::af::format_value(&f.actions.format, &values[0]); + // Month and weekday names in the date language that no font here can draw ("окт." in + // Helvetica) are drawn in English rather than as `?` (#5303). + let english = || crate::af::format_value_in(&f.actions.format, &values[0], &crate::af::ENGLISH_DATE_NAMES); + let dated = matches!(f.actions.format, crate::af::Format::Date(_) | crate::af::Format::Time(_)); + formatted = vec![if !dated || draws(doc, f, &wd, &shown) { shown } else { english() }]; &formatted[..] } else { values }; - let wobj = doc.get(w.obj); - let wd = wobj.as_dict().cloned().unwrap_or_default(); let (page_w, page_h) = ((w.rect[2] - w.rect[0]).max(1.0), (w.rect[3] - w.rect[1]).max(1.0)); let place = placement(mk_rotation(doc, &wd), page_w, page_h); let (width, height) = (place.layout_w, place.layout_h); diff --git a/crates/forms/src/author/edit.rs b/crates/forms/src/author/edit.rs index 6bb62cfb73..14659c4f52 100644 --- a/crates/forms/src/author/edit.rs +++ b/crates/forms/src/author/edit.rs @@ -234,6 +234,7 @@ pub fn set_props(doc: &mut Document, name: &str, props: &FieldProps) -> Result Result d.set(b"MaxLen".to_vec(), Object::Int(m as i64)), + // An absent key would inherit the parent's limit; zero is "no limit" (`read.rs`). + None if parent_limit => d.set(b"MaxLen".to_vec(), Object::Int(0)), None => { d.remove(b"MaxLen"); } diff --git a/crates/forms/src/scripting.rs b/crates/forms/src/scripting.rs index 7dd7bfde6c..31e6812e6b 100644 --- a/crates/forms/src/scripting.rs +++ b/crates/forms/src/scripting.rs @@ -60,6 +60,21 @@ pub struct ScriptResult { pub trait Scripts { /// Run `script` for `event` on `target`, whose value is `value`; `fields` is the whole form. fn run(&mut self, event: FieldEvent, script: &str, target: &Field, value: &str, fields: &[Field]) -> ScriptResult; + + /// [`Self::run`] again for an appearance, with dates written in English names (`util.printd`) + /// because no font of the field can draw the date language's (#5303); what it logs was + /// already logged by the first run. `None` when it can't: no English dates here, or the + /// script failed. What it returns is what the script made, even when that is the value. + fn rerun_with_english_dates( + &mut self, + _event: FieldEvent, + _script: &str, + _target: &Field, + _value: &str, + _fields: &[Field], + ) -> Option { + None + } } /// No script engine (or JavaScript turned off): scripts change nothing. @@ -98,6 +113,13 @@ pub(crate) fn formatted(doc: &Document, f: &Field, value: &str, scripts: &mut dy Some(r.value) } +/// [`formatted`] with dates in English names, for widgets whose fonts can't draw the date +/// language's; `None` when the field has no Format script or the English run didn't happen. +pub(crate) fn formatted_in_english(doc: &Document, f: &Field, value: &str, scripts: &mut dyn Scripts) -> Option { + let js = f.actions.scripts.format.as_ref()?; + scripts.rerun_with_english_dates(FieldEvent::Format, js, f, value, &fields(doc)).map(|r| r.value) +} + /// Apply script changes to other fields (values written without re-running their scripts; /// read-only, required and visibility as flags). `except` is the event's own field, whose value /// the event sets. diff --git a/crates/forms/src/tests.rs b/crates/forms/src/tests.rs index 38b5f5059f..3948e06e83 100644 --- a/crates/forms/src/tests.rs +++ b/crates/forms/src/tests.rs @@ -5487,6 +5487,29 @@ fn sorting_keeps_which_of_two_items_with_one_export_value_is_selected() { assert_eq!(selection_indices(&doc, &list), Some(vec![2])); } +#[test] +fn clearing_a_kids_length_limit_does_not_bring_back_its_parents() { + let mut doc = fixture(); + let city = field(&fields(&doc), "address.city").clone(); + let parent = doc.get(city.obj).as_dict().unwrap().reference(b"Parent").unwrap(); + doc.update_dict(parent, |d| d.set(b"MaxLen".to_vec(), Object::Int(3))).unwrap(); + set_props(&mut doc, "address.city", &FieldProps { max_len: Some(Some(8)), ..FieldProps::default() }).unwrap(); + assert_eq!(field(&fields(&doc), "address.city").max_len, Some(8)); + set_props(&mut doc, "address.city", &FieldProps { max_len: Some(None), ..FieldProps::default() }).unwrap(); + let mut doc = reopen(&doc); + let f = field(&fields(&doc), "address.city").clone(); + assert_eq!(f.max_len, None, "no limit, not the parent's 3"); + let own = doc.get(f.obj).as_dict().and_then(|d| d.get(b"MaxLen")).and_then(Object::as_int); + assert_eq!(own, Some(0), "the kid holds an explicit unlimited of its own"); + set_value(&mut doc, "address.city", &FieldValue::Text("Absalom".into())).unwrap(); + assert_eq!(field(&fields(&reopen(&doc)), "address.city").value, ["Absalom"], "not cut to 3 characters"); + // Without a parent limit, clearing still removes the key. + set_props(&mut doc, "name", &FieldProps { max_len: Some(Some(4)), ..FieldProps::default() }).unwrap(); + set_props(&mut doc, "name", &FieldProps { max_len: Some(None), ..FieldProps::default() }).unwrap(); + let name = field(&fields(&doc), "name").clone(); + assert!(doc.get(name.obj).as_dict().unwrap().get(b"MaxLen").is_none()); +} + #[test] fn clearing_a_kids_default_does_not_bring_back_its_parents() { let mut doc = fixture(); diff --git a/crates/forms/src/values.rs b/crates/forms/src/values.rs index cf191c2240..9d670e62db 100644 --- a/crates/forms/src/values.rs +++ b/crates/forms/src/values.rs @@ -426,9 +426,24 @@ fn redraw(doc: &mut Document, f: &Field, values: &[String], scripts: &mut dyn Sc }; let in_place = f.kind == FieldKind::Text && f.has(flags::PASSWORD); let mut undrawn = false; + // The Format script again with English month and weekday names, once, for widgets whose + // fonts can't draw the date language's (#5303). + let mut english: Option> = None; for w in &f.widgets { let (stream, missing) = match &shown { - Some(s) => appearance::field_appearance_drawn(doc, f, w, std::slice::from_ref(s), false), + Some(s) => { + let drawn = appearance::field_appearance_drawn(doc, f, w, std::slice::from_ref(s), false); + // Taken when the English run happened and draws whole, even where it is the + // stored value ("18-Oct-26" printed again from itself). + let again = match (&drawn.1, values) { + (Some(_), [one]) => english.get_or_insert_with(|| scripting::formatted_in_english(doc, f, one, scripts)).as_ref(), + _ => None, + }; + match again.map(|e| appearance::field_appearance_drawn(doc, f, w, std::slice::from_ref(e), false)) { + Some(english) if english.1.is_none() => english, + _ => drawn, + } + } None => appearance::field_appearance_drawn(doc, f, w, values, true), }; undrawn |= missing.is_some(); diff --git a/crates/js/fuzz/Cargo.toml b/crates/js/fuzz/Cargo.toml new file mode 100644 index 0000000000..373028192e --- /dev/null +++ b/crates/js/fuzz/Cargo.toml @@ -0,0 +1,24 @@ +[package] +name = "pdfcraft-js-fuzz" +version = "0.0.0" +publish = false +edition = "2024" +license = "MIT OR Apache-2.0" + +[package.metadata] +cargo-fuzz = true + +# Detached from the PdfCraft workspace: run with `cargo +nightly fuzz run run_script` +# from crates/js/. +[workspace] + +[dependencies] +libfuzzer-sys = "0.4" +pdfcraft-js = { path = ".." } + +[[bin]] +name = "run_script" +path = "fuzz_targets/run_script.rs" +test = false +doc = false +bench = false diff --git a/crates/js/fuzz/fuzz_targets/run_script.rs b/crates/js/fuzz/fuzz_targets/run_script.rs new file mode 100644 index 0000000000..57ccb9ea66 --- /dev/null +++ b/crates/js/fuzz/fuzz_targets/run_script.rs @@ -0,0 +1,26 @@ +//! Feeds arbitrary bytes, read as a script, to the JavaScript run entry with fixed event and document values. +//! +//! The run's result is discarded: the target only checks that no input panics or hangs past the limits. +#![no_main] + +use libfuzzer_sys::fuzz_target; +use pdfcraft_js::{DocInfo, Event, Limits, run}; + +/// The largest script the target runs (64 KiB); larger inputs are skipped. +const MAX_INPUT: usize = 64 << 10; + +fuzz_target!(|data: &[u8]| { + if data.len() > MAX_INPUT { + return; + } + let script = String::from_utf8_lossy(data).into_owned(); + let event = Event { + name: "Keystroke".into(), + kind: "Field".into(), + target: Some("field1".into()), + value: "value".into(), + change: String::new(), + will_commit: true, + }; + let _outcome = run(&script, &event, &DocInfo::default(), &[], &[], Limits::default()); +}); diff --git a/crates/js/src/format.rs b/crates/js/src/format.rs index 7f207b1916..26e2fad989 100644 --- a/crates/js/src/format.rs +++ b/crates/js/src/format.rs @@ -134,6 +134,177 @@ const MONTHS: [&str; 12] = ["January", "February", "March", "April", "May", "Jun const DAYS: [&str; 7] = ["Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday"]; +/// Month and weekday names (weekdays from Sunday): full, then short. +type Names = ([&'static str; 12], [&'static str; 12], [&'static str; 7], [&'static str; 7]); + +const ENGLISH: Names = ( + MONTHS, + ["Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"], + DAYS, + ["Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"], +); + +static NAMES: std::sync::RwLock = std::sync::RwLock::new(ENGLISH); + +/// Sets the month and weekday names `util.printd` writes for `mmmm`, `mmm`, `dddd` and `ddd` +/// (weekdays from Sunday), as a localized Acrobat writes its own language's (#5303). +/// `util.scand` reads them as well as English. +pub fn set_date_names(months: [&'static str; 12], months_short: [&'static str; 12], days: [&'static str; 7], days_short: [&'static str; 7]) { + *NAMES.write().unwrap_or_else(std::sync::PoisonError::into_inner) = (months, months_short, days, days_short); +} + +/// `text` with the month and weekday names of `lists` (in the order months, short months, +/// weekdays, short weekdays) written as their English names, so a parser that reads English +/// reads them. The longest name wins at each place, case ignored, and only a whole name: one +/// that doesn't continue a word, nor end in a digit that a number continues. A name may hold +/// digits ("10月"), spaces ("tháng 10"), dots ("ม.ค.") or combining vowel signs, which a +/// word-by-word reader would split (#5303). Where a weekday and a month share a name (Italian +/// "mar"), the order of the format's name fields (`mmm`, `ddd`) decides. English names are +/// found too, so that one typed ("Tue" before "Mar") fills its field of the format as a +/// localized one does. +fn english_names(text: &str, fmt: &str, lists: [&[&str]; 4]) -> String { + const EN: [&[&str]; 4] = [&MONTHS, &MONTHS, &DAYS, &DAYS]; + // No date is this long: hostile text is left as it is rather than scanned name by name. + if text.len() > MAX_DATE_TEXT_BYTES { + return text.to_string(); + } + let chars: Vec = text.chars().collect(); + let typed_english: [&[&str]; 4] = [&ENGLISH.0, &ENGLISH.1, &ENGLISH.2, &ENGLISH.3]; + // Each name as folded chars, whether it is a month, and its English name. + let mut names: Vec<(Vec, bool, &str)> = Vec::new(); + for (k, (list, en)) in lists.into_iter().zip(EN).chain(typed_english.into_iter().zip(EN)).enumerate() { + for (name, english) in list.iter().zip(en.iter()) { + // A short name ending in a dot ("mar.") is also read typed without it, as the same + // kind of name, so that where it equals another ("Mar", March) the format decides. + let bare = name.trim_end_matches('.'); + for name in std::iter::once(*name).chain((bare.len() < name.len()).then_some(bare)) { + let n: Vec = name.chars().map(fold).collect(); + if !n.is_empty() && n.len() <= MAX_NAME_CHARS { + names.push((n, k % 4 < 2, english)); + } + } + } + } + names.sort_by_key(|(n, ..)| std::cmp::Reverse(n.len())); + // The format's name fields in order: true for a month, false for a weekday. + let f: Vec = fmt.chars().collect(); + let mut expected = Vec::new(); + let mut j = 0; + while let Some(&c) = f.get(j) { + let run = f.get(j..).map_or(1, |r| r.iter().take_while(|x| **x == c).count()).max(1); + if run >= 3 && (c == 'm' || c == 'd') { + expected.push(c == 'm'); + } + j = j.saturating_add(run); + } + let mut slot = 0; + // A name's first letter may not continue a word; its last may not run on into a letter, nor + // (a digit) into a number. + let letter = |c: Option<&char>| c.is_some_and(|c| c.is_alphabetic()); + let mut out = String::with_capacity(text.len()); + let mut i = 0; + while let Some(&c) = chars.get(i) { + let fits = |n: &Vec| { + let end = i.saturating_add(n.len()); + let first = n.first().and_then(|e| e.chars().next()); + let last = n.last().and_then(|e| e.chars().next()); + chars.get(i..end).is_some_and(|t| t.iter().map(|c| fold(*c)).eq(n.iter().cloned())) + && !(first.is_some_and(char::is_alphabetic) && letter(i.checked_sub(1).and_then(|p| chars.get(p)))) + && !match last { + Some(l) if l.is_ascii_digit() => chars.get(end).is_some_and(char::is_ascii_digit), + Some(l) if l.is_alphabetic() => letter(chars.get(end)), + _ => false, + } + }; + let longest = names.iter().find(|(n, ..)| fits(n)).map(|(n, ..)| n.len()); + let hit = longest.and_then(|len| { + let mut same = names.iter().filter(|(n, ..)| n.len() == len && fits(n)); + let want = expected.get(slot).copied(); + let all: Vec<_> = same.by_ref().collect(); + all.iter().find(|(_, month, _)| Some(*month) == want).or_else(|| all.first()).copied() + }); + match hit { + Some((n, month, english)) => { + if expected.get(slot) == Some(month) { + slot += 1; + } + // Apart from what touches it ("10月18"), but no space added at either end. + if out.ends_with(char::is_alphanumeric) { + out.push(' '); + } + out.push_str(english); + i = i.saturating_add(n.len()); + if chars.get(i).is_some_and(|c| c.is_alphanumeric()) { + out.push(' '); + } + } + None => { + out.push(c); + i += 1; + } + } + } + out +} + +/// `c` as names are compared: lower case, with the capital I and the Turkish dotted and dotless +/// i all "i", and a combining dot above (U+0307) dropped, so that "EKİM", "SALI" and "MAYIS" are +/// Turkish "Ekim", "Salı" and "Mayıs", which lower-casing alone makes "ekim" with a dotted i, +/// "sali" and "mayis" (#5303). The Greek final sigma is a sigma, since a capital Σ, which +/// lower-cases to σ on its own, ends "ΙΑΝΟΥΆΡΙΟΣ" for "Ιανουάριος". One piece for each char, so +/// positions in the text stay aligned. +fn fold(c: char) -> String { + match c { + 'I' | 'ı' | 'İ' => "i".to_string(), + 'ς' => "σ".to_string(), + _ => c.to_lowercase().filter(|&l| l != '\u{307}').collect(), + } +} + +/// `s` [`fold`]ed char by char. +fn folded(s: &str) -> String { + s.chars().map(fold).collect() +} + +/// Typed dates longer than this are not searched for names. +const MAX_DATE_TEXT_BYTES: usize = 1024; + +/// Names longer than this are not looked for (settings are trusted, but a cap keeps the scan cheap). +const MAX_NAME_CHARS: usize = 32; + +fn names() -> Names { + *NAMES.read().unwrap_or_else(std::sync::PoisonError::into_inner) +} + +std::thread_local! { + /// Whether `util.printd` on this thread writes English names, whatever the date language. + static PRINTD_ENGLISH: std::cell::Cell = const { std::cell::Cell::new(false) }; +} + +/// Runs `run` with `util.printd` writing English month and weekday names when `english` (scripts +/// it starts carry this to their own thread): for an appearance whose fonts can't draw the date +/// language's names, which would come out as `?` (#5303). `util.scand` still reads both. +pub fn with_english_printd(english: bool, run: impl FnOnce() -> R) -> R { + struct Restore(bool); + impl Drop for Restore { + fn drop(&mut self) { + PRINTD_ENGLISH.with(|e| e.set(self.0)); + } + } + let _restore = Restore(PRINTD_ENGLISH.with(|e| e.replace(english))); + run() +} + +/// Whether [`with_english_printd`] is on for this thread. +pub(crate) fn printd_english() -> bool { + PRINTD_ENGLISH.with(std::cell::Cell::get) +} + +/// The names `util.printd` writes. +fn printd_names() -> Names { + if printd_english() { ENGLISH } else { names() } +} + /// `util.printd(cFormat, oDate)`: yyyy yy mmmm mmm mm m dddd ddd dd d HH H hh h MM M ss s tt, /// and the numeric formats 0 (D:yyyymmddHHMMss), 1 (yyyy.mm.dd HH:MM:ss) and 2 (m/d/yy h:MM:ss tt). pub(crate) fn printd(_: &JsValue, args: &[JsValue], ctx: &mut Context) -> JsResult { @@ -161,6 +332,7 @@ pub(crate) fn printd(_: &JsValue, args: &[JsValue], ctx: &mut Context) -> JsResu let (y, mo, day, wd, h, mi, sec) = (part("getFullYear")?, part("getMonth")?, part("getDate")?, part("getDay")?, part("getHours")?, part("getMinutes")?, part("getSeconds")?); let mut out = String::new(); + let (months, months_short, days, days_short) = printd_names(); let b = fmt.as_bytes(); let mut i = 0; while i < b.len() { @@ -186,13 +358,13 @@ pub(crate) fn printd(_: &JsValue, args: &[JsValue], ctx: &mut Context) -> JsResu } } b'm' => match run { - r if r >= 4 => out.push_str(MONTHS[mo as usize % 12]), - 3 => out.push_str(&MONTHS[mo as usize % 12][..3]), + r if r >= 4 => out.push_str(months.get(mo.rem_euclid(12) as usize).copied().unwrap_or_default()), + 3 => out.push_str(months_short.get(mo.rem_euclid(12) as usize).copied().unwrap_or_default()), n => push_num(&mut out, mo + 1, n), }, b'd' => match run { - r if r >= 4 => out.push_str(DAYS[wd as usize % 7]), - 3 => out.push_str(&DAYS[wd as usize % 7][..3]), + r if r >= 4 => out.push_str(days.get(wd.rem_euclid(7) as usize).copied().unwrap_or_default()), + 3 => out.push_str(days_short.get(wd.rem_euclid(7) as usize).copied().unwrap_or_default()), n => push_num(&mut out, day, n), }, b'H' => push_num(&mut out, h, run), @@ -435,11 +607,13 @@ fn scan_date(fmt: &str, text: &str, this_year: i32) -> Option { p += 1; } (ScanPart::Field { unit: DateUnit::MonthName, .. }, ScanToken::Letters(l)) => { - v.mo = Some(named(&MONTHS, l)?.saturating_add(1)); + let (months, months_short, ..) = names(); + v.mo = Some(named(&months, l).or_else(|| named(&months_short, l)).or_else(|| named(&MONTHS, l))?.saturating_add(1)); p += 1; } (ScanPart::Field { unit: DateUnit::Weekday, .. }, ScanToken::Letters(l)) => { - named(&DAYS, l)?; + let (.., days, days_short) = names(); + named(&days, l).or_else(|| named(&days_short, l)).or_else(|| named(&DAYS, l))?; p += 1; } (ScanPart::Field { unit: DateUnit::AmPm, .. }, ScanToken::Letters(l)) => { @@ -509,10 +683,11 @@ fn scan_date(fmt: &str, text: &str, this_year: i32) -> Option { /// The index of the name `word` is, in full or as its first three letters, in any case. fn named(names: &[&str], word: &str) -> Option { - let word = word.to_lowercase(); + let word = folded(word); let i = names.iter().position(|n| { - let n = n.to_lowercase(); - word == n || n.get(..3) == Some(word.as_str()) + let n = folded(n); + // A short name may end in a dot ("Okt."); three letters may be cut from any name. + word == n || word == n.trim_end_matches('.') || n.chars().take(3).collect::() == word })?; u32::try_from(i).ok() } @@ -524,6 +699,8 @@ pub(crate) fn scand(_: &JsValue, args: &[JsValue], ctx: &mut Context) -> JsResul return Ok(JsValue::null()); }; let fmt = text(&arg(args, 0), ctx)?; + let (months, months_short, days, days_short) = names(); + let typed = english_names(&typed, &fmt, [&months, &months_short, &days, &days_short]); let date = JsDate::new(ctx); let this_year = date.get_full_year(ctx)?.to_number(ctx)?; let this_year = if this_year.is_finite() { this_year as i32 } else { 2000 }; diff --git a/crates/js/src/lib.rs b/crates/js/src/lib.rs index 93f57e7528..0bf4032efd 100644 --- a/crates/js/src/lib.rs +++ b/crates/js/src/lib.rs @@ -44,6 +44,7 @@ mod run; #[cfg(test)] pub(crate) use format::PRINTD_SCANNED; use format::{printd, printf, printx, scand}; +pub use format::{set_date_names, with_english_printd}; pub(crate) use globals::menu_item; use globals::platform; use host::{arg, as_number, error, function, s, text}; diff --git a/crates/js/src/run.rs b/crates/js/src/run.rs index 07c0e984bb..ab7e32ed53 100644 --- a/crates/js/src/run.rs +++ b/crates/js/src/run.rs @@ -129,11 +129,13 @@ pub fn run_shared(script: &str, event: &Event, doc: &DocInfo, fields: Arc thread.join().unwrap_or_else(|_| failed("the script stopped with an internal error".into())), Err(e) => failed(format!("the script engine could not start: {e}")), diff --git a/crates/measure/src/snap.rs b/crates/measure/src/snap.rs index 930ffe3053..b26c389e96 100644 --- a/crates/measure/src/snap.rs +++ b/crates/measure/src/snap.rs @@ -189,6 +189,12 @@ pub fn constrain(last: Point, point: Point) -> Point { let angle = (v[1].atan2(v[0]) / std::f64::consts::FRAC_PI_4).round() * std::f64::consts::FRAC_PI_4; [last[0] + length * angle.cos(), last[1] + length * angle.sin()] } +/// Hold a proposed vertex level or plumb with the last vertex, whichever is nearer (a CAD +/// drawing's ortho mode): the vertex slides onto that axis rather than turning onto it. +pub fn ortho(last: Point, point: Point) -> Point { + let v = sub(point, last); + if v[0].abs() >= v[1].abs() { [point[0], last[1]] } else { [last[0], point[1]] } +} struct Walker<'a> { doc: &'a Document, geometry: Geometry, diff --git a/crates/measure/src/tests.rs b/crates/measure/src/tests.rs index 5b277e5546..88906910c7 100644 --- a/crates/measure/src/tests.rs +++ b/crates/measure/src/tests.rs @@ -449,6 +449,16 @@ fn listing_skips_repeated_refs_and_caps_what_it_keeps() { assert!(!listing.measurements.is_empty()); } #[test] +fn ortho_holds_a_vertex_level_or_plumb_with_the_last() { + let last = [10.0, 20.0]; + assert_eq!(snap::ortho(last, [50.0, 27.0]), [50.0, 20.0], "mostly across: level"); + assert_eq!(snap::ortho(last, [-30.0, 13.0]), [-30.0, 20.0], "to the left too"); + assert_eq!(snap::ortho(last, [14.0, 80.0]), [10.0, 80.0], "mostly up: plumb"); + assert_eq!(snap::ortho(last, [8.0, -40.0]), [10.0, -40.0], "and down"); + assert_eq!(snap::ortho(last, [30.0, 40.0]), [30.0, 20.0], "a tie stays level"); + assert_eq!(snap::ortho(last, last), last); +} +#[test] fn a_stream_over_the_byte_budget_uses_it_up_once() { // 50 references to one Flate stream that inflates past the 16 MiB budget decode once. let big = pdfcraft_cos::Stream::flate(Dict::new(), &vec![b' '; 17 << 20]); diff --git a/crates/ocr/Cargo.toml b/crates/ocr/Cargo.toml index 750e78b96d..1b04dbb522 100644 --- a/crates/ocr/Cargo.toml +++ b/crates/ocr/Cargo.toml @@ -16,5 +16,9 @@ ocrs = { version = "0.13.1", default-features = false, features = ["rten"] } rten = { version = "0.26.0", default-features = false, features = ["onnx_format", "rten_format"] } thiserror.workspace = true +[target.'cfg(not(target_arch = "wasm32"))'.dependencies] +# The page image is handed to the external tesseract process as a PNG. +png = "0.18" + [dev-dependencies] tempfile = "3.27" diff --git a/crates/ocr/README.md b/crates/ocr/README.md new file mode 100644 index 0000000000..cd33b186cf --- /dev/null +++ b/crates/ocr/README.md @@ -0,0 +1,104 @@ +# pdfcraft-ocr + +Layer L4 (may use `fonts` for the text layers' metrics): Scan & OCR — recognising the words in +a page raster and turning them back into page content (execution plan M10). + +```rust +// The caller renders a page to pixels and hands them over: +let image = OcrImage::new(w, h, rgba)?; // validated by construction +let lines = recognizer.recognize(&image, &options)?; // Vec: words in reading order +let placed: Vec = …; // PlacedWord::place maps boxes to user space +// The editable output draws real glyphs: plan the words, register the display faces into the +// plan (`pdfcraft_edit::add_type3_fonts`), then generate the layer. An unregistered plan falls +// back to WinAnsi, and characters no registered face has become '?'. +let plan = TextPlan::for_words(placed.iter().map(|w| w.text.as_str())); +add_type3_fonts(&mut doc, page, &mut plan)?; // names the plan's fonts on the page +let layer = visible_text_layer(&placed, &plan); // visible text: the editable output +// The searchable layer needs no display fonts: the engine's own layer (engine::ocr::layer) +// keeps every Unicode character with empty Type3 programs and a ToUnicode map beyond WinAnsi. +``` + +- **Engines behind one trait.** A [`Recognizer`] reports its `id`, whether it is `available`, + the languages it reads, and reads `Vec` from an image. Two exist: the pure-Rust + `OcrsRecognizer` (the `ocrs` crate, MIT/Apache-2.0, with its pre-trained models, + CC-BY-SA-4.0, fetched by `cargo xtask models`; found via `$PDFCRAFT_MODELS` or `models/` + beside the executable — see ATTRIBUTION.toml), and — off the web build — + `tesseract::TesseractCli`, which drives an installed `tesseract` binary as an external + process: never linked, bundled or downloaded. Implementations are shared across threads + (`Send + Sync`); nothing spawns on wasm32. +- **The tesseract process is fenced in.** The program comes from a user-configured location or + the PATH, never from document data (the engine validates a configured path as an absolute + path to an existing executable file before it is used). The language code is allow-listed + (`tesseract::language_code`, 12 codes) before anything reaches a process; the page image + goes to a private temp file (a fresh 0700 directory holding a 0600 file on Unix) that is + deleted afterwards; the run has a 120 s timeout, a 64 MiB output cap and a 50 000-word + reading cap (an incomplete reading is rejected rather than applied); every failure comes back through `OcrError` with its cause named. A configured + location that is invalid fails closed (it is never replaced by the PATH search). Availability + is probed with `--list-langs` (its list is read from stdout, or from stderr where older + versions print it); a failed probe is remembered for 30 s, so an install that appears later + is found without a restart. Known limit: at the timeout or the output cap only the direct + child is killed; a wrapper script's own children may outlive it. +- **Confidence drives review.** One classifier (`band_for`) puts a word in the High + (≥ 90), Medium (70–89) or Low (< 70) band — the review overlay's colours, the legend and the + counts; the batch flag rule (`low_confidence`, default threshold 60) is deliberately a + separate threshold. The `ocrs` models report no confidence (its words carry `None`); + a reading whose confidence is unknown is treated as unknown, never as good. +- **Two engines can be merged (ROVER).** Strategies: `PrimaryOnly` (the default), and two + ensembles — `ConfidenceWeighted` (the secondary engine joins when the primary's mean + confidence is under 70 or unknown; an empty primary keeps the fast path) and `RoverVote` + (both engines always run). The merge lets every primary word adopt the best-overlapping + unused secondary word at IoU ≥ 0.5; the higher confidence wins text, box and confidence (a + tie keeps the primary; against a primary word of unknown confidence — ocrs — the secondary + wins only at 70 or more, `UNKNOWN_CONFIDENCE_BAR`), the survivor is named `"ROVER"`, and + unclaimed secondary words are + appended under their engine's name. A merge is capped at 100 M word-pair comparisons (over + it, the primary reading stands alone), and a failing partner degrades to the primary with a + recorded note — a page is never skipped because a partner could not read it. +- **Preprocessing between the raster and recognition**, a pure function in fixed order: DPI + normalization to 300 (only when one of the steps below is on, since they are calibrated for + it; with all of them off the raster is read exactly as rendered, and a scan at or above 300 + is left alone), then the opt-in steps — + orientation (90°/180°/270° from projection-profile banding), deskew (± 5°, shear-scored, + below 0.1° is a no-op), denoise (3×3 median) and binarization (Sauvola, window 8, k 0.34). + Every geometric step composes one exact affine `InverseTransform`, and callers map every + word box back to the original raster before placing it. The heuristics are conservative and + panic-free: too small to judge, blank, or no clear winner is a no-op, never a guess. +- **Searchable output preserves Unicode without display fonts.** The engine uses empty + Type3 character programs and `/ToUnicode` maps for non-WinAnsi words, leaving the scan + unchanged. WinAnsi words use `/PCHelv`; the `/OCR` layer is fitted to each word's box. +- **Editable output uses display glyphs.** `visible_text_layer` writes visible text whose + size comes from the box height and whose horizontal scaling uses real font advances. + Beyond WinAnsi, attributed craft-fonts faces supply outline Type3 glyphs with `/ToUnicode`. + Uncovered characters become `?` and are reported in editable output notes; searchable + output retains them. The engine refuses nonfinite geometry or oversized input before + planning fonts: 4 MiB of text and 65,536 words, across all pages for editable export. + Empty geometry is skipped; rotated words carry `across`/`up` vectors and retain direction. +- **Hostile input is rejected, not panicked at.** `OcrImage` holds private, validated state: + non-zero sides, a 10 000-pixel side cap, a 64-megapixel area cap (checked arithmetic — the + counts come from documents, so they are untrusted), and an exact buffer length. + +## Where it is wired + +The engine's `ocr` module renders pages into `OcrImage`s, runs jobs (`OcrJob::run`) and applies +the words: searchable-image mode draws the invisible layer over the page's own image and leaves +that image untouched; editable mode writes a NEW `_ocr.pdf` of visible text (the scan +image is not carried over, the source is never overwritten). Skip probes keep pages — and +whole files — that already contain text out of the work unless Force OCR overrides them. A +searchable result applies as one undoable step; the Editable text output opens the OCR Verify +screen first, and Scan & OCR ▸ Correct recognized text reviews any recognition there before it +is applied (confidence-coloured boxes, corrections, a per-language user dictionary — a +plain-text file in the preferences folder, so words added from a document persist on disk +outside it until removed — Re-OCR of a region or a page, Accept/Reject), a +multiple-file run writes searchable copies as it goes, and the automation tools `ocr_recognize`, +`ocr_words`, `ocr_status` and `ocr_recognize_files` expose recognition headlessly. +`ocr_words` with `retain_review` returns a bounded session-local handle; `ocr_review_apply` +applies text corrections at the stored original boxes through its stale-document guard, +or discards the result. Empty corrections remove words. Searchable acceptance is one undo +step; editable acceptance writes a separate file. `ocr_dictionary` reads or atomically +updates the UI's per-language dictionary format through an explicitly root-confined path. + +Not yet: languages beyond the tested set (the `ocrs` models read only unaccented Latin; the +Tesseract side is bounded by its allow-list and whatever its installed language packs cover), +OCR in the browser (the tesseract module is compiled out for wasm32), the deep scan-cleanup +options (descreen, background removal, edge-shadow removal), and editable glyph coverage +beyond the available attributed faces. diff --git a/crates/ocr/src/confidence.rs b/crates/ocr/src/confidence.rs new file mode 100644 index 0000000000..1c0de605f5 --- /dev/null +++ b/crates/ocr/src/confidence.rs @@ -0,0 +1,106 @@ +//! Confidence: one classifier ([`band_for`]) drives the review overlay, the legend, the counts +//! and the uncertain-word navigation; the batch rule ([`low_confidence`]) is a separate +//! threshold with its own range check. The two rules are deliberately never merged. + +/// The three confidence bands (the review overlay's green / yellow / red). +#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)] +pub enum ConfidenceBand { + Low, + Medium, + High, +} + +impl ConfidenceBand { + /// The legend line for this band, so the meaning never rests on colour alone. + pub fn legend(self) -> &'static str { + match self { + ConfidenceBand::High => "HIGH (>= 90%)", + ConfidenceBand::Medium => "MEDIUM (70-89%)", + ConfidenceBand::Low => "LOW (< 70%)", + } + } +} + +/// The one classifier: `>= 90` is High, `70..=89` Medium, `< 70` Low. Out-of-range values +/// clamp (negative → Low, above 100 → High); NaN is Low. +pub fn band_for(confidence: f32) -> ConfidenceBand { + if confidence >= 90.0 { + ConfidenceBand::High + } else if confidence >= 70.0 { + ConfidenceBand::Medium + } else { + ConfidenceBand::Low + } +} + +/// The batch rule, separate from [`band_for`]: a word is flagged when `0 <= conf <= threshold` +/// (default 60). Negative or above-threshold values are not flagged; NaN is not. +pub fn low_confidence(confidence: f32, threshold: f32) -> bool { + (0.0..=threshold).contains(&confidence) +} + +/// A word whose confidence lands in the Low band — a suspect for review. A word without a +/// reported confidence has no signal and is never a suspect. +pub fn is_suspect(confidence: Option) -> bool { + confidence.is_some_and(|c| band_for(c) == ConfidenceBand::Low) +} + +#[cfg(test)] +mod tests { + use super::*; + + /// The boundary matrix: 89 and 90, 69 and 70 straddle the bands. + #[test] + fn bands_straddle_the_boundaries() { + assert_eq!(band_for(89.0), ConfidenceBand::Medium); + assert_eq!(band_for(90.0), ConfidenceBand::High); + assert_eq!(band_for(69.0), ConfidenceBand::Low); + assert_eq!(band_for(70.0), ConfidenceBand::Medium); + assert_eq!(band_for(0.0), ConfidenceBand::Low); + assert_eq!(band_for(100.0), ConfidenceBand::High); + } + + /// Out-of-range values clamp, and NaN lands in the Low band instead of comparing equal to + /// nothing. + #[test] + fn out_of_range_confidences_clamp() { + assert_eq!(band_for(-0.5), ConfidenceBand::Low, "negative → Low"); + assert_eq!(band_for(1000.0), ConfidenceBand::High, "above 100 → High"); + assert_eq!(band_for(f32::NAN), ConfidenceBand::Low); + assert_eq!(band_for(f32::NEG_INFINITY), ConfidenceBand::Low); + assert_eq!(band_for(f32::INFINITY), ConfidenceBand::High); + } + + /// The batch threshold is its own rule: everything in `0..=threshold` flags, including + /// words band_for calls Medium (60 < 70), and negatives never do. + #[test] + fn batch_threshold_is_separate_from_the_bands() { + assert!(low_confidence(60.0, 60.0), "at the threshold it flags"); + assert!(low_confidence(0.0, 60.0)); + assert!(!low_confidence(60.5, 60.0)); + assert!(!low_confidence(-1.0, 60.0), "negative confidences are out of the rule's range"); + assert!(!low_confidence(f32::NAN, 60.0)); + // The rules disagree on real values: 65 is inside the Low band yet outside the batch + // rule with the default threshold (60), which is why the two must not be merged. + assert!(!low_confidence(65.0, 60.0)); + assert_eq!(band_for(65.0), ConfidenceBand::Low); + } + + /// Unconfident words are not suspects; Low-band words are. + #[test] + fn suspects_need_a_low_band_confidence() { + assert!(!is_suspect(None)); + assert!(!is_suspect(Some(95.0))); + assert!(!is_suspect(Some(70.0))); + assert!(is_suspect(Some(69.9))); + assert!(is_suspect(Some(0.0))); + } + + /// The legend names the thresholds, so the overlay never rests on colour alone. + #[test] + fn legend_text_matches_the_classifier() { + assert_eq!(ConfidenceBand::High.legend(), "HIGH (>= 90%)"); + assert_eq!(ConfidenceBand::Medium.legend(), "MEDIUM (70-89%)"); + assert_eq!(ConfidenceBand::Low.legend(), "LOW (< 70%)"); + } +} diff --git a/crates/ocr/src/lib.rs b/crates/ocr/src/lib.rs index 7f58270e1a..ff5b920dca 100644 --- a/crates/ocr/src/lib.rs +++ b/crates/ocr/src/lib.rs @@ -1,35 +1,50 @@ //! pdfcraft-ocr — Scan & OCR ▸ Recognize text (L4). //! -//! Recognition runs the ocrs engine (MIT/Apache-2.0) with its pre-trained models (CC-BY-SA-4.0, -//! fetched by `cargo xtask models`; see ATTRIBUTION.toml). The caller renders a page to pixels; -//! [`Ocr::recognize`] finds the words in them, and [`text_layer`] turns words placed in user space -//! into page content: invisible text (rendering mode 3) over each word, so the page becomes a +//! The caller renders a page to pixels ([`OcrImage`]) and hands it to a [`Recognizer`]. Two +//! engines exist: the pure-Rust `OcrsRecognizer` (ocrs, MIT/Apache-2.0, with its pre-trained +//! models, CC-BY-SA-4.0, fetched by `cargo xtask models`; see ATTRIBUTION.toml) and — off the +//! web — `tesseract::TesseractCli`, which drives an installed `tesseract` binary as an external +//! process. `merge` combines two engines' readings (ROVER), `preprocess` cleans the raster up +//! between rendering and recognition, and [`text_layer`] turns words placed in user space into +//! page content: invisible text (rendering mode 3) over each word, so the page becomes a //! searchable image (Acrobat's "Searchable Image (Exact)": the image is left untouched). //! -//! The models read the Latin alphabet (English and other languages written without accents). +//! Visible editable output uses [`pdfcraft_fonts::TextPlan`] and optional craft-fonts glyph +//! outlines; unsupported visible characters become `?` and are counted. The engine's invisible +//! searchable layer uses empty Type3 programs and ToUnicode maps, preserving Unicode without +//! display fonts. WinAnsi-only invisible text keeps the standard Helvetica layout here. +//! +//! The ocrs models read the Latin alphabet (English and other languages written without +//! accents); Tesseract (the `tesseract` module, off the web build) reads whatever its installed +//! language packs cover. #![deny(clippy::unwrap_used, clippy::expect_used, clippy::panic, clippy::unimplemented, clippy::todo, clippy::unreachable)] -use std::path::{Path, PathBuf}; +pub mod confidence; +pub mod merge; +mod ocrs; +pub mod preprocess; +#[cfg(not(target_arch = "wasm32"))] +pub mod tesseract; +pub use confidence::{ConfidenceBand, band_for, is_suspect, low_confidence}; +pub use merge::{MergeStrategy, StrategyOutcome, mean_confidence, recognize_with_strategy, rover_merge}; +pub use ocrs::{ALPHABET, DETECTION_MODEL, LANGUAGES, Models, Ocr, OcrsRecognizer, RECOGNITION_MODEL}; + +pub use pdfcraft_fonts::TextPlan; pub use pdfcraft_fonts::helvetica_width; -/// The model files, as named in ATTRIBUTION.toml. -pub const DETECTION_MODEL: &str = "text-detection.onnx"; -pub const RECOGNITION_MODEL: &str = "text-recognition.onnx"; +/// Hard cap on either side of an image any code in this crate will touch, in pixels; larger +/// images are rejected with [`OcrError::ImageTooLarge`] instead of being processed. +pub const MAX_SIDE: u32 = 10_000; -/// The recognition model's alphabet, as it was trained (ocrs-models' `DEFAULT_ALPHABET`): its -/// output classes follow this order. The ocrs crate's own copy has an `E` where the training had -/// `€`, which made a scanned "€250" read as "E250". -pub const ALPHABET: &str = " 0123456789!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~€ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"; - -/// The languages the models read (ISO 639-1); all use the Latin alphabet without accents. -pub const LANGUAGES: &[(&str, &str)] = &[("en", "English")]; +/// Hard cap on an image's TOTAL area, in pixels (64 MP): sides under [`MAX_SIDE`] can still +/// multiply out to a raster no engine should have to read; larger images are rejected with +/// [`OcrError::ImageTooLarge`] instead of being processed. +pub const MAX_PIXELS: u32 = 64_000_000; #[derive(Debug, thiserror::Error)] pub enum OcrError { - /// Shown to end users (release packages ship the models; a source build fetches them with - /// `cargo xtask models`). The UI catalogs key this exact text. #[error("the text recognition models are not installed (reinstall PdfCraft, or set PDFCRAFT_MODELS to the folder that holds them)")] NoModels, #[error("loading {0}: {1}")] @@ -38,52 +53,93 @@ pub enum OcrError { Recognize(String), #[error("the image is empty")] EmptyImage, + #[error("the image is larger than the {MAX_SIDE}-pixel side or {MAX_PIXELS}-pixel limit")] + ImageTooLarge, + #[error("the image buffer holds {got} bytes, expected {want} for {width}x{height} RGBA")] + ImageSize { got: usize, want: usize, width: u32, height: u32 }, + #[error("the tesseract program is not available: {0}")] + NoTesseract(String), + #[error("tesseract did not finish within {0} seconds")] + Timeout(u64), + #[error("tesseract produced more than {0} bytes of output")] + OutputTooLarge(u64), + #[error("tesseract failed: {0}")] + Process(String), } -/// Where the two model files are. -#[derive(Clone, Debug, PartialEq)] -pub struct Models { - pub detection: PathBuf, - pub recognition: PathBuf, +/// An RGBA8 raster, row-major, top-left origin: what the engines read. The fields are private +/// so the invariants (non-zero, capped, exactly `width * height * 4` bytes) hold by +/// construction: an [`OcrImage`] in hand has always been validated, whatever produced it. +#[derive(Clone, Debug, PartialEq, Eq)] +pub struct OcrImage { + width: u32, + height: u32, + rgba: Vec, } -impl Models { - /// A complete model pair in `dir`, preferring current ONNX files over legacy RTEN files. - /// An incomplete update never mixes formats or hides an intact legacy pair. - pub fn in_dir(dir: &Path) -> Option { - [(DETECTION_MODEL, RECOGNITION_MODEL), ("text-detection.rten", "text-recognition.rten")].into_iter().find_map(|(detection, recognition)| { - let m = Models { detection: dir.join(detection), recognition: dir.join(recognition) }; - (m.detection.is_file() && m.recognition.is_file()).then_some(m) - }) - } - - /// Look for the models: `$PDFCRAFT_MODELS`, then where the release packages install them - /// beside the executable ([`Models::dirs_beside_exe`]), then the source tree's `assets/models/`. - pub fn find() -> Option { - Self::search_dirs().iter().find_map(|d| Self::in_dir(d)) - } - - /// The directories [`Models::find`] looks in, in order. - pub fn search_dirs() -> Vec { - let mut dirs = Vec::new(); - if let Some(d) = std::env::var_os("PDFCRAFT_MODELS") { - dirs.push(PathBuf::from(d)); +impl OcrImage { + /// Validate the sides ([`MAX_SIDE`]) and total area ([`MAX_PIXELS`]) and that `rgba` holds + /// exactly `width * height * 4` bytes (checked arithmetic; the counts come from documents, + /// so they are untrusted). + pub fn new(width: u32, height: u32, rgba: Vec) -> Result { + if width == 0 || height == 0 { + return Err(OcrError::EmptyImage); } - if let Some(exe) = std::env::current_exe().ok().and_then(|e| e.parent().map(Path::to_path_buf)) { - dirs.extend(Self::dirs_beside_exe(&exe)); + if width > MAX_SIDE || height > MAX_SIDE { + return Err(OcrError::ImageTooLarge); + } + let pixels = (width as usize).checked_mul(height as usize).ok_or(OcrError::ImageTooLarge)?; + if pixels > MAX_PIXELS as usize { + return Err(OcrError::ImageTooLarge); + } + match pixels.checked_mul(4) { + Some(want) if rgba.len() == want => Ok(OcrImage { width, height, rgba }), + Some(want) => Err(OcrError::ImageSize { got: rgba.len(), want, width, height }), + None => Err(OcrError::ImageTooLarge), } - dirs.push(Path::new(env!("CARGO_MANIFEST_DIR")).join("../../assets/models")); - dirs } - /// Where the release packages put the models, relative to the directory holding the - /// executable (packaging/*): - /// - `models/` beside it: the Windows MSI and portable zip; - /// - `../Resources/models`: the macOS app bundle (`Contents/MacOS` → `Contents/Resources`); - /// - `../share/pdfcraft/models`: the FHS-style trees (`bin/` → `share/`) of the Linux - /// deb/rpm/tar.gz/AppImage/Flatpak and the FreeBSD tarball, wherever they are installed. - pub fn dirs_beside_exe(exe_dir: &Path) -> Vec { - vec![exe_dir.join("models"), exe_dir.join("../Resources/models"), exe_dir.join("../share/pdfcraft/models")] + /// The image's width, in pixels. + pub fn width(&self) -> u32 { + self.width + } + + /// The image's height, in pixels. + pub fn height(&self) -> u32 { + self.height + } + + /// The RGBA bytes, row-major, exactly `width * height * 4` of them. + pub fn rgba(&self) -> &[u8] { + &self.rgba + } + + /// The smallest side, in pixels. + pub fn min_side(&self) -> u32 { + self.width.min(self.height) + } + + /// One RGBA byte at integer pixel (`x`, `y`), edges clamped; 255 (white) outside. Only + /// whole-pixel positions and channel 0–3 make sense; anything else reads clamped. + pub fn byte(&self, x: f64, y: f64, ch: usize) -> u8 { + let stride = self.width as usize * 4; + if stride == 0 || self.height == 0 || ch > 3 { + return 255; + } + let xi = (x.round().max(0.0) as usize).min(self.width as usize - 1); + let yi = (y.round().max(0.0) as usize).min(self.height as usize - 1); + self.rgba.get(yi * stride + xi * 4 + ch).copied().unwrap_or(255) + } + + /// The grayscale value of the pixel at (`x`, `y`), 0 (ink) … 255 (paper), with the borders + /// clamped (edges replicate). `None` only when the image is empty. + pub fn gray(&self, x: i64, y: i64) -> Option { + let x = x.clamp(0, self.width as i64 - 1); + let y = y.clamp(0, self.height as i64 - 1); + let row = y as usize * self.width as usize * 4; + let o = row + x as usize * 4; + let px = self.rgba.get(o..o + 4)?; + Some(((299 * px[0] as u32 + 587 * px[1] as u32 + 114 * px[2] as u32) / 1000) as u8) } } @@ -92,6 +148,26 @@ impl Models { pub struct Word { pub text: String, pub rect: [f32; 4], + /// How sure the engine is, 0–100, when it reports a confidence at all (`None`: the ocrs + /// engine does not). + pub confidence: Option, + /// Which engine read this word: the recogniser's [`Recognizer::id`], or `"ROVER"` when two + /// engines were merged and the surviving text is the merged choice. + pub source: String, +} + +impl Word { + pub fn new(text: impl Into, rect: [f32; 4], source: &str) -> Word { + Word { text: text.into(), rect, confidence: None, source: source.into() } + } + + /// Set the confidence (builder style), clamped to the documented 0–100 scale (a negative + /// value is 0, anything over 100 is 100). NaN means no confidence was reported (`None`), + /// the same unknown the engines that report nothing carry. + pub fn with_confidence(mut self, confidence: f32) -> Word { + self.confidence = if confidence.is_nan() { None } else { Some(confidence.clamp(0.0, 100.0)) }; + self + } } /// A recognised line of words, in reading order. @@ -106,63 +182,34 @@ impl Line { } } -/// A loaded recogniser. Loading takes a moment; keep one and reuse it. -pub struct Ocr { - engine: ocrs::OcrEngine, +/// What a recogniser needs to know beyond the pixels. +#[derive(Clone, Debug, PartialEq, Eq)] +pub struct RecognizeOptions { + /// Language code from [`LANGUAGES`] (`"en"`, `"de"`, …). Engines map it to their own codes + /// (Tesseract: `tesseract::language_code`, allow-listed before anything reaches a process); + /// unknown codes fall back to the engine's default. + pub language: String, } -impl Ocr { - pub fn load(models: &Models) -> Result { - let load = |p: &Path| rten::Model::load_file(p).map_err(|e| OcrError::Load(p.display().to_string(), e.to_string())); - let params = ocrs::OcrEngineParams { - detection_model: Some(load(&models.detection)?), - recognition_model: Some(load(&models.recognition)?), - alphabet: Some(ALPHABET.into()), - ..Default::default() - }; - let engine = ocrs::OcrEngine::new(params).map_err(|e| OcrError::Load("ocr engine".into(), e.to_string()))?; - Ok(Ocr { engine }) +impl Default for RecognizeOptions { + fn default() -> Self { + RecognizeOptions { language: "en".into() } } +} - /// Load the models found by [`Models::find`]. - pub fn find() -> Result { - Self::load(&Models::find().ok_or(OcrError::NoModels)?) - } - - /// Recognise the text in an RGBA (or RGB, or grey) image, `width` × `height` pixels. - pub fn recognize(&self, pixels: &[u8], width: u32, height: u32) -> Result, OcrError> { - if width == 0 || height == 0 || pixels.is_empty() { - return Err(OcrError::EmptyImage); - } - // ocrs wants 1 or 3 channels. - let rgb: std::borrow::Cow<[u8]> = if pixels.len() == (width * height * 4) as usize { - pixels.as_chunks::<4>().0.iter().flat_map(|p| [p[0], p[1], p[2]]).collect::>().into() - } else { - pixels.into() - }; - let err = |e: &dyn std::fmt::Display| OcrError::Recognize(e.to_string()); - let source = ocrs::ImageSource::from_bytes(&rgb, (width, height)).map_err(|e| err(&e))?; - let input = self.engine.prepare_input(source).map_err(|e| err(&e))?; - let found = self.engine.detect_words(&input).map_err(|e| err(&e))?; - let lines = self.engine.find_text_lines(&input, &found); - let read = self.engine.recognize_text(&input, &lines).map_err(|e| err(&e))?; - use ocrs::TextItem; - Ok(read - .into_iter() - .flatten() - .map(|line| Line { - words: line - .words() - .map(|w| { - let r = w.bounding_rect(); - Word { text: w.to_string(), rect: [r.left() as f32, r.top() as f32, r.right() as f32, r.bottom() as f32] } - }) - .filter(|w| !w.text.trim().is_empty()) - .collect(), - }) - .filter(|l| !l.words.is_empty()) - .collect()) - } +/// One text-recognition engine. Implementations are shared across threads +/// (`Arc`); `available` and `recognize` may spawn processes only off the web +/// build (see the `tesseract` module). +pub trait Recognizer: Send + Sync { + /// The engine's short name (`"ocrs"`, `"tesseract"`): what [`Word::source`] carries and + /// what settings and tools name it by. + fn id(&self) -> &'static str; + /// Whether this engine can run right now (models installed, binary found). + fn available(&self) -> bool; + /// The languages it reads, as `(code, name)` pairs (`"en"`, `"English"`). + fn languages(&self) -> Vec<(String, String)>; + /// Read the words in `image`, in reading order. + fn recognize(&self, image: &OcrImage, options: &RecognizeOptions) -> Result, OcrError>; } /// A word placed on a page, in PDF user space: the bottom-left corner of its box, and the @@ -173,6 +220,10 @@ pub struct PlacedWord { pub origin: [f64; 2], pub across: [f64; 2], pub up: [f64; 2], + /// Confidence 0–100, when the engine reported one. + pub confidence: Option, + /// Which engine read this word (see [`Word::source`]). + pub source: String, } impl PlacedWord { @@ -183,7 +234,14 @@ impl PlacedWord { let o = to_user(l, b); let br = to_user(r, b); let tl = to_user(l, t); - PlacedWord { text: word.text.clone(), origin: o, across: [br[0] - o[0], br[1] - o[1]], up: [tl[0] - o[0], tl[1] - o[1]] } + PlacedWord { + text: word.text.clone(), + origin: o, + across: [br[0] - o[0], br[1] - o[1]], + up: [tl[0] - o[0], tl[1] - o[1]], + confidence: word.confidence, + source: word.source.clone(), + } } } @@ -214,32 +272,193 @@ fn num(v: f64) -> String { if s == "-0" { "0".into() } else { s.into() } } -/// Page content that writes `words` as invisible text in `/PCHelv` (standard Helvetica, -/// WinAnsiEncoding), each word stretched to its box so selection and search highlight the -/// right place. Marked content `/OCR` so it can be told apart from the page's own text. -pub fn text_layer(words: &[PlacedWord]) -> Vec { +/// One run of a word's planned characters in one Type3 font: the resource name, the codes to +/// show (one byte per character), and the run's start as an offset in em units from the word's +/// origin (the accumulated advances of the runs before it). +struct PlannedRun { + font: String, + codes: Vec, + start: f64, +} + +/// The planned runs of word `i` (the same index [`pdfcraft_fonts::TextPlan::for_words`] planned +/// the words with), or `None` when the word is drawn in Helvetica: every character has a +/// WinAnsi code, or nothing could be planned/registered for it (the pre-Unicode output). +fn planned_runs(plan: &pdfcraft_fonts::TextPlan, word: usize) -> Option<(Vec, f64)> { + let enc = plan.encoding(word); + if enc.is_empty() || !plan.registered(enc) { + return None; + } + let mut runs: Vec = Vec::new(); + let mut total = 0.0; + for &(f, code) in enc { + let font = plan.fonts[f].name.as_str(); + match runs.last_mut() { + Some(last) if last.font == font => last.codes.push(code), + _ => runs.push(PlannedRun { font: font.to_string(), codes: vec![code], start: total }), + } + total += plan.advance(f, code); + } + (total.is_finite() && total > 0.0).then_some((runs, total)) +} + +/// Page content that writes `words` as VISIBLE text at their original boxes: the +/// editable-text output's page content. WinAnsi words are set in `/PCHelv` (standard +/// Helvetica); words with characters beyond WinAnsi are set in the plan's Type3 fonts, one run +/// per font. Each word fills its box — the font size comes from the box height and the +/// horizontal scaling is computed from the real advances of the font that draws the word +/// (Helvetica's approximated, the Type3 faces' exact) and emitted as an explicit `Tz` +/// percentage. Words whose box is empty (or whose text has no measurable width) are skipped; a +/// page that ends up with nothing simply shows nothing. +pub fn visible_text_layer(words: &[PlacedWord], plan: &pdfcraft_fonts::TextPlan) -> Vec { + visible_text_layer_at(words, plan, 0) +} + +/// Visible text for words starting at `word_offset` in a shared document plan. +pub fn visible_text_layer_at(words: &[PlacedWord], plan: &pdfcraft_fonts::TextPlan, word_offset: usize) -> Vec { + // The rendering mode and the font are graphics state, which `ET` doesn't reset: without + // `q … Q` they would stay set for every later content stream on the page, so text added + // after OCR (a header or footer) drew invisibly (#3029). + let mut out = b"/OCR BMC\nq\nBT\n0 Tr\n/PCHelv 1 Tf\n".to_vec(); + // The font and size the text state is set in, so a Tf is written only when they change. + // (The header's size-1 Helvetica covers layers of WinAnsi words only.) + let mut current: Option<(String, f64)> = Some(("PCHelv".to_string(), 1.0)); + for (i, w) in words.iter().enumerate() { + let height = w.up[0].hypot(w.up[1]); + let across = w.across[0].hypot(w.across[1]); + let (ux, uy) = (w.across[0] / across, w.across[1] / across); + let (vx, vy) = (w.up[0] / height, w.up[1] / height); + if !across.is_finite() + || !height.is_finite() + || across <= 0.0 + || height <= 0.0 + || !ux.is_finite() + || !vx.is_finite() + || !w.origin.iter().all(|v| v.is_finite()) + { + continue; + } + let size = height / BOX_EM; + let runs = planned_runs(plan, word_offset.saturating_add(i)); + // The word's natural advance at this size: Helvetica's approximation, or the exact sum + // of the planned glyphs' widths. + let natural = runs.as_ref().map_or_else(|| helvetica_width(&w.text, size), |(_, total)| total * size); + if natural.is_nan() || natural <= 0.0 { + continue; + } + // Horizontal scaling as a percentage of the natural advance at this size, so the word + // lands exactly on its box (100 = no scaling; a long box on a short word stretches it). + let tz = (across / natural * 100.0).max(1.0); + // The baseline sits above the box's bottom edge by the descender depth. + let o = [w.origin[0] + vx * DESCENT * size, w.origin[1] + vy * DESCENT * size]; + out.extend_from_slice(format!("{} Tz ", num(tz)).as_bytes()); + match &runs { + None => { + if current.as_ref().is_some_and(|(f, s)| f != "PCHelv" || *s != size) { + out.extend_from_slice(format!("/PCHelv {} Tf ", num(size)).as_bytes()); + current = Some(("PCHelv".to_string(), size)); + } + out.extend_from_slice(format!("{} {} {} {} {} {} Tm ", num(ux), num(uy), num(vx), num(vy), num(o[0]), num(o[1])).as_bytes()); + out.extend(pdfcraft_fonts::literal(&pdfcraft_fonts::win_ansi(&w.text))); + out.extend_from_slice(b" Tj\n"); + } + Some((planned, total)) => { + for run in planned { + if current.as_ref().is_some_and(|(f, s)| *f != run.font || *s != size) { + out.extend_from_slice(format!("/{} {} Tf ", run.font, num(size)).as_bytes()); + current = Some((run.font.clone(), size)); + } + // The run's origin: the word's baseline plus the across vector scaled by + // the advances before this run (the stretch applies to runs alike). + let frac = run.start / total; + out.extend_from_slice( + format!( + "{} {} {} {} {} {} Tm ", + num(ux), + num(uy), + num(vx), + num(vy), + num(o[0] + w.across[0] * frac), + num(o[1] + w.across[1] * frac) + ) + .as_bytes(), + ); + out.extend(pdfcraft_fonts::literal(&run.codes)); + out.extend_from_slice(b" Tj\n"); + } + } + } + } + out.extend_from_slice(b"ET\nQ\nEMC\n"); + out +} + +/// Page content that writes `words` as invisible text (rendering mode 3), each word stretched +/// to its box so selection and search highlight the right place. WinAnsi words are set in +/// `/PCHelv` (standard Helvetica); words with characters beyond WinAnsi are set in the plan's +/// Type3 fonts, one run per font, with the same exact box fit. Marked content `/OCR` so it can +/// be told apart from the page's own text. +pub fn text_layer(words: &[PlacedWord], plan: &pdfcraft_fonts::TextPlan) -> Vec { // `3 Tr` and the font are graphics state, which `ET` doesn't reset: without `q … Q` they // would stay set for every later content stream on the page, so text added after OCR (a // header or footer) drew invisibly (#3029). let mut out = b"/OCR BMC\nq\nBT\n3 Tr\n/PCHelv 1 Tf\n".to_vec(); - for w in words { - let width = helvetica_width(&w.text, 1.0); + let mut current: Option<(String, f64)> = Some(("PCHelv".to_string(), 1.0)); + for (i, w) in words.iter().enumerate() { + let runs = planned_runs(plan, i); + let width = runs.as_ref().map_or_else(|| helvetica_width(&w.text, 1.0), |(_, total)| *total); let height = w.up[0].hypot(w.up[1]); - if width <= 0.0 || height <= 0.0 { + let across = w.across[0].hypot(w.across[1]); + // Only positive, finite geometry reaches the content stream: a zero-sized box cannot + // be drawn, and a NaN or infinite one would put non-finite numbers on the page. (NaN + // fails every comparison, so the positivity checks already refuse it; the finiteness + // checks keep that property explicit.) + if width <= 0.0 || !(height > 0.0 && height.is_finite()) || !(across > 0.0 && across.is_finite()) || !w.origin.iter().all(|v| v.is_finite()) { continue; } - // The em square: the word's own ink (cap height or x-height, descenders or not) spans the - // box, so the text is as tall as the scanned word; its width is stretched to fit the word. - let (top, bottom) = ink_extent(&w.text); + // The em square. Helvetica words: the word's own ink (cap height or x-height, + // descenders or not) spans the box, so the text is as tall as the scanned word. Planned + // (Type3) words: those Helvetica letter heights do not describe the face, so the box spans + // the ascender-descender height. Either way the width is stretched to fit the word. + let (top, bottom) = if runs.is_some() { (BOX_EM - DESCENT, -DESCENT) } else { ink_extent(&w.text) }; let em = [w.up[0] / (top - bottom), w.up[1] / (top - bottom)]; - let ax = [w.across[0] / width, w.across[1] / width]; let o = [w.origin[0] - em[0] * bottom, w.origin[1] - em[1] * bottom]; - let m = [ax[0], ax[1], em[0], em[1], o[0], o[1]]; - let nums: Vec = m.iter().map(|v| num(*v)).collect(); - out.extend_from_slice(nums.join(" ").as_bytes()); - out.extend_from_slice(b" Tm "); - out.extend(pdfcraft_fonts::literal(&pdfcraft_fonts::win_ansi(&w.text))); - out.extend_from_slice(b" Tj\n"); + match &runs { + None => { + let ax = [w.across[0] / width, w.across[1] / width]; + if current.as_ref().is_some_and(|(f, s)| f != "PCHelv" || *s != 1.0) { + out.extend_from_slice(b"/PCHelv 1 Tf "); + current = Some(("PCHelv".to_string(), 1.0)); + } + let m = [ax[0], ax[1], em[0], em[1], o[0], o[1]]; + let nums: Vec = m.iter().map(|v| num(*v)).collect(); + out.extend_from_slice(nums.join(" ").as_bytes()); + out.extend_from_slice(b" Tm "); + out.extend(pdfcraft_fonts::literal(&pdfcraft_fonts::win_ansi(&w.text))); + out.extend_from_slice(b" Tj\n"); + } + Some((planned, total)) => { + for run in planned { + if current.as_ref().is_some_and(|(f, s)| *f != run.font || *s != 1.0) { + out.extend_from_slice(format!("/{} 1 Tf ", run.font).as_bytes()); + current = Some((run.font.clone(), 1.0)); + } + let m = [ + w.across[0] / total, + w.across[1] / total, + em[0], + em[1], + o[0] + w.across[0] * run.start / total, + o[1] + w.across[1] * run.start / total, + ]; + let nums: Vec = m.iter().map(|v| num(*v)).collect(); + out.extend_from_slice(nums.join(" ").as_bytes()); + out.extend_from_slice(b" Tm "); + out.extend(pdfcraft_fonts::literal(&run.codes)); + out.extend_from_slice(b" Tj\n"); + } + } + } } out.extend_from_slice(b"ET\nQ\nEMC\n"); out @@ -253,10 +472,22 @@ mod tests { fn places_words_through_the_mapping() { // 2 pixels per point, page 612 × 792, y flipped. let to_user = |x: f32, y: f32| [x as f64 / 2.0, 792.0 - y as f64 / 2.0]; - let w = PlacedWord::place(&Word { text: "Hello".into(), rect: [100.0, 200.0, 300.0, 240.0] }, to_user); + let w = PlacedWord::place(&Word::new("Hello", [100.0, 200.0, 300.0, 240.0], "ocrs"), to_user); assert_eq!(w.origin, [50.0, 672.0]); assert_eq!(w.across, [100.0, 0.0]); assert_eq!(w.up, [0.0, 20.0]); + assert_eq!(w.source, "ocrs"); + assert_eq!(w.confidence, None); + } + + /// Confidence and source survive placement. + #[test] + fn placed_words_carry_confidence_and_source() { + let to_user = |x: f32, y: f32| [x as f64, y as f64]; + let w = Word::new("Hi", [1.0, 2.0, 3.0, 4.0], "tesseract").with_confidence(87.5); + let p = PlacedWord::place(&w, to_user); + assert_eq!(p.confidence, Some(87.5)); + assert_eq!(p.source, "tesseract"); } /// The text layer is Helvetica in WinAnsiEncoding: every character the model can recognise @@ -267,10 +498,27 @@ mod tests { assert_eq!(pdfcraft_fonts::first_undrawable(ALPHABET), None); } + /// `with_confidence` is clamped to the documented 0–100 scale: negatives (and −inf) are 0, + /// anything over 100 (and +inf) is 100, in-scale values pass through, and NaN means no + /// confidence was reported — the same `None` unknown the unreporting engines carry. + #[test] + fn with_confidence_clamps_and_treats_nan_as_unknown() { + let c = |v: f32| Word::new("w", [0.0; 4], "ocrs").with_confidence(v).confidence; + assert_eq!(c(0.0), Some(0.0)); + assert_eq!(c(50.0), Some(50.0)); + assert_eq!(c(100.0), Some(100.0)); + assert_eq!(c(-1.0), Some(0.0), "negatives clamp to 0"); + assert_eq!(c(f32::NEG_INFINITY), Some(0.0)); + assert_eq!(c(150.0), Some(100.0), "over 100 clamps to 100"); + assert_eq!(c(f32::INFINITY), Some(100.0)); + assert_eq!(c(f32::NAN), None, "NaN is unknown, not a confidence"); + assert_eq!(c(f32::EPSILON), Some(f32::EPSILON), "in-scale values are untouched"); + } + #[test] fn text_layer_is_invisible_text_fitted_to_each_box() { - let w = PlacedWord { text: "Hi".into(), origin: [10.0, 20.0], across: [30.0, 0.0], up: [0.0, 9.3] }; - let s = String::from_utf8(text_layer(&[w])).unwrap(); + let w = PlacedWord { text: "Hi".into(), origin: [10.0, 20.0], across: [30.0, 0.0], up: [0.0, 9.3], confidence: None, source: "ocrs".into() }; + let s = String::from_utf8(text_layer(&[w], &TextPlan::default())).unwrap(); assert!(s.contains("3 Tr") && s.contains("/PCHelv 1 Tf"), "{s}"); let width = helvetica_width("Hi", 1.0); // "Hi" has no descender: its box spans the cap height, from the baseline up. @@ -283,8 +531,8 @@ mod tests { /// after it on the page starts from the graphics state it had before. #[test] fn the_layer_restores_the_graphics_state_it_changes() { - let w = PlacedWord { text: "Hi".into(), origin: [10.0, 20.0], across: [30.0, 0.0], up: [0.0, 9.3] }; - let s = String::from_utf8(text_layer(&[w])).unwrap(); + let w = PlacedWord { text: "Hi".into(), origin: [10.0, 20.0], across: [30.0, 0.0], up: [0.0, 9.3], confidence: None, source: "ocrs".into() }; + let s = String::from_utf8(text_layer(&[w], &TextPlan::default())).unwrap(); let mut depth = 0_i32; for token in s.split_ascii_whitespace() { match token { @@ -309,78 +557,200 @@ mod tests { assert_eq!(ink_extent("Gallus.Ze@camtel.cm"), (CAP_HEIGHT, 0.0)); assert_eq!(ink_extent("Corporation"), (CAP_HEIGHT, -DESCENT)); assert_eq!(ink_extent("--"), (BOX_EM - DESCENT, -DESCENT)); - let w = PlacedWord { text: "gap".into(), origin: [0.0, 100.0], across: [20.0, 0.0], up: [0.0, 7.3] }; - let s = String::from_utf8(text_layer(&[w])).unwrap(); + let w = PlacedWord { text: "gap".into(), origin: [0.0, 100.0], across: [20.0, 0.0], up: [0.0, 7.3], confidence: None, source: "ocrs".into() }; + let s = String::from_utf8(text_layer(&[w], &TextPlan::default())).unwrap(); let em = 7.3 / (X_HEIGHT + DESCENT); assert!(s.contains(&format!(" 0 0 {} 0 {} Tm (gap) Tj", num(em), num(100.0 + em * DESCENT))), "{s}"); } + /// A word whose box is not positive, finite geometry never reaches the content stream: + /// NaN, infinite and zero-sized boxes are dropped, so the layer never carries non-finite + /// numbers. + #[test] + fn text_layer_refuses_non_finite_and_degenerate_boxes() { + let word = |across: [f64; 2], up: [f64; 2], origin: [f64; 2]| PlacedWord { + text: "Hi".into(), + origin, + across, + up, + confidence: Some(90.0), + source: "tesseract".into(), + }; + for (why, w) in [ + ("NaN across", word([f64::NAN, 0.0], [0.0, 9.3], [0.0; 2])), + ("NaN up", word([30.0, 0.0], [f64::NAN, 1.0], [0.0; 2])), + ("inf up", word([30.0, 0.0], [0.0, f64::INFINITY], [0.0; 2])), + ("inf origin", word([30.0, 0.0], [0.0, 9.3], [0.0, f64::INFINITY])), + ("zero across", word([0.0, 0.0], [0.0, 9.3], [0.0; 2])), + ("NaN origin", word([30.0, 0.0], [0.0, 9.3], [0.0, f64::NAN])), + ] { + let s = String::from_utf8(text_layer(&[w], &TextPlan::default())).unwrap(); + assert_eq!(s, "/OCR BMC\nq\nBT\n3 Tr\n/PCHelv 1 Tf\nET\nQ\nEMC\n", "{why}: {s}"); + assert!(!s.contains("NaN") && !s.contains("inf"), "{why}: {s}"); + } + } + + /// The editable-text layer draws visible text whose horizontal scaling is an explicit + /// percentage computed from real Helvetica metrics: `Tz` × natural advance = box width. + #[test] + fn visible_text_layer_scales_by_a_percentage_of_real_metrics() { + let w = PlacedWord { text: "Hi".into(), origin: [10.0, 20.0], across: [40.0, 0.0], up: [0.0, 9.3], confidence: None, source: "ocrs".into() }; + let s = String::from_utf8(visible_text_layer(&[w], &TextPlan::default())).unwrap(); + assert!(s.contains("0 Tr"), "visible, not invisible: {s}"); + let size = 9.3 / BOX_EM; + let natural = helvetica_width("Hi", size); + let tz = 40.0 / natural * 100.0; + assert!(s.contains(&format!("{} Tz", num(tz))), "{s}"); + assert!(s.contains("(Hi) Tj"), "{s}"); + assert!(s.starts_with("/OCR BMC") && s.ends_with("EMC\n")); + } + + /// Degenerate boxes and words with no measurable width are skipped, never divide by zero. + #[test] + fn visible_text_layer_skips_degenerate_words() { + let word = |across: [f64; 2], up: [f64; 2], text: &str| PlacedWord { + text: text.into(), + origin: [0.0; 2], + across, + up, + confidence: None, + source: "ocrs".into(), + }; + let words = vec![ + word([0.0, 0.0], [0.0, 9.0], "flat"), + word([10.0, 0.0], [0.0, 0.0], "short"), + word([f64::NAN, 0.0], [0.0, 9.0], "nan"), + word([10.0, 0.0], [0.0, 9.0], ""), + word([10.0, 0.0], [0.0, 9.0], "Hi"), + ]; + let s = String::from_utf8(visible_text_layer(&words, &TextPlan::default())).unwrap(); + assert_eq!(s.matches("Tj").count(), 1, "only the one drawable word is written: {s}"); + } + + /// A word whose ORIGIN is not finite never reaches the visible layer either: a NaN or + /// infinite origin would put "NaN"/"inf" into the content stream. + #[test] + fn visible_text_layer_refuses_non_finite_origins() { + let word = + |origin: [f64; 2]| PlacedWord { text: "Hi".into(), origin, across: [10.0, 0.0], up: [0.0, 9.0], confidence: None, source: "ocrs".into() }; + for (why, origin) in [("NaN x", [f64::NAN, 100.0]), ("inf y", [10.0, f64::INFINITY]), ("NaN both", [f64::NAN; 2])] { + let s = String::from_utf8(visible_text_layer(&[word(origin)], &TextPlan::default())).unwrap(); + assert!(!s.contains("NaN") && !s.contains("inf"), "{why}: {s}"); + assert_eq!(s.matches("Tj").count(), 0, "{why}: {s}"); + } + } + + /// After a Type3 word the text state goes back to Helvetica: the Latin words around a + /// planned word must not inherit its font. + #[test] + fn helvetica_is_restored_after_a_planned_word() { + if pdfcraft_fonts::CRAFT_FONTS.is_empty() { + eprintln!("skipping: built without craft-fonts (set CRAFT_FONTS_DIR to run it)"); + return; + } + let word = |text: &str| PlacedWord { + text: text.into(), + origin: [10.0, 20.0], + across: [60.0, 0.0], + up: [0.0, 9.3], + confidence: None, + source: "ocrs".into(), + }; + let words = [word("Hi"), word("日本"), word("Ho")]; + let mut plan = TextPlan::for_words(["Hi", "日本", "Ho"]); + for (i, font) in plan.fonts.iter_mut().enumerate() { + font.name = format!("PCOcr{i}"); + } + let s = String::from_utf8(text_layer(&words, &plan)).unwrap(); + assert!(s.contains("/PCOcr0 1 Tf"), "{s}"); + // The last Helvetica selection comes after the Type3 run, and the Latin word after it + // is drawn in Helvetica, not left in the planned font. + assert!(s.rfind("/PCOcr0 1 Tf").unwrap_or(0) < s.rfind("/PCHelv 1 Tf").unwrap_or(usize::MAX), "{s}"); + // The `Q` closes the `q … Q` scope that keeps `3 Tr` from leaking into later content. + assert!(s.ends_with("Tm (Ho) Tj\nET\nQ\nEMC\n"), "{s}"); + // The visible layer restores Helvetica with the word's size. + let v = String::from_utf8(visible_text_layer(&words, &plan)).unwrap(); + assert!(v.rfind("/PCOcr0").unwrap_or(0) < v.rfind("/PCHelv").unwrap_or(usize::MAX), "{v}"); + assert!(v.contains("(Ho) Tj"), "{v}"); + } + + /// Words with characters beyond WinAnsi draw in the plan's Type3 font, one run, with the + /// same exact box fit as the Helvetica words (the matrix x-axis spans across / advance). + #[test] + fn planned_words_draw_in_their_font_with_the_same_exact_fit() { + if pdfcraft_fonts::CRAFT_FONTS.is_empty() { + eprintln!("skipping: built without craft-fonts (set CRAFT_FONTS_DIR to run it)"); + return; + } + let w = + PlacedWord { text: "日本".into(), origin: [10.0, 20.0], across: [60.0, 0.0], up: [0.0, 9.3], confidence: None, source: "ocrs".into() }; + let mut plan = TextPlan::for_words(["日本"]); + // The names the registrar (pdfcraft_edit::add_type3_fonts) fills in — this crate + // cannot reach it, so the test sets them where it reads them. + for (i, font) in plan.fonts.iter_mut().enumerate() { + font.name = format!("PCOcr{i}"); + } + let codes: Vec = plan.encoding(0).iter().map(|&(_, c)| c).collect(); + assert_eq!(codes.len(), 2, "{codes:?}"); + let total = plan.advance_of(plan.encoding(0)); + let s = String::from_utf8(text_layer(std::slice::from_ref(&w), &plan)).unwrap(); + assert!(s.contains("/PCOcr0 1 Tf"), "{s}"); + assert!(s.contains(&format!("{} 0 0 10 10 22.1 Tm ", num(60.0 / total))), "{s}"); + assert!(s.contains(&format!("Tm {} Tj\n", String::from_utf8_lossy(&pdfcraft_fonts::literal(&codes)))), "{s}"); + // The visible layer: size from the box height, Tz from the Type3's own advance. + let v = String::from_utf8(visible_text_layer(std::slice::from_ref(&w), &plan)).unwrap(); + let size = 9.3 / BOX_EM; + assert!(v.contains(&format!("/PCOcr0 {} Tf ", num(size))), "{v}"); + assert!(v.contains(&format!("{} Tz", num(60.0 / (total * size) * 100.0))), "{v}"); + } + + /// Without planned characters (an empty plan: built without craft-fonts, or before the + /// fonts are registered) the WinAnsi path stands, '?' included. + #[test] + fn unplanned_words_fall_back_to_win_ansi_question_marks() { + let w = + PlacedWord { text: "日本".into(), origin: [10.0, 20.0], across: [60.0, 0.0], up: [0.0, 9.3], confidence: None, source: "ocrs".into() }; + let s = String::from_utf8(text_layer(&[w], &TextPlan::default())).unwrap(); + assert!(s.contains("(??) Tj"), "{s}"); + assert!(!s.contains("PCOcr"), "{s}"); + } + + /// A character no face has a glyph for becomes the font's '?', still in the run, counted + /// by the plan. + #[test] + fn substitutions_are_written_as_the_font_s_question_mark() { + if pdfcraft_fonts::CRAFT_FONTS.is_empty() { + eprintln!("skipping: built without craft-fonts (set CRAFT_FONTS_DIR to run it)"); + return; + } + let w = PlacedWord { + text: "日\u{1f4a9}語".into(), + origin: [0.0, 0.0], + across: [60.0, 0.0], + up: [0.0, 9.3], + confidence: None, + source: "ocrs".into(), + }; + let mut plan = TextPlan::for_words(["日\u{1f4a9}語"]); + for (i, font) in plan.fonts.iter_mut().enumerate() { + font.name = format!("PCOcr{i}"); + } + assert_eq!(plan.missing(), 1, "the emoji has no glyph anywhere"); + assert_eq!(plan.encoding(0).len(), 3, "it still gets an encoding: the font's ?"); + let codes: Vec = plan.encoding(0).iter().map(|&(_, c)| c).collect(); + let s = String::from_utf8(text_layer(&[w], &plan)).unwrap(); + assert!(s.contains(&format!("Tm {} Tj\n", String::from_utf8_lossy(&pdfcraft_fonts::literal(&codes)))), "{s}"); + } + #[test] fn rotated_pages_keep_the_reading_direction() { // A page turned 90°: image x runs up the page. let to_user = |x: f32, y: f32| [y as f64, x as f64]; - let w = PlacedWord::place(&Word { text: "Up".into(), rect: [0.0, 0.0, 50.0, 10.0] }, to_user); + let w = PlacedWord::place(&Word::new("Up", [0.0, 0.0, 50.0, 10.0], "ocrs"), to_user); assert_eq!(w.across, [0.0, 50.0]); assert_eq!(w.up, [-10.0, 0.0]); } - /// The current ONNX pair wins when an installation also retains both legacy models. - #[test] - fn model_discovery_prefers_the_complete_onnx_pair() { - let dir = tempfile::tempdir().unwrap(); - for file in ["text-detection.onnx", "text-recognition.onnx"] { - std::fs::write(dir.path().join(file), b"model").unwrap(); - } - let expected = Some(Models { detection: dir.path().join("text-detection.onnx"), recognition: dir.path().join("text-recognition.onnx") }); - assert_eq!(Models::in_dir(dir.path()), expected, "a complete ONNX pair is sufficient"); - for file in ["text-detection.rten", "text-recognition.rten"] { - std::fs::write(dir.path().join(file), b"model").unwrap(); - } - assert_eq!(Models::in_dir(dir.path()), expected, "the complete ONNX pair takes precedence over legacy models"); - } - - /// An interrupted ONNX update still permits an intact legacy pair in the same directory. - #[test] - fn model_discovery_falls_back_to_legacy_when_the_onnx_pair_is_incomplete() { - for current in ["text-detection.onnx", "text-recognition.onnx"] { - let dir = tempfile::tempdir().unwrap(); - for file in [current, "text-detection.rten", "text-recognition.rten"] { - std::fs::write(dir.path().join(file), b"model").unwrap(); - } - assert_eq!( - Models::in_dir(dir.path()), - Some(Models { detection: dir.path().join("text-detection.rten"), recognition: dir.path().join("text-recognition.rten") }), - "{current} alone must not hide a complete legacy pair" - ); - } - } - - /// Detection and recognition must come from one complete format pair, never a mixed update. - #[test] - fn model_discovery_refuses_mixed_format_pairs() { - for pair in [["text-detection.onnx", "text-recognition.rten"], ["text-detection.rten", "text-recognition.onnx"]] { - let dir = tempfile::tempdir().unwrap(); - for file in pair { - std::fs::write(dir.path().join(file), b"model").unwrap(); - } - assert_eq!(Models::in_dir(dir.path()), None, "{pair:?} is not a complete model pair"); - } - } - - /// Existing source builds and PDFCRAFT_MODELS overrides retain their complete RTEN pair. - #[test] - fn model_discovery_keeps_legacy_only_installations() { - let dir = tempfile::tempdir().unwrap(); - assert_eq!(Models::in_dir(dir.path()), None); - std::fs::write(dir.path().join("text-detection.rten"), b"model").unwrap(); - assert_eq!(Models::in_dir(dir.path()), None, "one legacy model is incomplete"); - std::fs::write(dir.path().join("text-recognition.rten"), b"model").unwrap(); - assert_eq!( - Models::in_dir(dir.path()), - Some(Models { detection: dir.path().join("text-detection.rten"), recognition: dir.path().join("text-recognition.rten") }) - ); - } - /// #103: every release package's layout puts the models where the installed app looks. /// Mirrors packaging/{windows/package.ps1,macos/package.sh,linux/package.sh,freebsd/package.sh}. #[test] @@ -443,6 +813,47 @@ mod tests { assert!(word.rect[0] >= 10.0 && word.rect[0] < 60.0, "{word:?}"); } + /// The recognised words name their engine. + #[test] + fn ocrs_recognizer_names_its_words() { + let Ok(rec) = OcrsRecognizer::find() else { + eprintln!("skipped: OCR models not installed"); + return; + }; + assert_eq!(rec.id(), "ocrs"); + assert!(rec.languages().iter().any(|(c, n)| c == "en" && n == "English"), "{:?}", rec.languages()); + let (w, h, px) = test_image::hello(); + let image = OcrImage::new(w, h, px).unwrap(); + let lines = rec.recognize(&image, &RecognizeOptions::default()).unwrap(); + let words: Vec<&Word> = lines.iter().flat_map(|l| &l.words).collect(); + assert!(!words.is_empty()); + assert!(words.iter().all(|w| w.source == "ocrs"), "{words:?}"); + } + + /// Hostile images are rejected with errors, never panics: zero sides, sides over the cap, + /// totals over the area cap (even with legal sides), buffers that do not fit, counts that + /// overflow. + #[test] + fn hostile_image_dimensions_are_rejected_not_panicked_at() { + assert!(matches!(OcrImage::new(0, 10, vec![]), Err(OcrError::EmptyImage))); + assert!(matches!(OcrImage::new(10, 0, vec![]), Err(OcrError::EmptyImage))); + assert!(matches!(OcrImage::new(u32::MAX, u32::MAX, vec![]), Err(OcrError::ImageTooLarge))); + assert!(matches!(OcrImage::new(MAX_SIDE + 1, 1, vec![0; 4]), Err(OcrError::ImageTooLarge))); + // 10 000 × 7 000 = 70 MP: legal on both sides, over the 64 MP area cap (no allocation + // is needed to reject it — the buffer is never inspected). + assert!(matches!(OcrImage::new(MAX_SIDE, 7000, vec![]), Err(OcrError::ImageTooLarge))); + // 8 000 × 8 000 = exactly the area cap: accepted shape (the short buffer proves the + // cap check passed), while one more row is refused. + assert!(matches!(OcrImage::new(8000, 8000, vec![0; 16]), Err(OcrError::ImageSize { .. }))); + assert!(matches!(OcrImage::new(8000, 8001, vec![0; 16]), Err(OcrError::ImageTooLarge))); + assert!(matches!(OcrImage::new(2, 2, vec![0; 15]), Err(OcrError::ImageSize { .. }))); + let ok = OcrImage::new(2, 2, vec![255; 16]).unwrap(); + assert_eq!((ok.width(), ok.height()), (2, 2)); + assert_eq!(ok.rgba().len(), 16); + assert_eq!(ok.gray(5, 5), Some(255), "clamped reads stay inside"); + assert_eq!(ok.gray(-1, -1), Some(255)); + } + mod test_image { /// "HELLO WORLD" drawn with blocky 5×7 letters, 4 pixels per dot, black on white. pub fn hello() -> (u32, u32, Vec) { diff --git a/crates/ocr/src/merge.rs b/crates/ocr/src/merge.rs new file mode 100644 index 0000000000..58fc17a4e5 --- /dev/null +++ b/crates/ocr/src/merge.rs @@ -0,0 +1,583 @@ +//! Combining engines. "ROVER" merges two engines' readings of one page: every primary word +//! adopts the best-overlapping unused secondary word when their boxes agree enough (IoU, below); +//! the higher confidence wins the text, box and confidence (a tie keeps the primary; against a +//! primary of unknown confidence the secondary must reach [`UNKNOWN_CONFIDENCE_BAR`]) and the +//! surviving word is named `"ROVER"`; secondary words nothing claimed are appended, keeping +//! their own engine's name. + +use crate::{Line, OcrError, OcrImage, RecognizeOptions, Recognizer, Word}; + +/// How the chosen engines combine. The default is [`MergeStrategy::PrimaryOnly`] with the +/// default engine (ocrs), which is today's behavior. +#[derive(Clone, Copy, Debug, Default, PartialEq, Eq, Hash)] +pub enum MergeStrategy { + /// Only the primary engine runs. + #[default] + PrimaryOnly, + /// The secondary engine joins only when the primary's reading looks weak: its mean word + /// confidence is under 70, or its confidence is unknown (words were read but none reports + /// a confidence — the ocrs engine). An unknown reading is never counted as 100; an empty + /// primary keeps the fast path (no words, nothing to double-check). + ConfidenceWeighted, + /// Both engines always run and their readings merge (with no secondary engine this + /// degrades to primary only). + RoverVote, +} + +/// The mean of the confidences the words report, 0–100; `None` when the reading's confidence +/// is UNKNOWN — words were read but none reports a confidence (the ocrs engine's words). +/// Unknown is never counted as 100: there is no low-confidence signal, but a reading whose +/// confidence nobody knows is not therefore confident, so a confidence-weighted strategy +/// answers it by asking the secondary engine. An empty reading is `Some(100.0)`: no words +/// were read, so there is nothing to double-check and the fast path stands. +pub fn mean_confidence(lines: &[Line]) -> Option { + let mut total = 0usize; + let mut reported = (0.0f64, 0usize); + for word in lines.iter().flat_map(|l| &l.words) { + total += 1; + if let Some(c) = word.confidence { + reported.0 += c as f64; + reported.1 += 1; + } + } + match (total, reported.1) { + (0, _) => Some(100.0), + (_, 0) => None, + (_, n) => Some((reported.0 / n as f64) as f32), + } +} + +/// Intersection over union of two `[left, top, right, bottom]` boxes. An empty intersection is +/// 0; degenerate or NaN boxes give 0 rather than panicking or poisoning the comparison. +pub fn iou(a: &[f32; 4], b: &[f32; 4]) -> f64 { + let l = a[0].max(b[0]); + let t = a[1].max(b[1]); + let r = a[2].min(b[2]); + let btm = a[3].min(b[3]); + if !(r > l && btm > t) { + return 0.0; // disjoint, touching, inverted, or NaN (all comparisons false) + } + let inter = (r - l) as f64 * (btm - t) as f64; + let area = |x: &[f32; 4]| ((x[2] - x[0]).max(0.0) as f64) * ((x[3] - x[1]).max(0.0) as f64); + let union = area(a) + area(b) - inter; + if union <= 0.0 { + return 0.0; + } + inter / union +} + +/// The most primary×secondary word comparisons one ROVER merge will attempt. Known limit: a +/// merge over this cap is skipped — the primary reading stands alone, and the skip is noted, +/// never silent — so two enormous readings cannot turn the merge into quadratic work without +/// end. A real page is a few hundred words on each side, orders of magnitude under the cap. +pub const ROVER_MAX_COMPARISONS: usize = 100_000_000; + +/// Whether the two readings are too large to merge: their word counts multiply past +/// [`ROVER_MAX_COMPARISONS`]. Shared by [`rover_merge`] (which then keeps the primary) and +/// [`recognize_with_strategy`] (which then notes the skip). +fn over_cap(primary: &[Line], secondary: &[Line]) -> bool { + let primary_words: usize = primary.iter().map(|l| l.words.len()).sum(); + let secondary_words: usize = secondary.iter().map(|l| l.words.len()).sum(); + primary_words.saturating_mul(secondary_words) > ROVER_MAX_COMPARISONS +} + +/// The confidence a secondary word needs to replace a primary word whose confidence is +/// unknown (the ocrs engine reports none). It is the bar under which a reading counts as weak +/// ([`MergeStrategy::ConfidenceWeighted`] asks the second engine below it): a word the second +/// engine itself calls weak does not displace a reading nobody scored. +pub const UNKNOWN_CONFIDENCE_BAR: f32 = 70.0; + +/// True when the secondary word's confidence `a` beats the primary's `b`. Both known: the +/// strictly higher one wins (a tie keeps the primary). A known confidence against an unknown +/// primary wins only at or above [`UNKNOWN_CONFIDENCE_BAR`]: an unknown confidence is neither +/// 0 nor 100, so a 5% reading must not replace it, while a confident one may. NaN never wins, +/// and with no confidence on the secondary side the primary stands. +fn higher(a: Option, b: Option) -> bool { + match (a, b) { + (Some(a), Some(b)) => a.partial_cmp(&b) == Some(std::cmp::Ordering::Greater), + (Some(a), None) => a >= UNKNOWN_CONFIDENCE_BAR, + _ => false, + } +} + +/// Merge the two engines' readings. Primary words keep their order and lines; a word is +/// replaced by its overlapping secondary (IoU >= 0.5) only when that secondary is more +/// confident, and either way the merged word's [`Word::source`] is `"ROVER"`. Secondary words +/// no primary claimed are appended, in their own reading order and lines, under their own +/// engine's name (`secondary_id`). Known limit: over [`ROVER_MAX_COMPARISONS`] comparisons +/// (word counts multiplied) the primary reading is kept as is, unmerged; the strategy runner +/// names that skip in a note. +pub fn rover_merge(primary: Vec, secondary: Vec, secondary_id: &str) -> Vec { + if over_cap(&primary, &secondary) { + return primary; // over the comparison cap: the primary stands alone + } + let mut primary = primary; + // The secondary's words, marked `None` once a primary word claimed them, grouped by their + // own lines so the unclaimed ones can be appended in order. + let mut unused: Vec>> = secondary.into_iter().map(|l| l.words.into_iter().map(Some).collect()).collect(); + for line in &mut primary { + for word in &mut line.words { + // The best-overlapping unused secondary word, if any beats IoU 0.5. + let mut best: Option<(f64, usize, usize)> = None; + for (li, words) in unused.iter().enumerate() { + for (wi, slot) in words.iter().enumerate() { + let Some(w) = slot else { continue }; + let score = iou(&word.rect, &w.rect); + if score >= 0.5 && best.is_none_or(|(b, _, _)| score > b) { + best = Some((score, li, wi)); + } + } + } + let Some((_, li, wi)) = best else { continue }; + let claimed = unused.get_mut(li).and_then(|l| l.get_mut(wi)).and_then(|slot| slot.take()); + let Some(claim) = claimed else { continue }; + if higher(claim.confidence, word.confidence) { + *word = Word { source: "ROVER".into(), ..claim }; + } else { + // Tie (or the primary wins): keep the primary's text, box and confidence, but + // record that this reading is the engines' agreed one. + word.source = "ROVER".into(); + } + } + } + // Append the secondary words nothing claimed, line by line, under their engine's name + // (`secondary_id` backfills words that name no source of their own). + let mut merged = primary; + for line in unused { + let words: Vec = line + .into_iter() + .flatten() + .map(|mut w| { + if w.source.is_empty() { + w.source = secondary_id.into(); + } + w + }) + .collect(); + if !words.is_empty() { + merged.push(Line { words }); + } + } + merged +} + +/// What a strategy run produced: the merged (or primary-only) reading, whether the secondary +/// engine's reading is part of it, and the notes recorded along the way. +#[derive(Clone, Debug, PartialEq, Default)] +pub struct StrategyOutcome { + pub lines: Vec, + /// Whether the secondary engine's reading is part of `lines`: true only when the merge + /// actually happened. A secondary that failed, or whose reading was discarded over the + /// comparison cap, is `false`, with a note in [`StrategyOutcome::notes`] saying why. + pub secondary_ran: bool, + /// Why the page's reading is partial, if it is. The secondary engine is an enhancement, + /// never a gate: when it is configured but fails, the primary reading is kept and the + /// failure is noted here — a page is never skipped because a partner could not read it. + /// Only a primary failure is an [`OcrError`]. + pub notes: Vec, +} + +/// The note recorded when an ensemble partner fails: the page keeps the primary reading and +/// the cause is named for the report. +fn degraded_note(id: &str, error: &OcrError) -> String { + format!("the {id} engine failed ({error}) — the page keeps the primary reading alone") +} + +/// The note recorded when the two readings are too large to merge: the page keeps the primary +/// reading alone and the discarded partner reading is named, never silent. +fn cap_note(id: &str, primary_words: usize, secondary_words: usize) -> String { + format!( + "the readings were too large to merge ({primary_words} × {secondary_words} comparisons, over the cap of {ROVER_MAX_COMPARISONS}) — the page keeps the primary reading alone and the {id} reading is unused" + ) +} + +/// Run the engines per `strategy` and merge. See [`StrategyOutcome`] for what comes back. +pub fn recognize_with_strategy( + primary: &dyn Recognizer, + secondary: Option<&dyn Recognizer>, + strategy: MergeStrategy, + image: &OcrImage, + options: &RecognizeOptions, +) -> Result { + // The shared shape of the ensemble strategies: the primary has read, the secondary joins + // or fails (a failure degrades to the primary, with a note). + let join = |p: Vec, sec: &dyn Recognizer| match sec.recognize(image, options) { + // Over the comparison cap the merge is skipped: the secondary's reading was read but + // is not part of the result, so the outcome says so — primary only, honestly not + // merged, with the skip named. + Ok(s) if over_cap(&p, &s) => { + let (primary_words, secondary_words) = (p.iter().map(|l| l.words.len()).sum::(), s.iter().map(|l| l.words.len()).sum::()); + StrategyOutcome { lines: p, secondary_ran: false, notes: vec![cap_note(sec.id(), primary_words, secondary_words)] } + } + Ok(s) => StrategyOutcome { lines: rover_merge(p, s, sec.id()), secondary_ran: true, notes: Vec::new() }, + Err(e) => StrategyOutcome { lines: p, secondary_ran: false, notes: vec![degraded_note(sec.id(), &e)] }, + }; + match strategy { + MergeStrategy::PrimaryOnly => Ok(StrategyOutcome { lines: primary.recognize(image, options)?, secondary_ran: false, notes: Vec::new() }), + MergeStrategy::RoverVote => match secondary { + None => Ok(StrategyOutcome { lines: primary.recognize(image, options)?, secondary_ran: false, notes: Vec::new() }), + Some(sec) => { + let p = primary.recognize(image, options)?; + Ok(join(p, sec)) + } + }, + MergeStrategy::ConfidenceWeighted => { + let p = primary.recognize(image, options)?; + match mean_confidence(&p) { + // Confident enough — or an empty reading: the primary stands alone. + Some(mean) if mean >= UNKNOWN_CONFIDENCE_BAR => Ok(StrategyOutcome { lines: p, secondary_ran: false, notes: Vec::new() }), + // A weak mean, or an unknown confidence (words, none reported): the secondary + // runs. + _ => match secondary { + None => Ok(StrategyOutcome { lines: p, secondary_ran: false, notes: Vec::new() }), + Some(sec) => Ok(join(p, sec)), + }, + } + } + } +} + +#[cfg(test)] +mod tests { + use super::*; + use std::sync::atomic::{AtomicUsize, Ordering}; + + /// A fake engine: returns canned lines, counts recognize calls, never available. + struct Fake { + id: &'static str, + lines: Vec, + calls: AtomicUsize, + fail: bool, + } + + impl Fake { + fn new(id: &'static str, words: &[(&str, [f32; 4], f32)]) -> Fake { + Fake { + id, + lines: vec![Line { words: words.iter().map(|(t, r, c)| Word::new(*t, *r, id).with_confidence(*c)).collect() }], + calls: AtomicUsize::new(0), + fail: false, + } + } + + fn empty(id: &'static str) -> Fake { + Fake { id, lines: Vec::new(), calls: AtomicUsize::new(0), fail: false } + } + + /// An engine like ocrs: words, but no confidence reported on any of them. + fn unknown(id: &'static str, n: usize) -> Fake { + Fake { + id, + lines: vec![Line { + words: (0..n).map(|i| Word::new(format!("w{i}"), [i as f32 * 10.0, 0.0, i as f32 * 10.0 + 8.0, 10.0], id)).collect(), + }], + calls: AtomicUsize::new(0), + fail: false, + } + } + } + + impl Recognizer for Fake { + fn id(&self) -> &'static str { + self.id + } + fn available(&self) -> bool { + !self.fail + } + fn languages(&self) -> Vec<(String, String)> { + vec![] + } + fn recognize(&self, _image: &OcrImage, _options: &RecognizeOptions) -> Result, OcrError> { + self.calls.fetch_add(1, Ordering::SeqCst); + if self.fail { + return Err(OcrError::Recognize("the fake engine failed".into())); + } + Ok(self.lines.clone()) + } + } + + fn image() -> OcrImage { + OcrImage::new(4, 4, vec![255; 64]).unwrap() + } + + fn words(lines: &[Line]) -> Vec<&Word> { + lines.iter().flat_map(|l| &l.words).collect() + } + + #[test] + fn iou_of_disjoint_touching_and_nan_boxes_is_zero() { + let a = [0.0, 0.0, 10.0, 10.0]; + assert_eq!(iou(&a, &[20.0, 0.0, 30.0, 10.0]), 0.0, "disjoint"); + assert_eq!(iou(&a, &[10.0, 0.0, 20.0, 10.0]), 0.0, "touching edges are an empty intersection"); + assert_eq!(iou(&a, &[f32::NAN, 0.0, 10.0, 10.0]), 0.0, "NaN boxes score 0, not a panic"); + assert_eq!(iou(&a, &a), 1.0, "identical boxes"); + assert!((iou(&a, &[5.0, 0.0, 15.0, 10.0]) - 1.0 / 3.0).abs() < 1e-9, "half overlap is IoU 1/3"); + } + + /// With nothing on the primary side, every secondary word is appended under the secondary + /// engine's own name. + #[test] + fn rover_appends_everything_when_the_primary_is_empty() { + let primary = Vec::new(); + let secondary = vec![Line { words: vec![Word::new("sole", [0.0, 0.0, 9.0, 9.0], "tesseract").with_confidence(80.0)] }]; + let merged = rover_merge(primary, secondary, "tesseract"); + assert_eq!(words(&merged).len(), 1); + assert_eq!(merged[0].words[0].text, "sole"); + assert_eq!(merged[0].words[0].source, "tesseract", "unclaimed words keep their own engine"); + } + + /// Non-overlapping secondary words never displace primary words; they are appended. + #[test] + fn rover_appends_non_overlapping_secondary_words() { + let primary = vec![Line { words: vec![Word::new("kept", [0.0, 0.0, 10.0, 10.0], "ocrs").with_confidence(50.0)] }]; + let secondary = vec![Line { words: vec![Word::new("extra", [100.0, 100.0, 120.0, 110.0], "tesseract").with_confidence(99.0)] }]; + let merged = rover_merge(primary, secondary, "tesseract"); + let w = words(&merged); + assert_eq!(w.len(), 2); + assert_eq!((w[0].text.as_str(), w[0].source.as_str()), ("kept", "ocrs"), "the more confident but disjoint word does not win"); + assert_eq!((w[1].text.as_str(), w[1].source.as_str()), ("extra", "tesseract")); + } + + /// The higher-confidence word of an overlapping pair wins text, box and confidence, and is + /// named ROVER. + #[test] + fn rover_higher_confidence_wins_and_is_named_rover() { + let primary = vec![Line { words: vec![Word::new("helo", [0.0, 0.0, 10.0, 10.0], "ocrs").with_confidence(60.0)] }]; + let secondary = vec![Line { words: vec![Word::new("hello", [1.0, 0.0, 11.0, 10.0], "tesseract").with_confidence(95.0)] }]; + let merged = rover_merge(primary, secondary, "tesseract"); + let w = words(&merged); + assert_eq!(w.len(), 1, "the secondary word was claimed, not appended"); + assert_eq!(w[0].text, "hello"); + assert_eq!(w[0].rect, [1.0, 0.0, 11.0, 10.0]); + assert_eq!(w[0].confidence, Some(95.0)); + assert_eq!(w[0].source, "ROVER"); + } + + /// A tie keeps the primary word (still named ROVER, as an agreed reading). + #[test] + fn rover_tie_keeps_the_primary() { + let primary = vec![Line { words: vec![Word::new("primary", [0.0, 0.0, 10.0, 10.0], "ocrs").with_confidence(80.0)] }]; + let secondary = vec![Line { words: vec![Word::new("secondary", [0.0, 0.0, 10.0, 10.0], "tesseract").with_confidence(80.0)] }]; + let merged = rover_merge(primary, secondary, "tesseract"); + let w = words(&merged); + assert_eq!(w.len(), 1); + assert_eq!(w[0].text, "primary"); + assert_eq!(w[0].source, "ROVER"); + } + + /// Boxes under the 0.5 IoU bar never merge, even when the secondary is more confident. + #[test] + fn rover_below_half_iou_does_not_merge() { + let primary = vec![Line { words: vec![Word::new("kept", [0.0, 0.0, 10.0, 10.0], "ocrs").with_confidence(50.0)] }]; + // IoU 10·10 / (100 + 169 - 25) = 100/244 ≈ 0.41 < 0.5. + let secondary = vec![Line { words: vec![Word::new("other", [5.0, 0.0, 23.0, 20.0], "tesseract").with_confidence(99.0)] }]; + let merged = rover_merge(primary, secondary, "tesseract"); + let w = words(&merged); + assert_eq!(w.len(), 2); + assert_eq!(w[0].text, "kept"); + assert_eq!(w[0].source, "ocrs"); + } + + /// Merged words keep their line; the merge never panics on NaN boxes. + #[test] + fn rover_survives_nan_boxes() { + let primary = vec![Line { words: vec![Word::new("nan", [f32::NAN, 0.0, 10.0, 10.0], "ocrs").with_confidence(50.0)] }]; + let secondary = vec![Line { words: vec![Word::new("real", [0.0, 0.0, 10.0, 10.0], "tesseract").with_confidence(90.0)] }]; + let merged = rover_merge(primary, secondary, "tesseract"); + let w = words(&merged); + assert_eq!(w.len(), 2, "a NaN primary box merges with nothing"); + assert_eq!(w[0].text, "nan"); + assert_eq!(w[1].text, "real"); + } + + /// Only reported confidences count. An empty reading is Some(100) (nothing to + /// double-check); words without a single confidence are UNKNOWN (`None`), never 100. + #[test] + fn mean_confidence_distinguishes_empty_from_unknown() { + assert_eq!(mean_confidence(&[]), Some(100.0)); + assert_eq!(mean_confidence(&[Line { words: vec![Word::new("a", [0.0; 4], "ocrs")] }]), None, "words, no confidences: unknown"); + let lines = vec![Line { + words: vec![ + Word::new("a", [0.0; 4], "e").with_confidence(40.0), + Word::new("b", [0.0; 4], "e").with_confidence(80.0), + Word::new("c", [0.0; 4], "e"), + ], + }]; + assert_eq!(mean_confidence(&lines), Some(60.0), "reported confidences are averaged"); + } + + /// ROVER's preference needs a meaningful comparison: a known confidence beats an unknown + /// one (the ocrs primary yields to a confident tesseract reading), and None vs None keeps + /// the primary. + #[test] + fn rover_known_confidence_beats_unknown_and_none_vs_none_keeps_the_primary() { + let primary = vec![Line { words: vec![Word::new("helo", [0.0, 0.0, 10.0, 10.0], "ocrs")] }]; + let secondary = vec![Line { words: vec![Word::new("hello", [0.0, 0.0, 10.0, 10.0], "tesseract").with_confidence(95.0)] }]; + let merged = rover_merge(primary, secondary, "tesseract"); + assert_eq!( + (merged[0].words[0].text.as_str(), merged[0].words[0].confidence, merged[0].words[0].source.as_str()), + ("hello", Some(95.0), "ROVER") + ); + let primary = vec![Line { words: vec![Word::new("primary", [0.0, 0.0, 10.0, 10.0], "ocrs")] }]; + let secondary = vec![Line { words: vec![Word::new("secondary", [0.0, 0.0, 10.0, 10.0], "tesseract")] }]; + let merged = rover_merge(primary, secondary, "tesseract"); + assert_eq!( + (merged[0].words[0].text.as_str(), merged[0].words[0].source.as_str()), + ("primary", "ROVER"), + "unknown vs unknown is not a comparison: the primary stands" + ); + } + + /// Over the comparison cap the merge is skipped and the primary reading stands alone (a + /// documented known limit, bounding the merge's quadratic core). + #[test] + fn rover_over_the_comparison_cap_keeps_the_primary() { + // 11 000 × 11 000 = 121 M comparisons > the 100 M cap. + let words = |id: &str| (0..11_000).map(|i| Word::new(format!("w{i}"), [i as f32, 0.0, i as f32 + 5.0, 10.0], id)).collect::>(); + let primary = vec![Line { words: words("ocrs") }]; + let secondary = vec![Line { words: words("tesseract") }]; + let merged = rover_merge(primary, secondary, "tesseract"); + assert_eq!(merged.len(), 1); + assert!(merged[0].words.iter().all(|w| w.source == "ocrs"), "every primary word stands, unmerged"); + assert_eq!(merged[0].words.len(), 11_000); + // Under the cap, merging still happens (a small overlap is replaced and named ROVER). + let small_primary = vec![Line { words: vec![Word::new("kept", [0.0, 0.0, 10.0, 10.0], "ocrs").with_confidence(50.0)] }]; + let small_secondary = vec![Line { words: vec![Word::new("wins", [0.0, 0.0, 10.0, 10.0], "tesseract").with_confidence(95.0)] }]; + let merged = rover_merge(small_primary, small_secondary, "tesseract"); + assert_eq!((merged[0].words[0].text.as_str(), merged[0].words[0].source.as_str()), ("wins", "ROVER")); + } + + /// Over the comparison cap the merge is skipped honestly at the strategy level: the page + /// keeps the primary reading alone, the skip is named in a note, and `secondary_ran` does + /// not claim the discarded reading is part of the result (though the secondary really did + /// run — its work was spent, its reading unused). + #[test] + fn rover_cap_hit_reports_primary_only_with_a_note_and_an_honest_flag() { + let mut primary = Fake::empty("ocrs"); + primary.lines = + vec![Line { words: (0..11_000).map(|i| Word::new(format!("w{i}"), [i as f32, 0.0, i as f32 + 5.0, 10.0], "ocrs")).collect() }]; + let mut secondary = Fake::empty("tesseract"); + secondary.lines = + vec![Line { words: (0..11_000).map(|i| Word::new(format!("w{i}"), [i as f32, 0.0, i as f32 + 5.0, 10.0], "tesseract")).collect() }]; + let outcome = recognize_with_strategy(&primary, Some(&secondary), MergeStrategy::RoverVote, &image(), &RecognizeOptions::default()).unwrap(); + assert_eq!(secondary.calls.load(Ordering::SeqCst), 1, "the secondary really ran"); + assert_eq!(outcome.lines.len(), 1); + assert_eq!(outcome.lines[0].words.len(), 11_000, "primary only"); + assert!(outcome.lines[0].words.iter().all(|w| w.source == "ocrs"), "every primary word stands, unmerged"); + assert!(!outcome.secondary_ran, "the secondary's reading is not part of the result"); + assert_eq!(outcome.notes.len(), 1, "{:?}", outcome.notes); + assert!(outcome.notes[0].contains("merge"), "{:?}", outcome.notes[0]); + assert!(outcome.notes[0].contains("tesseract"), "{:?}", outcome.notes[0]); + // Under the cap the same strategy merges and reports the secondary honestly. + let primary = Fake::new("ocrs", &[("kept", [0.0, 0.0, 10.0, 10.0], 50.0)]); + let secondary = Fake::new("tesseract", &[("wins", [0.0, 0.0, 10.0, 10.0], 95.0)]); + let outcome = recognize_with_strategy(&primary, Some(&secondary), MergeStrategy::RoverVote, &image(), &RecognizeOptions::default()).unwrap(); + assert!(outcome.secondary_ran && outcome.notes.is_empty()); + assert_eq!((outcome.lines[0].words[0].text.as_str(), outcome.lines[0].words[0].source.as_str()), ("wins", "ROVER")); + } + + #[test] + fn primary_only_never_runs_the_secondary() { + let primary = Fake::new("ocrs", &[("w", [0.0; 4], 90.0)]); + let secondary = Fake::empty("tesseract"); + let outcome = + recognize_with_strategy(&primary, Some(&secondary), MergeStrategy::PrimaryOnly, &image(), &RecognizeOptions::default()).unwrap(); + assert_eq!(primary.calls.load(Ordering::SeqCst), 1); + assert_eq!(secondary.calls.load(Ordering::SeqCst), 0); + assert!(!outcome.secondary_ran); + assert!(outcome.notes.is_empty()); + assert_eq!(words(&outcome.lines).len(), 1); + } + + #[test] + fn confidence_weighted_runs_the_secondary_only_on_a_weak_primary() { + let weak = Fake::new("ocrs", &[("w", [0.0; 4], 40.0)]); + let strong = Fake::new("ocrs", &[("w", [0.0; 4], 95.0)]); + let secondary = Fake::empty("tesseract"); + let outcome = + recognize_with_strategy(&weak, Some(&secondary), MergeStrategy::ConfidenceWeighted, &image(), &RecognizeOptions::default()).unwrap(); + assert!(outcome.secondary_ran, "mean 40 < 70 → the secondary joins"); + let outcome = + recognize_with_strategy(&strong, Some(&secondary), MergeStrategy::ConfidenceWeighted, &image(), &RecognizeOptions::default()).unwrap(); + assert!(!outcome.secondary_ran, "mean 95 >= 70 → primary alone"); + // An empty primary counts as 100: no signal, no secondary run. + let empty = Fake::empty("ocrs"); + let outcome = + recognize_with_strategy(&empty, Some(&secondary), MergeStrategy::ConfidenceWeighted, &image(), &RecognizeOptions::default()).unwrap(); + assert!(!outcome.secondary_ran); + assert_eq!(secondary.calls.load(Ordering::SeqCst), 1, "only the weak page ran the secondary"); + } + + /// The ocrs engine reports no confidences: unknown is not 100, so the secondary runs — + /// and its known reading wins the overlapping word (the ensemble's whole point). + #[test] + fn confidence_weighted_runs_the_secondary_on_an_unknown_primary() { + let ocrs_like = Fake::unknown("ocrs", 1); + let secondary = Fake::new("tesseract", &[("hello", [0.0, 0.0, 10.0, 10.0], 95.0)]); + let outcome = + recognize_with_strategy(&ocrs_like, Some(&secondary), MergeStrategy::ConfidenceWeighted, &image(), &RecognizeOptions::default()).unwrap(); + assert!(outcome.secondary_ran, "an unknown confidence asks the secondary"); + assert_eq!((outcome.lines[0].words[0].text.as_str(), outcome.lines[0].words[0].confidence), ("hello", Some(95.0))); + assert_eq!(outcome.lines[0].words[0].source, "ROVER"); + assert_eq!(secondary.calls.load(Ordering::SeqCst), 1); + } + + /// Against a primary word of unknown confidence, a secondary word wins only when it is + /// confident (at or above the weak-reading bar): a 5% reading does not replace a word nobody + /// scored, a 70% one does. The merged word is named ROVER either way. + #[test] + fn an_unknown_primary_word_yields_only_to_a_confident_secondary() { + let merge = |conf: f32| { + let secondary = vec![Line { words: vec![Word::new("other", [0.0, 0.0, 10.0, 10.0], "tesseract").with_confidence(conf)] }]; + let primary = Fake::unknown("ocrs", 1).lines; + rover_merge(primary, secondary, "tesseract").remove(0).words.remove(0) + }; + let weak = merge(5.0); + assert_eq!((weak.text.as_str(), weak.confidence, weak.source.as_str()), ("w0", None, "ROVER"), "a 5% word keeps the primary"); + let below = merge(UNKNOWN_CONFIDENCE_BAR - 0.5); + assert_eq!(below.text, "w0"); + let at_bar = merge(UNKNOWN_CONFIDENCE_BAR); + assert_eq!((at_bar.text.as_str(), at_bar.confidence), ("other", Some(UNKNOWN_CONFIDENCE_BAR)), "a confident word wins"); + let nan = merge(f32::NAN); + assert_eq!(nan.text, "w0", "NaN (unknown after clamping) never wins"); + } + + #[test] + fn rover_vote_always_runs_both_and_degrades_without_a_secondary() { + let primary = Fake::new("ocrs", &[("w", [0.0; 4], 95.0)]); + let secondary = Fake::empty("tesseract"); + let outcome = recognize_with_strategy(&primary, Some(&secondary), MergeStrategy::RoverVote, &image(), &RecognizeOptions::default()).unwrap(); + assert!(outcome.secondary_ran); + assert_eq!(secondary.calls.load(Ordering::SeqCst), 1); + // No secondary: primary only. + let outcome = recognize_with_strategy(&primary, None, MergeStrategy::RoverVote, &image(), &RecognizeOptions::default()).unwrap(); + assert!(!outcome.secondary_ran); + } + + /// A configured secondary that fails degrades to the primary reading, with the failure + /// recorded as a note — never a skipped page, never a silently partial reading without a + /// trace. Both ensemble strategies degrade; a primary failure is still an error. + #[test] + fn a_failing_secondary_degrades_to_the_primary_with_a_note() { + let mut secondary = Fake::empty("tesseract"); + secondary.fail = true; + let primary = Fake::new("ocrs", &[("w", [0.0; 4], 95.0)]); + let outcome = recognize_with_strategy(&primary, Some(&secondary), MergeStrategy::RoverVote, &image(), &RecognizeOptions::default()).unwrap(); + assert_eq!(words(&outcome.lines).len(), 1, "the primary reading stands"); + assert_eq!(outcome.lines[0].words[0].text, "w"); + assert!(!outcome.secondary_ran); + assert_eq!(outcome.notes.len(), 1); + assert!(outcome.notes[0].contains("tesseract"), "{:?}", outcome.notes[0]); + // The confidence-weighted strategy degrades the same way on its weak path. + let weak = Fake::new("ocrs", &[("w", [0.0; 4], 40.0)]); + let outcome = + recognize_with_strategy(&weak, Some(&secondary), MergeStrategy::ConfidenceWeighted, &image(), &RecognizeOptions::default()).unwrap(); + assert_eq!(words(&outcome.lines).len(), 1); + assert_eq!(outcome.notes.len(), 1, "{:?}", outcome.notes); + // A failing primary is still an error: there is nothing to degrade to. + let mut failing_primary = Fake::new("ocrs", &[("w", [0.0; 4], 95.0)]); + failing_primary.fail = true; + let r = recognize_with_strategy(&failing_primary, Some(&secondary), MergeStrategy::RoverVote, &image(), &RecognizeOptions::default()); + assert!(r.is_err()); + } +} diff --git a/crates/ocr/src/ocrs.rs b/crates/ocr/src/ocrs.rs new file mode 100644 index 0000000000..7f8feb47c4 --- /dev/null +++ b/crates/ocr/src/ocrs.rs @@ -0,0 +1,251 @@ +//! The pure-Rust engine: the ocrs crate with its pre-trained models, wrapped as a +//! [`Recognizer`] (`OcrsRecognizer`). Reading is offline and local; the models do not report a +//! confidence, so its words have `confidence: None`. + +use std::path::{Path, PathBuf}; + +use super::{Line, MAX_PIXELS, MAX_SIDE, OcrError, OcrImage, RecognizeOptions, Recognizer, Word}; + +/// The validated byte length of a `width` × `height` RGBA raster, with the same caps +/// [`OcrImage::new`] applies: zero sides are empty, sides over [`MAX_SIDE`] or totals over +/// [`MAX_PIXELS`] are too large. `Ocr::recognize` takes raw dimensions (they come from +/// callers, ultimately from documents, so they are untrusted), and the naive +/// `width * height * 4` overflows `u32` for absurd values — the caps are checked first, in +/// `usize`, so nothing overflows and nothing huge is ever touched. +fn checked_len(width: u32, height: u32) -> Result { + if width == 0 || height == 0 { + return Err(OcrError::EmptyImage); + } + if width > MAX_SIDE || height > MAX_SIDE { + return Err(OcrError::ImageTooLarge); + } + let pixels = (width as usize).checked_mul(height as usize).ok_or(OcrError::ImageTooLarge)?; + if pixels > MAX_PIXELS as usize { + return Err(OcrError::ImageTooLarge); + } + pixels.checked_mul(4).ok_or(OcrError::ImageTooLarge) +} + +/// The languages the models read (ISO 639-1); all use the Latin alphabet without accents. +pub const LANGUAGES: &[(&str, &str)] = &[("en", "English")]; + +/// The model files, as named in ATTRIBUTION.toml. +pub const DETECTION_MODEL: &str = "text-detection.onnx"; +pub const RECOGNITION_MODEL: &str = "text-recognition.onnx"; + +pub const ALPHABET: &str = " 0123456789!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~€ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"; + +/// Where the two model files are. +#[derive(Clone, Debug, PartialEq)] +pub struct Models { + pub detection: PathBuf, + pub recognition: PathBuf, +} + +impl Models { + pub fn dirs_beside_exe(exe_dir: &Path) -> Vec { + vec![exe_dir.join("models"), exe_dir.join("../Resources/models"), exe_dir.join("../share/pdfcraft/models")] + } + + /// The models in `dir`, preferring current ONNX files over legacy RTEN files. An incomplete + /// update never mixes formats or hides an intact legacy pair. + pub fn in_dir(dir: &Path) -> Option { + [(DETECTION_MODEL, RECOGNITION_MODEL), ("text-detection.rten", "text-recognition.rten")].into_iter().find_map(|(detection, recognition)| { + let m = Models { detection: dir.join(detection), recognition: dir.join(recognition) }; + (m.detection.is_file() && m.recognition.is_file()).then_some(m) + }) + } + + /// Look for the models: `$PDFCRAFT_MODELS`, then `models/` beside the executable (and + /// `Resources/models` in a macOS bundle), then the source tree's `assets/models/`. + pub fn find() -> Option { + Self::search_dirs().iter().find_map(|d| Self::in_dir(d)) + } + + /// The directories [`Models::find`] looks in, in order. + pub fn search_dirs() -> Vec { + let mut dirs = Vec::new(); + if let Some(d) = std::env::var_os("PDFCRAFT_MODELS") { + dirs.push(PathBuf::from(d)); + } + if let Some(exe) = std::env::current_exe().ok().and_then(|e| e.parent().map(Path::to_path_buf)) { + dirs.extend(Self::dirs_beside_exe(&exe)); + } + dirs.push(Path::new(env!("CARGO_MANIFEST_DIR")).join("../../assets/models")); + dirs + } +} + +/// A loaded recogniser. Loading takes a moment; keep one and reuse it. +pub struct Ocr { + engine: ocrs::OcrEngine, +} + +impl Ocr { + pub fn load(models: &Models) -> Result { + let load = |p: &Path| rten::Model::load_file(p).map_err(|e| OcrError::Load(p.display().to_string(), e.to_string())); + let params = ocrs::OcrEngineParams { + detection_model: Some(load(&models.detection)?), + recognition_model: Some(load(&models.recognition)?), + alphabet: Some(ALPHABET.into()), + ..Default::default() + }; + let engine = ocrs::OcrEngine::new(params).map_err(|e| OcrError::Load("ocr engine".into(), e.to_string()))?; + Ok(Ocr { engine }) + } + + /// Load the models found by [`Models::find`]. + pub fn find() -> Result { + Self::load(&Models::find().ok_or(OcrError::NoModels)?) + } + + /// Recognise the text in an RGBA (or RGB, or grey) image, `width` × `height` pixels. + pub fn recognize(&self, pixels: &[u8], width: u32, height: u32) -> Result, OcrError> { + if pixels.is_empty() { + return Err(OcrError::EmptyImage); + } + let want = checked_len(width, height)?; + // ocrs wants 1 or 3 channels. + let rgb: std::borrow::Cow<[u8]> = if pixels.len() == want { + pixels.as_chunks::<4>().0.iter().flat_map(|p| [p[0], p[1], p[2]]).collect::>().into() + } else { + pixels.into() + }; + let err = |e: &dyn std::fmt::Display| OcrError::Recognize(e.to_string()); + let source = ocrs::ImageSource::from_bytes(&rgb, (width, height)).map_err(|e| err(&e))?; + let input = self.engine.prepare_input(source).map_err(|e| err(&e))?; + let found = self.engine.detect_words(&input).map_err(|e| err(&e))?; + let lines = self.engine.find_text_lines(&input, &found); + let read = self.engine.recognize_text(&input, &lines).map_err(|e| err(&e))?; + use ocrs::TextItem; + Ok(read + .into_iter() + .flatten() + .map(|line| Line { + words: line + .words() + .map(|w| { + let r = w.bounding_rect(); + Word::new(w.to_string(), [r.left() as f32, r.top() as f32, r.right() as f32, r.bottom() as f32], "ocrs") + }) + .filter(|w| !w.text.trim().is_empty()) + .collect(), + }) + .filter(|l| !l.words.is_empty()) + .collect()) + } +} + +/// The ocrs engine as a [`Recognizer`] (the default engine). +pub struct OcrsRecognizer { + engine: Ocr, +} + +impl OcrsRecognizer { + /// Load the models from `models`. + pub fn load(models: &Models) -> Result { + Ok(OcrsRecognizer { engine: Ocr::load(models)? }) + } + + /// Load the models found by [`Models::find`]. + pub fn find() -> Result { + Self::load(&Models::find().ok_or(OcrError::NoModels)?) + } +} + +impl Recognizer for OcrsRecognizer { + fn id(&self) -> &'static str { + "ocrs" + } + + fn available(&self) -> bool { + Models::find().is_some() + } + + fn languages(&self) -> Vec<(String, String)> { + LANGUAGES.iter().map(|(c, n)| ((*c).to_string(), (*n).to_string())).collect() + } + + fn recognize(&self, image: &OcrImage, _options: &RecognizeOptions) -> Result, OcrError> { + self.engine.recognize(image.rgba(), image.width(), image.height()) + } +} + +#[cfg(test)] +mod tests { + use super::*; + + /// Raw dimensions are untrusted (they come from callers, ultimately from documents): zero + /// sides, sides or pixel totals over the caps, and products that would overflow the naive + /// `width * height * 4` are errors, never panics, before anything is touched. + #[test] + fn raw_dimensions_are_capped_and_checked_not_panicked_at() { + assert!(matches!(checked_len(0, 10), Err(OcrError::EmptyImage))); + assert!(matches!(checked_len(10, 0), Err(OcrError::EmptyImage))); + assert!(matches!(checked_len(MAX_SIDE + 1, 1), Err(OcrError::ImageTooLarge))); + // 65536² would wrap u32 in the old `width * height * 4`; the side cap refuses it first. + assert!(matches!(checked_len(u32::MAX, u32::MAX), Err(OcrError::ImageTooLarge))); + // Legal sides whose product is over the 64 MP area cap. + assert!(matches!(checked_len(MAX_SIDE, 7000), Err(OcrError::ImageTooLarge))); + assert_eq!(checked_len(2, 2).unwrap(), 16); + // Exactly the area cap fits: 8000 × 8000 = 64 MP. + assert_eq!(checked_len(8000, 8000).unwrap(), 8000usize * 8000 * 4); + } + + /// The current ONNX pair wins when an installation also retains both legacy models. + #[test] + fn model_discovery_prefers_the_complete_onnx_pair() { + let dir = tempfile::tempdir().unwrap(); + for file in ["text-detection.onnx", "text-recognition.onnx"] { + std::fs::write(dir.path().join(file), b"model").unwrap(); + } + let expected = Some(Models { detection: dir.path().join("text-detection.onnx"), recognition: dir.path().join("text-recognition.onnx") }); + assert_eq!(Models::in_dir(dir.path()), expected, "a complete ONNX pair is sufficient"); + for file in ["text-detection.rten", "text-recognition.rten"] { + std::fs::write(dir.path().join(file), b"model").unwrap(); + } + assert_eq!(Models::in_dir(dir.path()), expected, "the complete ONNX pair takes precedence over legacy models"); + } + + /// An interrupted ONNX update still permits an intact legacy pair in the same directory. + #[test] + fn model_discovery_falls_back_to_legacy_when_the_onnx_pair_is_incomplete() { + for current in ["text-detection.onnx", "text-recognition.onnx"] { + let dir = tempfile::tempdir().unwrap(); + for file in [current, "text-detection.rten", "text-recognition.rten"] { + std::fs::write(dir.path().join(file), b"model").unwrap(); + } + assert_eq!( + Models::in_dir(dir.path()), + Some(Models { detection: dir.path().join("text-detection.rten"), recognition: dir.path().join("text-recognition.rten") }), + "{current} alone must not hide a complete legacy pair" + ); + } + } + + /// Detection and recognition must come from one complete format pair, never a mixed update. + #[test] + fn model_discovery_refuses_mixed_format_pairs() { + for pair in [["text-detection.onnx", "text-recognition.rten"], ["text-detection.rten", "text-recognition.onnx"]] { + let dir = tempfile::tempdir().unwrap(); + for file in pair { + std::fs::write(dir.path().join(file), b"model").unwrap(); + } + assert_eq!(Models::in_dir(dir.path()), None, "{pair:?} is not a complete model pair"); + } + } + + /// Existing source builds and PDFCRAFT_MODELS overrides retain their complete RTEN pair. + #[test] + fn model_discovery_keeps_legacy_only_installations() { + let dir = tempfile::tempdir().unwrap(); + assert_eq!(Models::in_dir(dir.path()), None); + std::fs::write(dir.path().join("text-detection.rten"), b"model").unwrap(); + assert_eq!(Models::in_dir(dir.path()), None, "one legacy model is incomplete"); + std::fs::write(dir.path().join("text-recognition.rten"), b"model").unwrap(); + assert_eq!( + Models::in_dir(dir.path()), + Some(Models { detection: dir.path().join("text-detection.rten"), recognition: dir.path().join("text-recognition.rten") }) + ); + } +} diff --git a/crates/ocr/src/preprocess.rs b/crates/ocr/src/preprocess.rs new file mode 100644 index 0000000000..280ee73fec --- /dev/null +++ b/crates/ocr/src/preprocess.rs @@ -0,0 +1,973 @@ +//! Preprocessing: a pure function between the page raster and recognition, in fixed order — +//! (a) DPI normalization, then the four opt-in steps (b) orientation, (c) deskew, (d) denoise, +//! (e) binarization. Normalization exists for those steps (their windows and thresholds are +//! calibrated for 300 dpi), so it runs only when at least one of them is on: with every step +//! off the raster is handed to the engines exactly as it was rendered. (Enlarging a render +//! on its own changes what the models read: a 72 dpi scan of "due: €250" read as "ue: ?250" +//! after a 4× bilinear enlargement.) Everything is plain Rust on the RGBA raster: no +//! Leptonica, no process, no new dependencies. +//! +//! Every geometric step records its inverse; the result carries one exact +//! [`InverseTransform`] (an affine map, composed in f64), and callers map every word box +//! through it back to the *original* raster coordinates before placing the words. +//! +//! The heuristics are deliberately conservative and panic-free: an image too small to judge, +//! a blank page, or a score without a clear margin is a no-op, never a guess. + +use crate::{MAX_PIXELS, MAX_SIDE, OcrError, OcrImage}; + +/// The four opt-in steps. All off by default: a scan is passed through as rendered. +#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)] +pub struct PreprocessOptions { + /// Detect pages scanned at 90°/180°/270° and turn them upright. + pub auto_rotate: bool, + /// Straighten skewed text lines. + pub deskew: bool, + /// 3×3 median filter (salt-and-pepper noise). + pub denoise: bool, + /// Sauvola binarization (window 8, k 0.34). + pub binarize: bool, +} + +impl PreprocessOptions { + /// Whether any step is on (and so the raster is normalized to 300 dpi first). + pub fn any(&self) -> bool { + self.auto_rotate || self.deskew || self.denoise || self.binarize + } +} + +/// A preprocessed raster: the image to recognize, the exact way back to the input raster, and +/// the resolution the image now stands for. +#[derive(Debug)] +pub struct Preprocessed { + pub image: OcrImage, + /// Maps pixel coordinates in [`Preprocessed::image`] back to the input raster's. + pub inverse_transform: InverseTransform, + /// The image's resolution in dots per inch: 300 after normalization (only when a step is + /// on), or the source's own + /// when it already rendered at 300 or more. + pub dpi: f32, +} + +/// Preprocess `image` (rendered at `source_dpi`) per `options`, in the fixed order. The image +/// is consumed: the common path (every step off, or a no-op) hands the raster back without +/// copying it, and a step that rewrites pixels replaces it. +pub fn preprocess(image: OcrImage, source_dpi: f32, options: &PreprocessOptions) -> Result { + let mut current = image; + let mut inverse = InverseTransform::identity(); + // (a) DPI normalization: for the steps below, a scan under 300 dpi is enlarged by 300/src so + // they see letter-sized strokes. With every step off nothing is resampled. A broken dpi + // (zero, negative, NaN) is left alone. + let mut dpi = if source_dpi.is_finite() && source_dpi >= 1.0 { source_dpi } else { 300.0 }; + if options.any() && dpi < 300.0 { + let scale = f64::from(300.0 / dpi); + current = resize_bilinear(¤t, scale)?; + inverse = compose(&inverse, &InverseTransform::scale(1.0 / scale)); + dpi = 300.0; + } + // (b) Orientation: turn 90/180/270 scans upright (opt-in, skipped for small images). + if options.auto_rotate { + let (w, h) = (current.width, current.height); + let turns = detect_orientation(¤t); + if turns != 0 { + current = rotate_quarter_turns(¤t, turns); + inverse = compose(&inverse, &inverse_quarter_turns(turns, w, h)); + } + } + // (c) Deskew: straighten skewed lines (opt-in; no-op under 64 px, blank, or without a + // clear margin or below 0.1°). + if options.deskew + && let Some(angle) = estimate_skew(¤t) + { + let (straightened, back) = deskew_rotate(¤t, angle); + current = straightened; + inverse = compose(&inverse, &back); + } + // (d) Denoise: 3×3 median with edge clamping (opt-in; dimensions never change). + if options.denoise { + current = median3(¤t); + } + // (e) Binarize: Sauvola (window 8, k 0.34) via an integral image (opt-in); dark ink stays + // dark and paper stays light. + if options.binarize { + current = sauvola(¤t); + } + Ok(Preprocessed { image: current, inverse_transform: inverse, dpi }) +} + +/// An affine map `(x, y) → (a·x + c·y + e, b·x + d·y + f)`, composed in f64 so mapping a +/// word box back is exact up to float rounding. +#[derive(Clone, Copy, Debug, PartialEq)] +pub struct InverseTransform { + a: f64, + b: f64, + c: f64, + d: f64, + e: f64, + f: f64, +} + +impl InverseTransform { + pub fn identity() -> InverseTransform { + InverseTransform { a: 1.0, b: 0.0, c: 0.0, d: 1.0, e: 0.0, f: 0.0 } + } + + fn scale(s: f64) -> InverseTransform { + InverseTransform { a: s, b: 0.0, c: 0.0, d: s, e: 0.0, f: 0.0 } + } + + /// `compose(outer, inner)` maps `p` to `outer(inner(p))`: the accumulated inverse of the + /// earlier steps is the outer map (their spaces come later in the round trip), so each + /// step calls `compose(&inverse, &this_step_inverse)`. The scale step, for instance, + /// happens first, so its inverse is applied last: `scale⁻¹(rotate⁻¹(p))`. + fn compose_(outer: &InverseTransform, inner: &InverseTransform) -> InverseTransform { + InverseTransform { + a: outer.a * inner.a + outer.c * inner.b, + b: outer.b * inner.a + outer.d * inner.b, + c: outer.a * inner.c + outer.c * inner.d, + d: outer.b * inner.c + outer.d * inner.d, + e: outer.a * inner.e + outer.c * inner.f + outer.e, + f: outer.b * inner.e + outer.d * inner.f + outer.f, + } + } + + /// Map a point from the processed image back to the input raster. + pub fn apply(&self, x: f64, y: f64) -> (f64, f64) { + (self.a * x + self.c * y + self.e, self.b * x + self.d * y + self.f) + } + + /// Map a word box `[left, top, right, bottom]` back to the input raster: the four corners + /// go through [`Self::apply`] and their bounding box is rounded outward to whole pixels. + /// A box with no finite corners comes back unchanged (garbage in, garbage out — never a + /// panic). + pub fn map_rect(&self, rect: [f32; 4]) -> [f32; 4] { + let corners = [(rect[0], rect[1]), (rect[2], rect[1]), (rect[0], rect[3]), (rect[2], rect[3])]; + let mut lo = (f64::INFINITY, f64::INFINITY); + let mut hi = (f64::NEG_INFINITY, f64::NEG_INFINITY); + let mut finite = false; + for (x, y) in corners { + if !(x.is_finite() && y.is_finite()) { + continue; + } + let (mx, my) = self.apply(f64::from(x), f64::from(y)); + if mx.is_finite() && my.is_finite() { + finite = true; + lo = (lo.0.min(mx), lo.1.min(my)); + hi = (hi.0.max(mx), hi.1.max(my)); + } + } + if !finite { + return rect; + } + [lo.0.floor() as f32, lo.1.floor() as f32, hi.0.ceil() as f32, hi.1.ceil() as f32] + } +} + +/// The step inverses compose in pipeline order: each new step's inverse is applied *after* +/// the inverses accumulated so far. +fn compose(outer: &InverseTransform, inner: &InverseTransform) -> InverseTransform { + InverseTransform::compose_(outer, inner) +} + +/// The size `width` x `height` becomes under `scale`, refused (before any buffer exists) when +/// a side is under one pixel or over [`MAX_SIDE`], or the area is over [`MAX_PIXELS`]: two +/// sides under the side cap can still ask for hundreds of megabytes, which `OcrImage::new` +/// would only refuse after they were allocated. +fn scaled_size(width: u32, height: u32, scale: f64) -> Result<(u32, u32), OcrError> { + let new_w = (f64::from(width) * scale).ceil(); + let new_h = (f64::from(height) * scale).ceil(); + if !(1.0..=f64::from(MAX_SIDE)).contains(&new_w) || !(1.0..=f64::from(MAX_SIDE)).contains(&new_h) { + return Err(OcrError::ImageTooLarge); + } + let (nw, nh) = (new_w as u32, new_h as u32); + if (nw as usize).checked_mul(nh as usize).is_none_or(|pixels| pixels > MAX_PIXELS as usize) { + return Err(OcrError::ImageTooLarge); + } + Ok((nw, nh)) +} + +/// Bilinear resize by `scale`, clamped at the edges; the result stays inside [`scaled_size`]'s +/// limits (the side cap and the area cap). +fn resize_bilinear(image: &OcrImage, scale: f64) -> Result { + let (nw, nh) = scaled_size(image.width, image.height, scale)?; + let mut out = vec![255u8; (nw as usize).checked_mul(nh as usize).and_then(|p| p.checked_mul(4)).ok_or(OcrError::ImageTooLarge)?]; + for (y, row) in out.chunks_exact_mut(4 * nw as usize).enumerate() { + for (x, px) in row.as_chunks_mut::<4>().0.iter_mut().enumerate() { + // The source position of the destination pixel's centre, and the four + // neighbours bilinear interpolation blends (positions clamp at the edges). + let sx = (x as f64 + 0.5) / scale - 0.5; + let sy = (y as f64 + 0.5) / scale - 0.5; + let (x0, y0) = (sx.floor().max(0.0), sy.floor().max(0.0)); + let (fx, fy) = (sx - x0, sy - y0); + for (ch, o) in px.iter_mut().enumerate() { + let blend = |dx: f64, dy: f64| f64::from(image.byte(x0 + dx, y0 + dy, ch)); + let v = blend(0.0, 0.0) * (1.0 - fx) * (1.0 - fy) + + blend(1.0, 0.0) * fx * (1.0 - fy) + + blend(0.0, 1.0) * (1.0 - fx) * fy + + blend(1.0, 1.0) * fx * fy; + *o = v.round().clamp(0.0, 255.0) as u8; + } + } + } + OcrImage::new(nw, nh, out) +} + +/// Grayscale (Rec.601 luma) of every pixel. +fn grayscale(image: &OcrImage) -> Vec { + image.rgba.as_chunks::<4>().0.iter().map(|p| ((299 * p[0] as u32 + 587 * p[1] as u32 + 114 * p[2] as u32) / 1000).min(255) as u8).collect() +} + +/// How the page must be turned (quarter turns, clockwise) to read upright. +type QuarterTurns = u32; + +/// Detect the correction for pages scanned sideways or upside down. Method (documented, +/// panic-free, no OCR): ink is everything darker than the midpoint between the darkest and +/// lightest pixel; for each candidate correction (0/90/180/270 clockwise) the ink's row +/// profile is taken and scored by how sharply it bands (Σ Δ², the projection-profile score: +/// upright text lines band sharply, turned text does not). The best-banding axis wins when it +/// beats the other axis by a factor of 1.3; within the winning axis the line bands' +/// bottom-heaviness (real text carries more ink below each line's middle — baselines and +/// descenders) picks the quarter turn, with a 4%-of-ink margin. Anything less clear is a +/// no-op (`0`), and images under 32 px are never judged. +fn detect_orientation(image: &OcrImage) -> QuarterTurns { + if image.min_side() < 32 { + return 0; + } + let gray = grayscale(image); + let (Some(lo), Some(hi)) = (gray.iter().min(), gray.iter().max()) else { return 0 }; + let t = (*lo as u32 + *hi as u32) / 2; + let ink: Vec = gray.iter().map(|&g| (g as u32) < t).collect(); + let (w, h) = (image.width as usize, image.height as usize); + let total: u64 = ink.iter().filter(|i| **i).count() as u64; + if total == 0 { + return 0; + } + // The ink grid rotated by k quarter turns (clockwise), as row profiles. + let mut scores = [0.0f64; 4]; + let mut asyms = [0.0f64; 4]; + for k in 0u32..4 { + let k = k as usize; + let profile = rotated_row_profile(&ink, w, h, k as u32); + scores[k] = profile.windows(2).map(|p| (p[1] as f64 - p[0] as f64).powi(2)).sum(); + asyms[k] = band_bottom_heaviness(&profile); + } + // 0 and 180 share one profile shape (reversed), as do 90 and 270: pick the axis first. + let (vertical_pair, horizontal_pair) = ((scores[0], scores[2]), (scores[1], scores[3])); + let row_axis = vertical_pair.0.max(vertical_pair.1); + let column_axis = horizontal_pair.0.max(horizontal_pair.1); + if row_axis >= column_axis { + if row_axis < column_axis * 1.3 { + return 0; // the axes are too close to call + } + if asyms[0] > 0.15 * total as f64 { + 0 + } else if asyms[2] > 0.15 * total as f64 { + 2 + } else { + 0 // the bands are not clearly bottom-heavy either way: no-op + } + } else { + if column_axis < row_axis * 1.3 { + return 0; + } + if asyms[1] > 0.15 * total as f64 { + 1 + } else if asyms[3] > 0.15 * total as f64 { + 3 + } else { + 0 + } + } +} + +/// The row profile of the ink grid after `k` quarter turns clockwise. +fn rotated_row_profile(ink: &[bool], w: usize, h: usize, k: u32) -> Vec { + // The rotated image is `w` pixels tall for odd k and `h` for even k; its row index is + // always the new y of the source pixel. + let len = if k.is_multiple_of(2) { h } else { w }; + let mut profile = vec![0u64; len]; + for (y, row) in ink.chunks_exact(w).enumerate() { + for (x, on) in row.iter().enumerate() { + if !on { + continue; + } + // Where does (x, y) land after k quarter turns clockwise? + let ny = match k { + 1 => x, + 2 => h - 1 - y, + 3 => w - 1 - x, + _ => y, + }; + if let Some(p) = profile.get_mut(ny) { + *p += 1; + } + } + } + profile +} + +/// Sum over text-line bands of (ink in the band's lower half − ink in its upper half): +/// positive when the bands carry more ink near their bottoms, as real text does. +fn band_bottom_heaviness(profile: &[u64]) -> f64 { + let mut asym = 0.0; + let mut start: Option = None; + for (i, &p) in profile.iter().chain(std::iter::once(&0)).enumerate() { + match (start, p > 0) { + (None, true) => start = Some(i), + (Some(s), false) => { + let mid = (s + i) / 2; + let lower: u64 = profile.get(s..i).map_or(0, |b| b[mid.saturating_sub(s)..].iter().sum()); + let upper: u64 = profile.get(s..i).map_or(0, |b| b[..mid.saturating_sub(s)].iter().sum()); + asym += lower as f64 - upper as f64; // signed: a top-heavy band scores below zero + start = None; + } + _ => {} + } + } + asym +} + +/// Rotate the raster `k` quarter turns clockwise (an exact pixel permutation). +fn rotate_quarter_turns(image: &OcrImage, k: QuarterTurns) -> OcrImage { + let (w, h) = (image.width as usize, image.height as usize); + let (nw, nh) = if k.is_multiple_of(2) { (image.width, image.height) } else { (image.height, image.width) }; + let mut out = vec![255u8; nw as usize * nh as usize * 4]; + for (y, row) in image.rgba.chunks_exact(w * 4).enumerate() { + for (x, px) in row.as_chunks::<4>().0.iter().enumerate() { + let (nx, ny) = match k { + 1 => (h - 1 - y, x), + 2 => (w - 1 - x, h - 1 - y), + 3 => (y, w - 1 - x), + _ => (x, y), + }; + if let Some(dst) = out.get_mut((ny * nw as usize + nx) * 4..(ny * nw as usize + nx) * 4 + 4) { + dst.copy_from_slice(px); + } + } + } + OcrImage::new(nw, nh, out).unwrap_or_else(|_| image.clone()) +} + +/// The inverse of a `k`-quarter-turn correction: where each corrected pixel came from. +fn inverse_quarter_turns(k: QuarterTurns, old_w: u32, old_h: u32) -> InverseTransform { + let (w, h) = (f64::from(old_w), f64::from(old_h)); + match k { + // Forward (content moves): (x, y) ↦ (h−1−y, x) for one turn clockwise. + 1 => InverseTransform { a: 0.0, b: -1.0, c: 1.0, d: 0.0, e: 0.0, f: h - 1.0 }, + 2 => InverseTransform { a: -1.0, b: 0.0, c: 0.0, d: -1.0, e: w - 1.0, f: h - 1.0 }, + 3 => InverseTransform { a: 0.0, b: 1.0, c: -1.0, d: 0.0, e: w - 1.0, f: 0.0 }, + _ => InverseTransform::identity(), + } +} + +/// Estimate the skew of the text lines, in degrees (−5…5, positive: lines run downhill to the +/// right). Method (documented, panic-free): for each candidate angle the ink is sheared into a +/// row profile (`r = y − (x − cx)·tan φ`); the projection-profile score (Σ profile²) peaks +/// where the shear realigns the lines. `None` — leave the page alone — when the image is under +/// 64 px on a side, blank, the winner is under 0.1°, or the winner does not clearly beat +/// leaving the page as is (score < 1.02 × the 0° score). +fn estimate_skew(image: &OcrImage) -> Option { + if image.min_side() < 64 { + return None; + } + let gray = grayscale(image); + let (Some(lo), Some(hi)) = (gray.iter().min(), gray.iter().max()) else { return None }; + let t = (*lo as u32 + *hi as u32) / 2; + let ink: Vec = gray.iter().map(|&g| (g as u32) < t).collect(); + if !ink.iter().any(|i| *i) { + return None; // a blank page has no skew + } + let (w, h) = (image.width, image.height); + let cx = f64::from(w) / 2.0; + let max_shift = (f64::from(w) / 2.0 * 5f64.to_radians().tan()).ceil() as i64 + 2; + let score = |phi_deg: f64| -> f64 { + let slope = phi_deg.to_radians().tan(); + let mut profile = vec![0u64; (h as i64 + 2 * max_shift + 1) as usize]; + for (y, row) in ink.chunks_exact(w as usize).enumerate() { + for (x, on) in row.iter().enumerate() { + if !on { + continue; + } + let r = y as f64 - (x as f64 - cx) * slope; + let idx = r.round() as i64 + max_shift; + if let Some(p) = profile.get_mut(idx as usize) { + *p += 1; + } + } + } + profile.iter().map(|p| (*p as f64).powi(2)).sum() + }; + let mut best = (0.0f64, score(0.0)); + let mut phi = -5.0; + while phi <= 5.0 + 1e-9 { + let s = score(phi); + if s > best.1 { + best = (phi, s); + } + phi += 0.25; + } + let (angle, best_score) = best; + if angle.abs() < 0.1 || best_score < 1.02 * score(0.0) { + return None; // straight enough, or no clear winner: no-op + } + Some(angle) +} + +/// Rotate the content by `−angle` about the centre (the deskew correction), nearest-neighbour +/// with white fill; also returns the exact dest→src map (the correction's inverse). +fn deskew_rotate(image: &OcrImage, angle_deg: f64) -> (OcrImage, InverseTransform) { + let phi = angle_deg.to_radians(); + let (cos, sin) = (phi.cos(), phi.sin()); + let (w, h) = (image.width, image.height); + let (cx, cy) = (f64::from(w) / 2.0, f64::from(h) / 2.0); + let back = InverseTransform { a: cos, b: sin, c: -sin, d: cos, e: cx - cos * cx + sin * cy, f: cy - sin * cx - cos * cy }; + let mut out = vec![255u8; w as usize * h as usize * 4]; + for (y, row) in out.chunks_exact_mut(w as usize * 4).enumerate() { + for (x, px) in row.as_chunks_mut::<4>().0.iter_mut().enumerate() { + let (sx, sy) = back.apply(x as f64, y as f64); + let (xi, yi) = (sx.round() as i64, sy.round() as i64); + if xi < 0 || yi < 0 || xi >= w as i64 || yi >= h as i64 { + continue; // outside the scan: white fill + } + if let Some(src) = image.rgba.get((yi as usize * w as usize + xi as usize) * 4..(yi as usize * w as usize + xi as usize) * 4 + 4) { + px.copy_from_slice(src); + } + } + } + (OcrImage::new(w, h, out).unwrap_or_else(|_| image.clone()), back) +} + +/// 3×3 median filter on each colour channel, edges clamped (border pixels replicate); +/// dimensions and alpha are unchanged. +fn median3(image: &OcrImage) -> OcrImage { + let (w, h) = (image.width as usize, image.height as usize); + let mut out = image.rgba.clone(); + for (i, dst) in out.as_chunks_mut::<4>().0.iter_mut().enumerate() { + let (x, y) = (i % w, i / w); + let mut rgba = [255u8; 3]; + for (ch, v) in rgba.iter_mut().enumerate() { + let mut window = [0u8; 9]; + for (n, (dx, dy)) in [(-1i64, -1i64), (0, -1), (1, -1), (-1, 0), (0, 0), (1, 0), (-1, 1), (0, 1), (1, 1)].iter().enumerate() { + let nx = ((x as i64) + dx).clamp(0, w as i64 - 1) as usize; + let ny = ((y as i64) + dy).clamp(0, h as i64 - 1) as usize; + window[n] = image.byte(nx as f64, ny as f64, ch); + } + window.sort_unstable(); + *v = window[4]; + } + let alpha = dst[3]; + dst.copy_from_slice(&[rgba[0], rgba[1], rgba[2], alpha]); + } + OcrImage::new(image.width, image.height, out).unwrap_or_else(|_| image.clone()) +} + +/// The Sauvola ink mask: local thresholding with the documented window — 8 pixels wide, the +/// 8 pixels from −4 to +3 (the end is exclusive, so `x0..x1` is exactly `x−4..=x+3`), k 0.34 — +/// through an integral image. A pixel is ink when its grey is at or under the local +/// threshold. +fn sauvola_mask(image: &OcrImage) -> Vec { + let (w, h) = (image.width as usize, image.height as usize); + let gray = grayscale(image); + // Integral sums and square sums, (w+1)×(h+1), so any window sum is four lookups. + let stride = w + 1; + let mut sums = vec![0u64; stride * (h + 1)]; + let mut squares = vec![0u64; stride * (h + 1)]; + for (y, row) in gray.chunks_exact(w).enumerate() { + let (mut row_sum, mut row_sq) = (0u64, 0u64); + for (x, &g) in row.iter().enumerate() { + row_sum += u64::from(g); + row_sq += u64::from(g) * u64::from(g); + let above = sums.get(y * stride + x + 1).copied().unwrap_or(0); + let above_sq = squares.get(y * stride + x + 1).copied().unwrap_or(0); + if let (Some(s), Some(sq)) = (sums.get_mut((y + 1) * stride + x + 1), squares.get_mut((y + 1) * stride + x + 1)) { + *s = above + row_sum; + *sq = above_sq + row_sq; + } + } + } + let (k, r) = (0.34f64, 128.0f64); + let half = 4usize; + let mut ink = vec![false; w * h]; + for (y, row) in ink.chunks_exact_mut(w).enumerate() { + for (x, dst) in row.iter_mut().enumerate() { + let (x0, y0) = (x.saturating_sub(half), y.saturating_sub(half)); + // Exclusive end: the window covers −4..=+3, 8 pixels on a side (an end of + // `x + half + 1` would make it 9 and disagree with the documented window). + let (x1, y1) = ((x + half).min(w), (y + half).min(h)); + let n = ((x1 - x0) * (y1 - y0)) as f64; + let sum = rect_sum(&sums, stride, x0, y0, x1, y1); + let sq = rect_sum(&squares, stride, x0, y0, x1, y1); + let mean = sum / n; + let var = (sq / n - mean * mean).max(0.0); + let threshold = mean * (1.0 + k * (var.sqrt() / r - 1.0)); + *dst = matches!(gray.get(y * w + x), Some(&g) if f64::from(g) <= threshold); + } + } + ink +} + +/// Sauvola binarization through [`sauvola_mask`], followed by a one-pixel dilation of the ink: +/// ink goes to black, paper to white, alpha preserved. +fn sauvola(image: &OcrImage) -> OcrImage { + let (w, h) = (image.width as usize, image.height as usize); + // One dilation keeps anti-aliased strokes whole: a thresholded edge pixel that flaked off + // rejoins its stroke (slightly bold beats broken, for reading). + let ink = dilate1(&sauvola_mask(image), w, h); + let mut out = image.rgba.clone(); + for (i, px) in out.as_chunks_mut::<4>().0.iter_mut().enumerate() { + let v = if ink.get(i).copied().unwrap_or(false) { 0u8 } else { 255u8 }; + if let Some(dst) = px.get_mut(..3) { + dst.copy_from_slice(&[v, v, v]); + } + } + OcrImage::new(image.width, image.height, out).unwrap_or_else(|_| image.clone()) +} + +/// Grow the ink mask by one pixel (8-neighbourhood, edges clamped by the bounds checks). +fn dilate1(ink: &[bool], w: usize, h: usize) -> Vec { + let mut out = ink.to_vec(); + for y in 0..h { + for x in 0..w { + if !ink.get(y * w + x).copied().unwrap_or(false) { + continue; + } + for dy in [-1isize, 0, 1] { + for dx in [-1isize, 0, 1] { + let (nx, ny) = (x as isize + dx, y as isize + dy); + if nx < 0 || ny < 0 || nx >= w as isize || ny >= h as isize { + continue; + } + if let Some(o) = out.get_mut(ny as usize * w + nx as usize) { + *o = true; + } + } + } + } + } + out +} + +fn rect_sum(table: &[u64], stride: usize, x0: usize, y0: usize, x1: usize, y1: usize) -> f64 { + let a = table.get(y1 * stride + x1).copied().unwrap_or(0); + let b = table.get(y0 * stride + x1).copied().unwrap_or(0); + let c = table.get(y1 * stride + x0).copied().unwrap_or(0); + let d = table.get(y0 * stride + x0).copied().unwrap_or(0); + (a + d - b - c) as f64 +} + +#[cfg(test)] +mod tests { + use super::*; + + /// Sides under MAX_SIDE whose area passes MAX_PIXELS (9500 × 9500, 90 MP) are refused + /// when the size is worked out, before the 360 MB buffer is allocated rather than after. + #[test] + fn an_upscale_over_the_area_cap_is_refused_before_allocating() { + assert!(matches!(scaled_size(1000, 1000, 9.5), Err(OcrError::ImageTooLarge))); + assert_eq!(scaled_size(1000, 1000, 2.0).unwrap(), (2000, 2000), "under the caps it still scales"); + assert_eq!(scaled_size(8000, 8000, 1.0).unwrap(), (8000, 8000), "exactly 64 MP is allowed"); + assert!(matches!(scaled_size(8001, 8000, 1.0), Err(OcrError::ImageTooLarge)), "one row over is not"); + assert!(matches!(scaled_size(1000, 1000, 10.5), Err(OcrError::ImageTooLarge)), "a side over MAX_SIDE"); + assert!(matches!(scaled_size(1000, 1000, 0.0), Err(OcrError::ImageTooLarge)), "a side under one pixel"); + assert!(matches!(scaled_size(1000, 1000, f64::NAN), Err(OcrError::ImageTooLarge))); + let small = OcrImage::new(100, 100, vec![255; 100 * 100 * 4]).unwrap(); + assert!(matches!(resize_bilinear(&small, 100.5), Err(OcrError::ImageTooLarge))); + assert_eq!(resize_bilinear(&small, 2.0).unwrap().width, 200); + } + + fn white(w: u32, h: u32) -> OcrImage { + OcrImage::new(w, h, vec![255; (w * h * 4) as usize]).unwrap() + } + + /// Ink bounding box of a raster (None when blank). + fn ink_bbox(image: &OcrImage) -> Option<[f32; 4]> { + let gray = grayscale(image); + let (Some(lo), Some(hi)) = (gray.iter().min(), gray.iter().max()) else { return None }; + let t = (*lo as u32 + *hi as u32) / 2; + let (mut l, mut tp, mut r, mut b) = (u32::MAX, u32::MAX, 0u32, 0u32); + let mut any = false; + for (i, &g) in gray.iter().enumerate() { + if (g as u32) >= t { + continue; + } + any = true; + let (x, y) = ((i as u32) % image.width, (i as u32) / image.width); + l = l.min(x); + r = r.max(x); + tp = tp.min(y); + b = b.max(y); + } + any.then_some([l as f32, tp as f32, r as f32, b as f32]) + } + + /// A horizontal line of three bottom-heavy block glyphs (L shapes) on white — the fixture + /// for orientation detection (240 × 64 at 300 dpi). + fn l_line() -> OcrImage { + let (w, h) = (240u32, 64u32); + let mut img = white(w, h); + for gx in [20u32, 100, 180] { + fill(&mut img, gx, 12, 12, 40); + fill(&mut img, gx, 40, 36, 12); + } + img + } + + fn fill(img: &mut OcrImage, x0: u32, y0: u32, w: u32, h: u32) { + for y in y0..y0 + h { + for x in x0..x0 + w { + let o = ((y * img.width + x) * 4) as usize; + if let Some(p) = img.rgba.get_mut(o..o + 3) { + p.copy_from_slice(&[0, 0, 0]); + } + } + } + } + + /// The same fixture pre-rotated by k quarter turns clockwise, built independently of the + /// crate's rotation (so the test cannot pass by a shared bug). + fn rotated(k: u32) -> OcrImage { + let src = l_line(); + let (w, h) = (src.width as usize, src.height as usize); + let (nw, nh) = if k.is_multiple_of(2) { (src.width, src.height) } else { (src.height, src.width) }; + let mut out = white(nw, nh); + for (y, row) in src.rgba.chunks_exact(w * 4).enumerate() { + for (x, px) in row.as_chunks::<4>().0.iter().enumerate() { + if px[0] != 0 { + continue; + } + let (nx, ny) = match k { + 1 => (h - 1 - y, x), + 2 => (w - 1 - x, h - 1 - y), + 3 => (y, w - 1 - x), + _ => (x, y), + }; + fill(&mut out, nx as u32, ny as u32, 1, 1); + } + } + out + } + + /// A ruled fixture sheared by `degrees` (lines running downhill to the right for positive + /// angles): four 8-px bars across a 300 × 120 page. + fn sheared(degrees: f64) -> OcrImage { + let (w, h) = (300u32, 120u32); + let mut img = white(w, h); + let slope = degrees.to_radians().tan(); + for (base, height) in [(20u32, 8u32), (45, 8), (70, 8), (95, 8)] { + for x in 20..280u32 { + let y = (base as f64 + (x as f64 - 150.0) * slope).round().clamp(0.0, f64::from(h - height - 1)) as u32; + fill(&mut img, x, y, 1, height); + } + } + img + } + + /// With the four opt-in flags off and a dpi at or above 300, the raster passes through + /// untouched and the inverse transform is exact. + #[test] + fn flags_off_and_adequate_dpi_is_an_identity_pass_through() { + let src = l_line(); + let p = preprocess(src.clone(), 300.0, &PreprocessOptions::default()).unwrap(); + assert_eq!(p.image.rgba, src.rgba); + assert_eq!(p.dpi, 300.0); + assert_eq!(p.inverse_transform, InverseTransform::identity()); + assert_eq!(p.inverse_transform.map_rect([3.0, 7.0, 41.0, 19.0]), [3.0, 7.0, 41.0, 19.0]); + } + + /// With a step on, a scan below 300 dpi is enlarged by 300/src first, and its boxes map + /// back to the original raster. + #[test] + fn low_dpi_scans_are_upscaled_for_the_steps_and_boxes_map_back() { + let src = white(100, 50); + let on = PreprocessOptions { denoise: true, ..Default::default() }; + let p = preprocess(src.clone(), 150.0, &on).unwrap(); + assert_eq!((p.image.width, p.image.height), (200, 100)); + assert_eq!(p.dpi, 300.0); + assert_eq!(p.inverse_transform.map_rect([0.0, 0.0, 200.0, 100.0]), [0.0, 0.0, 100.0, 50.0]); + assert_eq!(p.inverse_transform.apply(2.0, 4.0), (1.0, 2.0)); + // 300 and 600 dpi are not resampled; a broken dpi is left alone. + assert_eq!(preprocess(src.clone(), 600.0, &on).unwrap().image.width, 100); + assert_eq!(preprocess(src, f32::NAN, &on).unwrap().image.width, 100); + } + + /// With every step off a low-dpi render is read exactly as rendered: not resampled, with + /// the identity as its way back. (Reading a 72 dpi scan through a 4x enlargement changed + /// what the models read: "due: \€250" came back as "ue: ?250".) + #[test] + fn with_every_step_off_a_low_dpi_render_is_not_resampled() { + let src = l_line(); + let p = preprocess(src.clone(), 72.0, &PreprocessOptions::default()).unwrap(); + assert_eq!(p.image.rgba, src.rgba); + assert_eq!((p.image.width, p.image.height), (src.width, src.height)); + assert_eq!(p.dpi, 72.0, "the render's own resolution"); + assert_eq!(p.inverse_transform, InverseTransform::identity()); + } + + /// The upscale refuses to break the side cap. + #[test] + fn upscaling_respects_the_side_cap() { + let src = white(9999, 10); + let r = preprocess(src, 72.0, &PreprocessOptions { denoise: true, ..Default::default() }); + assert!(matches!(r, Err(OcrError::ImageTooLarge)), "upscale past the cap: {r:?}"); + } + + /// An upright page is left alone by the orientation detector. + #[test] + fn upright_pages_are_not_rotated() { + let p = preprocess(l_line(), 300.0, &PreprocessOptions { auto_rotate: true, ..Default::default() }).unwrap(); + assert_eq!(p.inverse_transform, InverseTransform::identity()); + } + + /// Pages scanned at 90/180/270 are detected, turned upright, and their word boxes map back + /// onto the scanned raster's ink. + #[test] + fn quarter_turns_are_detected_corrected_and_mapped_back() { + for k in 1..=3 { + let scanned = rotated(k); + let p = preprocess(scanned.clone(), 300.0, &PreprocessOptions { auto_rotate: true, ..Default::default() }).unwrap(); + let corrected = ink_bbox(&p.image).unwrap_or_else(|| panic!("k={k}: the corrected page still has ink")); + // The corrected reading is a horizontal line again: wider than tall. + assert!(corrected[2] - corrected[0] > corrected[3] - corrected[1], "k={k}: {corrected:?}"); + // And the box maps back onto the scanned raster's own ink box. + let back = p.inverse_transform.map_rect(corrected); + let original = ink_bbox(&scanned).unwrap(); + for i in 0..4 { + assert!((back[i] - original[i]).abs() <= 2.0, "k={k}: {back:?} vs {original:?}"); + } + } + } + + /// Images under 32 px on a side are never judged; the step is a no-op. + #[test] + fn tiny_images_skip_orientation() { + let src = white(31, 31); + let p = preprocess(src, 300.0, &PreprocessOptions { auto_rotate: true, ..Default::default() }).unwrap(); + assert_eq!(p.inverse_transform, InverseTransform::identity()); + } + + /// Blank pages are inconclusive: every step leaves them alone. + #[test] + fn blank_pages_are_a_noop() { + let src = white(200, 200); + let p = preprocess(src, 300.0, &PreprocessOptions { auto_rotate: true, deskew: true, denoise: true, binarize: true }).unwrap(); + assert_eq!(p.inverse_transform, InverseTransform::identity()); + } + + /// The skew estimator sees a 3° shear (either direction) and leaves a straight page alone. + #[test] + fn skew_is_estimated_and_straight_pages_are_left_alone() { + for degrees in [3.0, -3.0] { + let angle = estimate_skew(&sheared(degrees)).unwrap_or_else(|| panic!("{degrees}° should be found")); + assert!((angle - degrees).abs() <= 0.5, "{degrees}°: got {angle}"); + } + assert!(estimate_skew(&sheared(0.0)).is_none(), "a straight page is a no-op"); + assert!(estimate_skew(&l_line()).is_none(), "glyph lines without a shear are a no-op"); + } + + /// Deskew straightens ±3° lines (the vertical span tightens) and the centre maps back to + /// itself. + #[test] + fn deskew_corrects_plus_and_minus_three_degrees() { + for degrees in [3.0, -3.0] { + let src = sheared(degrees); + let before = ink_bbox(&src).unwrap(); + let p = preprocess(src.clone(), 300.0, &PreprocessOptions { deskew: true, ..Default::default() }).unwrap(); + let after = ink_bbox(&p.image).unwrap(); + assert!(after[3] - after[1] < before[3] - before[1], "{degrees}°: {before:?} → {after:?}"); + // The exact inverse maps the corrected centre back to the scanned centre. + let (cx, cy) = (f64::from(src.width) / 2.0, f64::from(src.height) / 2.0); + let (bx, by) = p.inverse_transform.apply(cx, cy); + assert!((bx - cx).abs() < 1.0 && (by - cy).abs() < 1.0, "{degrees}°: ({bx}, {by})"); + } + } + + /// Deskew no-ops on small images (under 64 px) — the detector is never consulted. + #[test] + fn small_images_skip_deskew() { + let src = white(63, 63); + let p = preprocess(src, 300.0, &PreprocessOptions { deskew: true, ..Default::default() }).unwrap(); + assert_eq!(p.inverse_transform, InverseTransform::identity()); + } + + /// The median filter removes isolated specks, keeps solid strokes, and never changes the + /// dimensions. + #[test] + fn denoise_removes_specks_and_keeps_dimensions() { + let mut src = white(32, 32); + fill(&mut src, 15, 15, 1, 1); // salt + fill(&mut src, 4, 4, 5, 5); // a solid stroke + let p = preprocess(src, 300.0, &PreprocessOptions { denoise: true, ..Default::default() }).unwrap(); + assert_eq!((p.image.width, p.image.height), (32, 32)); + assert_eq!(p.image.byte(15.0, 15.0, 0), 255, "an isolated speck is gone"); + assert_eq!(p.image.byte(6.0, 6.0, 0), 0, "the inside of a solid block stays"); + assert_eq!(p.inverse_transform, InverseTransform::identity()); + } + + /// Sauvola keeps the polarity (dark ink stays dark, paper stays light) and the metadata. + /// The fixture is a thin grey stroke: its windows mix ink and paper, so the local + /// threshold separates them (a perfectly flat mid-tone area has no contrast and whitens, + /// which is Sauvola working as designed). + #[test] + fn binarize_keeps_polarity_and_metadata() { + let (w, h) = (64u32, 64u32); + let mut src = white(w, h); + for y in 26..30 { + for x in 8..56 { + let o = ((y * w + x) * 4) as usize; + if let Some(p) = src.rgba.get_mut(o..o + 3) { + p.copy_from_slice(&[70, 70, 70]); // a grey scan's ink, far below paper + } + } + } + let p = preprocess(src, 300.0, &PreprocessOptions { binarize: true, ..Default::default() }).unwrap(); + assert_eq!((p.image.width, p.image.height), (w, h)); + assert_eq!(p.dpi, 300.0); + assert_eq!(p.image.byte(30.0, 28.0, 0), 0, "ink stays dark"); + assert_eq!(p.image.byte(2.0, 2.0, 0), 255, "paper stays light"); + for px in p.image.rgba.as_chunks::<4>().0.iter() { + assert!((px[0] == 0 || px[0] == 255) && px[0] == px[1] && px[1] == px[2], "{px:?}"); + } + } + + /// The inverse transform composes: an upscaled, turned, deskewed page maps corrected + /// points back inside the original raster. + #[test] + fn inverse_transform_maps_points_back_through_every_step() { + let scanned = rotated(1); // a page photographed on its side + let p = preprocess(scanned.clone(), 150.0, &PreprocessOptions { auto_rotate: true, deskew: true, ..Default::default() }).unwrap(); + let (w, h) = (f64::from(p.image.width), f64::from(p.image.height)); + for (x, y) in [(0.0, 0.0), (w - 1.0, 0.0), (0.0, h - 1.0), (w - 1.0, h - 1.0), (w / 2.0, h / 2.0)] { + let (bx, by) = p.inverse_transform.apply(x, y); + assert!((-2.0..=f64::from(scanned.width) + 2.0).contains(&bx), "x {x} → {bx}"); + assert!((-2.0..=f64::from(scanned.height) + 2.0).contains(&by), "y {y} → {by}"); + } + // The scale step, alone, is exact. + assert_eq!(InverseTransform::scale(0.5).apply(4.0, 10.0), (2.0, 5.0)); + } + + /// A realistic letter page (one line of type on white, at 300 dpi) reads as upright: the + /// row axis bands decisively sharper than the column axis, and the near-balanced + /// bottom-heaviness of real mixed-case text is below the 15% flip bar, so the page is left + /// alone rather than risk flipping a good scan upside down. + #[test] + fn realistic_letter_pages_are_left_upright() { + let (w, h) = (1200u32, 1650u32); + let mut img = white(w, h); + let mut x = 120u32; + let base = 300u32; + let mut seed = 7u64; + let mut rnd = || { + seed = seed.wrapping_mul(6364136223846793005).wrapping_add(1442695040888963407); + (seed >> 33) as u32 + }; + while x < 1000 { + let word = rnd() % 5 + 2; + for _ in 0..word { + let kind = rnd() % 4; + let wl = rnd() % 8 + 3; + match kind { + 0 => fill(&mut img, x, base - 20, wl, 20), // x-height + 1 => fill(&mut img, x, base - 34, wl, 34), // ascender + 2 => fill(&mut img, x, base - 20, wl, 28), // descender + _ => { + fill(&mut img, x, base - 20, 2, 20); + fill(&mut img, x + 3, base - 14, 2, 14); // stems + } + } + x += wl + 2; + } + x += 12; // space + } + assert_eq!(detect_orientation(&img), 0); + let p = preprocess(img, 300.0, &PreprocessOptions { auto_rotate: true, ..Default::default() }).unwrap(); + assert_eq!(p.inverse_transform, InverseTransform::identity()); + } + + /// Hostile inputs are errors or no-ops, never panics: 1×1 rasters, all-black pages, NaN + /// boxes through the inverse map. + #[test] + fn hostile_inputs_never_panic() { + let tiny = white(1, 1); + let all = PreprocessOptions { auto_rotate: true, deskew: true, denoise: true, binarize: true }; + let p = preprocess(tiny, 300.0, &all).unwrap(); + assert_eq!((p.image.width, p.image.height), (1, 1)); + // All-black: no paper, nothing to judge. + let mut black = white(80, 80); + for px in black.rgba.as_chunks_mut::<4>().0 { + px.copy_from_slice(&[0, 0, 0, 255]); + } + let p = preprocess(black, 300.0, &all).unwrap(); + assert_eq!(p.inverse_transform, InverseTransform::identity(), "all-black is inconclusive"); + // NaN boxes never poison downstream math: the finite corners bound the mapped box, + // and a box with no finite corners comes back unchanged. + let mapped = p.inverse_transform.map_rect([f32::NAN, 0.0, 5.0, 5.0]); + assert!(mapped.iter().all(|v| v.is_finite()), "{mapped:?}"); + let unchanged = InverseTransform::identity().map_rect([f32::NAN; 4]); + assert!(unchanged.iter().all(|v| v.is_nan()), "{unchanged:?}"); + } + + /// The one-pixel dilation is load-bearing, pinned with a fixture that fragments without + /// it: a thin anti-aliased-looking stroke on a grey scan (alternating dark and light + /// pixels) — the raw Sauvola mask drops the light runs (the local window sits on a grey + /// page, so the threshold falls below them), while the binarized image keeps the stroke + /// whole. + #[test] + fn binarize_without_the_dilation_the_stroke_fragments_and_with_it_stays_whole() { + let (w, h) = (64u32, 64u32); + let mut src = white(w, h); + let set = |img: &mut OcrImage, x: u32, y: u32, v: u8| { + let o = ((y * w + x) * 4) as usize; + if let Some(p) = img.rgba.get_mut(o..o + 3) { + p.copy_from_slice(&[v, v, v]); + } + }; + for px in src.rgba.as_chunks_mut::<4>().0.iter_mut() { + px[..3].copy_from_slice(&[210, 210, 210]); // a grey scan's paper + } + // The stroke: one pixel tall, alternating dark (reads as ink) and light (an + // anti-aliased shoulder the threshold alone drops) in runs of two. + for x in 8..56u32 { + let v = if (x / 2) % 2 == 0 { 110 } else { 170 }; + set(&mut src, x, 32, v); + } + let mask = sauvola_mask(&src); + let at = |x: u32, y: u32| mask[y as usize * w as usize + x as usize]; + let dropped: Vec = (12..52).filter(|&x| !at(x, 32)).collect(); + assert!(!dropped.is_empty(), "without the dilation the stroke fragments: these stroke pixels fall under the threshold at {dropped:?}"); + let p = preprocess(src, 300.0, &PreprocessOptions { binarize: true, ..Default::default() }).unwrap(); + // With the dilation the stroke is whole: every position on it has ink within its + // 8-neighbourhood. + for x in 12..=51u32 { + let ink_near = (-1i32..=1).any(|dy| { + (-1i32..=1).any(|dx| { + let (nx, ny) = (x as i32 + dx, 32 + dy); + p.image.byte(f64::from(nx), f64::from(ny), 0) == 0 + }) + }); + assert!(ink_near, "the stroke is broken at x={x}"); + } + } + + /// The deskew inverse, checked at OFF-CENTRE points (the centre maps to itself under any + /// rotation, so it cannot catch a wrong sign): a point on a straightened line maps back + /// onto the sheared source line, uphill and downhill. + #[test] + fn deskew_inverse_maps_off_centre_points_back_onto_the_source_line() { + for degrees in [3.0, -3.0] { + let src = sheared(degrees); + let p = preprocess(src, 300.0, &PreprocessOptions { deskew: true, ..Default::default() }).unwrap(); + let slope = degrees.to_radians().tan(); + // The corrected first line sits at y = 20 (the fixture's base); the same ink was + // at 20 + (x − 150)·slope in the source. Well off centre on purpose. The + // rotation moves x a little too (it mixes in y: at most sin(5°)·40 ≈ 3.5 px + // here), so the line is checked at the mapped x — a wrong sign still misses it + // by ±(x−150)·slope, 6+ px at these offsets. + for x in [30.0f64, 150.0, 270.0] { + let (sx, sy) = p.inverse_transform.apply(x, 20.0); + assert!((sx - x).abs() < 4.0, "{degrees}° x={x}: mapped to ({sx}, {sy})"); + let expected = 20.0 + (sx - 150.0) * slope; + assert!((sy - expected).abs() < 2.0, "{degrees}° x={x}: ({sx}, {sy}), the source line is at {expected}"); + } + } + } +} diff --git a/crates/ocr/src/tesseract.rs b/crates/ocr/src/tesseract.rs new file mode 100644 index 0000000000..9fbcc574d8 --- /dev/null +++ b/crates/ocr/src/tesseract.rs @@ -0,0 +1,1022 @@ +//! The Tesseract engine: an installed `tesseract` binary driven as an optional external +//! process (`tesseract stdout -l tsv`), never linked, never bundled, never +//! downloaded. The whole module is compiled out for wasm32, where nothing may spawn. +//! +//! Security posture (hard rules): the program path comes from the PATH or a user-configured +//! location, never from document data; the language code is allow-listed before it reaches a +//! process; the page image goes to a private temp file (a fresh 0700 directory holding a 0600 +//! file on Unix) that is deleted afterwards; the run has a timeout and an output size cap; and +//! every failure comes back through [`OcrError`] with its cause named. + +use std::io::Write; +use std::path::{Path, PathBuf}; +use std::sync::Mutex; +use std::sync::atomic::{AtomicU8, Ordering}; +use std::time::{Duration, Instant}; + +use super::{Line, OcrError, OcrImage, RecognizeOptions, Recognizer, Word}; + +/// How long one recognition may run before the process is killed. +const TIMEOUT: Duration = Duration::from_secs(120); +/// How much TSV output is accepted before the run is abandoned. +const MAX_OUTPUT: u64 = 64 * 1024 * 1024; +/// How much of the program's stderr is read back (its end): enough for a `--list-langs` list +/// printed there, small enough that error text stays a message, not a memory DoS. +const MAX_STDERR: u64 = 256 * 1024; +/// The most words one TSV reading may hold. Known limit: a reading with more words than this +/// is truncated at the cap — corrupt or runaway output must not become unbounded memory in +/// the word list, the text layer and the review UI. +pub const MAX_WORDS: usize = 50_000; +/// Tesseract's code for each language PdfCraft offers (ISO 639-1 → tessdata name). Only these +/// codes ever reach the process, whatever the settings say. +pub const LANGUAGE_CODES: &[(&str, &str)] = &[ + ("en", "eng"), + ("de", "deu"), + ("fr", "fra"), + ("es", "spa"), + ("it", "ita"), + ("pt", "por"), + ("ru", "rus"), + ("zh", "chi_sim"), + ("ja", "jpn"), + ("ko", "kor"), + ("ar", "ara"), + ("nl", "nld"), +]; + +/// English names for [`LANGUAGE_CODES`] (dropdowns show code + name). +pub const LANGUAGE_NAMES: &[(&str, &str)] = &[ + ("en", "English"), + ("de", "German"), + ("fr", "French"), + ("es", "Spanish"), + ("it", "Italian"), + ("pt", "Portuguese"), + ("ru", "Russian"), + ("zh", "Chinese (Simplified)"), + ("ja", "Japanese"), + ("ko", "Korean"), + ("ar", "Arabic"), + ("nl", "Dutch"), +]; + +/// Map a settings language code to a Tesseract one: trimmed, case-insensitive, allow-listed; +/// anything unknown falls back to `"eng"`. +pub fn language_code(code: &str) -> &'static str { + let c = code.trim().to_ascii_lowercase(); + LANGUAGE_CODES.iter().find(|(k, _)| *k == c).map(|(_, v)| *v).unwrap_or("eng") +} + +/// The user-configured program location, when valid: `$PDFCRAFT_TESSERACT` (a user +/// preference, never taken from document data). An invalid value — relative, missing, a +/// directory, not executable — means unavailable: an explicit setting is never silently +/// replaced by a PATH search. +pub fn configured_path() -> Option { + let value = std::env::var_os("PDFCRAFT_TESSERACT")?; + let program = PathBuf::from(value); + program_is_executable(&program).then_some(program) +} + +/// The program [`TesseractCli::find`] would run: a set `configured` location (the value of +/// `PDFCRAFT_TESSERACT`) decides alone, so an invalid one yields `None` and `search` (the PATH) +/// is not even consulted; only when nothing is configured does `search` decide. +fn locate(configured: Option, search: impl FnOnce() -> Option) -> Option { + match configured { + Some(value) => Some(PathBuf::from(value)).filter(|program| program_is_executable(program)), + None => search(), + } +} + +/// The first `tesseract` on the PATH that is an existing executable file. Entries that are not +/// absolute are refused: a relative PATH entry resolves against the caller's working +/// directory, which is neither stable nor a safe place to find a program. Also named by +/// `ocr_status` (the engine reports where it would look, without probing). +pub fn path_search() -> Option { + path_search_in(std::env::split_paths(&std::env::var_os("PATH")?)) +} + +/// [`path_search`] over an explicit directory list (so tests can pass one). +fn path_search_in(dirs: impl Iterator) -> Option { + for dir in dirs.filter(|d| d.is_absolute()) { + for name in program_file_names() { + let candidate = dir.join(name); + if program_is_executable(&candidate) { + return Some(candidate); + } + } + } + None +} + +/// The file names an installed `tesseract` binary goes by. +fn program_file_names() -> &'static [&'static str] { + #[cfg(windows)] + { + &["tesseract.exe"] + } + #[cfg(not(windows))] + { + &["tesseract"] + } +} + +/// Whether `program` is a usable program location (see [`program_is_executable`]); callers +/// that run a validated program by path (the engine's `--version` query) check it first. +pub fn program_is_usable(program: &Path) -> bool { + program_is_executable(program) +} + +/// Whether `program` is a usable configured location: an absolute path to an existing, +/// executable FILE (on Unix the executable bits must be set; elsewhere being a file is what +/// the OS requires). A directory or a missing file is simply unavailable — never a panic. +fn program_is_executable(program: &Path) -> bool { + if !program.is_absolute() || !program.is_file() { + return false; + } + #[cfg(unix)] + { + use std::os::unix::fs::PermissionsExt; + std::fs::metadata(program).is_ok_and(|m| m.permissions().mode() & 0o111 != 0) + } + #[cfg(not(unix))] + { + true + } +} + +/// One row of Tesseract's TSV output. `level` 5 rows are the words; `conf` is −1 on the header +/// rows (page, block, paragraph, line), which are not words. +struct TsvRow<'a> { + level: &'a str, + block: u32, + par: u32, + line: u32, + left: f32, + top: f32, + width: f32, + height: f32, + conf: f32, + text: &'a str, +} + +/// Parse one TSV row; `None` for anything unreadable (short rows, non-numeric fields), which +/// is skipped rather than trusted. A row with non-finite numbers or a negative width or height +/// is not a word either: geometry like that is output corruption, and trusting it puts NaN +/// into word boxes and on to the page. +fn tsv_row(row: &str) -> Option> { + let c: Vec<&str> = row.split('\t').collect(); + if c.len() < 12 { + return None; + } + let (block, par, line) = (c[2].trim().parse().ok()?, c[3].trim().parse().ok()?, c[4].trim().parse().ok()?); + let (left, top): (f32, f32) = (c[6].trim().parse().ok()?, c[7].trim().parse().ok()?); + let (width, height, conf): (f32, f32, f32) = (c[8].trim().parse().ok()?, c[9].trim().parse().ok()?, c[10].trim().parse().ok()?); + if !left.is_finite() || !top.is_finite() || !width.is_finite() || !height.is_finite() || !conf.is_finite() { + return None; // NaN/inf anywhere in the numbers + } + if width < 0.0 || height < 0.0 { + return None; // a word box has a non-negative size + } + Some(TsvRow { level: c[0], block, par, line, left, top, width, height, conf, text: c[11] }) +} + +/// Parse `tesseract … tsv` output into lines of words. Rows with confidence −1 are the +/// structural header rows and never become words; consecutive word rows that share their +/// (block, paragraph, line) form one [`Line`], in output (reading) order. Confidence clamps +/// into 0–100. Known limit: past [`MAX_WORDS`] the rest of the reading is dropped (truncated, +/// not an error — the words before the cap are still placed). +pub fn parse_tsv(tsv: &str) -> Vec { + parse_tsv_bounded(tsv).0 +} + +/// The parser utility keeps its bounded-prefix API; recognition also checks completeness. +fn parse_tsv_bounded(tsv: &str) -> (Vec, bool) { + let mut lines: Vec = Vec::new(); + let mut key: Option<(u32, u32, u32)> = None; + let mut words = 0usize; + for row in tsv.lines() { + let Some(r) = tsv_row(row) else { continue }; + if r.level != "5" || r.conf < 0.0 || r.text.trim().is_empty() { + continue; + } + if words >= MAX_WORDS { + return (lines, true); + } + let this = (r.block, r.par, r.line); + if key != Some(this) { + lines.push(Line::default()); + key = Some(this); + } + let word = Word::new(r.text, [r.left, r.top, r.left + r.width, r.top + r.height], "tesseract").with_confidence(r.conf.clamp(0.0, 100.0)); + if let Some(last) = lines.last_mut() { + last.words.push(word); + words += 1; + } + } + (lines, false) +} + +/// Parse `tesseract --list-langs` output: the header line ("List of available languages…"), +/// then one code per line. Lines with spaces are not codes (a header, a warning) and are left +/// out. +pub fn parse_list_langs(out: &str) -> Vec { + out.lines() + .skip_while(|l| !l.trim().is_empty() && l.trim().starts_with("List of available")) + .map(|l| l.trim().to_string()) + .filter(|l| !l.is_empty() && !l.contains(char::is_whitespace)) + .collect() +} + +/// The languages a `--list-langs` run reported: its stdout, or, when that holds none, its +/// stderr (Tesseract 3.x prints the list there). +pub fn list_langs_from(stdout: &str, stderr: &str) -> Vec { + let codes = parse_list_langs(stdout); + if codes.is_empty() { parse_list_langs(stderr) } else { codes } +} + +/// The first non-empty line of what `tesseract --version` printed: on stdout, or on stderr +/// where older versions print it. +fn version_line(stdout: &str, stderr: &str) -> Option { + let first = |text: &str| text.lines().map(str::trim).find(|line| !line.is_empty()).map(str::to_owned); + first(stdout).or_else(|| first(stderr)) +} + +/// The Tesseract program. Build one (via [`TesseractCli::find`] for the PATH copy, or +/// [`TesseractCli::new`] for a user-configured path) and reuse it: `--list-langs` runs at most +/// once. +pub struct TesseractCli { + program: PathBuf, + /// The timeout and the output cap are fields so tests can shorten them. + timeout: Duration, + max_output: u64, + /// Whether `--list-langs` has succeeded yet: 0 untried, 1 yes, 2 no. + probed: AtomicU8, + langs: Mutex>, +} + +impl TesseractCli { + /// The program at `path` (a user preference; never taken from document data). + pub fn new(program: PathBuf) -> TesseractCli { + TesseractCli { program, timeout: TIMEOUT, max_output: MAX_OUTPUT, probed: AtomicU8::new(0), langs: Mutex::new(Vec::new()) } + } + + /// The program this engine runs (`ocr_status` names it). + pub fn program(&self) -> &std::path::Path { + &self.program + } + + /// The program at a user-configured `path`, if it is usable there: only an absolute path + /// to an existing, executable file is accepted (a directory, a missing file or a relative + /// path is unavailable, never a panic), and it must answer `--list-langs`. + pub fn at(program: PathBuf) -> Option { + if !program_is_executable(&program) { + return None; + } + let t = TesseractCli::new(program); + let langs = t.list_langs().ok()?; + *t.langs.lock().unwrap_or_else(|e| e.into_inner()) = langs; + t.probed.store(1, Ordering::Relaxed); + Some(t) + } + + /// The program this machine offers, if it answers `--list-langs`: the user-configured + /// location ([`configured_path`] / `$PDFCRAFT_TESSERACT`) when valid, else the first + /// absolute PATH entry that holds an executable `tesseract`. + /// + /// # The PATH-lookup risk + /// + /// Resolving through the PATH trusts every directory on it: a binary planted earlier on + /// the PATH runs as a subprocess with this application's privileges — it sees the page + /// image and anything the process can reach. The language code and the page image still + /// never come from document data, but a user who handles untrusted documents should pin + /// the exact program with `PDFCRAFT_TESSERACT`, which is validated (absolute, existing, + /// executable file) before any process starts. + /// + /// A `PDFCRAFT_TESSERACT` that is set but invalid fails closed: it is never replaced by + /// the PATH search. + pub fn find() -> Option { + TesseractCli::at(locate(std::env::var_os("PDFCRAFT_TESSERACT"), path_search)?) + } + + /// Run `tesseract --list-langs` and parse it; also decides [`TesseractCli::available`]. + fn list_langs(&self) -> Result, OcrError> { + let (out, err) = self.run_capture(&["--list-langs"], self.timeout, self.max_output)?; + let codes = list_langs_from(&out, &err); + if codes.is_empty() { + return Err(OcrError::NoTesseract("--list-langs reported no languages".into())); + } + Ok(codes + .into_iter() + .map(|c| { + let name = LANGUAGE_NAMES.iter().find(|(k, _)| *k == c.as_str()).map(|(_, n)| (*n).to_string()).unwrap_or_else(|| c.clone()); + (c, name) + }) + .collect()) + } + + /// The program's first `--version` line (named by `ocr_status`); `None` when the program + /// does not answer. Not cached: status asks rarely, and a re-install mid-session should + /// show up without a restart. + pub fn version(&self) -> Option { + let (out, err) = self.run_capture(&["--version"], self.timeout, self.max_output).ok()?; + version_line(&out, &err) + } + + /// Run the program with fixed argv, its stdout in a private temp file, under `timeout` + /// seconds and an output cap of `max_output` bytes. Returns the (capped) stdout. + fn run(&self, args: &[&str], timeout: Duration, max_output: u64) -> Result { + self.run_capture(args, timeout, max_output).map(|(out, _)| out) + } + + /// [`Self::run`], also returning what the program wrote to stderr (up to + /// [`MAX_STDERR`] bytes, its end kept): some Tesseract versions print the `--list-langs` + /// list there instead of on stdout. + fn run_capture(&self, args: &[&str], timeout: Duration, max_output: u64) -> Result<(String, String), OcrError> { + let dir = TempDir::create()?; + let out_path = dir.path.join("out.txt"); + let err_path = dir.path.join("err.txt"); + let out_file = std::fs::File::create(&out_path).map_err(|e| OcrError::Process(format!("creating the output file failed: {e}")))?; + // stderr goes to a file, not a pipe: an undrained pipe fills (around 64 KiB) and then + // blocks the child mid-write, which a watcher can only read as a timeout. A file never + // blocks; it is read back as a short tail after the run. + let err_file = std::fs::File::create(&err_path).map_err(|e| OcrError::Process(format!("creating the error file failed: {e}")))?; + // Known limit: only the direct child is killed at the timeout or on an early cap. A + // wrapper script's children (grandchildren of this process) may survive the kill; the + // cap and the timeout still bound what this run reads and how long it waits. + let mut child = + std::process::Command::new(&self.program) + .args(args) + .stdin(std::process::Stdio::null()) + .stdout(out_file) + .stderr(err_file) + .current_dir(&dir.path) + // A minimal environment: nothing from the document or the caller's shell beyond + // what an installed tesseract needs — its data prefix, the PATH (bare-name lookup + // and DLL loading), the Windows system root, and a temp dir it may write to. + .env_clear() + .envs(std::env::vars().filter(|(k, _)| { + k.eq_ignore_ascii_case("TESSDATA_PREFIX") || k.eq_ignore_ascii_case("PATH") || k.eq_ignore_ascii_case("SystemRoot") + })) + .env("TMP", &dir.path) + .env("TEMP", &dir.path) + .spawn() + .map_err(|e| OcrError::NoTesseract(format!("{} could not start: {e}", self.program.display())))?; + let started = Instant::now(); + let status = loop { + match child.try_wait() { + Ok(Some(status)) => break Ok(status), + Ok(None) => { + // The output caps are enforced while the child runs: a flood is killed as + // soon as a file passes its cap, not after it has finished writing. The + // read-back below caps what this run KEEPS of stderr; this cap bounds what + // the child can write there in the first place. + if std::fs::metadata(&out_path).map(|m| m.len()).unwrap_or(0) > max_output + || std::fs::metadata(&err_path).map(|m| m.len()).unwrap_or(0) > MAX_STDERR + { + let _ = child.kill(); + let _ = child.wait(); + break Err(OcrError::OutputTooLarge(max_output)); + } + if started.elapsed() >= timeout { + let _ = child.kill(); + let _ = child.wait(); + break Err(OcrError::Timeout(timeout.as_secs())); + } + std::thread::sleep(Duration::from_millis(20)); + } + Err(e) => { + let _ = child.kill(); + let _ = child.wait(); + break Err(OcrError::Process(format!("waiting for {} failed: {e}", self.program.display()))); + } + } + }; + let status = status?; + // Whatever the exit, the stderr tail names the cause (capped: error text stays a + // message, not a memory DoS). + let stderr = std::fs::File::open(&err_path) + .and_then(|mut f| { + use std::io::{Read, Seek, SeekFrom}; + let len = f.metadata()?.len().min(MAX_STDERR); + f.seek(SeekFrom::End(-(len as i64)))?; + let mut buf = Vec::with_capacity(len as usize); + f.take(len).read_to_end(&mut buf)?; + Ok::<_, std::io::Error>(buf) + }) + .unwrap_or_default(); + let stderr = String::from_utf8_lossy(&stderr).into_owned(); + // The message names only the end of it. + let tail = stderr.char_indices().rev().nth(2047).map_or(0, |(i, _)| i); + let note = stderr.get(tail..).unwrap_or_default().trim(); + if !status.success() { + return Err(OcrError::Process(format!( + "{} exited with {}: {}", + self.program.display(), + status.code().map(|c| c.to_string()).unwrap_or_else(|| "a signal".into()), + if note.is_empty() { "no error message" } else { note } + ))); + } + let len = std::fs::metadata(&out_path).map(|m| m.len()).unwrap_or(0); + if len > max_output { + return Err(OcrError::OutputTooLarge(max_output)); + } + let bytes = std::fs::read(&out_path).map_err(|e| OcrError::Process(format!("reading the output failed: {e}")))?; + Ok((String::from_utf8_lossy(&bytes).into_owned(), stderr)) + } + + /// Retry `--list-langs` once (an install may have appeared since the last try). + fn refresh(&self) -> Result<(), OcrError> { + let langs = self.list_langs()?; + *self.langs.lock().unwrap_or_else(|e| e.into_inner()) = langs; + self.probed.store(1, Ordering::Relaxed); + Ok(()) + } +} + +impl Recognizer for TesseractCli { + fn id(&self) -> &'static str { + "tesseract" + } + + /// Whether the program answers `--list-langs` (remembered; one process at most after the + /// first call). + fn available(&self) -> bool { + match self.probed.load(Ordering::Relaxed) { + 1 => true, + 2 => false, + _ => self.refresh().is_ok(), + } + } + + fn languages(&self) -> Vec<(String, String)> { + if self.probed.load(Ordering::Relaxed) == 2 { + return Vec::new(); + } + let langs = self.langs.lock().unwrap_or_else(|e| e.into_inner()); + if langs.is_empty() { + drop(langs); + return self.refresh().map_or(Vec::new(), |()| self.langs.lock().unwrap_or_else(|e| e.into_inner()).clone()); + } + langs.clone() + } + + /// Write the page as a PNG and read it back as TSV. + fn recognize(&self, image: &OcrImage, options: &RecognizeOptions) -> Result, OcrError> { + let lang = language_code(&options.language); + let png = encode_png(image)?; + let dir = TempDir::create()?; + let png_path = dir.path.join("page.png"); + { + let mut f = std::fs::File::create(&png_path).map_err(|e| OcrError::Process(format!("creating the image file failed: {e}")))?; + set_private(&mut f); + f.write_all(&png).map_err(|e| OcrError::Process(format!("writing the image failed: {e}")))?; + } + // Fixed argv, no shell: tesseract stdout -l tsv. + let tsv = self.run(&[&png_path.to_string_lossy(), "stdout", "-l", lang, "tsv"], self.timeout, self.max_output)?; + let (lines, truncated) = parse_tsv_bounded(&tsv); + if truncated { + return Err(OcrError::Recognize(format!("the {MAX_WORDS}-word limit was exceeded; the incomplete reading was rejected"))); + } + Ok(lines) + } +} + +/// Encode an RGBA raster as PNG (the interchange format Tesseract reads best). +fn encode_png(image: &OcrImage) -> Result, OcrError> { + let mut out = Vec::new(); + let mut enc = png::Encoder::new(&mut out, image.width, image.height); + enc.set_color(png::ColorType::Rgba); + enc.set_depth(png::BitDepth::Eight); + let mut writer = enc.write_header().map_err(|e| OcrError::Process(format!("encoding the page image failed: {e}")))?; + writer.write_image_data(&image.rgba).map_err(|e| OcrError::Process(format!("encoding the page image failed: {e}")))?; + writer.finish().map_err(|e| OcrError::Process(format!("encoding the page image failed: {e}")))?; + Ok(out) +} + +/// Make a just-created file owner-only, where the OS supports it (Unix). Windows relies on the +/// user's private %TEMP% ACLs; the fresh directory keeps other sessions' files out of sight. +fn set_private(f: &mut std::fs::File) { + #[cfg(unix)] + { + use std::os::unix::fs::PermissionsExt; + let _ = f.set_permissions(std::fs::Permissions::from_mode(0o600)); + } + #[cfg(not(unix))] + let _ = f; +} + +/// A fresh directory under the system temp dir, deleted (with its contents) when dropped. +/// On Unix it is created 0700, so the page image inside is private. +struct TempDir { + path: PathBuf, +} + +impl TempDir { + fn create() -> Result { + // The recognizer is shared across threads (`Arc`), so two `recognize` calls can run at + // once, and on a coarse clock they can read the same nanosecond. One counter value per + // attempt keeps the names of one process apart without trusting the clock. + static SEQ: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0); + let nanos = std::time::SystemTime::now().duration_since(std::time::UNIX_EPOCH).map(|d| d.as_nanos()).unwrap_or(0); + // Created owner-only in one step on Unix (no window in which the directory exists with + // the default, wider mode); `create` refuses a path that already exists. + let mut builder = std::fs::DirBuilder::new(); + #[cfg(unix)] + { + use std::os::unix::fs::DirBuilderExt; + builder.mode(0o700); + } + // Elsewhere the default mode stands (Windows relies on the user's private %TEMP% ACLs). + #[cfg(not(unix))] + let _ = &mut builder; + // A leftover directory of an earlier process with the same pid can still take a name; + // that one collision is retried with the next counter value, anything else is fatal. + for _ in 0..4 { + let seq = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed); + let path = std::env::temp_dir().join(format!("pdfcraft-ocr-{}-{nanos}-{seq}", std::process::id())); + match builder.create(&path) { + Ok(()) => return Ok(TempDir { path }), + Err(e) if e.kind() == std::io::ErrorKind::AlreadyExists => continue, + Err(e) => return Err(OcrError::Process(format!("creating the private temp directory failed: {e}"))), + } + } + Err(OcrError::Process("creating the private temp directory failed: every candidate name already exists".into())) + } +} + +impl Drop for TempDir { + fn drop(&mut self) { + let _ = std::fs::remove_dir_all(&self.path); + } +} + +#[cfg(test)] +mod tests { + use super::*; + // The Unix process tests below build fake scripts and name their paths. + #[cfg(unix)] + use std::path::Path; + + /// Several directories alive at once never share a name: the recognizer is shared across + /// threads (`Arc`), so two `recognize` calls can create their directories concurrently, + /// and the counter in the name — not the clock alone — tells them apart. + #[test] + fn temp_directories_alive_at_once_have_distinct_names() { + let dirs: Vec = (0..32).map(|_| TempDir::create().expect("create")).collect(); + let distinct: std::collections::BTreeSet = dirs.iter().map(|d| d.path.clone()).collect(); + assert_eq!(distinct.len(), dirs.len(), "{:?}", dirs.iter().map(|d| &d.path).collect::>()); + } + + /// A tiny image with one dark square (a synthetic "word"); the Unix process tests use it. + #[cfg(unix)] + fn blot(w: u32, h: u32, x0: u32, y0: u32, bw: u32, bh: u32) -> OcrImage { + let mut rgba = vec![255u8; (w * h * 4) as usize]; + for y in y0..(y0 + bh).min(h) { + for x in x0..(x0 + bw).min(w) { + let o = ((y * w + x) * 4) as usize; + rgba[o..o + 3].copy_from_slice(&[0, 0, 0]); + } + } + OcrImage::new(w, h, rgba).unwrap() + } + + /// The literal TSV fixture: two header rows (conf −1) and two lines of words. + const TSV: &str = "level\tpage\tblock\tpar\tline\tword\tleft\ttop\twidth\theight\tconf\ttext +1\t1\t0\t0\t0\t0\t0\t0\t100\t50\t-1\t +5\t1\t1\t1\t1\t1\t10\t10\t40\t12\t96.5\tHello +5\t1\t1\t1\t1\t2\t55\t10\t30\t12\t91.0\tworld +5\t1\t1\t2\t2\t1\t10\t30\t20\t12\t42.123\tsecond"; + + #[test] + fn tsv_parses_words_into_lines() { + let lines = parse_tsv(TSV); + assert_eq!(lines.len(), 2, "two (block, par, line) groups: {lines:?}"); + let first = &lines[0].words; + assert_eq!(first.len(), 2); + assert_eq!(first[0].text, "Hello"); + assert_eq!(first[0].rect, [10.0, 10.0, 50.0, 22.0], "left + width, top + height"); + assert_eq!(first[0].confidence, Some(96.5)); + assert_eq!(first[0].source, "tesseract"); + assert_eq!(first[1].text, "world"); + assert_eq!(lines[1].words.len(), 1, "a new par starts a new line"); + assert_eq!(lines[1].words[0].text, "second"); + assert_eq!(lines[1].words[0].confidence, Some(42.123)); + } + + /// Conf −1 rows and level ≠ 5 rows are structure, not words; blank text is dropped. + #[test] + fn tsv_treats_header_rows_and_blank_text_as_not_words() { + let tsv = "5\t1\t1\t1\t1\t1\t0\t0\t10\t10\t-1\tskipped\n\ + 4\t1\t1\t1\t1\t0\t0\t0\t10\t10\t95.0\tline row\n\ + 5\t1\t1\t1\t1\t1\t0\t0\t10\t10\t95.0\t \n\ + 5\t1\t1\t1\t1\t2\t0\t0\t10\t10\t88\tkept"; + let lines = parse_tsv(tsv); + assert_eq!(lines.len(), 1, "{lines:?}"); + assert_eq!(lines[0].words.len(), 1); + assert_eq!(lines[0].words[0].text, "kept"); + } + + /// Garbage rows (short, non-numeric) are skipped; CRLF endings survive; boxes saturate + /// instead of overflowing. + #[test] + fn tsv_skips_garbage_and_never_panics() { + let lines = parse_tsv("not a tsv row\n\n5\t1\t1\n5\t1\t1\t1\t1\t1\tx\ty\tw\th\tzzz\ttext\r\n"); + assert!(lines.is_empty(), "{lines:?}"); + let big = parse_tsv(&format!("5\t1\t1\t1\t1\t1\t{}\t0\t{}\t10\t50\tbig", u32::MAX, u32::MAX)); + let rect = big[0].words[0].rect; + assert!(rect[2].is_finite() && rect[3].is_finite(), "no overflow: {rect:?}"); + // A row whose text field is empty yields no words. + assert!(parse_tsv("5\t1\t1\t1\t1\t1\t1\t1\t2\t2\t50\t\t").is_empty()); + } + + /// Non-finite numbers and negative sizes are not words: a row with NaN/inf coordinates or + /// a negative width/height never reaches the output (and so never reaches the content + /// stream). `1e39` overflows f32 to infinity, so it is refused the same way. + #[test] + fn tsv_rejects_non_finite_and_negative_geometry() { + let row = |left: &str, top: &str, w: &str, h: &str, conf: &str| format!("5\t1\t1\t1\t1\t1\t{left}\t{top}\t{w}\t{h}\t{conf}\tword"); + for (why, r) in [ + ("NaN left", row("NaN", "0", "10", "10", "50")), + ("NaN width", row("0", "0", "NaN", "10", "50")), + ("NaN height", row("0", "0", "10", "NaN", "50")), + ("NaN confidence", row("0", "0", "10", "10", "NaN")), + ("inf left", row("inf", "0", "10", "10", "50")), + ("inf width", row("0", "0", "inf", "10", "50")), + ("1e39 width overflows to inf", row("0", "0", "1e39", "10", "50")), + ("negative width", row("0", "0", "-10", "10", "50")), + ("negative height", row("0", "0", "10", "-10", "50")), + ] { + let lines = parse_tsv(&r); + assert!(lines.is_empty(), "{why}: {lines:?}"); + } + // A sane row still parses. + assert_eq!(parse_tsv(&row("0", "0", "10", "10", "50"))[0].words[0].text, "word"); + } + + /// A reading longer than the word cap is truncated at [`MAX_WORDS`] (a known limit, not an + /// error): the words before the cap are kept, the rest is dropped. + #[test] + fn tsv_truncates_readings_past_the_word_cap() { + let mut tsv = String::new(); + for i in 0..MAX_WORDS + 5 { + tsv.push_str(&format!("5\t1\t1\t1\t1\t{}\t{}\t10\t40\t12\t96.5\tw{i}\n", i % 7 + 1, i % 500)); + } + let lines = parse_tsv(&tsv); + let count: usize = lines.iter().map(|l| l.words.len()).sum(); + assert_eq!(count, MAX_WORDS, "truncated at the cap"); + assert_eq!(lines.last().and_then(|l| l.words.last()).map(|w| w.text.as_str()), Some(&*format!("w{}", MAX_WORDS - 1))); + } + + #[test] + fn language_table_maps_and_falls_back() { + assert_eq!(language_code("en"), "eng"); + assert_eq!(language_code(" EN "), "eng", "trimmed, case-insensitive"); + assert_eq!(language_code("de"), "deu"); + assert_eq!(language_code("ZH"), "chi_sim"); + assert_eq!(language_code("ja"), "jpn"); + assert_eq!(language_code("ko"), "kor"); + assert_eq!(language_code("nl"), "nld"); + assert_eq!(language_code("klingon"), "eng", "unknown → eng"); + assert_eq!(language_code(""), "eng"); + // The security property: whatever the settings say, only allow-listed codes reach a + // process. + for code in ["en", "de", "fr", "es", "it", "pt", "ru", "zh", "ja", "ko", "ar", "nl", "klingon", "", " eng ", "EN/;rm"] { + let mapped = language_code(code); + assert!(LANGUAGE_CODES.iter().any(|(_, v)| *v == mapped), "{code:?} → {mapped:?} is not allow-listed"); + } + } + + #[test] + fn list_langs_come_from_stderr_when_stdout_has_none() { + let listing = "List of available languages in \"/usr/share/tessdata/\" (2):\neng\nosd\n"; + assert_eq!(list_langs_from(listing, ""), ["eng", "osd"]); + assert_eq!(list_langs_from("", listing), ["eng", "osd"], "3.x prints the list on stderr"); + assert_eq!(list_langs_from("", "Error opening data file\n"), Vec::::new(), "a warning is not a language"); + assert_eq!(list_langs_from("eng\n", "deu\n"), ["eng"], "stdout wins when it has the list"); + } + + /// The private temp directory is created owner-only (Unix) and removed on drop. + #[test] + fn temp_dir_is_private_and_removed() { + let dir = TempDir::create().unwrap(); + let path = dir.path.clone(); + assert!(path.is_dir()); + #[cfg(unix)] + { + use std::os::unix::fs::PermissionsExt; + assert_eq!(std::fs::metadata(&path).unwrap().permissions().mode() & 0o777, 0o700); + } + drop(dir); + assert!(!path.exists()); + } + + /// A set `PDFCRAFT_TESSERACT` decides alone: when it is invalid, [`TesseractCli::find`] does + /// not fall back to a PATH copy (which would run a program the user did not name); the PATH + /// search only runs when nothing is configured. + #[test] + fn an_invalid_configured_program_fails_closed() { + let Ok(real) = std::env::current_exe() else { return }; + let searched = std::cell::Cell::new(false); + let on_path = || { + searched.set(true); + Some(real.clone()) + }; + assert_eq!(locate(None, on_path), Some(real.clone()), "nothing configured: the PATH search decides"); + assert!(searched.get()); + searched.set(false); + assert_eq!(locate(Some(real.clone().into_os_string()), on_path), Some(real.clone()), "a valid location is used as is"); + assert!(!searched.get(), "and the PATH is not searched"); + let dir = real.parent().map(Path::to_path_buf).unwrap_or_default().into_os_string(); + for bad in [std::ffi::OsString::from("relative/tesseract"), std::ffi::OsString::from(""), dir, "/no/such/program-0123456789".into()] { + assert_eq!(locate(Some(bad.clone()), on_path), None, "{bad:?} is invalid, so there is no program"); + assert!(!searched.get(), "{bad:?}: an invalid location never falls back to the PATH"); + } + } + + #[test] + fn the_version_comes_from_stdout_or_stderr() { + assert_eq!(version_line("tesseract 5.4.0\n leptonica-1.84\n", ""), Some("tesseract 5.4.0".into())); + assert_eq!(version_line("", "\ntesseract 4.0.0\n"), Some("tesseract 4.0.0".into()), "older versions print it on stderr"); + assert_eq!(version_line(" \n", ""), None); + } + + #[test] + fn list_langs_output_parses() { + let out = "List of available languages in \"C:\\Program Files\\Tesseract-OCR\\tessdata/\" (2):\neng\nosd\n"; + assert_eq!(parse_list_langs(out), vec!["eng".to_string(), "osd".to_string()]); + assert_eq!(parse_list_langs("List of available languages (1):\neng\n"), vec!["eng".to_string()]); + assert!(parse_list_langs("").is_empty()); + } + + /// The real binary, when it is installed: a run end to end, with words that carry a + /// confidence and the engine's name (skipped where `tesseract` is missing). + #[test] + fn recognize_runs_the_installed_binary() { + let Some(t) = TesseractCli::find() else { + eprintln!("skipped: the tesseract program is not installed"); + return; + }; + assert_eq!(t.id(), "tesseract"); + assert!(t.available()); + // 300 dpi-ish block capitals, black on white. + let (w, h, rgba) = hello_world_image(); + let image = OcrImage::new(w, h, rgba).unwrap(); + let lines = t.recognize(&image, &RecognizeOptions::default()).unwrap(); + let text: Vec = lines.iter().map(Line::text).collect(); + assert!(text.iter().any(|t| t.to_uppercase().contains("HELL")), "{text:?}"); + let words: Vec<&Word> = lines.iter().flat_map(|l| &l.words).collect(); + assert!(words.iter().all(|w| w.source == "tesseract")); + assert!(words.iter().all(|w| matches!(w.confidence, Some(c) if (0.0..=100.0).contains(&c))), "{words:?}"); + } + + /// `available`/`languages` come from `--list-langs` (skipped without the binary). + #[test] + fn languages_come_from_list_langs() { + let Some(t) = TesseractCli::find() else { + eprintln!("skipped: the tesseract program is not installed"); + return; + }; + let langs = t.languages(); + assert!(!langs.is_empty()); + assert!(langs.iter().all(|(c, n)| !c.is_empty() && !n.is_empty()), "{langs:?}"); + } + + #[cfg(unix)] + #[test] + fn recognize_refuses_a_truncated_reading() { + let dir = TempDir::create().unwrap(); + let script = dir.path.join("synthetic-recognizer"); + let mut tsv = String::new(); + for i in 0..=MAX_WORDS { + use std::fmt::Write as _; + writeln!(tsv, "5\t1\t1\t1\t1\t{i}\t0\t0\t1\t1\t95\tword{i}").unwrap(); + } + let exact = tsv.lines().take(MAX_WORDS).collect::>().join("\n"); + std::fs::write(dir.path.join("reading.tsv"), tsv).unwrap(); + std::fs::write(&script, "#!/bin/sh\ncat \"$(dirname \"$0\")/reading.tsv\"\n").unwrap(); + make_executable(&script); + let recognizer = TesseractCli::new(script); + let result = recognizer.recognize(&blot(8, 8, 0, 0, 4, 4), &RecognizeOptions::default()); + let status = result.as_ref().map(|lines| lines.iter().map(|line| line.words.len()).sum::()).map_err(ToString::to_string); + assert!( + matches!(result, Err(OcrError::Recognize(ref reason)) if reason.contains("word limit")), + "a capped reading must be reported, never accepted as complete: {status:?}" + ); + std::fs::write(dir.path.join("reading.tsv"), exact).unwrap(); + let complete = recognizer.recognize(&blot(8, 8, 0, 0, 4, 4), &RecognizeOptions::default()).unwrap(); + assert_eq!(complete.iter().map(|line| line.words.len()).sum::(), MAX_WORDS); + } + + /// A hung binary is killed at the timeout and reported (Unix: a fake `tesseract` script; + /// this test needs a real process, so it is off the web and off Windows). + #[cfg(unix)] + #[test] + fn a_hung_program_is_killed_at_the_timeout() { + let dir = std::env::temp_dir().join(format!("pdfcraft-ocr-test-{}", std::process::id())); + let _ = std::fs::remove_dir_all(&dir); + std::fs::create_dir_all(&dir).unwrap(); + let script = dir.join("tesseract-hang"); + std::fs::write(&script, "#!/bin/sh\nsleep 30\n").unwrap(); + make_executable(&script); + let mut t = TesseractCli::new(script); + t.timeout = Duration::from_millis(500); + let started = Instant::now(); + let r = t.recognize(&blot(8, 8, 0, 0, 4, 4), &RecognizeOptions::default()); + assert!(matches!(r, Err(OcrError::Timeout(_))), "{r:?}"); + assert!(started.elapsed() < Duration::from_secs(10), "killed, not waited out"); + let _ = std::fs::remove_dir_all(&dir); + } + + /// Output past the cap kills the run EARLY, while the child is still writing (Unix only, + /// same fake-script trick): the script floods far past the cap and would sleep for 30 s + /// after; a cap checked only after the exit would come back as a timeout instead. + #[cfg(unix)] + #[test] + fn flooding_output_is_capped_and_killed_early() { + let dir = std::env::temp_dir().join(format!("pdfcraft-ocr-test-cap-{}", std::process::id())); + let _ = std::fs::remove_dir_all(&dir); + std::fs::create_dir_all(&dir).unwrap(); + let script = dir.join("tesseract-flood"); + std::fs::write(&script, "#!/bin/sh\nhead -c 100000000 /dev/zero | tr '\\0' 'x'\nsleep 30\n").unwrap(); + make_executable(&script); + let mut t = TesseractCli::new(script); + t.max_output = 1024; + t.timeout = Duration::from_secs(20); + let started = Instant::now(); + let r = t.recognize(&blot(8, 8, 0, 0, 4, 4), &RecognizeOptions::default()); + assert!(matches!(r, Err(OcrError::OutputTooLarge(_))), "{r:?}"); + assert!(started.elapsed() < Duration::from_secs(10), "killed early, at {:?}", started.elapsed()); + let _ = std::fs::remove_dir_all(&dir); + } + + /// An stderr flood does not block the child, so it does not cause a false timeout (Unix + /// only): the script writes megabytes of stderr — far more than a pipe would swallow — + /// then exits successfully with a small output, which must come back as a reading. + #[cfg(unix)] + #[test] + fn an_stderr_flood_does_not_cause_a_false_timeout() { + let dir = std::env::temp_dir().join(format!("pdfcraft-ocr-test-err-{}", std::process::id())); + let _ = std::fs::remove_dir_all(&dir); + std::fs::create_dir_all(&dir).unwrap(); + let script = dir.join("tesseract-err-flood"); + // (below MAX_STDERR, past which the child is stopped: `an_stderr_flood_is_killed_while_the_child_runs`) + std::fs::write(&script, "#!/bin/sh\nhead -c 200000 /dev/zero | tr '\\0' 'e' >&2\necho flooded away\n").unwrap(); + make_executable(&script); + let mut t = TesseractCli::new(script); + t.timeout = Duration::from_secs(10); + let r = t.recognize(&blot(8, 8, 0, 0, 4, 4), &RecognizeOptions::default()); + assert!(r.is_ok(), "stderr flood must not fail the run: {r:?}"); + let _ = std::fs::remove_dir_all(&dir); + } + + /// The child runs with a minimal environment (its data prefix, the PATH, the Windows + /// system root, and a temp dir), never the caller's whole environ. This is checked with a + /// script that prints its own environ as the "TSV" (Unix only). + #[cfg(unix)] + #[test] + fn the_child_gets_a_minimal_environment() { + let dir = std::env::temp_dir().join(format!("pdfcraft-ocr-test-env-{}", std::process::id())); + let _ = std::fs::remove_dir_all(&dir); + std::fs::create_dir_all(&dir).unwrap(); + let script = dir.join("tesseract-env"); + std::fs::write(&script, "#!/bin/sh\nenv | grep -v '^_='\n").unwrap(); + make_executable(&script); + let t = TesseractCli::new(script); + let png = encode_png(&blot(8, 8, 0, 0, 4, 4)).unwrap(); + std::fs::write(dir.join("page.png"), png).unwrap(); + let png_arg = dir.join("page.png").to_string_lossy().into_owned(); + let tsv = t.run(&[png_arg.as_str(), "stdout", "-l", "eng", "tsv"], t.timeout, t.max_output).unwrap(); + for (why, name) in [("the child's temp dir", "TMP="), ("temp", "TEMP="), ("lookup path", "PATH=")] { + assert!(tsv.contains(name), "{why} must be passed: {tsv}"); + } + // cargo always sets CARGO_* for the test process; env_clear must drop them. + assert!(!tsv.contains("CARGO_"), "the caller's environment must not leak: {tsv}"); + let _ = std::fs::remove_dir_all(&dir); + } + + #[cfg(unix)] + fn make_executable(p: &Path) { + use std::os::unix::fs::PermissionsExt; + let _ = std::fs::set_permissions(p, std::fs::Permissions::from_mode(0o755)); + } + + /// The temp directory and its contents are gone after a run (private files do not linger). + #[cfg(unix)] + #[test] + fn temp_files_are_deleted_and_private() { + let dir = TempDir::create().unwrap(); + assert!(dir.path.is_dir()); + std::fs::write(dir.path.join("x"), b"secret").unwrap(); + let p = dir.path.clone(); + drop(dir); + assert!(!p.exists(), "the temp directory is deleted after use"); + } + + /// A configured program location is only usable as an absolute path to an existing, + /// executable file: a relative path, a missing file and a directory are unavailable (and + /// [`TesseractCli::at`] refuses them without spawning anything). + #[test] + fn a_configured_program_must_be_an_absolute_executable_file() { + assert!(!program_is_executable(Path::new(".")), "relative"); + assert!(!program_is_executable(Path::new("definitely/missing/tesseract")), "missing"); + let temp = std::env::temp_dir(); + assert!(!program_is_executable(&temp), "a directory is not a program"); + let file = temp.join(format!("pdfcraft-ocr-not-a-program-{}", std::process::id())); + std::fs::write(&file, b"").unwrap(); + #[cfg(unix)] + { + use std::os::unix::fs::PermissionsExt; + let _ = std::fs::set_permissions(&file, std::fs::Permissions::from_mode(0o644)); + assert!(!program_is_executable(&file), "a file without executable bits is not a program"); + } + #[cfg(not(unix))] + assert!(program_is_executable(&file), "an existing file is usable on this platform"); + let _ = std::fs::remove_file(&file); + // `at` validates before any process starts, so these are None and not panics. + assert!(TesseractCli::at(PathBuf::from(".")).is_none()); + assert!(TesseractCli::at(temp).is_none()); + assert!(TesseractCli::at(PathBuf::from("definitely/missing/tesseract")).is_none()); + } + + /// The PATH search resolves through absolute entries only and skips relative ones: a + /// relative PATH entry would resolve against the caller's working directory. + #[test] + fn path_search_skips_relative_entries() { + let dir = std::env::temp_dir().join(format!("pdfcraft-ocr-path-{}", std::process::id())); + let _ = std::fs::remove_dir_all(&dir); + std::fs::create_dir_all(&dir).unwrap(); + let program = dir.join(program_file_names()[0]); + std::fs::write(&program, b"").unwrap(); + #[cfg(unix)] + { + use std::os::unix::fs::PermissionsExt; + let _ = std::fs::set_permissions(&program, std::fs::Permissions::from_mode(0o755)); + } + let found = path_search_in([PathBuf::from("relative/dir"), PathBuf::from("definitely/missing"), dir.clone()].into_iter()); + assert_eq!(found.as_deref(), Some(program.as_path()), "the absolute entry with the program wins, relative entries are skipped"); + let none = path_search_in([PathBuf::from("relative/dir")].into_iter()); + assert_eq!(none, None, "only relative entries: nothing is found"); + let _ = std::fs::remove_dir_all(&dir); + } + + /// "HELLO" as blocky 5×7 glyphs, 8 pixels per dot with wide tracking, at roughly 300 dpi + /// scale (each dot ~10 px): big enough for Tesseract's word detector. + fn hello_world_image() -> (u32, u32, Vec) { + const GLYPHS: &[&str] = &[ + "X...X", "X...X", "X...X", "XXXXX", "X...X", "X...X", "X...X", // H + ]; + // Draw the five glyphs of HELLO with per-letter shapes for E, L, L, O. + const E: &[&str] = &["XXXXX", "X....", "X....", "XXXX.", "X....", "X....", "XXXXX"]; + const L: &[&str] = &["X....", "X....", "X....", "X....", "X....", "X....", "XXXXX"]; + const O: &[&str] = &[".XXX.", "X...X", "X...X", "X...X", "X...X", "X...X", ".XXX."]; + let letters: [&[&str]; 5] = [GLYPHS, E, L, L, O]; + let (dot, x0, y0, advance) = (8u32, 40u32, 40u32, 7u32 * 8); + let w = x0 * 2 + advance * 5; + let h = y0 * 2 + 7 * dot; + let mut rgba = vec![255u8; (w * h * 4) as usize]; + for (i, rows) in letters.iter().enumerate() { + for (ry, row) in rows.iter().enumerate() { + for (rx, b) in row.bytes().enumerate() { + if b != b'X' { + continue; + } + for dy in 0..dot { + for dx in 0..dot { + let x = x0 + i as u32 * advance + rx as u32 * dot + dx; + let y = y0 + ry as u32 * dot + dy; + let o = ((y * w + x) * 4) as usize; + rgba[o..o + 3].copy_from_slice(&[0, 0, 0]); + } + } + } + } + } + (w, h, rgba) + } +} + +#[cfg(all(test, unix))] +mod process_tests { + use super::*; + + /// A stderr flood is killed at the cap while the child still runs, not accepted until it + /// exits (an endless writer would otherwise run out the whole timeout and fill the disk). + /// The fixture needs a Unix shell, like the other process fixtures. + #[test] + fn a_stderr_flood_is_killed_while_the_child_runs() { + let dir = std::env::temp_dir().join(format!("pdfcraft-tesseract-flood-{}", std::process::id())); + std::fs::create_dir_all(&dir).unwrap(); + let program = dir.join("flood-tesseract"); + std::fs::write(&program, "#!/bin/sh\nwhile true; do echo aaaaaaaaaaaaaaaaaaaaaaaa >&2; done\n").unwrap(); + use std::os::unix::fs::PermissionsExt; + let mut perms = std::fs::metadata(&program).unwrap().permissions(); + perms.set_mode(0o755); + std::fs::set_permissions(&program, perms).unwrap(); + let t = TesseractCli::new(program); + let started = std::time::Instant::now(); + let err = t.run_capture(&["--list-langs"], Duration::from_secs(60), 1024 * 1024).unwrap_err(); + assert!(matches!(err, OcrError::OutputTooLarge(_)), "{err}"); + assert!(started.elapsed() < Duration::from_secs(30), "killed at the cap, not at the timeout"); + let _ = std::fs::remove_dir_all(&dir); + } +} diff --git a/crates/optimize/src/images.rs b/crates/optimize/src/images.rs index 41f893c71b..159ba04e8c 100644 --- a/crates/optimize/src/images.rs +++ b/crates/optimize/src/images.rs @@ -9,7 +9,7 @@ use pdfcraft_cos::{DecodeBudget, Dict, Document, ImageBudget, ObjRef, Object, St use crate::{Compression, ImageSettings, MonoCompression, MonoSettings, OptimizeError, Report, Settings, Stage}; /// Images with more pixels than this are left alone (memory). -const MAX_PIXELS: u64 = 80_000_000; +pub(crate) const MAX_PIXELS: u64 = 80_000_000; /// The resources of a page, inherited from the page tree if the page has none. pub(crate) fn page_resources(doc: &Document, page: ObjRef) -> Option { @@ -419,7 +419,7 @@ pub(crate) fn components(doc: &Document, cs: &Object) -> Option { } } -fn is_dct(s: &Stream) -> Option { +pub(crate) fn is_dct(s: &Stream) -> Option { match s.dict.get(b"Filter") { None => Some(false), Some(Object::Name(n)) => match n.as_slice() { @@ -506,7 +506,7 @@ pub(crate) fn pixels(doc: &Document, s: &Stream, n: usize, budget: &mut ImageBud /// to three bytes for each of them and of a PNG predictor's bytes a row (ASCIIHex text under /// Flate, its two digits a byte broken into lines or spaced). More can only be a decompression /// bomb, and the image is left alone. -fn allowance(bytes: usize, rows: u32) -> Option { +pub(crate) fn allowance(bytes: usize, rows: u32) -> Option { bytes.checked_add(usize::try_from(rows).ok()?)?.checked_mul(3)?.checked_add(4096) } @@ -518,7 +518,7 @@ fn allowance(bytes: usize, rows: u32) -> Option { /// /ColorTransform contradict each other (PdfCraft's renderer lets /ColorTransform 0 win where /// the standard lets the marker win), either holds a value other than 0 or 1, or the parameters /// or the JPEG's segments can't be read. -fn ycc_encoded(doc: &Document, s: &Stream) -> Option { +pub(crate) fn ycc_encoded(doc: &Document, s: &Stream) -> Option { let entry = color_transform(doc, s)?; match adobe_transform(&s.raw)? { Some(t @ (0 | 1)) => entry.is_none_or(|c| c == i64::from(t)).then_some(t == 1), @@ -658,7 +658,7 @@ fn scaled + 'static>(img: &image::ImageBuffer Option> { +pub(crate) fn jpeg(px: &[u8], n: usize, w: u32, h: u32, quality: u8) -> Option> { let mut out = Vec::new(); let mut enc = image::codecs::jpeg::JpegEncoder::new_with_quality(&mut out, quality.clamp(1, 100)); let ty = if n == 1 { image::ExtendedColorType::L8 } else { image::ExtendedColorType::Rgb8 }; @@ -681,7 +681,7 @@ pub(crate) fn decodable(doc: &Document, s: &Stream) -> Option { /// Whether `decode` is the default decode array for `n` components ([0 1] each), which changes /// nothing. Scanners write it on page images (#490). -fn is_identity_decode(doc: &Document, decode: &Object, n: usize) -> bool { +pub(crate) fn is_identity_decode(doc: &Document, decode: &Object, n: usize) -> bool { let decode = doc.resolve(decode); let Some(a) = decode.as_array() else { return false }; a.len() == 2 * n && a.chunks(2).all(|p| matches!(p, [lo, hi] if doc.resolve(lo).as_f64() == Some(0.0) && doc.resolve(hi).as_f64() == Some(1.0))) diff --git a/crates/optimize/src/lib.rs b/crates/optimize/src/lib.rs index e70aa5e71d..ef6423ab7d 100644 --- a/crates/optimize/src/lib.rs +++ b/crates/optimize/src/lib.rs @@ -32,6 +32,7 @@ mod audit; pub mod fragments; mod images; mod links; +mod samples; mod unused; pub use audit::{SpaceCategory, SpaceUse, audit_space}; @@ -40,6 +41,7 @@ use pdfcraft_cos::{Dict, Document, ImageBudget, MAX_PAGE_DECODED, ObjRef, Object use pdfcraft_filters::{Filter, Params}; pub use images::effective_resolutions; +pub use samples::{NotDecoded, Samples, decode_samples, encode_samples}; #[derive(Debug, thiserror::Error)] pub enum OptimizeError { diff --git a/crates/optimize/src/samples.rs b/crates/optimize/src/samples.rs new file mode 100644 index 0000000000..4800da6fd7 --- /dev/null +++ b/crates/optimize/src/samples.rs @@ -0,0 +1,312 @@ +//! An image's samples read out and written back at the same size, for tools that change the +//! pixels of an image in place (Scan & OCR ▸ Enhance ▸ Scanned document). The image dictionary +//! stays as it was apart from the encoding: its size, colour space, masks, `/Interpolate`, +//! `/Intent`, `/Metadata` and every other entry are kept. + +use pdfcraft_cos::{Document, ImageBudget, Object, Stream}; + +use crate::images::{MAX_PIXELS, allowance, components, is_dct, is_identity_decode, jpeg, pixels, ycc_encoded}; + +/// An image's samples, one byte each, row-major. +#[derive(Clone, Debug, PartialEq)] +pub struct Samples { + pub width: u32, + pub height: u32, + /// 1 (grey) or 3 (RGB). + pub components: usize, + /// The samples. A bi-level image's are 0 (black) or 255 (white) as drawn, whatever its + /// decode array. + pub data: Vec, + /// Stored as a JPEG (written back as one); otherwise written back losslessly (Flate). + pub jpeg: bool, + /// One bit per pixel (written back as one). + pub bilevel: bool, + /// A bi-level image whose 1 bits are black (decode array `[1 0]`): written back the same way, + /// so its decode array stays true. + pub inverted: bool, +} + +/// Why an image's samples were not read. Each says what the image is, so a report can name it. +#[derive(Clone, Copy, Debug, PartialEq, Eq, thiserror::Error)] +pub enum NotDecoded { + #[error("it is a stencil mask, not a picture")] + StencilMask, + #[error("its colour space is not grey or RGB (CMYK, indexed and special colours are not supported yet)")] + ColourSpace, + #[error("it has {0} bits per sample (8, and 1 for black and white, are supported)")] + BitDepth(i64), + #[error("its decode array remaps the samples")] + Decode, + #[error("it is compressed in a format PdfCraft cannot read back yet (JPEG 2000, JBIG2 or CCITT fax)")] + Filter, + #[error("its JPEG colour transform is ambiguous, so readers may disagree about its colours")] + Ambiguous, + #[error("it is larger than 80 million pixels")] + TooLarge, + #[error("its data could not be decoded, or is shorter than its size says")] + Corrupt, +} + +/// Read the samples of image stream `s` (a dictionary with `/Subtype /Image`), decoding it within +/// `budget` (an operation over many images shares one). +pub fn decode_samples(doc: &Document, s: &Stream, budget: &mut ImageBudget) -> Result { + let d = &s.dict; + if matches!(d.get(b"ImageMask").map(|v| doc.resolve(v)).as_deref(), Some(Object::Bool(true))) { + return Err(NotDecoded::StencilMask); + } + // The filter first: a JPEG 2000 image may leave its colour space and bit depth to the data, + // and the report should name the format, not the missing entries. + let jpeg = is_dct(s).ok_or(NotDecoded::Filter)?; + let n = d.get(b"ColorSpace").and_then(|cs| components(doc, cs)).ok_or(NotDecoded::ColourSpace)?; + let bpc = d.get(b"BitsPerComponent").map(|v| doc.resolve(v)).and_then(|v| v.as_int()).unwrap_or(0); + let bilevel = match bpc { + 8 => false, + 1 if n == 1 => true, + other => return Err(NotDecoded::BitDepth(other)), + }; + let width = d.get(b"Width").map(|v| doc.resolve(v)).and_then(|v| v.as_int()).and_then(|v| u32::try_from(v).ok()).unwrap_or(0); + let height = d.get(b"Height").map(|v| doc.resolve(v)).and_then(|v| v.as_int()).and_then(|v| u32::try_from(v).ok()).unwrap_or(0); + if width == 0 || height == 0 { + return Err(NotDecoded::Corrupt); + } + if u64::from(width) * u64::from(height) > MAX_PIXELS { + return Err(NotDecoded::TooLarge); + } + if bilevel { + if jpeg { + // A JPEG has 8 bits per sample: the dictionary contradicts the data. + return Err(NotDecoded::Corrupt); + } + let inverted = match d.get(b"Decode") { + None => false, + Some(a) if is_identity_decode(doc, a, 1) => false, + Some(a) if is_inverted_decode(doc, a) => true, + Some(_) => return Err(NotDecoded::Decode), + }; + let data = unpack_bits(doc, s, width, height, inverted, budget).ok_or(NotDecoded::Corrupt)?; + return Ok(Samples { width, height, components: 1, data, jpeg: false, bilevel: true, inverted }); + } + if d.get(b"Decode").is_some_and(|a| !is_identity_decode(doc, a, n)) { + return Err(NotDecoded::Decode); + } + if jpeg && n == 3 && ycc_encoded(doc, s).is_none() { + return Err(NotDecoded::Ambiguous); + } + let data = pixels(doc, s, n, budget).ok_or(NotDecoded::Corrupt)?; + Ok(Samples { width, height, components: n, data, jpeg, bilevel: false, inverted: false }) +} + +/// `[1 0]`: a bi-level image whose 1 bits are black. +fn is_inverted_decode(doc: &Document, decode: &Object) -> bool { + let decode = doc.resolve(decode); + matches!(decode.as_array().map(Vec::as_slice), Some([a, b]) if doc.resolve(a).as_f64() == Some(1.0) && doc.resolve(b).as_f64() == Some(0.0)) +} + +/// The bytes a `w` × `h` bi-level image takes: each row padded to whole bytes. +fn packed_len(w: u32, h: u32) -> Option { + usize::try_from(w).ok()?.div_ceil(8).checked_mul(usize::try_from(h).ok()?) +} + +/// A bi-level image's pixels, 0 (black) or 255 (white) as drawn, decoded within `budget` with +/// its `/DecodeParms` resolved, as the 8-bit images are. +fn unpack_bits(doc: &Document, s: &Stream, w: u32, h: u32, inverted: bool, budget: &mut ImageBudget) -> Option> { + let stride = usize::try_from(w).ok()?.div_ceil(8); + let need = packed_len(w, h)?; + let bits = budget.decode_filters(doc, s, allowance(need, h)?).ok()?; + let bits = bits.get(..need)?; + let (w, h) = (usize::try_from(w).ok()?, usize::try_from(h).ok()?); + let mut out = Vec::with_capacity(w.checked_mul(h)?); + for row in bits.chunks(stride).take(h) { + for x in 0..w { + let byte = row.get(x / 8).copied().unwrap_or(0); + let one = byte & (0x80 >> (x % 8)) != 0; + // Without a decode array 0 is black and 1 white; [1 0] swaps them. + out.push(if one != inverted { 255 } else { 0 }); + } + } + Some(out) +} + +/// The pixels (0 black … 255 white, threshold at half) packed back to bits. +fn pack_bits(px: &[u8], w: u32, h: u32, inverted: bool) -> Option> { + let stride = usize::try_from(w).ok()?.div_ceil(8); + let mut out = vec![0u8; packed_len(w, h)?]; + let (w, h) = (usize::try_from(w).ok()?, usize::try_from(h).ok()?); + if px.len() != w.checked_mul(h)? { + return None; + } + for (y, row) in px.chunks(w.max(1)).enumerate() { + for (x, &v) in row.iter().enumerate() { + let white = v >= 128; + if white != inverted + && let Some(b) = out.get_mut(y * stride + x / 8) + { + *b |= 0x80 >> (x % 8); + } + } + } + Some(out) +} + +/// Image stream `old` with its samples replaced by `samples` (same size and components): a JPEG +/// stays a JPEG (at `quality`, 1–100), anything else becomes Flate, and a bi-level image stays +/// one bit per pixel with its decode array. Filter parameters go (the new data needs none); every +/// other entry stays. `None` if the samples don't fit the image. +pub fn encode_samples(old: &Stream, samples: &Samples, quality: u8) -> Option { + let Samples { width, height, components: n, data, jpeg: as_jpeg, bilevel, inverted } = samples; + let (w, h, n) = (*width, *height, *n); + if !(n == 1 || n == 3) || data.len() != usize::try_from(w).ok()?.checked_mul(usize::try_from(h).ok()?)?.checked_mul(n)? { + return None; + } + let mut d = old.dict.clone(); + for k in [&b"Filter"[..], b"DecodeParms", b"Length", b"DL"] { + d.remove(k); + } + if *bilevel { + if n != 1 { + return None; + } + d.set(b"BitsPerComponent".to_vec(), Object::Int(1)); + return Some(Stream::flate(d, &pack_bits(data, w, h, *inverted)?)); + } + d.set(b"BitsPerComponent".to_vec(), Object::Int(8)); + if *as_jpeg { + let bytes = jpeg(data, n, w, h, quality)?; + d.set(b"Filter".to_vec(), Object::name("DCTDecode")); + Some(Stream::from_raw(d, bytes)) + } else { + Some(Stream::flate(d, data)) + } +} + +#[cfg(test)] +mod tests { + use super::*; + use pdfcraft_cos::Dict; + + fn image(dict: &[(&str, Object)], raw: Vec) -> Stream { + let mut d = Dict::new(); + d.set(b"Type".to_vec(), Object::name("XObject")); + d.set(b"Subtype".to_vec(), Object::name("Image")); + for (k, v) in dict { + d.set(k.as_bytes().to_vec(), v.clone()); + } + Stream::from_raw(d, raw) + } + + fn doc() -> Document { + let pdf = b"%PDF-1.7\n1 0 obj\n<< /Type /Catalog /Pages 2 0 R >>\nendobj\n2 0 obj\n<< /Type /Pages /Kids [] /Count 0 >>\nendobj\ntrailer\n<< /Root 1 0 R >>\n%%EOF\n"; + Document::open(std::sync::Arc::new(pdf.to_vec())).unwrap() + } + + #[test] + fn grey_samples_round_trip_losslessly_and_keep_the_other_entries() { + let px: Vec = (0..12 * 5).map(|i| (i * 4) as u8).collect(); + let s = image( + &[ + ("Width", Object::Int(12)), + ("Height", Object::Int(5)), + ("ColorSpace", Object::name("DeviceGray")), + ("BitsPerComponent", Object::Int(8)), + ("Interpolate", Object::Bool(true)), + ], + px.clone(), + ); + let got = decode_samples(&doc(), &s, &mut ImageBudget::with_work(u64::MAX)).unwrap(); + assert_eq!((got.width, got.height, got.components, got.jpeg, got.bilevel), (12, 5, 1, false, false)); + assert_eq!(got.data, px); + let back = encode_samples(&s, &got, 90).unwrap(); + assert_eq!(back.dict.name(b"Filter"), Some(&b"FlateDecode"[..])); + assert_eq!(back.dict.get(b"Interpolate"), Some(&Object::Bool(true))); + assert_eq!(decode_samples(&doc(), &back, &mut ImageBudget::with_work(u64::MAX)).unwrap().data, px); + } + + #[test] + fn bilevel_samples_keep_their_decode_array_and_bits() { + // 10 × 2: first row black on the left half, second row all white; [1 0] makes 1 black. + let raw = vec![0b1111_1000, 0b0000_0000, 0, 0]; + let s = image( + &[ + ("Width", Object::Int(10)), + ("Height", Object::Int(2)), + ("ColorSpace", Object::name("DeviceGray")), + ("BitsPerComponent", Object::Int(1)), + ("Decode", Object::Array(vec![Object::Int(1), Object::Int(0)])), + ], + raw.clone(), + ); + let got = decode_samples(&doc(), &s, &mut ImageBudget::with_work(u64::MAX)).unwrap(); + assert!(got.bilevel); + assert_eq!(&got.data[..10], &[0, 0, 0, 0, 0, 255, 255, 255, 255, 255]); + assert!(got.data[10..].iter().all(|&v| v == 255)); + assert!(got.inverted); + let back = encode_samples(&s, &got, 90).unwrap(); + assert_eq!(back.dict.int(b"BitsPerComponent"), Some(1)); + assert!(back.dict.get(b"Decode").is_some()); + assert_eq!(back.decoded().unwrap(), raw); + } + + #[test] + fn bilevel_decode_parms_by_reference_decode_like_direct_ones() { + // 10 × 2 with a TIFF predictor: each 1 bit flips the pixel from the one to its left. + let packed = vec![0b1000_0000, 0b0000_0000, 0b0100_0000, 0b0000_0000]; + let parms = || { + let mut p = pdfcraft_cos::Dict::new(); + for (k, v) in [("Predictor", 2), ("Colors", 1), ("BitsPerComponent", 1), ("Columns", 10)] { + p.set(k.as_bytes().to_vec(), Object::Int(v)); + } + Object::Dict(p) + }; + let dict = [ + ("Width", Object::Int(10)), + ("Height", Object::Int(2)), + ("ColorSpace", Object::name("DeviceGray")), + ("BitsPerComponent", Object::Int(1)), + ]; + let mut d = doc(); + let mut direct = Stream::flate(image(&dict, vec![]).dict, &packed); + direct.dict.set(b"DecodeParms".to_vec(), parms()); + let mut indirect = direct.clone(); + let r = d.add(parms()); + indirect.dict.set(b"DecodeParms".to_vec(), Object::Ref(r)); + let a = decode_samples(&d, &direct, &mut ImageBudget::with_work(u64::MAX)).unwrap(); + let b = decode_samples(&d, &indirect, &mut ImageBudget::with_work(u64::MAX)).unwrap(); + // Row 1: white from x = 0 on; row 2: black at x = 0, white from x = 1 on. + assert_eq!(&a.data[..10], &[255; 10]); + assert_eq!(&a.data[10..], &[0, 255, 255, 255, 255, 255, 255, 255, 255, 255]); + assert_eq!(b, a); + } + + #[test] + fn what_is_not_supported_says_why_and_never_panics() { + let base = |extra: &[(&str, Object)], raw: Vec| { + let mut d: Vec<(&str, Object)> = vec![ + ("Width", Object::Int(4)), + ("Height", Object::Int(4)), + ("ColorSpace", Object::name("DeviceGray")), + ("BitsPerComponent", Object::Int(8)), + ]; + d.extend(extra.iter().cloned()); + decode_samples(&doc(), &image(&d, raw), &mut ImageBudget::with_work(u64::MAX)) + }; + assert_eq!(base(&[("ColorSpace", Object::name("DeviceCMYK"))], vec![0; 64]), Err(NotDecoded::ColourSpace)); + assert_eq!(base(&[("BitsPerComponent", Object::Int(4))], vec![0; 8]), Err(NotDecoded::BitDepth(4))); + assert_eq!(base(&[("Filter", Object::name("JBIG2Decode"))], vec![0; 8]), Err(NotDecoded::Filter)); + // JPEG 2000 may leave the colour space and bit depth to its data: the format is the reason. + let jpx = |dict: &[(&str, Object)]| decode_samples(&doc(), &image(dict, vec![0; 8]), &mut ImageBudget::with_work(u64::MAX)); + let size = [("Width", Object::Int(4)), ("Height", Object::Int(4)), ("Filter", Object::name("JPXDecode"))]; + assert_eq!(jpx(&size), Err(NotDecoded::Filter)); + assert_eq!(jpx(&[size[0].clone(), size[1].clone(), size[2].clone(), ("ColorSpace", Object::name("DeviceRGB"))]), Err(NotDecoded::Filter)); + assert_eq!(base(&[("ImageMask", Object::Bool(true))], vec![0; 8]), Err(NotDecoded::StencilMask)); + assert_eq!(base(&[("Decode", Object::Array(vec![Object::Int(1), Object::Int(0)]))], vec![0; 16]), Err(NotDecoded::Decode)); + // Too short, a huge size, a broken JPEG, no size at all. + assert_eq!(base(&[], vec![0; 5]), Err(NotDecoded::Corrupt)); + assert_eq!(base(&[("Width", Object::Int(1 << 30)), ("Height", Object::Int(1 << 30))], vec![0; 16]), Err(NotDecoded::TooLarge)); + assert_eq!(base(&[("Filter", Object::name("DCTDecode"))], vec![0xFF, 0xD8, 1, 2, 3]), Err(NotDecoded::Corrupt)); + assert_eq!(base(&[("Width", Object::Int(-3))], vec![0; 16]), Err(NotDecoded::Corrupt)); + // Samples that don't fit the image are refused when written back. + let s = image(&[("Width", Object::Int(4)), ("Height", Object::Int(4))], vec![0; 16]); + let wrong = Samples { width: 4, height: 4, components: 1, data: vec![0; 15], jpeg: false, bilevel: false, inverted: false }; + assert!(encode_samples(&s, &wrong, 90).is_none()); + } +} diff --git a/crates/organize/src/lib.rs b/crates/organize/src/lib.rs index b890965d0c..3b3d56cd83 100644 --- a/crates/organize/src/lib.rs +++ b/crates/organize/src/lib.rs @@ -20,6 +20,7 @@ mod dedupe; mod import; mod labels; mod outline; +mod outline_io; mod pdfx; mod prune; mod transition; @@ -40,6 +41,7 @@ pub use outline::{ Bookmark, BookmarkPage, OutlineEntry, OutlineError, add_bookmark, add_bookmark_at, add_bookmark_tree, bookmark_page, bookmarks, delete_bookmark, move_bookmark, rename_bookmark, set_bookmark_color, set_bookmark_open, set_bookmark_page, set_bookmark_style, top_level_bookmarks, }; +pub use outline_io::{MAX_IMPORT_BOOKMARKS, OutlineNode, OutlineTree, import_outline, outline_tree, unmapped, view_arity}; pub use transition::{TRANSITION_DIRECTIONS, TRANSITION_STYLES, Transition, TransitionSpec, set_transition}; pub use view::{InitialView, displays_doc_title, initial_view, set_initial_view}; diff --git a/crates/organize/src/outline.rs b/crates/organize/src/outline.rs index ff7d3d8da7..13c02eb662 100644 --- a/crates/organize/src/outline.rs +++ b/crates/organize/src/outline.rs @@ -19,7 +19,7 @@ use crate::{OrganizeError, walk}; /// Deepest level a listing enters (the top level is 0). Bookmarks deeper than this are not /// listed; edits still reach them by path. -const LIST_DEPTH: usize = 32; +pub(crate) const LIST_DEPTH: usize = 32; /// Deepest name-tree node read when looking up named destinations. const NAME_TREE_DEPTH: usize = 32; @@ -72,6 +72,14 @@ pub enum OutlineError { BadColor, #[error("this document has no tagged headings to make bookmarks from")] NoEntries, + #[error("there are no bookmarks to add")] + NothingToImport, + #[error("the bookmark at {0:?} has a view that is not a known destination type with its values")] + BadView(Vec), + #[error("bookmarks may nest at most {0} levels deep")] + TooDeep(usize), + #[error("at most {0} bookmarks can be added at once")] + TooManyBookmarks(usize), #[error("{0}")] Organize(#[from] OrganizeError), #[error("{0}")] @@ -449,7 +457,7 @@ impl From for OrganizeError { /// Make `kids` the children of `parent`, in order: rewires `/Parent`, `/Prev`, `/Next`, /// `/First` and `/Last`. -fn relink(doc: &mut Document, parent: ObjRef, kids: &[ObjRef]) -> Result<()> { +pub(crate) fn relink(doc: &mut Document, parent: ObjRef, kids: &[ObjRef]) -> Result<()> { for (i, k) in kids.iter().enumerate() { put(doc, *k, b"Parent", Some(Object::Ref(parent)))?; put(doc, *k, b"Prev", i.checked_sub(1).map(|p| Object::Ref(kids[p])))?; @@ -473,7 +481,7 @@ struct Counting { /// Recompute every `/Count` (§12.3.3: open items count their visible descendants, closed items /// the negative of what opening them would show; the root counts all visible items). -fn recount(doc: &mut Document) -> Result<()> { +pub(crate) fn recount(doc: &mut Document) -> Result<()> { let Some(root) = outline_root(doc) else { return Ok(()) }; let mut seen = HashSet::from([root]); let kids = children_of(doc, root, &mut seen); @@ -714,7 +722,7 @@ pub fn move_bookmark(doc: &mut Document, from: &[usize], to_parent: &[usize], in /// Open `parent` so a bookmark just added or moved into it shows (the root is always open). /// Only the sign of `/Count` matters here: [`recount`] sets its value. -fn show_children(doc: &mut Document, parent: ObjRef, parent_path: &[usize]) -> Result<()> { +pub(crate) fn show_children(doc: &mut Document, parent: ObjRef, parent_path: &[usize]) -> Result<()> { if !parent_path.is_empty() && doc.get(parent).as_dict().and_then(|d| d.int(b"Count")).is_none_or(|c| c <= 0) { put(doc, parent, b"Count", Some(Object::Int(1)))?; } diff --git a/crates/organize/src/outline_io.rs b/crates/organize/src/outline_io.rs new file mode 100644 index 0000000000..1fc3afc4f3 --- /dev/null +++ b/crates/organize/src/outline_io.rs @@ -0,0 +1,261 @@ +//! Moving a whole outline (or one bookmark's subtree) out of a document and into another +//! (ISO 32000-2 §12.3.3 outlines, §12.3.2.2 explicit destinations). +//! +//! [`outline_tree`] reads bookmarks as a portable tree of [`OutlineNode`]s: titles, nesting, open +//! state, colour, style and destinations as page index plus view in the file's own user space, +//! so writing a tree back with [`import_outline`] gives the same structure. A node whose page the +//! target lacks keeps its place in the tree without a destination, and [`unmapped`] names it, so +//! nothing is dropped silently. Not carried: actions other than a GoTo, named destinations (they +//! travel resolved to their page), and structure links (`/SE`). + +use pdfcraft_cos::{Dict, Document, ObjRef, Object, PdfString}; + +use crate::outline::{LIST_DEPTH, children, container, recount, relink, show_children}; +use crate::{OrganizeError, OutlineError, walk}; + +/// Most bookmarks one import writes. +pub const MAX_IMPORT_BOOKMARKS: usize = 100_000; + +/// The destination types (Table 149) and how many numbers follow each. +const VIEWS: [(&[u8], usize); 8] = [(b"XYZ", 3), (b"Fit", 0), (b"FitH", 1), (b"FitV", 1), (b"FitR", 4), (b"FitB", 0), (b"FitBH", 1), (b"FitBV", 1)]; + +/// One bookmark and the bookmarks under it, independent of any document. +#[derive(Clone, Debug, PartialEq)] +pub struct OutlineNode { + pub title: String, + /// The 0-based page it goes to; `None` for a bookmark that goes nowhere in its document. + pub page: Option, + /// Where on the page: a destination type from Table 149 and exactly as many values as that + /// type takes, in PDF user space (`None` keeps the reader's value). `None` goes to the page at + /// the reader's position and zoom. + pub view: Option<(Vec, Vec>)>, + /// The title's colour (`/C`), each component from 0 to 1; `None` is the default black. + pub color: Option<[f32; 3]>, + pub italic: bool, + pub bold: bool, + /// Shown expanded, when it has children. + pub open: bool, + pub children: Vec, +} + +/// The tree [`outline_tree`] read. +#[derive(Clone, Debug, PartialEq)] +pub struct OutlineTree { + pub nodes: Vec, + /// Bookmarks nested deeper than the listing reaches (32 levels) were left out. + pub truncated: bool, +} + +/// The bookmark at `path` with everything under it, or the whole outline for `[]`. +pub fn outline_tree(doc: &Document, path: &[usize]) -> Result { + // A whole outline's top level sits at depth 0; a subtree's root is the bookmark at `path`. + let base = path.len().max(1); + let mut truncated = false; + // The open nodes, outermost first, each with its depth. + let mut open: Vec<(usize, OutlineNode)> = Vec::new(); + let mut roots = Vec::new(); + for b in crate::bookmarks(doc).into_iter().filter(|b| b.path.starts_with(path) && !b.path.is_empty()) { + let depth = b.path.len().saturating_sub(base); + if b.path.len() > LIST_DEPTH && has_children(doc, b.obj) { + truncated = true; + } + while open.last().is_some_and(|(d, _)| *d >= depth) { + close(&mut open, &mut roots); + } + let view = b.view.and_then(|(kind, values)| portable_view(kind, values)); + let node = + OutlineNode { title: b.title, page: b.page, view, color: b.color, italic: b.italic, bold: b.bold, open: b.open, children: Vec::new() }; + open.push((depth, node)); + } + while !open.is_empty() { + close(&mut open, &mut roots); + } + if !path.is_empty() && roots.is_empty() { + return Err(OutlineError::NoSuchBookmark(path.to_vec())); + } + Ok(OutlineTree { nodes: roots, truncated }) +} + +fn close(open: &mut Vec<(usize, OutlineNode)>, roots: &mut Vec) { + if let Some((_, node)) = open.pop() { + match open.last_mut() { + Some((_, parent)) => parent.children.push(node), + None => roots.push(node), + } + } +} + +fn has_children(doc: &Document, r: ObjRef) -> bool { + doc.get(r).as_dict().is_some_and(|d| d.reference(b"First").is_some()) +} + +/// A destination's view as [`import_outline`] takes it: a known type with exactly its number of +/// values (extra ones dropped, missing ones `None`). An unknown type, or a FitR missing one of +/// its four values (the rectangle can't be kept), gives `None`. +fn portable_view(kind: Vec, mut values: Vec>) -> Option<(Vec, Vec>)> { + let (_, arity) = VIEWS.iter().find(|(k, _)| *k == kind.as_slice())?; + values.resize(*arity, None); + if kind.as_slice() == b"FitR" && values.iter().any(Option::is_none) { + return None; + } + Some((kind, values)) +} + +/// The nodes of `nodes` whose page `page_count` pages don't have, as paths into `nodes` +/// (0-based child indices), in tree order. [`import_outline`] keeps these without a destination. +pub fn unmapped(nodes: &[OutlineNode], page_count: usize) -> Vec> { + let mut out = Vec::new(); + visit(nodes, |path, node| { + if node.page.is_some_and(|p| p >= page_count) { + out.push(path.to_vec()); + } + }); + out +} + +/// How many values a view of destination type `kind` takes (Table 149), or `None` for a type +/// that is not one of them. +pub fn view_arity(kind: &[u8]) -> Option { + VIEWS.iter().find(|(k, _)| *k == kind).map(|(_, arity)| *arity) +} + +/// Call `f` with each node's path and the node, parents first. Iterative. +fn visit<'a>(nodes: &'a [OutlineNode], mut f: impl FnMut(&[usize], &'a OutlineNode)) { + // One entry per open level: its nodes and how many of them have been visited. + let mut stack: Vec<(&'a [OutlineNode], usize)> = vec![(nodes, 0)]; + // The visited node's path, kept in step with the stack rather than rebuilt per node. + let mut path: Vec = Vec::new(); + while let Some(level) = stack.last_mut() { + let list = level.0; + let index = level.1; + let Some(node) = list.get(index) else { + stack.pop(); + continue; + }; + level.1 = index.saturating_add(1); + path.truncate(stack.len().saturating_sub(1)); + path.push(index); + f(&path, node); + if !node.children.is_empty() { + stack.push((&node.children, 0)); + } + } +} + +/// Check every node before anything is written: depth (counting the `base` levels above the +/// tree), count, colour and view. +fn check(nodes: &[OutlineNode], base: usize) -> Result<(), OutlineError> { + let mut problem = None; + let mut count = 0usize; + visit(nodes, |path, node| { + count = count.saturating_add(1); + if problem.is_some() { + return; + } + if base.saturating_add(path.len()) > LIST_DEPTH.saturating_add(1) { + problem = Some(OutlineError::TooDeep(LIST_DEPTH.saturating_add(1))); + } else if node.color.is_some_and(|c| !c.iter().all(|v| (0.0..=1.0).contains(v))) { + problem = Some(OutlineError::BadColor); + } else if node.view.as_ref().is_some_and(|v| !view_is_valid(v)) { + problem = Some(OutlineError::BadView(path.to_vec())); + } + }); + if count > MAX_IMPORT_BOOKMARKS { + return Err(OutlineError::TooManyBookmarks(MAX_IMPORT_BOOKMARKS)); + } + problem.map_or(Ok(()), Err) +} + +/// A known type with its number of values, each finite; `/FitR` needs all four. +fn view_is_valid((kind, values): &(Vec, Vec>)) -> bool { + VIEWS.iter().any(|(k, arity)| *k == kind.as_slice() && *arity == values.len()) + && values.iter().flatten().all(|v| v.is_finite()) + && (kind.as_slice() != b"FitR" || values.iter().all(Option::is_some)) +} + +/// Write `nodes` as children of the bookmark at `parent_path` (`[]` = top level), starting at +/// child `index` (past the end appends), or, with `replace`, in place of the outline's +/// top-level bookmarks (then `parent_path` must be `[]`). Nodes whose page this document lacks +/// are written without a destination (see [`unmapped`]). Everything is checked before anything +/// is written. Returns the path of the first node written. +pub fn import_outline( + doc: &mut Document, + parent_path: &[usize], + index: usize, + nodes: &[OutlineNode], + replace: bool, +) -> Result, OutlineError> { + if nodes.is_empty() { + return Err(OutlineError::NothingToImport); + } + if replace && !parent_path.is_empty() { + return Err(OrganizeError::Invalid("replace applies to the whole outline: leave the parent out".into()).into()); + } + check(nodes, parent_path.len())?; + let pages: Vec = walk(doc)?.into_iter().map(|(r, _)| r).collect(); + let parent = container(doc, parent_path)?; + let top = write_nodes(doc, &pages, nodes)?; + let mut kids = if replace { Vec::new() } else { children(doc, parent) }; + let at = index.min(kids.len()); + kids.splice(at..at, top); + relink(doc, parent, &kids)?; + show_children(doc, parent, parent_path)?; + recount(doc)?; + let mut path = parent_path.to_vec(); + path.push(at); + Ok(path) +} + +/// Add one item per node, linked under each other, and return the top-level items in order. +fn write_nodes(doc: &mut Document, pages: &[ObjRef], nodes: &[OutlineNode]) -> Result, OutlineError> { + // Each written item with its node, then its children's items once they are written. + let mut written: Vec<(ObjRef, Vec)> = Vec::new(); + let mut top = Vec::new(); + // (the nodes of one level, the next to write, the index in `written` of their parent) + let mut stack: Vec<(&[OutlineNode], usize, Option)> = vec![(nodes, 0, None)]; + while let Some(level) = stack.last_mut() { + let (list, parent) = (level.0, level.2); + let Some(node) = list.get(level.1) else { + stack.pop(); + continue; + }; + level.1 = level.1.saturating_add(1); + let item = doc.add(Object::Dict(item_dict(node, pages))); + match parent.and_then(|p| written.get_mut(p)) { + Some((_, kids)) => kids.push(item), + None => top.push(item), + } + written.push((item, Vec::new())); + if !node.children.is_empty() { + stack.push((&node.children, 0, Some(written.len().saturating_sub(1)))); + } + } + for (item, kids) in &written { + relink(doc, *item, kids)?; + } + Ok(top) +} + +/// The item's own entries; the links and `/Count`'s value come from [`relink`] and [`recount`] +/// (only the sign of the `/Count` written here matters: it keeps the open state). +fn item_dict(node: &OutlineNode, pages: &[ObjRef]) -> Dict { + let mut d = Dict::new(); + d.set(b"Title".to_vec(), Object::String(PdfString::text(&node.title))); + if let Some(page) = node.page.and_then(|p| pages.get(p)) { + let (kind, values) = node.view.clone().unwrap_or_else(|| (b"XYZ".to_vec(), vec![None; 3])); + let mut dest = vec![Object::Ref(*page), Object::Name(kind)]; + dest.extend(values.into_iter().map(|v| v.map_or(Object::Null, Object::Real))); + d.set(b"Dest".to_vec(), Object::Array(dest)); + } + if let Some(c) = node.color.filter(|c| *c != [0.0; 3]) { + d.set(b"C".to_vec(), Object::Array(c.iter().map(|v| Object::Real(f64::from(*v))).collect())); + } + let flags = i64::from(node.italic) | i64::from(node.bold) << 1; + if flags != 0 { + d.set(b"F".to_vec(), Object::Int(flags)); + } + if !node.children.is_empty() { + d.set(b"Count".to_vec(), Object::Int(if node.open { 1 } else { -1 })); + } + d +} diff --git a/crates/organize/src/tests.rs b/crates/organize/src/tests.rs index 33dc262fce..d5ff2116ff 100644 --- a/crates/organize/src/tests.rs +++ b/crates/organize/src/tests.rs @@ -3813,3 +3813,175 @@ fn contents_pieces_join_with_a_space_across_a_trailing_comment() { ctrl_names.sort(); assert_eq!(ctrl_names, [b"ImHidden".to_vec(), b"ImShown".to_vec()]); } + +/// Intro (closed, red bold, at a point) > [Part (italic, FitH), Figure (FitR)], then Results. +fn outline_source() -> Document { + let mut d = doc_a(); + crate::add_bookmark_at(&mut d, &[], 0, "Intro", 0, Some(20.0), Some(380.0)).unwrap(); + crate::add_bookmark(&mut d, &[0], 0, "Part", 1).unwrap(); + crate::add_bookmark(&mut d, &[0], 1, "Figure", 2).unwrap(); + crate::add_bookmark(&mut d, &[], 1, "Results", 2).unwrap(); + crate::set_bookmark_color(&mut d, &[0], Some([1.0, 0.0, 0.0])).unwrap(); + crate::set_bookmark_style(&mut d, &[0], Some(true), None).unwrap(); + crate::set_bookmark_style(&mut d, &[0, 0], None, Some(true)).unwrap(); + let pages = crate::walk(&d).unwrap(); + let (part, figure) = (crate::bookmarks(&d)[1].obj, crate::bookmarks(&d)[2].obj); + let dest = |p: usize, rest: Vec| Object::Array([vec![Object::Ref(pages[p].0)], rest].concat()); + d.update_dict(part, |x| x.set(b"Dest".to_vec(), dest(1, vec![Object::name("FitH"), Object::Real(300.0)]))).unwrap(); + let fit_r = vec![Object::name("FitR"), Object::Int(10), Object::Int(20), Object::Int(110), Object::Int(220)]; + d.update_dict(figure, |x| x.set(b"Dest".to_vec(), dest(2, fit_r))).unwrap(); + crate::set_bookmark_open(&mut d, &[0], false).unwrap(); + d +} + +fn node(title: &str, page: Option, children: Vec) -> crate::OutlineNode { + crate::OutlineNode { title: title.into(), page, view: None, color: None, italic: false, bold: false, open: true, children } +} + +/// The `/Prev`, `/Last` and `/Parent` links agree with `/First`–`/Next` at every level. +fn assert_back_links(doc: &Document) { + let root = doc.get(doc.root().unwrap()).as_dict().unwrap().reference(b"Outlines").unwrap(); + let mut stack = vec![root]; + while let Some(parent) = stack.pop() { + let kids = crate::outline::children(doc, parent); + let d = doc.get(parent).as_dict().cloned().unwrap(); + assert_eq!(d.reference(b"Last"), kids.last().copied()); + for (i, k) in kids.iter().enumerate() { + let kd = doc.get(*k).as_dict().cloned().unwrap(); + assert_eq!(kd.reference(b"Prev"), i.checked_sub(1).map(|p| kids[p])); + assert_eq!(kd.reference(b"Parent"), Some(parent)); + } + stack.extend(kids); + } +} + +#[test] +fn an_exported_outline_imports_into_another_document_unchanged() { + let source = outline_source(); + let tree = crate::outline_tree(&source, &[]).unwrap(); + assert!(!tree.truncated); + let intro = &tree.nodes[0]; + assert_eq!((intro.title.as_str(), intro.page, intro.open, intro.bold, intro.color), ("Intro", Some(0), false, true, Some([1.0, 0.0, 0.0]))); + assert_eq!(intro.view, Some((b"XYZ".to_vec(), vec![Some(20.0), Some(380.0), None]))); + assert_eq!(intro.children[0].view, Some((b"FitH".to_vec(), vec![Some(300.0)]))); + assert!(intro.children[0].italic); + assert_eq!(intro.children[1].view, Some((b"FitR".to_vec(), vec![Some(10.0), Some(20.0), Some(110.0), Some(220.0)]))); + + let mut target = doc_a(); + assert_eq!(crate::import_outline(&mut target, &[], 0, &tree.nodes, false).unwrap(), vec![0]); + assert_back_links(&target); + let back = full_roundtrip(&target); + assert_eq!(crate::outline_tree(&back, &[]).unwrap(), tree); + // The closed parent stays closed: its count is negative and the root counts only what shows. + let first = crate::bookmarks(&back)[0].obj; + let root = back.get(back.root().unwrap()).as_dict().unwrap().reference(b"Outlines").unwrap(); + assert_eq!((count_of(&back, first), count_of(&back, root)), (Some(-2), Some(2))); +} + +#[test] +fn a_subtree_exports_alone_and_a_missing_path_is_refused() { + let source = outline_source(); + let tree = crate::outline_tree(&source, &[0]).unwrap(); + assert_eq!(tree.nodes.len(), 1); + assert_eq!(tree.nodes[0].title, "Intro"); + assert_eq!(tree.nodes[0].children.iter().map(|n| n.title.as_str()).collect::>(), ["Part", "Figure"]); + let leaf = crate::outline_tree(&source, &[0, 1]).unwrap(); + assert_eq!((leaf.nodes.len(), leaf.nodes[0].title.as_str()), (1, "Figure")); + assert_eq!(crate::outline_tree(&source, &[5]), Err(crate::OutlineError::NoSuchBookmark(vec![5]))); + assert_eq!(crate::outline_tree(&doc_a(), &[]).unwrap().nodes, []); +} + +#[test] +fn a_page_the_target_lacks_keeps_its_bookmark_without_a_destination() { + let tree = crate::outline_tree(&outline_source(), &[]).unwrap(); + // doc_b has two pages: Figure and Results go to the third. + assert_eq!(crate::unmapped(&tree.nodes, 2), [vec![0, 1], vec![1]]); + let mut target = doc_b(); + crate::import_outline(&mut target, &[], 0, &tree.nodes, false).unwrap(); + let listed = crate::bookmarks(&target); + assert_eq!(titles(&listed), ["Intro[Part,Figure]", "Results"]); + assert_eq!(listed.iter().map(|b| b.page).collect::>(), [Some(0), Some(1), None, None]); + assert!(target.get(listed[2].obj).as_dict().unwrap().get(b"Dest").is_none()); +} + +#[test] +fn a_fit_r_missing_a_value_copies_with_its_page_and_no_view() { + let mut src = outline_source(); + let pages = crate::walk(&src).unwrap(); + let figure = crate::bookmarks(&src)[2].obj; + let short = vec![Object::Ref(pages[2].0), Object::name("FitR"), Object::Int(10), Object::Int(20), Object::Int(110)]; + src.update_dict(figure, |x| x.set(b"Dest".to_vec(), Object::Array(short))).unwrap(); + let tree = crate::outline_tree(&src, &[]).unwrap(); + let copied = &tree.nodes[0].children[1]; + assert_eq!((copied.page, copied.view.clone()), (Some(2), None)); + let mut target = doc_a(); + crate::import_outline(&mut target, &[], 0, &tree.nodes, false).unwrap(); + assert_eq!(titles(&crate::bookmarks(&target)), ["Intro[Part,Figure]", "Results"]); +} + +#[test] +fn an_import_appends_inserts_or_replaces() { + let mut d = doc_a(); + crate::add_bookmark(&mut d, &[], 0, "Old A", 0).unwrap(); + crate::add_bookmark(&mut d, &[], 1, "Old B", 1).unwrap(); + let new = [node("New", Some(2), vec![node("Child", Some(1), vec![])])]; + assert_eq!(crate::import_outline(&mut d, &[], 1, &new, false).unwrap(), vec![1]); + assert_eq!(titles(&crate::bookmarks(&d)), ["Old A", "New[Child]", "Old B"]); + assert_eq!(crate::import_outline(&mut d, &[2], 9, &new, false).unwrap(), vec![2, 0]); + assert_eq!(titles(&crate::bookmarks(&d)), ["Old A", "New[Child]", "Old B[New[Child]]"]); + assert_back_links(&d); + assert_eq!(crate::import_outline(&mut d, &[], 5, &new, true).unwrap(), vec![0]); + assert_eq!(titles(&crate::bookmarks(&d)), ["New[Child]"]); + assert_back_links(&d); + let refused = crate::import_outline(&mut d, &[0], 0, &new, true); + assert!(matches!(refused, Err(crate::OutlineError::Organize(crate::OrganizeError::Invalid(_))))); + assert_eq!(crate::import_outline(&mut d, &[], 0, &[], false), Err(crate::OutlineError::NothingToImport)); + assert_eq!(crate::import_outline(&mut d, &[7], 0, &new, false), Err(crate::OutlineError::NoSuchBookmark(vec![7]))); +} + +#[test] +fn a_bad_node_anywhere_refuses_the_whole_import() { + let mut d = doc_a(); + crate::add_bookmark(&mut d, &[], 0, "Kept", 0).unwrap(); + let before = crate::bookmarks(&d); + let with = |f: &dyn Fn(&mut crate::OutlineNode)| { + let mut child = node("Child", Some(0), vec![]); + f(&mut child); + vec![node("Fine", Some(0), vec![]), node("Parent", Some(0), vec![child])] + }; + let bad_views: [(&[u8], Vec>); 4] = + [(b"XYZ", vec![Some(1.0)]), (b"Zoom", vec![]), (b"FitH", vec![Some(f64::NAN)]), (b"FitR", vec![Some(0.0), None, Some(1.0), Some(1.0)])]; + for (kind, values) in bad_views { + let nodes = with(&|n| n.view = Some((kind.to_vec(), values.clone()))); + assert_eq!(crate::import_outline(&mut d, &[], 0, &nodes, false), Err(crate::OutlineError::BadView(vec![1, 0])), "{kind:?}"); + } + let nodes = with(&|n| n.color = Some([0.5, 1.5, 0.0])); + assert_eq!(crate::import_outline(&mut d, &[], 0, &nodes, false), Err(crate::OutlineError::BadColor)); + let nodes = with(&|n| n.color = Some([f32::NAN, 0.0, 0.0])); + assert_eq!(crate::import_outline(&mut d, &[], 0, &nodes, false), Err(crate::OutlineError::BadColor)); + let mut deep = node("Leaf", None, vec![]); + for _ in 0..33 { + deep = node("Level", None, vec![deep]); + } + assert_eq!(crate::import_outline(&mut d, &[], 0, &[deep], false), Err(crate::OutlineError::TooDeep(33))); + // 33 levels fit at the top, but not under a bookmark: the deepest would be past the listing. + let mut chain = node("Leaf", None, vec![]); + for _ in 0..32 { + chain = node("Level", None, vec![chain]); + } + assert_eq!(crate::import_outline(&mut d, &[0], 0, std::slice::from_ref(&chain), false), Err(crate::OutlineError::TooDeep(33))); + let many = vec![node("Item", None, vec![]); crate::MAX_IMPORT_BOOKMARKS + 1]; + assert_eq!(crate::import_outline(&mut d, &[], 0, &many, false), Err(crate::OutlineError::TooManyBookmarks(crate::MAX_IMPORT_BOOKMARKS))); + assert_eq!(crate::bookmarks(&d), before); +} + +#[test] +fn an_outline_deeper_than_the_listing_is_exported_as_truncated() { + let tree = crate::outline_tree(&open(deep_outline(40)), &[]).unwrap(); + assert!(tree.truncated); + assert!(!crate::outline_tree(&open(deep_outline(33)), &[]).unwrap().truncated); + // What was exported imports back: 33 levels is the deepest an import takes. + let mut target = doc_a(); + crate::import_outline(&mut target, &[], 0, &tree.nodes, false).unwrap(); + assert_eq!(crate::bookmarks(&target).len(), 33); +} diff --git a/crates/prc-render/Cargo.toml b/crates/prc-render/Cargo.toml new file mode 100644 index 0000000000..11ccd3346f --- /dev/null +++ b/crates/prc-render/Cargo.toml @@ -0,0 +1,18 @@ +[package] +name = "pdfcraft-prc-render" +description = "Headless CPU renderer for PRC 3D models: camera, depth-buffered triangle fill, headlight shading" +version.workspace = true +edition.workspace = true +license.workspace = true +rust-version.workspace = true +repository.workspace = true + +[lints] +workspace = true + +[dependencies] +pdfcraft-prc.workspace = true +thiserror.workspace = true + +[dev-dependencies] +flate2.workspace = true diff --git a/crates/prc-render/README.md b/crates/prc-render/README.md new file mode 100644 index 0000000000..000d673179 --- /dev/null +++ b/crates/prc-render/README.md @@ -0,0 +1,29 @@ +# pdfcraft-prc-render + +A headless CPU renderer for PRC 3D models read by `pdfcraft-prc`: placed meshes in world space, a fitted camera, a +depth-buffered triangle fill and headlight shading, into a straight-RGBA image. + +- **Layer:** L3. Depends on `pdfcraft-prc` only. +- **Licence:** MIT OR Apache-2.0. +- `#![forbid(unsafe_code)]`. Never panics on input: image size, triangle count and every index are checked, and + damage returns `RenderError`. + +## API + +```rust +assemble(&PrcFile) -> Result, RenderError> // every placement's mesh moved into the world and coloured +place(&TriangleMesh, &Matrix, colour, triangle_colours) -> WorldMesh +Camera::fit(&Bounds, direction, up, perspective, aspect) / Camera::fit_meshes(&[WorldMesh], ...) +render(&[WorldMesh], &Camera, &RenderOptions) -> Result +``` + +- With no product tree, each tessellated mesh is drawn where it lies, in the default colour. +- A triangle's own colour wins over its mesh's, which wins over `RenderOptions::colour`. +- Opaque triangles are drawn first, then translucent ones blended over them; alpha 0 is not drawn. +- `Lighting::Headlight` (default) shades by the angle to the view direction; `Lighting::Unlit` draws flat colour. +- Perspective triangles are clipped at a near plane 1/10,000 of the eye-to-target distance from the eye; orthographic views are not clipped. +- `MAX_RENDER_PIXELS` (2^24, about 16 MP) bounds the image: at most 64 MB of RGBA plus 128 MB of depth. + `MAX_ASSEMBLED_TRIANGLES` and `MAX_ASSEMBLED_VERTICES` (2^24 each) bound one `assemble` call. `MAX_RENDER_WORK` (2^31) bounds + the pixel centres one `render` tests, summed over each triangle's clipped bounding box; past it the render returns + `RenderError::TooMuchWork`. That is still seconds of CPU at the top, so callers on a UI thread should render small or off-thread. +- Not yet: shadows, ambient occlusion, specular light, scene lights, textures, wires and markups. diff --git a/crates/prc-render/src/assemble.rs b/crates/prc-render/src/assemble.rs new file mode 100644 index 0000000000..c9264d620e --- /dev/null +++ b/crates/prc-render/src/assemble.rs @@ -0,0 +1,164 @@ +//! A parsed PRC file's tessellations placed in the world, ready for [`crate::render`]. + +use crate::RenderError; +use crate::raster::channel; +use pdfcraft_prc::{Matrix, PrcFile, Rgba, Tessellation, TriangleMesh}; + +/// The most triangles [`assemble`] places, over every placement: a tessellation many +/// occurrences draw is copied once per occurrence. +pub const MAX_ASSEMBLED_TRIANGLES: usize = 1 << 24; + +/// The most positions, normals and texture coordinates [`assemble`] copies, over every placement. +pub const MAX_ASSEMBLED_VERTICES: usize = 1 << 24; + +/// A triangle mesh in world coordinates with the colours it draws in. +#[derive(Debug, Clone, Default, PartialEq)] +pub struct WorldMesh { + /// The triangles, already placed. + pub mesh: TriangleMesh, + /// The mesh's straight-RGBA colour; `None` takes [`crate::RenderOptions::colour`], opaque. + pub colour: Option<[u8; 4]>, + /// Per triangle, its own colour when its face is styled; `None` (the whole list or an entry) + /// takes [`Self::colour`]. + pub triangle_colours: Option>>, +} + +impl WorldMesh { + /// `mesh`, uncoloured. + #[must_use] + pub fn new(mesh: TriangleMesh) -> Self { + WorldMesh { mesh, colour: None, triangle_colours: None } + } + + /// The colour triangle `t` draws in, before the render options' default. + pub(crate) fn colour_of(&self, t: usize) -> Option<[u8; 4]> { + self.triangle_colours.as_ref().and_then(|c| c.get(t).copied().flatten()).or(self.colour) + } +} + +/// `mesh` moved into the world by `matrix`, drawn in `colour` and, per triangle, `triangle_colours`. +#[must_use] +pub fn place(mesh: &TriangleMesh, matrix: &Matrix, colour: Option<[u8; 4]>, triangle_colours: Option>>) -> WorldMesh { + WorldMesh { mesh: mesh.transformed(matrix), colour, triangle_colours } +} + +/// Every triangle mesh `prc` draws, in world coordinates and coloured as its product tree says +/// ([`PrcFile::placements`]). A file with no tree draws each file structure's meshes where they +/// are, uncoloured. Wires, markups and compressed tessellations are not drawn. +/// +/// # Errors +/// [`RenderError::Model`] when a tessellation section or the tree cannot be read, or the tree +/// places a tessellation its file structure does not have; [`RenderError::TooManyTriangles`] past +/// [`MAX_ASSEMBLED_TRIANGLES`] and [`RenderError::TooManyVertices`] past [`MAX_ASSEMBLED_VERTICES`], +/// counting both what is decoded and what is placed. +pub fn assemble(prc: &PrcFile) -> Result, RenderError> { + let tessellations = decode(prc.file_structures.iter().map(|fs| fs.tessellations()), Budget::default())?; + let mut budget = Budget::default(); + let mut out = Vec::new(); + if !prc.stores_tree() { + for mesh in tessellations.iter().flatten().filter_map(as_mesh) { + budget.spend(mesh)?; + out.push(WorldMesh::new(mesh.clone())); + } + return Ok(out); + } + let items = prc.placements()?.into_iter().map(|p| Item { + file_structure: p.file_structure, + tessellation: p.tessellation, + matrix: p.matrix, + colour: p.colour, + triangle_colours: Box::new(move |mesh: &TriangleMesh| p.triangle_colours(mesh)), + }); + place_items(&tessellations, items, &mut budget, &mut out)?; + Ok(out) +} + +/// Each file structure's tessellations, decoded one structure at a time and charged to `budget` +/// as they are, so decoding stops at the structure that passes a ceiling. +pub(crate) fn decode( + structures: impl IntoIterator, pdfcraft_prc::PrcError>>, + mut budget: Budget, +) -> Result>, RenderError> { + let mut out = Vec::new(); + for tessellations in structures { + let tessellations = tessellations?; + for mesh in tessellations.iter().filter_map(as_mesh) { + budget.spend(mesh)?; + } + out.push(tessellations); + } + Ok(out) +} + +/// One placement as [`place_items`] reads it: a tessellation by file structure and index, where +/// it goes and how it is coloured. +pub(crate) struct Item<'a> { + pub(crate) file_structure: usize, + pub(crate) tessellation: usize, + pub(crate) matrix: Matrix, + pub(crate) colour: Option, + pub(crate) triangle_colours: Box>, +} + +/// Per triangle of a mesh, the colour its face's style resolves to. +type FaceColours<'a> = dyn Fn(&TriangleMesh) -> Option>> + 'a; + +/// Each item's mesh moved into the world and coloured, appended to `out`. Items naming a wire, +/// markup or compressed tessellation are skipped; an item naming no tessellation at all is +/// [`PrcError::Malformed`](pdfcraft_prc::PrcError::Malformed). +pub(crate) fn place_items<'a>( + tessellations: &[Vec], + items: impl IntoIterator>, + budget: &mut Budget, + out: &mut Vec, +) -> Result<(), RenderError> { + for p in items { + let Some(tessellation) = tessellations.get(p.file_structure).and_then(|t| t.get(p.tessellation)) else { + let what = format!("tessellation {} of file structure {}, which has none by that index", p.tessellation, p.file_structure); + return Err(pdfcraft_prc::PrcError::Malformed(what).into()); + }; + let Some(mesh) = as_mesh(tessellation) else { + continue; + }; + budget.spend(mesh)?; + let triangle_colours = (p.triangle_colours)(mesh).map(|c| c.into_iter().map(|c| c.map(bytes)).collect()); + out.push(place(mesh, &p.matrix, p.colour.map(bytes), triangle_colours)); + } + Ok(()) +} + +/// Triangles and vertex entries (positions, normals, texture coordinates) placed so far. +#[derive(Debug, Default)] +pub(crate) struct Budget { + pub(crate) triangles: usize, + pub(crate) vertices: usize, +} + +impl Budget { + fn spend(&mut self, mesh: &TriangleMesh) -> Result<(), RenderError> { + let triangles = + self.triangles.checked_add(mesh.triangles.len()).filter(|&n| n <= MAX_ASSEMBLED_TRIANGLES).ok_or(RenderError::TooManyTriangles)?; + let vertices = [mesh.positions.len(), mesh.normals.len(), mesh.uvs.len()] + .into_iter() + .try_fold(self.vertices, usize::checked_add) + .filter(|&n| n <= MAX_ASSEMBLED_VERTICES) + .ok_or(RenderError::TooManyVertices)?; + (self.triangles, self.vertices) = (triangles, vertices); + Ok(()) + } +} + +fn as_mesh(t: &Tessellation) -> Option<&TriangleMesh> { + match t { + Tessellation::Mesh(m) => Some(m), + _ => None, + } +} + +/// A 0–1 colour as straight-RGBA bytes, each channel clamped. +fn bytes(c: Rgba) -> [u8; 4] { + [c.r, c.g, c.b, c.a].map(|v| channel(v.clamp(0.0, 1.0) * 255.0)) +} + +#[cfg(test)] +mod tests; diff --git a/crates/prc-render/src/assemble/tests.rs b/crates/prc-render/src/assemble/tests.rs new file mode 100644 index 0000000000..5018bdbe8c --- /dev/null +++ b/crates/prc-render/src/assemble/tests.rs @@ -0,0 +1,127 @@ +#![allow(clippy::unwrap_used)] + +use super::*; + +fn triangle() -> TriangleMesh { + let mut m = TriangleMesh::default(); + m.positions = vec![[0.0, 0.0, 0.0], [1.0, 0.0, 0.0], [0.0, 1.0, 0.0]]; + m.triangles = vec![[0, 1, 2]]; + m +} + +#[test] +fn a_placement_moves_its_mesh_into_the_world() { + let mut shift = Matrix::IDENTITY; + shift.0[0][3] = 5.0; + shift.0[2][3] = -2.0; + let placed = place(&triangle(), &shift, Some([1, 2, 3, 255]), None); + assert_eq!(placed.mesh.positions, [[5.0, 0.0, -2.0], [6.0, 0.0, -2.0], [5.0, 1.0, -2.0]]); + assert_eq!(placed.colour, Some([1, 2, 3, 255])); + // A mirror keeps the triangle facing the same way round. + let mut mirror = Matrix::IDENTITY; + mirror.0[0][0] = -1.0; + assert_eq!(place(&triangle(), &mirror, None, None).mesh.triangles, [[0, 2, 1]]); +} + +#[test] +fn a_triangle_colour_wins_over_the_mesh_colour() { + let mut m = place(&triangle(), &Matrix::IDENTITY, Some([9, 9, 9, 255]), Some(vec![Some([1, 1, 1, 255]), None])); + m.mesh.triangles.push([0, 2, 1]); + m.mesh.triangles.push([1, 2, 0]); + assert_eq!(m.colour_of(0), Some([1, 1, 1, 255])); + assert_eq!(m.colour_of(1), Some([9, 9, 9, 255]), "an unstyled face takes the mesh's colour"); + assert_eq!(m.colour_of(2), Some([9, 9, 9, 255]), "past the list too"); + assert_eq!(WorldMesh::new(triangle()).colour_of(0), None); +} + +#[test] +fn colours_are_clamped_into_bytes() { + let [r, g, b, a] = [0.0, 1.0, 0.5, 2.0]; + let mut c = Rgba::default(); + (c.r, c.g, c.b, c.a) = (r, g, b, a); + assert_eq!(bytes(c), [0, 255, 128, 255]); + c.r = f64::NAN; + c.g = -3.0; + assert_eq!(bytes(c)[..2], [0, 0]); +} + +#[test] +fn the_triangle_budget_is_shared_and_refuses_past_the_ceiling() { + let mut b = Budget { triangles: MAX_ASSEMBLED_TRIANGLES - 1, vertices: 0 }; + assert!(b.spend(&triangle()).is_ok()); + assert_eq!(b.spend(&triangle()), Err(RenderError::TooManyTriangles)); + let mut b = Budget { triangles: usize::MAX, vertices: 0 }; + assert_eq!(b.spend(&triangle()), Err(RenderError::TooManyTriangles)); +} + +#[test] +fn the_vertex_budget_counts_positions_normals_and_uvs() { + let mut heavy = triangle(); + heavy.normals = vec![[0.0, 0.0, 1.0]; 2]; + heavy.uvs = vec![[0.0; 2]; 4]; + let mut b = Budget::default(); + b.spend(&heavy).unwrap(); + assert_eq!((b.triangles, b.vertices), (1, 9)); + // One triangle over many vertices: the triangle ceiling alone would let it through. + let mut b = Budget { triangles: 0, vertices: MAX_ASSEMBLED_VERTICES - 8 }; + assert_eq!(b.spend(&heavy), Err(RenderError::TooManyVertices)); + assert_eq!((b.triangles, b.vertices), (0, MAX_ASSEMBLED_VERTICES - 8), "a refused mesh spends nothing"); + let mut b = Budget { triangles: 0, vertices: usize::MAX }; + assert_eq!(b.spend(&triangle()), Err(RenderError::TooManyVertices)); +} + +fn item<'a>(fs: usize, tess: usize, x: f64, colour: Option) -> Item<'a> { + let mut matrix = Matrix::IDENTITY; + matrix.0[0][3] = x; + Item { file_structure: fs, tessellation: tess, matrix, colour, triangle_colours: Box::new(|_| None) } +} + +#[test] +fn each_placement_draws_its_tessellation_where_and_in_the_colour_the_tree_says() { + let tessellations = vec![vec![Tessellation::Mesh(triangle())]]; + let mut red = Rgba::default(); + red.r = 1.0; + red.a = 1.0; + // Two occurrences of one tessellation at different places, one coloured. + let items = [item(0, 0, 3.0, Some(red)), item(0, 0, -4.0, None)]; + let (mut budget, mut out) = (Budget::default(), Vec::new()); + place_items(&tessellations, items, &mut budget, &mut out).unwrap(); + assert_eq!(out.len(), 2); + assert_eq!(out[0].mesh.positions, [[3.0, 0.0, 0.0], [4.0, 0.0, 0.0], [3.0, 1.0, 0.0]]); + assert_eq!(out[0].colour, Some([255, 0, 0, 255])); + assert_eq!(out[1].mesh.positions, [[-4.0, 0.0, 0.0], [-3.0, 0.0, 0.0], [-4.0, 1.0, 0.0]]); + assert_eq!(out[1].colour, None); + assert_eq!(budget.triangles, 2, "each occurrence spends its triangles"); +} + +#[test] +fn a_placement_naming_a_missing_tessellation_is_malformed() { + let tessellations = vec![vec![Tessellation::Mesh(triangle())]]; + for (fs, tess) in [(1, 0), (0, 7)] { + let (mut budget, mut out) = (Budget::default(), Vec::new()); + let r = place_items(&tessellations, [item(fs, tess, 0.0, None)], &mut budget, &mut out); + assert!(matches!(r, Err(RenderError::Model(pdfcraft_prc::PrcError::Malformed(_)))), "{fs}/{tess}: {r:?}"); + } +} + +#[test] +fn a_placement_past_the_triangle_budget_is_an_error() { + let tessellations = vec![vec![Tessellation::Mesh(triangle())]]; + let (mut budget, mut out) = (Budget { triangles: MAX_ASSEMBLED_TRIANGLES, vertices: 0 }, Vec::new()); + let r = place_items(&tessellations, [item(0, 0, 0.0, None)], &mut budget, &mut out); + assert!(matches!(r, Err(RenderError::TooManyTriangles))); + assert!(out.is_empty()); +} + +#[test] +fn decoding_stops_at_the_structure_that_passes_the_ceiling() { + let pulled = std::cell::Cell::new(0); + let structures = std::iter::repeat_with(|| { + pulled.set(pulled.get() + 1); + Ok(vec![Tessellation::Mesh(triangle())]) + }) + .take(3); + let nearly_full = Budget { triangles: MAX_ASSEMBLED_TRIANGLES - 1, vertices: 0 }; + assert_eq!(decode(structures, nearly_full).err(), Some(RenderError::TooManyTriangles)); + assert_eq!(pulled.get(), 2, "the third structure is never decoded"); +} diff --git a/crates/prc-render/src/lib.rs b/crates/prc-render/src/lib.rs new file mode 100644 index 0000000000..9cfd49efbe --- /dev/null +++ b/crates/prc-render/src/lib.rs @@ -0,0 +1,18 @@ +//! Headless CPU rendering of PRC 3D models (ISO 14739-1) to an RGBA image. +//! +//! Layer L3: depends only on `pdfcraft-prc`. [`assemble`] turns a parsed [`pdfcraft_prc::PrcFile`] +//! into world-space meshes; [`render`] draws them from a [`Camera`] with one depth-tested sample per +//! pixel centre and a light at the camera. Single-threaded and GUI-free, so it runs unchanged on wasm32. +//! The image size is capped by [`MAX_RENDER_PIXELS`], the drawing work by [`MAX_RENDER_WORK`] and the +//! assembled model by [`MAX_ASSEMBLED_TRIANGLES`] and [`MAX_ASSEMBLED_VERTICES`]; bad input is an +//! error, never a panic. +#![forbid(unsafe_code)] +#![deny(clippy::unwrap_used, clippy::expect_used, clippy::panic, clippy::unimplemented, clippy::todo, clippy::unreachable)] + +mod assemble; +mod raster; +mod render; +mod vec3; + +pub use assemble::{MAX_ASSEMBLED_TRIANGLES, MAX_ASSEMBLED_VERTICES, WorldMesh, assemble, place}; +pub use render::{Bounds, Camera, Image, Lighting, MAX_RENDER_PIXELS, MAX_RENDER_WORK, Projection, RenderError, RenderOptions, render}; diff --git a/crates/prc-render/src/raster.rs b/crates/prc-render/src/raster.rs new file mode 100644 index 0000000000..4427df5973 --- /dev/null +++ b/crates/prc-render/src/raster.rs @@ -0,0 +1,321 @@ +//! Projection, near-plane clipping and the depth-tested triangle fill behind [`crate::render`]. + +use crate::render::{Camera, Projection, RenderError}; +use crate::vec3::{length, sub}; + +/// One view-space corner: x, y, z (z along the view direction), then its red, green and blue shade. +pub(crate) type ViewCorner = [f64; 6]; + +/// One screen-space corner: pixel x, pixel y and depth key, and its shade. +#[derive(Debug, Clone, Copy)] +pub(crate) struct Corner { + pub(crate) at: [f64; 3], + pub(crate) shade: [f64; 3], +} + +/// The perspective near plane, as a fraction of the eye-to-target distance. +pub(crate) const NEAR_FRACTION: f64 = 1e-4; + +pub(crate) enum View { + /// `focal` = 1 / tan(fov_y / 2); `near` = the clip distance. + Perspective { + focal: f64, + aspect: f64, + near: f64, + }, + Orthographic { + half_height: f64, + aspect: f64, + }, +} + +impl View { + /// The projection for `camera` at width/height `aspect`; refuses a field of view outside + /// (0, 180) degrees or a non-positive orthographic height. + pub(crate) fn new(camera: &Camera, aspect: f64) -> Result { + match camera.projection { + Projection::Perspective { fov_y } => { + if !(fov_y > 0.0 && fov_y < 180.0) { + return Err(RenderError::Camera("the field of view is not between 0 and 180 degrees")); + } + let focal = 1.0 / (fov_y.to_radians() / 2.0).tan(); + // A field of view too small to measure leaves no finite focal length. + if !focal.is_finite() { + return Err(RenderError::Camera("the field of view is too small")); + } + let distance = length(sub(camera.target, camera.eye)); + Ok(View::Perspective { focal, aspect, near: (distance * NEAR_FRACTION).max(f64::MIN_POSITIVE) }) + } + Projection::Orthographic { height } => { + if !(height.is_finite() && height > 0.0) { + return Err(RenderError::Camera("the orthographic height is not positive")); + } + Ok(View::Orthographic { half_height: height / 2.0, aspect }) + } + } + } + + /// The part of `tri` in front of the near plane (Sutherland–Hodgman against one plane): + /// 0, 3 or 4 corners, the shade interpolated along each cut edge. + pub(crate) fn clip(&self, tri: [ViewCorner; 3]) -> Vec { + let View::Perspective { near, .. } = *self else { + return tri.to_vec(); + }; + let mut out = Vec::with_capacity(4); + for (k, &p) in tri.iter().enumerate() { + let q = tri.get((k + 1) % 3).copied().unwrap_or(p); + let z = |v: ViewCorner| v.get(2).copied().unwrap_or(0.0); + let (p_in, q_in) = (z(p) >= near, z(q) >= near); + if p_in { + out.push(p); + } + if p_in != q_in { + // Cut from the same end whichever way round the edge is met, so two triangles + // sharing it get the same cut point. + let (s, e) = if first(&p, &q) { (p, q) } else { (q, p) }; + let t = (near - z(s)) / (z(e) - z(s)); + let mut cut = s; + for (c, d) in cut.iter_mut().zip(e) { + *c += t * (d - *c); + } + out.push(cut); + } + } + out + } + + /// The screen corner of view corner `v`: pixel x, pixel y and a depth key that is linear + /// in screen space and larger for nearer points. + pub(crate) fn project(&self, [x, y, z, r, g, b]: ViewCorner, width: u32, height: u32) -> Corner { + let (nx, ny, key) = match *self { + // Divided before multiplied, so an on-screen point's intermediate products stay finite. + View::Perspective { focal, aspect, .. } => (x / z * focal / aspect, y / z * focal, 1.0 / z), + View::Orthographic { half_height, aspect } => (x / half_height / aspect, y / half_height, -z), + }; + Corner { at: [(nx + 1.0) / 2.0 * f64::from(width), (1.0 - ny) / 2.0 * f64::from(height), key], shade: [r, g, b] } + } +} + +/// Whether view corner `a`'s position comes before `b`'s, ordering x, then y, then z. +fn first(a: &ViewCorner, b: &ViewCorner) -> bool { + a.iter().zip(b).take(3).map(|(p, q)| p.total_cmp(q)).find(|o| o.is_ne()).is_none_or(std::cmp::Ordering::is_lt) +} + +/// The image being drawn: `rgba` holds `width * height` straight-RGBA pixels and `depth` their +/// depth keys, row by row from the top. +pub(crate) struct Target<'b> { + /// Depth keys are 1/z (a perspective view), so the shade is interpolated perspective-correct. + pub(crate) perspective: bool, + pub(crate) width: u32, + pub(crate) height: u32, + pub(crate) rgba: &'b mut [u8], + pub(crate) depth: &'b mut [f64], + /// Pixel centres the fills may still test. + pub(crate) work: u64, +} + +impl Target<'_> { + /// Fill the screen triangle of `corners` with straight-RGBA `colour`, depth-tested at pixel + /// centres, the shade interpolated across it. A pixel centre on an edge two triangles share + /// belongs to one of them only. An opaque pixel replaces the pixel and its depth; a + /// translucent one is blended over it, leaving the depth. The clipped bounding box's pixel + /// centres are charged to [`Self::work`] first: [`RenderError::TooMuchWork`], with nothing + /// drawn, when too few are left. + pub(crate) fn fill(&mut self, corners: [Corner; 3], colour: [u8; 4]) -> Result<(), RenderError> { + let [a, b, c] = corners; + let area = edge(a.at, b.at, c.at); + let edges = [(b.at, c.at), (c.at, a.at), (a.at, b.at)]; + if !area.is_finite() || area.abs() < 1e-12 { + return Ok(()); + } + let Some((x0, x1, y0, y1)) = self.bounds(&corners) else { + return Ok(()); + }; + let span = |lo: u32, hi: u32| u64::from(hi.saturating_sub(lo)).saturating_add(1); + self.work = self.work.checked_sub(span(x0, x1).saturating_mul(span(y0, y1))).ok_or(RenderError::TooMuchWork)?; + for y in y0..=y1 { + for x in x0..=x1 { + let p = [f64::from(x) + 0.5, f64::from(y) + 0.5, 0.0]; + let w = edges.map(|(from, to)| edge(from, to, p) / area); + if w.iter().zip(edges).any(|(&k, (from, to))| k < 0.0 || (k == 0.0 && !owns(from, to, area))) { + continue; + } + let mix = |f: &dyn Fn(&Corner) -> f64| w.iter().zip(&corners).map(|(k, c)| k * f(c)).sum::(); + let key = mix(&|c| at(c.at, 2)); + let Some(i) = pixel_index(self.width, x, y) else { + continue; + }; + if self.depth.get(i).is_none_or(|&d| key <= d) { + continue; + } + let shade = if a.shade == b.shade && b.shade == c.shade { + a.shade + } else if self.perspective && key > 0.0 { + // Screen weights times 1/z, renormalised, are the surface's own weights. + [0, 1, 2].map(|j| mix(&|c| at(c.shade, j) * at(c.at, 2)) / key) + } else { + [0, 1, 2].map(|j| mix(&|c| at(c.shade, j))) + }; + self.blend(i, key, shaded(colour, shade)); + } + } + Ok(()) + } + + /// The pixel rows and columns `corners` may cover, clipped to the image; `None` when off it. + fn bounds(&self, corners: &[Corner; 3]) -> Option<(u32, u32, u32, u32)> { + let lo = |i: usize| corners.iter().map(|c| at(c.at, i)).fold(f64::INFINITY, f64::min).floor().max(0.0); + let hi = |i: usize, limit: u32| corners.iter().map(|c| at(c.at, i)).fold(f64::NEG_INFINITY, f64::max).ceil().min(f64::from(limit) - 1.0); + let (x0, x1) = (lo(0), hi(0, self.width)); + let (y0, y1) = (lo(1), hi(1, self.height)); + // Each bound is finite and within 0..limit here, so the casts are exact. + (x0 <= x1 && y0 <= y1).then_some((x0 as u32, x1 as u32, y0 as u32, y1 as u32)) + } + + /// Write straight-RGBA `rgba` at pixel `i`: replacing it and its depth when opaque, blended + /// over it otherwise. + fn blend(&mut self, i: usize, key: f64, rgba: [u8; 4]) { + let opaque = rgba[3] == 255; + if opaque && let Some(d) = self.depth.get_mut(i) { + *d = key; + } + let Some(px) = i.checked_mul(4).and_then(|s| self.rgba.get_mut(s..s.checked_add(4)?)) else { + return; + }; + if opaque { + px.copy_from_slice(&rgba); + } else if let [pr, pg, pb, pa] = px { + let [sr, sg, sb, sa] = rgba; + let src = f64::from(sa) / 255.0; + let dst = f64::from(*pa) / 255.0 * (1.0 - src); + let out = src + dst; + if out <= 0.0 { + return; + } + for (p, s) in [pr, pg, pb].into_iter().zip([sr, sg, sb]) { + *p = channel((f64::from(s) * src + f64::from(*p) * dst) / out); + } + *pa = channel(out * 255.0); + } + } +} + +/// Pixel `x, y`'s index in an image `width` wide; `None` on overflow. +pub(crate) fn pixel_index(width: u32, x: u32, y: u32) -> Option { + usize::try_from(y).ok()?.checked_mul(usize::try_from(width).ok()?)?.checked_add(usize::try_from(x).ok()?) +} + +/// `v` rounded and clamped to a colour channel; NaN is 0. +pub(crate) fn channel(v: f64) -> u8 { + // Clamped to 0..=255 first, so the cast is exact (NaN casts to 0). + v.round().clamp(0.0, 255.0) as u8 +} + +/// `colour`'s RGB scaled channel by channel by `shade`, saturating at 255; alpha kept. +fn shaded([r, g, b, a]: [u8; 4], [sr, sg, sb]: [f64; 3]) -> [u8; 4] { + let [r, g, b] = [(r, sr), (g, sg), (b, sb)].map(|(v, s)| channel(f64::from(v) * s)); + [r, g, b, a] +} + +/// Twice the signed area of (a, b, p) in the xy plane. +pub(crate) fn edge(a: [f64; 3], b: [f64; 3], p: [f64; 3]) -> f64 { + // Two triangles sharing an edge meet it in opposite orders; evaluating it in one canonical + // order makes their values exact negatives, so a pixel centre on the edge cannot round out + // of both and leave a crack. + if (at(a, 0), at(a, 1)) > (at(b, 0), at(b, 1)) { + return -edge(b, a, p); + } + (at(b, 0) - at(a, 0)) * (at(p, 1) - at(a, 1)) - (at(b, 1) - at(a, 1)) * (at(p, 0) - at(a, 0)) +} + +/// Whether a pixel centre exactly on the edge `from`-`to` of a triangle of signed doubled area +/// `area` is the triangle's: the side the triangle grows towards, +x first, then +y. Two +/// triangles sharing the edge lie on opposite sides of it, so exactly one owns it. +fn owns(from: [f64; 3], to: [f64; 3], area: f64) -> bool { + let s = area.signum(); + let (gx, gy) = (-(at(to, 1) - at(from, 1)) * s, (at(to, 0) - at(from, 0)) * s); + gx > 0.0 || (gx == 0.0 && gy > 0.0) +} + +/// Component `i` of `v`, 0.0 out of range. +pub(crate) fn at(v: [f64; 3], i: usize) -> f64 { + v.get(i).copied().unwrap_or(0.0) +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn an_edge_seen_from_either_triangle_is_exactly_negated() { + // So no pixel centre on a shared edge is rejected by both. + let vals = [0.1, 0.7, 1.3, 2.9, 3.3, 5.5]; + for (a, b, p) in [([0.1, 0.1], [0.7, 0.7], [0.1, 2.9]), ([5.5, 0.7], [1.3, 3.3], [2.9, 0.1])] + .into_iter() + .chain(vals.iter().map(|&v| ([v, 0.1], [0.7, v], [v, v]))) + { + let [a, b, p] = [a, b, p].map(|[x, y]| [x, y, 0.0]); + assert_eq!(edge(a, b, p), -edge(b, a, p), "{a:?} {b:?} {p:?}"); + } + } + + #[test] + fn pixel_indices_and_channels_never_overflow() { + assert_eq!(pixel_index(10, 3, 2), Some(23)); + assert_eq!(channel(f64::NAN), 0); + assert_eq!(channel(1e300), 255); + assert_eq!(channel(-4.0), 0); + assert_eq!(channel(127.5), 128); + } + + #[test] + fn a_shared_edge_is_cut_at_the_same_point_from_either_triangle() { + // Cutting this edge from P towards Q and from Q towards P rounds differently. + let view = View::Perspective { focal: 1.0, aspect: 1.0, near: 0.5 }; + let (p, q) = ([0.1, 0.1, 0.2, 1.0, 1.0, 1.0], [0.7, 0.2, 1.6, 1.0, 1.0, 1.0]); + let one = view.clip([p, q, [0.0, 3.0, 2.0, 1.0, 1.0, 1.0]]); + let other = view.clip([q, p, [3.0, 0.0, 2.0, 1.0, 1.0, 1.0]]); + let shared = one.iter().filter(|a| other.iter().any(|b| a[..3] == b[..3])).count(); + assert_eq!(shared, 2, "Q and the cut on PQ: {one:?} {other:?}"); + } + + fn corner(x: f64, y: f64, key: f64, shade: f64) -> Corner { + Corner { at: [x, y, key], shade: [shade; 3] } + } + + #[test] + fn a_pixel_on_an_edge_two_triangles_share_is_blended_once() { + let (mut rgba, mut depth) = (vec![0; 64], vec![f64::NEG_INFINITY; 16]); + let mut t = Target { perspective: false, width: 4, height: 4, rgba: &mut rgba, depth: &mut depth, work: u64::MAX }; + let [a, b, c, d] = [(0.0, 0.0), (4.0, 0.0), (4.0, 4.0), (0.0, 4.0)].map(|(x, y)| corner(x, y, 1.0, 1.0)); + t.fill([a, b, c], [255, 0, 0, 128]).unwrap(); + t.fill([a, c, d], [255, 0, 0, 128]).unwrap(); + // The diagonal runs through four pixel centres; each must look like the rest. + for px in rgba.chunks(4) { + assert_eq!(px, [255, 0, 0, 128]); + } + } + + #[test] + fn a_perspective_shade_is_interpolated_on_the_surface_not_the_screen() { + let (mut rgba, mut depth) = (vec![0; 64], vec![f64::NEG_INFINITY; 16]); + let mut t = Target { perspective: true, width: 4, height: 4, rgba: &mut rgba, depth: &mut depth, work: u64::MAX }; + // The x = 4 corner is three times as far away as the others, and the only lit one. + t.fill([corner(0.0, 0.0, 1.0, 0.0), corner(4.0, 0.0, 1.0 / 3.0, 1.0), corner(0.0, 4.0, 1.0, 0.0)], [240, 240, 240, 255]).unwrap(); + // At screen (1.5, 0.5) the weights are 0.5, 0.375, 0.125: on the surface the far corner + // counts 0.125 / 0.75 = 1/6, where on the screen it would count 0.375. + assert_eq!(rgba.get(4..8), Some(&[40, 40, 40, 255][..])); + } + + #[test] + fn fills_stop_when_the_work_allowance_is_spent() { + let (mut rgba, mut depth) = (vec![0; 64], vec![f64::NEG_INFINITY; 16]); + // Each fill's box is the whole 4x4 image: 16 pixel centres. + let mut t = Target { perspective: false, width: 4, height: 4, rgba: &mut rgba, depth: &mut depth, work: 40 }; + let tri = [corner(0.0, 0.0, 1.0, 1.0), corner(4.0, 0.0, 1.0, 1.0), corner(0.0, 4.0, 1.0, 1.0)]; + assert_eq!(t.fill(tri, [255, 0, 0, 255]), Ok(())); + assert_eq!(t.fill(tri, [255, 0, 0, 255]), Ok(())); + assert_eq!(t.fill(tri, [255, 0, 0, 255]), Err(RenderError::TooMuchWork)); + assert_eq!(t.work, 8); + } +} diff --git a/crates/prc-render/src/render.rs b/crates/prc-render/src/render.rs new file mode 100644 index 0000000000..bcaf8acced --- /dev/null +++ b/crates/prc-render/src/render.rs @@ -0,0 +1,388 @@ +//! The camera, the render options and [`render`]: world meshes to an RGBA image. +//! +//! One sample per pixel centre, no antialiasing. Faces are lit on both sides because a +//! producer's winding is not trusted to face outward. + +use crate::WorldMesh; +use crate::raster::{Corner, Target, View, ViewCorner, at}; +use crate::vec3::{cross, dot, length, normalize, scale, span, sub, unit_cross}; +use pdfcraft_prc::TriangleMesh; + +/// The largest image [`render`] draws, in pixels (width × height). +pub const MAX_RENDER_PIXELS: u64 = 16 * 1024 * 1024; + +/// The most pixel centres one [`render`] tests, summed over each drawn triangle's bounding box +/// clipped to the image: 128 full layers of the largest image. +pub const MAX_RENDER_WORK: u64 = 1 << 31; + +/// [`Camera::fit`]'s framed extent over the model's: 5% clear on each side. +const FIT_MARGIN: f64 = 1.1; + +/// How the camera projects the scene onto the image. +#[derive(Debug, Clone, Copy, PartialEq)] +pub enum Projection { + /// A pinhole camera with this vertical field of view, in degrees (exclusive range 0–180). + Perspective { + /// Vertical field of view, degrees. + fov_y: f64, + }, + /// A parallel projection showing this much of the scene vertically, in model units. + Orthographic { + /// Visible height, model units. + height: f64, + }, +} + +/// Where the camera is and where it looks, in world coordinates. +#[derive(Debug, Clone, Copy, PartialEq)] +pub struct Camera { + /// The camera position. + pub eye: [f64; 3], + /// The point at the image centre. + pub target: [f64; 3], + /// The direction that appears upward; must not be parallel to the view direction. + pub up: [f64; 3], + /// The projection. + pub projection: Projection, +} + +/// An axis-aligned box around every point a set of meshes draws. +#[derive(Debug, Clone, Copy, PartialEq)] +pub struct Bounds { + /// The smallest x, y and z. + pub min: [f64; 3], + /// The largest x, y and z. + pub max: [f64; 3], +} + +impl Bounds { + /// The box around every finite vertex the meshes' drawn triangles use, or `None` when there + /// is none. + #[must_use] + pub fn of(meshes: &[WorldMesh]) -> Option { + let mut out: Option = None; + for p in vertices(meshes) { + let b = out.get_or_insert(Bounds { min: p, max: p }); + for ((lo, hi), c) in b.min.iter_mut().zip(b.max.iter_mut()).zip(p) { + *lo = lo.min(c); + *hi = hi.max(c); + } + } + out + } + + /// The box's centre, finite for any finite box. + #[must_use] + pub fn centre(&self) -> [f64; 3] { + [0, 1, 2].map(|i| at(self.min, i) / 2.0 + at(self.max, i) / 2.0) + } + + /// Half the box's diagonal: the radius of a sphere holding it. + #[must_use] + pub fn radius(&self) -> f64 { + length(sub(self.max, self.min)) / 2.0 + } +} + +/// Every corner of the triangles [`render`] draws: those not fully transparent whose three +/// corners exist and are finite. +fn vertices(meshes: &[WorldMesh]) -> impl Iterator + Clone + '_ { + meshes + .iter() + .flat_map(|m| m.mesh.triangles.iter().enumerate().filter(|&(t, _)| !invisible(m, t)).filter_map(|(_, tri)| corners(&m.mesh, tri)).flatten()) +} + +/// Whether triangle `t` of `m` is fully transparent, so [`render`] does not draw it. +fn invisible(m: &WorldMesh, t: usize) -> bool { + m.colour_of(t).is_some_and(|c| c[3] == 0) +} + +impl Camera { + /// A camera looking along `direction` that fits `bounds` in an image of the given aspect + /// ratio (width / height), with a 30° perspective or an orthographic projection: the box's + /// eight corners framed from this direction, about 5% clear on each side of the tighter axis. + /// + /// # Errors + /// [`RenderError::Camera`] when `direction` is zero or parallel to `up`, or `aspect` is not positive. + pub fn fit(bounds: &Bounds, direction: [f64; 3], up: [f64; 3], perspective: bool, aspect: f64) -> Result { + let corners = (0..8).map(|k| [0, 1, 2].map(|i| if k >> i & 1 == 0 { at(bounds.min, i) } else { at(bounds.max, i) })); + Camera::frame(bounds, corners, direction, up, (perspective, aspect)) + } + + /// As [`Camera::fit`], framing every vertex the meshes' drawn triangles use rather than their box, + /// so an oblique view is not padded by empty box corners. + /// + /// # Errors + /// As [`Camera::fit`], and when the meshes have no finite vertex. + pub fn fit_meshes(meshes: &[WorldMesh], direction: [f64; 3], up: [f64; 3], perspective: bool, aspect: f64) -> Result { + let bounds = Bounds::of(meshes).ok_or(RenderError::Camera("the model has no vertex"))?; + Camera::frame(&bounds, vertices(meshes), direction, up, (perspective, aspect)) + } + + /// Frames `points` (which `bounds` holds) from `direction`, aimed at the middle of their projection. + fn frame( + bounds: &Bounds, + points: impl Iterator + Clone, + direction: [f64; 3], + up: [f64; 3], + (perspective, aspect): (bool, f64), + ) -> Result { + if !(aspect.is_finite() && aspect > 0.0) { + return Err(RenderError::Camera("the aspect ratio is not positive")); + } + let dir = normalize(direction).ok_or(RenderError::Camera("the view direction is zero"))?; + let centre = bounds.centre(); + let radius = match bounds.radius() { + r if r.is_finite() && r > 0.0 => r, + _ => 1.0, + }; + let [right, upward, forward] = axes(dir, up)?; + let view = |p: [f64; 3]| { + let r = sub(p, centre); + [dot(r, right), dot(r, upward), dot(r, forward)] + }; + let ((x0, x1), (y0, y1)) = points + .clone() + .map(view) + .fold(((f64::INFINITY, f64::NEG_INFINITY), (f64::INFINITY, f64::NEG_INFINITY)), |((a0, a1), (b0, b1)), [x, y, _]| { + ((a0.min(x), a1.max(x)), (b0.min(y), b1.max(y))) + }); + if !(x0 <= x1 && y0 <= y1) { + return Err(RenderError::Camera("the model has no vertex")); + } + let (mx, my) = ((x0 + x1) / 2.0, (y0 + y1) / 2.0); + let target = [0, 1, 2].map(|i| at(centre, i) + mx * at(right, i) + my * at(upward, i)); + let floor = radius * 1e-3; + let (projection, distance) = if perspective { + let fov_y: f64 = 30.0; + let ty = (fov_y.to_radians() / 2.0).tan() / FIT_MARGIN; + let tx = ty * aspect; + // A point at lateral (x, y) from the target and depth f past it is in view when + // |x| <= (d + f)·tx and |y| <= (d + f)·ty. + let d = points.map(view).fold(floor, |d, [x, y, f]| { + let need = ((x - mx).abs() / tx).max((y - my).abs() / ty); + d.max(need - f).max(floor - f) + }); + (Projection::Perspective { fov_y }, d) + } else { + let height = ((y1 - y0).max((x1 - x0) / aspect) * FIT_MARGIN).max(floor); + (Projection::Orthographic { height }, 2.0 * radius) + }; + let camera = Camera { eye: sub(target, scale(dir, distance)), target, up, projection }; + camera.basis()?; + Ok(camera) + } + + /// Right, up and forward unit vectors. + pub(crate) fn basis(&self) -> Result<[[f64; 3]; 3], RenderError> { + let forward = normalize(sub(self.target, self.eye)).ok_or(RenderError::Camera("the eye and the target coincide"))?; + axes(forward, self.up) + } +} + +/// Right, up and `forward` (a unit vector) for a camera whose up direction is `up`. +fn axes(forward: [f64; 3], up: [f64; 3]) -> Result<[[f64; 3]; 3], RenderError> { + let right = normalize(cross(forward, up)).ok_or(RenderError::Camera("the up direction is zero or parallel to the view direction"))?; + Ok([right, cross(right, forward), forward]) +} + +/// How surfaces are lit. +#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)] +#[non_exhaustive] +pub enum Lighting { + /// One light at the camera: a surface's colour times `0.3 + 0.7·|n·v|`, where `v` points from + /// the surface to the eye (orthographic: against the view direction). + #[default] + Headlight, + /// No shading: every surface in its own colour. + Unlit, +} + +/// What [`render`] draws. +#[derive(Debug, Clone, Copy, PartialEq, Eq)] +pub struct RenderOptions { + /// Image width, pixels. + pub width: u32, + /// Image height, pixels. + pub height: u32, + /// Background colour, straight RGBA. + pub background: [u8; 4], + /// The colour of a surface with none of its own. + pub colour: [u8; 3], + /// How surfaces are lit. + pub lighting: Lighting, +} + +impl Default for RenderOptions { + /// 1024 × 768, opaque white background, light grey surface, the headlight. + fn default() -> Self { + RenderOptions { width: 1024, height: 768, background: [255, 255, 255, 255], colour: [190, 192, 200], lighting: Lighting::Headlight } + } +} + +/// A rendered image: `rgba` holds `width * height` straight-alpha pixels, row by row from the top. +#[derive(Debug, Clone, PartialEq, Eq)] +pub struct Image { + /// Width, pixels. + pub width: u32, + /// Height, pixels. + pub height: u32, + /// The pixels. + pub rgba: Vec, +} + +/// Why [`render`], [`Camera::fit`] or [`crate::assemble`] refused. +#[derive(Debug, Clone, PartialEq, Eq, thiserror::Error)] +#[non_exhaustive] +pub enum RenderError { + /// The image is empty or larger than [`MAX_RENDER_PIXELS`]. + #[error("a {width}x{height} image is empty or larger than the render ceiling")] + Size { + /// Requested width. + width: u32, + /// Requested height. + height: u32, + }, + /// The camera cannot form a view. + #[error("invalid camera: {0}")] + Camera(&'static str), + /// The model could not be read. + #[error("unreadable model: {0}")] + Model(#[from] pdfcraft_prc::PrcError), + /// The placed model has more than [`crate::MAX_ASSEMBLED_TRIANGLES`] triangles. + #[error("the placed model has more triangles than the assembly ceiling")] + TooManyTriangles, + /// The placed model has more than [`crate::MAX_ASSEMBLED_VERTICES`] positions, normals and + /// texture coordinates. + #[error("the placed model has more vertices than the assembly ceiling")] + TooManyVertices, + /// Drawing would test more than [`MAX_RENDER_WORK`] pixel centres. + #[error("the render would test more pixels than the work ceiling")] + TooMuchWork, +} + +/// Draw `meshes` as seen by `camera`. +/// +/// Each triangle is filled in its own colour ([`WorldMesh::triangle_colours`]), else its mesh's +/// ([`WorldMesh::colour`]), else [`RenderOptions::colour`], and lit per [`RenderOptions::lighting`]; +/// nearer surfaces hide farther ones. Opaque triangles are drawn first; translucent ones are then +/// blended over them, hidden by nearer opaque surfaces but not by each other, in mesh order; fully +/// transparent ones are not drawn. Triangles with a non-finite or out-of-range vertex are skipped. +/// A perspective camera clips at a near plane 1/10,000 of the way from the eye to the target, so +/// what lies behind the eye, or in front of it but nearer than that, is not drawn; an orthographic +/// camera does not clip. +/// +/// # Errors +/// [`RenderError::Size`] for an empty or over-ceiling image; [`RenderError::Camera`] for a camera +/// that cannot form a view (eye on the target, `up` parallel to the view, a field of view outside +/// 0–180° or too small for a finite focal length, a non-positive orthographic height); +/// [`RenderError::TooMuchWork`] past [`MAX_RENDER_WORK`]. +pub fn render(meshes: &[WorldMesh], camera: &Camera, options: &RenderOptions) -> Result { + let (width, height) = (options.width, options.height); + let size = RenderError::Size { width, height }; + let pixels = u64::from(width) * u64::from(height); + if pixels == 0 || pixels > MAX_RENDER_PIXELS { + return Err(size); + } + let pixels = usize::try_from(pixels).map_err(|_| size)?; + let view = View::new(camera, f64::from(width) / f64::from(height))?; + let perspective = matches!(view, View::Perspective { .. }); + let scene = Scene { eye: camera.eye, basis: camera.basis()?, view, lighting: options.lighting }; + let mut rgba = options.background.repeat(pixels); + let mut depth = vec![f64::NEG_INFINITY; pixels]; + let mut target = Target { perspective, width, height, rgba: &mut rgba, depth: &mut depth, work: MAX_RENDER_WORK }; + let [r, g, b] = options.colour; + for opaque in [true, false] { + for m in meshes { + for t in 0..m.mesh.triangles.len() { + let colour = m.colour_of(t).unwrap_or([r, g, b, 255]); + if (colour[3] == 255) == opaque && colour[3] != 0 { + scene.draw(&mut target, &m.mesh, t, colour)?; + } + } + } + } + Ok(Image { width, height, rgba }) +} + +/// The camera a frame is drawn from, with its basis and projection. +struct Scene { + eye: [f64; 3], + /// Right, up and forward. + basis: [[f64; 3]; 3], + view: View, + lighting: Lighting, +} + +impl Scene { + /// The shade at world point `p` with normal `n`. + fn shade(&self, p: [f64; 3], n: [f64; 3]) -> [f64; 3] { + if self.lighting == Lighting::Unlit { + return [1.0; 3]; + } + let towards = match self.view { + View::Perspective { .. } => normalize(sub(self.eye, p)), + View::Orthographic { .. } => Some(scale(at3(&self.basis, 2), -1.0)), + }; + [0.3 + 0.7 * towards.map_or(1.0, |v| dot(n, v).abs().min(1.0)); 3] + } + + /// Draw triangle `t` of `mesh` into `target` in straight-RGBA `colour`; refused once the + /// target's work allowance is spent. + fn draw(&self, target: &mut Target, mesh: &TriangleMesh, t: usize, colour: [u8; 4]) -> Result<(), RenderError> { + let Some(world) = mesh.triangles.get(t).and_then(|tri| corners(mesh, tri)) else { + return Ok(()); + }; + let Some(face) = unit_cross(span(at3(&world, 0), at3(&world, 1)), span(at3(&world, 0), at3(&world, 2))) else { + return Ok(()); + }; + let stored = stored_normals(mesh, t); + let normal = |k: usize| stored.and_then(|n| n.get(k).copied().flatten()).unwrap_or(face); + // Without stored normals the triangle is flat: one shade, lit along the view ray to its first corner. + let flat = stored.is_none().then(|| self.shade(at3(&world, 0), normal(0))); + let [right, up, forward] = self.basis; + let view_corners = [0, 1, 2].map(|k| -> ViewCorner { + let p = at3(&world, k); + let [r, g, b] = flat.unwrap_or_else(|| self.shade(p, normal(k))); + let d = sub(p, self.eye); + [dot(d, right), dot(d, up), dot(d, forward), r, g, b] + }); + let screen: Vec = self.view.clip(view_corners).into_iter().map(|v| self.view.project(v, target.width, target.height)).collect(); + if let Some((&a, rest)) = screen.split_first() { + for pair in rest.windows(2) { + if let [b, c] = pair { + target.fill([a, *b, *c], colour)?; + } + } + } + Ok(()) + } +} + +/// Triangle `t`'s stored corner normals, unit length; a corner whose normal is zero or not finite +/// is `None`, and so is the whole when the mesh stores none. +fn stored_normals(mesh: &TriangleMesh, t: usize) -> Option<[Option<[f64; 3]>; 3]> { + let slots = mesh.triangle_normals.get(t)?; + Some(slots.map(|i| mesh.normals.get(usize::try_from(i).ok()?).copied().and_then(normalize))) +} + +/// The three corners of a triangle, or `None` when one is missing or not finite. +fn corners(mesh: &TriangleMesh, tri: &[u32; 3]) -> Option<[[f64; 3]; 3]> { + let mut out = [[0.0; 3]; 3]; + for (slot, &i) in out.iter_mut().zip(tri) { + let p = *mesh.positions.get(usize::try_from(i).ok()?)?; + if !p.iter().all(|c| c.is_finite()) { + return None; + } + *slot = p; + } + Some(out) +} + +/// Corner `i` of a triangle, the origin when out of range. +fn at3(v: &[[f64; 3]; 3], i: usize) -> [f64; 3] { + v.get(i).copied().unwrap_or([0.0; 3]) +} + +#[cfg(test)] +mod tests; diff --git a/crates/prc-render/src/render/tests.rs b/crates/prc-render/src/render/tests.rs new file mode 100644 index 0000000000..586292ef80 --- /dev/null +++ b/crates/prc-render/src/render/tests.rs @@ -0,0 +1,347 @@ +#![allow(clippy::unwrap_used, clippy::panic, clippy::unreachable)] + +use super::*; + +fn mesh(positions: Vec<[f64; 3]>, triangles: Vec<[u32; 3]>) -> TriangleMesh { + let mut m = TriangleMesh::default(); + m.positions = positions; + m.triangles = triangles; + m +} + +fn quad_mesh(z: f64, half: f64, tilt: f64) -> TriangleMesh { + mesh(vec![[-half, -half, z - tilt], [half, -half, z + tilt], [half, half, z + tilt], [-half, half, z - tilt]], vec![[0, 1, 2], [0, 2, 3]]) +} + +fn quad(z: f64, half: f64, tilt: f64) -> WorldMesh { + WorldMesh::new(quad_mesh(z, half, tilt)) +} + +fn painted(m: WorldMesh, colour: Option<[u8; 4]>) -> WorldMesh { + WorldMesh { colour, ..m } +} + +fn ortho(height: f64) -> Camera { + Camera { eye: [0.0, 0.0, 10.0], target: [0.0, 0.0, 0.0], up: [0.0, 1.0, 0.0], projection: Projection::Orthographic { height } } +} + +fn small() -> RenderOptions { + RenderOptions { width: 40, height: 40, ..RenderOptions::default() } +} + +fn pixel(image: &Image, x: u32, y: u32) -> [u8; 4] { + let i = (y * image.width + x) as usize * 4; + image.rgba[i..i + 4].try_into().unwrap() +} + +#[test] +fn a_facing_square_fills_the_middle_and_leaves_the_corners() { + let image = render(&[quad(0.0, 1.0, 0.0)], &ortho(4.0), &small()).unwrap(); + assert_eq!(pixel(&image, 20, 20), [190, 192, 200, 255]); + assert_eq!(pixel(&image, 1, 1), [255, 255, 255, 255]); + // Half the view height is covered: pixels 10..30. + assert_eq!(pixel(&image, 10, 20)[0], 190); + assert_eq!(pixel(&image, 9, 20)[0], 255); +} + +#[test] +fn stored_normals_shade_smoothly_across_a_flat_square() { + let mut m = quad_mesh(0.0, 1.0, 0.0); + let s = 3f64.sqrt() / 2.0; + // The left corners lean 60° away from the viewer; the right face it. + m.normals = vec![[s, 0.0, 0.5], [0.0, 0.0, 2.0]]; + m.triangle_normals = vec![[0, 1, 1], [0, 1, 0]]; + let image = render(&[WorldMesh::new(m.clone())], &ortho(4.0), &small()).unwrap(); + let red = |x| pixel(&image, x, 20)[0]; + assert!(red(11) < red(20) && red(20) < red(29), "{} {} {}", red(11), red(20), red(29)); + // Shade 0.3 + 0.7 cos θ at the edges: 0.65 left, 1.0 right. + assert!((i32::from(red(10)) - 128).abs() <= 4, "{}", red(10)); + assert!(red(29) >= 186); + m.normals.clear(); + m.triangle_normals.clear(); + let flat = render(&[WorldMesh::new(m)], &ortho(4.0), &small()).unwrap(); + assert_eq!(pixel(&flat, 11, 20), pixel(&flat, 29, 20)); +} + +#[test] +fn the_nearer_surface_hides_the_farther_whatever_the_order() { + let far = quad(-1.0, 2.0, 1.0); + let near = quad(1.0, 0.5, 0.0); + for meshes in [[far.clone(), near.clone()], [near, far]] { + let image = render(&meshes, &ortho(8.0), &small()).unwrap(); + assert_eq!(pixel(&image, 20, 20), [190, 192, 200, 255]); + assert!(pixel(&image, 20, 12)[0] < 190, "the tilted far square is darker"); + } +} + +const RED: Option<[u8; 4]> = Some([255, 0, 0, 255]); +const GLASS: Option<[u8; 4]> = Some([0, 0, 255, 102]); + +/// Glass (40% blue) in front of red blends whichever is listed first, glass behind red is hidden, +/// a clear mesh draws nothing, and a mesh with no colour takes the options' colour. +#[test] +fn meshes_draw_in_their_own_colours_and_glass_blends() { + let unlit = RenderOptions { lighting: Lighting::Unlit, ..small() }; + let at_middle = |meshes: &[(f64, Option<[u8; 4]>)]| { + let meshes: Vec = meshes.iter().map(|&(z, c)| painted(quad(z, 1.0, 0.0), c)).collect(); + pixel(&render(&meshes, &ortho(4.0), &unlit).unwrap(), 20, 20) + }; + assert_eq!(at_middle(&[(1.0, GLASS), (-1.0, RED)]), [153, 0, 102, 255]); + assert_eq!(at_middle(&[(-1.0, RED), (1.0, GLASS)]), [153, 0, 102, 255]); + assert_eq!(at_middle(&[(-1.0, GLASS), (1.0, RED)]), [255, 0, 0, 255]); + assert_eq!(at_middle(&[(1.0, Some([0, 0, 255, 0])), (-1.0, RED)]), [255, 0, 0, 255]); + assert_eq!(at_middle(&[(1.0, None)]), [190, 192, 200, 255]); + assert_eq!(at_middle(&[(2.0, GLASS), (1.0, GLASS), (-1.0, RED)]), [92, 0, 163, 255], "glass does not hide glass"); + let clear = RenderOptions { background: [0, 0, 0, 0], ..unlit }; + let image = render(&[painted(quad(1.0, 1.0, 0.0), GLASS)], &ortho(4.0), &clear).unwrap(); + assert_eq!(pixel(&image, 20, 20), [0, 0, 255, 102], "glass over nothing"); +} + +#[test] +fn a_triangle_colour_overrides_its_mesh_colour() { + let mut m = painted(quad(0.0, 1.0, 0.0), RED); + // The lower-right triangle is green; the upper-left keeps the mesh's red. + m.triangle_colours = Some(vec![Some([0, 255, 0, 255]), None]); + let unlit = RenderOptions { lighting: Lighting::Unlit, ..small() }; + let image = render(&[m], &ortho(4.0), &unlit).unwrap(); + assert_eq!(pixel(&image, 25, 25), [0, 255, 0, 255]); + assert_eq!(pixel(&image, 14, 14), [255, 0, 0, 255]); +} + +#[test] +fn each_edge_bounds_a_triangle() { + let tri = WorldMesh::new(mesh(vec![[-1.0, -1.0, 0.0], [1.0, -1.0, 0.0], [0.0, 1.0, 0.0]], vec![[0, 1, 2]])); + let image = render(&[tri], &ortho(4.0), &small()).unwrap(); + assert_ne!(pixel(&image, 20, 25)[0], 255); + // Just past each edge: below the base, left and right of the apex. + for (x, y) in [(20, 31), (13, 15), (27, 15)] { + assert_eq!(pixel(&image, x, y)[0], 255, "({x},{y})"); + } +} + +#[test] +fn a_face_turned_away_is_still_drawn() { + let mut back = quad_mesh(0.0, 1.0, 0.0); + for t in &mut back.triangles { + t.swap(1, 2); + } + let image = render(&[WorldMesh::new(back)], &ortho(4.0), &small()).unwrap(); + assert_eq!(pixel(&image, 20, 20), [190, 192, 200, 255]); +} + +#[test] +fn perspective_clips_what_is_behind_the_camera() { + let camera = Camera { projection: Projection::Perspective { fov_y: 60.0 }, ..ortho(1.0) }; + let image = render(&[quad(20.0, 1.0, 0.0)], &camera, &small()).unwrap(); + assert!(image.rgba.chunks(4).all(|p| p == [255, 255, 255, 255])); + // A floor running from behind the camera to beyond the target. + let floor = WorldMesh::new(mesh(vec![[-5.0, -1.0, 20.0], [5.0, -1.0, 20.0], [0.0, -1.0, -20.0]], vec![[0, 1, 2]])); + let image = render(&[floor], &camera, &small()).unwrap(); + assert_ne!(pixel(&image, 20, 38), [255, 255, 255, 255]); + assert_eq!(pixel(&image, 20, 2), [255, 255, 255, 255]); +} + +#[test] +fn a_fitted_camera_frames_the_model() { + // Depth matters in perspective: the nearer square looks larger. + let meshes = [quad(3.0, 3.0, 0.0), quad(-3.0, 3.0, 0.0)]; + let bounds = Bounds::of(&meshes).unwrap(); + for perspective in [true, false] { + let camera = Camera::fit(&bounds, [0.0, 0.0, -1.0], [0.0, 1.0, 0.0], perspective, 2.0).unwrap(); + let image = render(&meshes, &camera, &RenderOptions { width: 80, height: 40, ..RenderOptions::default() }).unwrap(); + assert_ne!(pixel(&image, 40, 20)[0], 255); + // Nothing drawn touches the image border. + for x in 0..80 { + assert_eq!(pixel(&image, x, 0)[0], 255); + assert_eq!(pixel(&image, x, 39)[0], 255); + } + } +} + +#[test] +fn a_mesh_fit_ignores_empty_box_corners() { + // A thin strip along the box's diagonal: seen across the diagonal, the box is as tall as it + // is wide, the strip is not. + let meshes = [WorldMesh::new(mesh(vec![[0.0, 0.0, 0.0], [10.0, 10.0, 0.0], [10.0, 10.01, 0.0]], vec![[0, 1, 2]]))]; + let (dir, up) = ([0.0, 0.0, -1.0], [1.0, -1.0, 0.0]); + let height = |c: Camera| match c.projection { + Projection::Orthographic { height } => height, + Projection::Perspective { .. } => unreachable!(), + }; + let boxed = height(Camera::fit(&Bounds::of(&meshes).unwrap(), dir, up, false, 2.0).unwrap()); + let tight = height(Camera::fit_meshes(&meshes, dir, up, false, 2.0).unwrap()); + assert!((boxed / tight - 2.0).abs() < 0.01, "{boxed} vs {tight}"); + let near = Camera::fit_meshes(&meshes, dir, up, true, 2.0).unwrap(); + let far = Camera::fit(&Bounds::of(&meshes).unwrap(), dir, up, true, 2.0).unwrap(); + assert!(length(sub(near.eye, near.target)) < 0.6 * length(sub(far.eye, far.target))); + + // Lopsided in its box: the limiting vertex sits exactly on the margin, measured from the + // re-centred target. Narrow (width limits) and wide (height limits). + let tri = WorldMesh::new(mesh(vec![[0.0, 0.0, 0.0], [2.0, 0.0, 0.0], [1.0, 2.0, 1.0]], vec![[0, 1, 2]])); + for aspect in [0.5, 3.0] { + let c = Camera::fit_meshes(std::slice::from_ref(&tri), [0.0, 0.3, -1.0], [1.0, 1.0, 0.0], true, aspect).unwrap(); + let [right, upward, forward] = c.basis().unwrap(); + let ty = 15f64.to_radians().tan() / FIT_MARGIN; + let fill = tri + .mesh + .positions + .iter() + .map(|&p| { + let r = sub(p, c.eye); + (dot(r, right).abs() / (ty * aspect)).max(dot(r, upward).abs() / ty) / dot(r, forward) + }) + .fold(0.0, f64::max); + assert!((fill - 1.0).abs() < 1e-6, "aspect {aspect}: {fill}"); + } +} + +#[test] +fn a_fitted_camera_fills_the_tighter_axis() { + let drawn = |image: &Image, horizontal: bool, line: u32| { + let n = if horizontal { image.width } else { image.height }; + (0..n) + .filter(|&i| { + let (x, y) = if horizontal { (i, line) } else { (line, i) }; + pixel(image, x, y)[0] != 255 + }) + .count() + }; + // Height-limited: a 6 x 6 square in an 80 x 40 image. + let square = [quad(0.0, 3.0, 0.0)]; + // Width-limited and off-origin: a 10 x 1 strip in a 40 x 40 image. + let strip = [WorldMesh::new(mesh(vec![[0.0, 0.0, 0.0], [10.0, 0.0, 0.0], [10.0, 1.0, 0.0], [0.0, 1.0, 0.0]], vec![[0, 1, 2], [0, 2, 3]]))]; + for perspective in [true, false] { + let fit = + |meshes: &[WorldMesh], aspect| Camera::fit(&Bounds::of(meshes).unwrap(), [0.0, 0.0, -1.0], [0.0, 1.0, 0.0], perspective, aspect).unwrap(); + let wide = RenderOptions { width: 80, height: 40, ..RenderOptions::default() }; + let rows = drawn(&render(&square, &fit(&square, 2.0), &wide).unwrap(), false, 40); + assert!((34..=37).contains(&rows), "{perspective}: {rows} of 40 rows"); + let image = render(&strip, &fit(&strip, 1.0), &small()).unwrap(); + let cols = drawn(&image, true, 20); + assert!((34..=37).contains(&cols), "{perspective}: {cols} of 40 columns"); + // Centred: the margins differ by at most a pixel. + let left = (0..40).take_while(|&x| pixel(&image, x, 20)[0] == 255).count(); + assert!(left.abs_diff(40 - cols - left) <= 1, "{perspective}: left {left}"); + } +} + +#[test] +fn bad_sizes_and_cameras_are_refused() { + let m = [quad(0.0, 1.0, 0.0)]; + let size = |width, height| RenderOptions { width, height, ..RenderOptions::default() }; + assert!(matches!(render(&m, &ortho(1.0), &size(0, 10)), Err(RenderError::Size { .. }))); + assert!(matches!(render(&m, &ortho(1.0), &size(10_000, 10_000)), Err(RenderError::Size { .. }))); + let bad = [ + Camera { eye: [0.0; 3], target: [0.0; 3], ..ortho(1.0) }, + Camera { up: [0.0, 0.0, 1.0], ..ortho(1.0) }, + Camera { projection: Projection::Perspective { fov_y: 180.0 }, ..ortho(1.0) }, + Camera { projection: Projection::Perspective { fov_y: f64::NAN }, ..ortho(1.0) }, + Camera { projection: Projection::Perspective { fov_y: 1e-310 }, ..ortho(1.0) }, + ortho(0.0), + ortho(f64::INFINITY), + ]; + for camera in bad { + assert!(matches!(render(&m, &camera, &small()), Err(RenderError::Camera(_))), "{camera:?}"); + } + let b = Bounds::of(&m).unwrap(); + assert!(Camera::fit(&b, [0.0; 3], [0.0, 1.0, 0.0], true, 1.0).is_err()); + assert!(Camera::fit(&b, [0.0, 0.0, 1.0], [0.0, 1.0, 0.0], true, 0.0).is_err()); + assert!(Camera::fit_meshes(&[], [0.0, 0.0, 1.0], [0.0, 1.0, 0.0], true, 1.0).is_err()); +} + +#[test] +fn non_finite_and_missing_vertices_are_skipped() { + let m = WorldMesh::new(mesh(vec![[0.0, 0.0, 0.0], [f64::NAN, 0.0, 0.0], [0.0, 1.0, 0.0]], vec![[0, 1, 2], [0, 2, 9], [0, 2, u32::MAX]])); + let image = render(std::slice::from_ref(&m), &ortho(4.0), &small()).unwrap(); + assert!(image.rgba.chunks(4).all(|p| p == [255, 255, 255, 255])); + // No triangle can be drawn, so none frames the model. + assert_eq!(Bounds::of(&[m]), None); +} + +#[test] +fn unlit_draws_the_plain_colour_and_the_headlight_shades_by_angle() { + let lit = |lighting, tilt| { + let image = render(&[quad(0.0, 1.0, tilt)], &ortho(4.0), &RenderOptions { lighting, ..small() }).unwrap(); + pixel(&image, 20, 20) + }; + assert_eq!(lit(Lighting::Headlight, 0.0), [190, 192, 200, 255]); + // Tilted 45°: 0.3 + 0.7 cos 45° ≈ 0.795 of the colour. + assert_eq!(lit(Lighting::Headlight, 1.0), [151, 153, 159, 255]); + assert_eq!(lit(Lighting::Unlit, 1.0), [190, 192, 200, 255]); +} + +#[test] +fn a_huge_square_is_drawn() { + // Its edges' cross product (about 1e320) overflows unless they are scaled first. + let image = render(&[quad(0.0, 1e160, 0.0)], &ortho(4e160), &small()).unwrap(); + assert_eq!(pixel(&image, 20, 20), [190, 192, 200, 255]); + assert_eq!(pixel(&image, 1, 1), [255, 255, 255, 255]); +} + +#[test] +fn a_camera_very_far_away_still_draws() { + let camera = Camera { eye: [0.0, 0.0, 1e200], target: [0.0; 3], up: [0.0, 1.0, 0.0], projection: Projection::Perspective { fov_y: 30.0 } }; + let image = render(&[quad(0.0, 1e199, 0.0)], &camera, &small()).unwrap(); + // The headlight meets the square almost head-on at the centre pixel. + let [r, _, _, a] = pixel(&image, 20, 20); + assert!((185..=190).contains(&r) && a == 255, "{:?}", pixel(&image, 20, 20)); +} + +#[test] +fn the_centre_of_a_box_near_the_largest_float_is_finite() { + let b = Bounds { min: [1e308; 3], max: [1.5e308; 3] }; + assert!(b.centre().iter().all(|c| c.is_finite()), "{:?}", b.centre()); +} + +#[test] +fn a_triangle_that_is_not_drawn_does_not_widen_the_box() { + // The second triangle has a non-finite corner, so its finite far corner is not drawn either. + let m = mesh(vec![[0.0; 3], [1.0, 0.0, 0.0], [0.0, 1.0, 0.0], [f64::NAN, 0.0, 0.0], [1e6, 0.0, 0.0]], vec![[0, 1, 2], [3, 4, 0]]); + let b = Bounds::of(&[WorldMesh::new(m)]).unwrap(); + assert_eq!(b.max, [1.0, 1.0, 0.0]); +} + +#[test] +fn a_fully_transparent_triangle_does_not_widen_the_box_or_the_fit() { + // The far triangle is clear, so the camera frames the near one as if it were alone. + let m = mesh(vec![[0.0; 3], [1.0, 0.0, 0.0], [0.0, 1.0, 0.0], [1e6, 0.0, 0.0], [1e6, 1.0, 0.0]], vec![[0, 1, 2], [3, 4, 1]]); + let clear = WorldMesh { triangle_colours: Some(vec![None, Some([0, 0, 0, 0])]), ..WorldMesh::new(m.clone()) }; + assert_eq!(Bounds::of(std::slice::from_ref(&clear)).unwrap().max, [1.0, 1.0, 0.0]); + let alone = WorldMesh::new(mesh(m.positions, vec![[0, 1, 2]])); + let fit = |w: &WorldMesh| Camera::fit_meshes(std::slice::from_ref(w), [0.0, 0.0, -1.0], [0.0, 1.0, 0.0], true, 1.0).unwrap(); + assert_eq!(fit(&clear), fit(&alone)); + // A whole mesh painted clear is not framed either. + assert_eq!( + Bounds::of(&[painted(WorldMesh::new(mesh(vec![[0.0; 3], [9.0, 0.0, 0.0], [0.0, 9.0, 0.0]], vec![[0, 1, 2]])), Some([1, 2, 3, 0]))]), + None + ); +} + +#[test] +fn a_small_model_far_from_the_origin_can_be_fitted() { + let (o, d) = (1e16, 4.0); + let m = mesh(vec![[o, o, o], [o + d, o, o], [o, o + d, o]], vec![[0, 1, 2]]); + let camera = Camera::fit_meshes(&[WorldMesh::new(m)], [1.0, 1.0, 1.0], [0.0, 1.0, 0.0], false, 1.0); + assert!(camera.is_ok(), "{camera:?}"); +} + +#[test] +fn a_triangle_whose_edges_overflow_is_drawn() { + // Its corners are finite but 2e308 apart, so the plain edge differences are infinite. + let m = mesh(vec![[-1e308, -1e308, 0.0], [1e308, -1e308, 0.0], [0.0, 1e308, 0.0]], vec![[0, 1, 2]]); + let image = render(&[WorldMesh::new(m)], &ortho(1e308), &small()).unwrap(); + assert_eq!(pixel(&image, 20, 20), [190, 192, 200, 255]); +} + +#[test] +fn an_on_screen_triangle_is_drawn_when_projecting_it_would_overflow() { + let wide = RenderOptions { width: 30, height: 10, ..RenderOptions::default() }; + let drawn = |image: &Image| image.rgba.chunks(4).filter(|px| px != &[255, 255, 255, 255]).count(); + // A tiny field of view: x times the focal length (about 1.1e300) overflows, though x over z does not. + let camera = Camera { eye: [0.0; 3], target: [0.0, 0.0, 1e308], up: [0.0, 1.0, 0.0], projection: Projection::Perspective { fov_y: 1e-298 } }; + let far = mesh(vec![[1.5e8, -3e7, 1e308], [2.5e8, -3e7, 1e308], [2e8, 3e7, 1e308]], vec![[0, 1, 2]]); + assert!(drawn(&render(&[WorldMesh::new(far)], &camera, &wide).unwrap()) > 0); + // Half the orthographic height times the aspect ratio overflows. + let big = mesh(vec![[-5e307, -5e307, 0.0], [5e307, -5e307, 0.0], [0.0, 5e307, 0.0]], vec![[0, 1, 2]]); + assert_eq!(pixel(&render(&[WorldMesh::new(big)], &ortho(1.5e308), &wide).unwrap(), 15, 5), [190, 192, 200, 255]); +} diff --git a/crates/prc-render/src/vec3.rs b/crates/prc-render/src/vec3.rs new file mode 100644 index 0000000000..bdbed70aed --- /dev/null +++ b/crates/prc-render/src/vec3.rs @@ -0,0 +1,119 @@ +//! Three-component vector arithmetic for the camera and the rasterizer. + +/// A point or direction, `[x, y, z]`. +pub(crate) type Vec3 = [f64; 3]; + +/// `a - b`. +pub(crate) fn sub([ax, ay, az]: Vec3, [bx, by, bz]: Vec3) -> Vec3 { + [ax - bx, ay - by, az - bz] +} + +/// The direction from `from` to `to`: `to - from`, halved when the full difference overflows. +pub(crate) fn span(from: Vec3, to: Vec3) -> Vec3 { + let d = sub(to, from); + if d.iter().all(|c| c.is_finite()) { d } else { sub(scale(to, 0.5), scale(from, 0.5)) } +} + +/// `a` times `s`. +pub(crate) fn scale(a: Vec3, s: f64) -> Vec3 { + a.map(|c| c * s) +} + +/// The dot product `a · b`. +pub(crate) fn dot([ax, ay, az]: Vec3, [bx, by, bz]: Vec3) -> f64 { + ax * bx + ay * by + az * bz +} + +/// The right-handed cross product `a × b`. +pub(crate) fn cross([ax, ay, az]: Vec3, [bx, by, bz]: Vec3) -> Vec3 { + [ay * bz - az * by, az * bx - ax * bz, ax * by - ay * bx] +} + +/// The Euclidean length of `a`, taken with `a` divided by its largest component so squaring +/// cannot overflow or underflow; infinite or NaN when a component is. +pub(crate) fn length(a: Vec3) -> f64 { + let big = largest(a); + if !(big.is_finite() && big > 0.0) { + return dot(a, a).sqrt(); + } + let a = a.map(|c| c / big); + big * dot(a, a).sqrt() +} + +/// The largest component of `a` by magnitude; NaN components are passed over. +fn largest(a: Vec3) -> f64 { + a.iter().fold(0.0_f64, |m, c| m.max(c.abs())) +} + +/// `a` scaled to unit length; `None` when its length is zero or not finite. `a` is first divided +/// by its largest component, so squaring cannot overflow or underflow. +pub(crate) fn normalize(a: Vec3) -> Option { + let big = largest(a); + if !(big.is_finite() && big > 0.0) { + return None; + } + let a = a.map(|c| c / big); + let l = dot(a, a).sqrt(); + (l.is_finite() && l > 0.0).then(|| scale(a, 1.0 / l)) +} + +/// The unit normal of the plane spanned by `a` and `b` (`a × b`, normalized); `None` when they are +/// parallel, zero or not finite. Each is normalized first, so neither huge edges nor edges of +/// very different sizes overflow or underflow the cross product. +pub(crate) fn unit_cross(a: Vec3, b: Vec3) -> Option { + normalize(cross(normalize(a)?, normalize(b)?)) +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn cross_follows_the_right_hand_rule() { + assert_eq!(cross([1.0, 0.0, 0.0], [0.0, 1.0, 0.0]), [0.0, 0.0, 1.0]); + assert_eq!(cross([0.0, 1.0, 0.0], [1.0, 0.0, 0.0]), [0.0, 0.0, -1.0]); + } + + #[test] + fn normalize_rejects_degenerate_vectors() { + assert_eq!(normalize([0.0, 4.0, 0.0]), Some([0.0, 1.0, 0.0])); + assert_eq!(normalize([0.0; 3]), None); + assert_eq!(normalize([f64::NAN, 0.0, 0.0]), None); + assert_eq!(normalize([f64::INFINITY, 0.0, 0.0]), None); + } + + #[test] + fn normalize_keeps_huge_and_tiny_vectors() { + assert_eq!(normalize([1e200, 0.0, 0.0]), Some([1.0, 0.0, 0.0])); + assert_eq!(normalize([0.0, -3e-320, 0.0]), Some([0.0, -1.0, 0.0])); + } + + #[test] + fn unit_cross_survives_edges_whose_product_overflows() { + assert_eq!(unit_cross([1e160, 0.0, 0.0], [0.0, 1e160, 0.0]), Some([0.0, 0.0, 1.0])); + assert_eq!(unit_cross([1e-170, 0.0, 0.0], [0.0, 1e-170, 0.0]), Some([0.0, 0.0, 1.0])); + assert_eq!(unit_cross([2.0, 0.0, 0.0], [4.0, 0.0, 0.0]), None); + assert_eq!(unit_cross([f64::INFINITY, 0.0, 0.0], [0.0, 1.0, 0.0]), None); + } + + #[test] + fn unit_cross_survives_edges_of_very_different_sizes() { + assert_eq!(unit_cross([1e-170, 0.0, 0.0], [0.0, 1e160, 0.0]), Some([0.0, 0.0, 1.0])); + } + + #[test] + fn span_keeps_the_direction_of_a_difference_that_overflows() { + assert_eq!(span([1.0, 2.0, 3.0], [4.0, 0.0, 3.0]), [3.0, -2.0, 0.0]); + assert_eq!(span([-1e308, 0.0, 0.0], [1e308, 1e308, 0.0]), [1e308, 5e307, 0.0]); + } + + #[test] + fn length_survives_huge_and_tiny_vectors() { + let near = |l: f64, want: f64| (l - want).abs() <= want * 1e-15; + assert!(near(length([3e200, 4e200, 0.0]), 5e200)); + assert!(near(length([0.0, 3e-200, 4e-200]), 5e-200)); + assert_eq!(length([0.0; 3]), 0.0); + assert!(length([f64::NAN, 1.0, 0.0]).is_nan()); + assert_eq!(length([f64::INFINITY, 1.0, 0.0]), f64::INFINITY); + } +} diff --git a/crates/prc-render/tests/model.rs b/crates/prc-render/tests/model.rs new file mode 100644 index 0000000000..80e84252c6 --- /dev/null +++ b/crates/prc-render/tests/model.rs @@ -0,0 +1,145 @@ +//! End to end: a synthetic PRC stream parsed, assembled and rendered to pixels. +#![allow(clippy::unwrap_used, clippy::expect_used, clippy::panic)] + +use pdfcraft_prc::PrcFile; +use pdfcraft_prc_render::{Camera, Image, RenderOptions, assemble, render}; +use std::io::Write as _; + +/// An MSB-first bit sink: the inverse of the PRC bit reader, for the few values these tests write. +#[derive(Default)] +struct W { + bits: Vec, +} + +impl W { + fn put(&mut self, v: u64, n: u32) -> &mut Self { + for i in (0..n).rev() { + self.bits.push((v >> i) & 1 == 1); + } + self + } + + fn bit(&mut self, b: bool) -> &mut Self { + self.bits.push(b); + self + } + + /// An `UnsignedInteger`: bytes low first, each after a 1 bit, then a 0 bit. + fn uint(&mut self, mut v: u32) -> &mut Self { + while v != 0 { + self.bit(true).put(u64::from(v & 0xff), 8); + v >>= 8; + } + self.bit(false) + } + + fn uints(&mut self, v: &[u32]) -> &mut Self { + self.uint(u32::try_from(v.len()).unwrap()); + for &x in v { + self.uint(x); + } + self + } + + /// A `Double` from the few the code table stores whole: 0, ±0.5, ±1, ±2. + fn double(&mut self, v: f64) -> &mut Self { + let negative = v < 0.0; + match v.abs() { + 0.0 => self.put(0b01, 2), + 0.5 => self.put(0x777, 11).bit(negative), + 1.0 => self.put(0b0000, 4).bit(negative), + // The exponent row of [2, 4) with no mantissa bytes. + 2.0 => self.put(0x1b, 5).bit(negative).bit(false), + _ => panic!("{v} has no short code here"), + } + } + + fn bytes(&self) -> Vec { + self.bits.chunks(8).map(|c| c.iter().enumerate().fold(0u8, |a, (i, &b)| a | (u8::from(b) << (7 - i)))).collect() + } +} + +/// A tessellation section of one `TESS_3D` (authoring version 8137) with explicit normals and +/// triangles of `(normal, point)` coordinate indices. +fn tess_section(points: &[[f64; 3]], normals: &[[f64; 3]], corners: &[(u32, u32)]) -> Vec { + let mut w = W::default(); + w.uint(305).uint(0).bit(true).uint(1); + w.uint(172).bit(false).uint(u32::try_from(points.len() * 3).unwrap()); + points.iter().flatten().for_each(|&c| _ = w.double(c)); + // Faces present; no loop flag; normals not recalculated. + w.bit(true).bit(false).bit(false); + w.uint(u32::try_from(normals.len() * 3).unwrap()); + normals.iter().flatten().for_each(|&c| _ = w.double(c)); + w.uint(0).uints(&corners.iter().flat_map(|&(n, p)| [n, p]).collect::>()); + // One face of triangles with a normal per corner. + w.uint(1).uint(174).uint(0).uint(0).uint(0).uint(0x2).uint(0).uints(&[u32::try_from(corners.len() / 3).unwrap()]); + w.uint(0).bit(false); + w.uint(0).uint(0); + w.bytes() +} + +/// A one-file-structure PRC stream whose tessellation section is `tess`, with no product tree. +fn prc_file(tess: &[u8]) -> Vec { + const HEADER_LEN: usize = 107; + let zlib = |data: &[u8]| { + let mut e = flate2::write::ZlibEncoder::new(Vec::new(), flate2::Compression::default()); + e.write_all(data).unwrap(); + e.finish().unwrap() + }; + let u32s = |out: &mut Vec, vs: &[u32]| vs.iter().for_each(|v| out.extend_from_slice(&v.to_le_bytes())); + let mut globals = W::default(); + globals.uint(0); + let sections: [&[u8]; 5] = [&globals.bytes(), b"", tess, b"", b""]; + let mut body = b"PRC".to_vec(); + u32s(&mut body, &[8137, 8137, 5, 6, 7, 8, 0, 0, 0, 0, 0]); + let mut offs = vec![HEADER_LEN]; + for s in sections { + offs.push(HEADER_LEN + body.len()); + body.extend(zlib(s)); + } + let mf_start = HEADER_LEN + body.len(); + body.extend(zlib(b"")); + let mf_end = HEADER_LEN + body.len(); + let mut out = b"PRC".to_vec(); + u32s(&mut out, &[8137, 8137, 1, 2, 3, 4, 0, 0, 0, 0, 1]); + u32s(&mut out, &[5, 6, 7, 8, 0, 6]); + u32s(&mut out, &offs.iter().map(|&o| u32::try_from(o).unwrap()).collect::>()); + u32s(&mut out, &[u32::try_from(mf_start).unwrap(), u32::try_from(mf_end).unwrap(), 0]); + out.extend(body); + out +} + +fn pixel(image: &Image, x: u32, y: u32) -> [u8; 4] { + let i = (y * image.width + x) as usize * 4; + image.rgba[i..i + 4].try_into().unwrap() +} + +#[test] +fn a_synthetic_prc_renders_its_nearer_triangle_over_the_farther() { + // Listed first, a small triangle at z = 1 facing the viewer; behind it at z = -1, a larger + // one whose stored normal lies across the view, so the headlight shades it darkest. + let points = [[-1., -1., 1.], [1., -1., 1.], [0., 1., 1.], [-2., -2., -1.], [2., -2., -1.], [0., 2., -1.]]; + let normals = [[0., 0., 1.], [1., 0., 0.]]; + let corners = [(0, 0), (0, 3), (0, 6), (3, 9), (3, 12), (3, 15)]; + let prc = PrcFile::parse(&prc_file(&tess_section(&points, &normals, &corners))).unwrap(); + let meshes = assemble(&prc).unwrap(); + assert_eq!(meshes.len(), 1); + assert_eq!(meshes[0].mesh.triangles.len(), 2); + let camera = Camera::fit_meshes(&meshes, [0.0, 0.0, -1.0], [0.0, 1.0, 0.0], false, 1.0).unwrap(); + let image = render(&meshes, &camera, &RenderOptions { width: 40, height: 40, ..RenderOptions::default() }).unwrap(); + assert_eq!(pixel(&image, 1, 1), [255, 255, 255, 255], "background corner"); + assert_eq!(pixel(&image, 20, 20), [190, 192, 200, 255], "the nearer triangle, lit head on"); + // Below the near triangle only the far one: 0.3 of the surface colour. + assert_eq!(pixel(&image, 20, 34), [57, 58, 60, 255], "the farther triangle, lit edge on"); +} + +#[test] +fn damaged_streams_are_errors_not_panics() { + let points = [[0., 0., 0.], [1., 0., 0.], [0., 1., 0.]]; + let bytes = prc_file(&tess_section(&points, &[[0., 0., 1.]], &[(0, 0), (0, 3), (0, 6)])); + for end in 0..bytes.len() { + if let Ok(prc) = PrcFile::parse(&bytes[..end]) { + let _ = assemble(&prc); + } + } +} diff --git a/crates/prc/Cargo.toml b/crates/prc/Cargo.toml new file mode 100644 index 0000000000..741c0ac5bd --- /dev/null +++ b/crates/prc/Cargo.toml @@ -0,0 +1,18 @@ +[package] +name = "pdfcraft-prc" +description = "PRC 3D model reader (ISO 14739-1): container, bit-level primitives, schema" +version.workspace = true +edition.workspace = true +license.workspace = true +rust-version.workspace = true +repository.workspace = true + +[lints] +workspace = true + +[dependencies] +flate2.workspace = true +thiserror.workspace = true + +[dev-dependencies] +proptest.workspace = true diff --git a/crates/prc/README.md b/crates/prc/README.md new file mode 100644 index 0000000000..40bdab335b --- /dev/null +++ b/crates/prc/README.md @@ -0,0 +1,43 @@ +# pdfcraft-prc + +A reader for PRC (ISO 14739-1), the 3D model format a PDF `/3D` stream of `/Subtype /PRC` carries (ISO 32000-2 §13.6.3). + +- **Layer:** L0, standalone. No dependency on any other PdfCraft crate. +- **Licence:** MIT OR Apache-2.0. Inflation through `flate2` with the pure-Rust `miniz_oxide` backend. +- `#![forbid(unsafe_code)]`. Never panics on input: inflated bytes, decoded elements, schema steps and walk depth are capped, and damage returns `PrcError`. + +## API + +```rust +PrcFile::parse(data) -> Result // header, file structures, model file +PrcFile::parse_with_limit(data, max_inflated) -> Result +FileStructure::section(SectionKind) -> &[u8] // inflated globals / tree / tessellation / geometry / extra geometry +FileStructure::schema() -> Result // the globals section's leading schema +FileStructure::tessellations() -> Result, PrcError> // meshes, wires, markups, compressed meshes +PrcFile::placements() / placements_with(StyleAlpha, EntityOverrides) -> Result, PrcError> +PrcFile::model_tree() -> Result, PrcError> // the assembly tree as a model-tree panel lists it +Placement::triangle_colours(&TriangleMesh) -> Option>> +TriangleMesh::transformed(&Matrix) -> TriangleMesh +Schema::read(&mut BitReader) / Schema::skip_added_fields(entity_type, &mut BitReader) +BitReader: bit, bits, character, unsigned_integer, integer, integer_vbn, double_vbn, nbits_then_unsigned, string, float_as_bytes, double +``` + +- `MAX_INFLATED_BYTES` (512 MiB) bounds every section together; `TooLarge` reports the limit hit. +- `MAX_DECODED_ELEMENTS` (2^24) bounds the array elements one pass decodes (a tree walk, or one `tessellations` call), since a single bit can decode to a 4-8 byte element; `TooManyElements` reports it. +- Schema programs (how a newer producer describes fields it added) run under a nesting ceiling (64) and `MAX_SCHEMA_STEPS` (2^22), shared like `MAX_DECODED_ELEMENTS` by every program one pass runs. +- `Unsupported` names constructs the reader cannot step over (schema pointer fields, `Value_CurveIs3D`). +- `Tessellation::Mesh` is a `TriangleMesh`: positions, triangles, one triangle range per face, the stored normals and + texture coordinates with each triangle's corner indices into them, and per-triangle line-attribute styles. + `Wire` is polylines; `Markup` is read past. +- `Tessellation::Compressed` is read to its end so later entities decode; its triangles are not rebuilt yet. +- A face may carry at most 8 texture-coordinate sets; more is `Malformed`. Optimised vertex colours are `Unsupported`. +- The product tree: each drawable tessellation placed by its occurrence path's locations and the item's coordinate + systems (a 4x4 `Matrix`), coloured by the style inheritance rules, with occurrences' entity references applied + (`EntityOverrides`: subtree, everywhere or ignored) and style transparency combined with material alpha by `StyleAlpha`. + A prototype's part is coloured by its own file structure's styles. Each `ModelNode` lists its name, where the name + came from, its visibility and the range of placements it drew. +- Occurrence and prototype cycles are `Malformed`; the walk is capped at depth 64, 2^20 visits and `MAX_PLACEMENTS`. +- A tree's markups (PMI), views, scene lights and clipping planes are read past, not modelled; so are textures. +- Not yet in the crate: rebuilding compressed meshes, textures. +- The `Double` code table's rows come from [prc-rs](https://github.com/ralovich/prc-rs) (MIT); its notice is in `src/acof/table.rs`. + A few reading choices also follow prc-rs; each is marked `// prc-rs:` in the source. diff --git a/crates/prc/src/acof.rs b/crates/prc/src/acof.rs new file mode 100644 index 0000000000..fc11a663a6 --- /dev/null +++ b/crates/prc/src/acof.rs @@ -0,0 +1,93 @@ +//! The prefix-code table behind the PRC `Double` [WD 11.17.2]. +//! +//! The table has 2077 rows: 29 frequent doubles and one row per IEEE binary64 exponent +//! (0..=2047). Each row is a code of 1 to 22 bits and the value it decodes to. +//! The rows come from prc-rs under its MIT licence (see [`table`]). + +use std::collections::HashMap; +use std::sync::OnceLock; + +mod table; + +/// Rows a complete table has. +const ROWS: usize = 2077; + +/// The longest code, in bits. +pub(crate) const MAX_CODE_BITS: u32 = 22; + +/// What a code decodes to. +#[derive(Debug, Clone, Copy, PartialEq)] +pub(crate) enum Entry { + /// A frequent value, returned with the sign bit applied. + Double(f64), + /// The upper 32 bits of a binary64 (exponent set, mantissa zero); the mantissa follows + /// in the stream. + Exponent(u32), +} + +/// One row: code length, code, and what it decodes to. +type Row = (u32, u32, Entry); + +fn key(nbits: u32, code: u32) -> u32 { + (nbits << 24) | code +} + +/// The rows keyed by `(nbits, code)`, or `None` when the table is incomplete or not prefix +/// free (then no `Double` decodes, rather than some decoding wrongly). +fn build(rows: &[Row]) -> Option> { + let mut map = HashMap::with_capacity(rows.len()); + for &(nbits, code, entry) in rows { + if nbits == 0 || nbits > MAX_CODE_BITS || code >> nbits != 0 || map.insert(key(nbits, code), entry).is_some() { + return None; + } + } + let prefix_free = map.keys().all(|&k| { + let (n, c) = (k >> 24, k & 0xff_ffff); + (1..n).all(|m| !map.contains_key(&key(m, c >> (n - m)))) + }); + (map.len() == ROWS && prefix_free).then_some(map) +} + +fn table() -> Option<&'static HashMap> { + static MAP: OnceLock>> = OnceLock::new(); + MAP.get_or_init(|| build(table::TABLE)).as_ref() +} + +/// The entry for an `nbits`-long code, if one exists. +pub(crate) fn lookup(nbits: u32, code: u32) -> Option { + table()?.get(&key(nbits, code)).copied() +} + +/// The `(nbits, code)` of IEEE exponent `e`'s row, the test writer's half of [`lookup`]. +#[cfg(test)] +pub(crate) fn exponent_code(e: u32) -> Option<(u32, u32)> { + table()?.iter().find(|(_, v)| matches!(v, Entry::Exponent(hi) if (hi >> 20) & 0x7ff == e)).map(|(&k, _)| (k >> 24, k & 0xff_ffff)) +} + +#[cfg(test)] +#[allow(clippy::unwrap_used)] +mod tests { + use super::*; + + #[test] + fn an_incomplete_or_ambiguous_table_decodes_nothing() { + assert!(build(&[(2, 0b01, Entry::Double(0.0))]).is_none(), "one row is not the table"); + let rows: Vec = (0..ROWS as u32).map(|i| (22, i, Entry::Exponent(i))).collect(); + assert!(build(&rows).is_some(), "2077 distinct 22-bit codes are prefix free"); + let mut ambiguous = rows.clone(); + ambiguous[0] = (1, 0, Entry::Double(1.0)); // a prefix of every code starting 0 + assert!(build(&ambiguous).is_none()); + let mut too_long = rows; + too_long[0] = (2, 0b111, Entry::Double(1.0)); // a code wider than its length + assert!(build(&too_long).is_none()); + } + + #[test] + fn the_table_is_complete() { + let t = table().unwrap(); + assert_eq!(t.len(), ROWS); + assert_eq!(t.values().filter(|e| matches!(e, Entry::Double(_))).count(), 29); + assert!((0..2048).all(|e| exponent_code(e).is_some())); + assert_eq!(lookup(2, 0b01), Some(Entry::Double(0.0))); + } +} diff --git a/crates/prc/src/acof/table.rs b/crates/prc/src/acof/table.rs new file mode 100644 index 0000000000..28cff50a2e --- /dev/null +++ b/crates/prc/src/acof/table.rs @@ -0,0 +1,2113 @@ +//! The 2077 rows of the PRC `Double` prefix-code table [WD 11.17.2], as `(code length, code, entry)`. +//! +//! The rows are taken from prc-rs (, `src/double.rs` at commit +//! 287b671bb5b31b05efb5ec1b98f7c359b9e4dd7d), used under its MIT licence: +//! +//! Copyright (c) 2025,2026 Kristof Ralovich +//! +//! Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated +//! documentation files (the "Software"), to deal in the Software without restriction, including without limitation +//! the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and +//! to permit persons to whom the Software is furnished to do so, subject to the following conditions: +//! +//! The above copyright notice and this permission notice shall be included in all copies or substantial portions of +//! the Software. +//! +//! THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO +//! THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +//! AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +//! TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +//! SOFTWARE. + +use super::{Entry, Row}; + +/// A frequent value: the binary64 with upper and lower words `hi`, `lo`. +const fn d(hi: u32, lo: u32) -> Entry { + Entry::Double(f64::from_bits(((hi as u64) << 32) | lo as u64)) +} + +/// An exponent row: the upper word of a binary64 whose mantissa follows in the stream. +const fn e(hi: u32) -> Entry { + Entry::Exponent(hi) +} + +#[rustfmt::skip] +pub(super) const TABLE: &[Row] = &[ + (2, 0x1, d(0x00000000, 0x00000000)), + (22, 0xd1d32, e(0x00000000)), + (22, 0xd1d33, e(0x00100000)), + (22, 0xf78d8, e(0x00200000)), + (22, 0xf78d9, e(0x00300000)), + (22, 0xf78da, e(0x00400000)), + (22, 0xf78db, e(0x00500000)), + (22, 0xf78dc, e(0x00600000)), + (22, 0xf78dd, e(0x00700000)), + (22, 0xf78de, e(0x00800000)), + (22, 0xf78df, e(0x00900000)), + (22, 0xf78e0, e(0x00a00000)), + (22, 0xf78e1, e(0x00b00000)), + (22, 0xf78e2, e(0x00c00000)), + (22, 0xf78e3, e(0x00d00000)), + (22, 0xf78e4, e(0x00e00000)), + (22, 0xf78e5, e(0x00f00000)), + (22, 0xf78e6, e(0x01000000)), + (22, 0xf78e7, e(0x01100000)), + (22, 0xf78e8, e(0x01200000)), + (22, 0xf78e9, e(0x01300000)), + (22, 0xf78ea, e(0x01400000)), + (22, 0xf78eb, e(0x01500000)), + (22, 0xf78ec, e(0x01600000)), + (22, 0xf78ed, e(0x01700000)), + (22, 0xf78ee, e(0x01800000)), + (22, 0xf78ef, e(0x01900000)), + (22, 0xf78f0, e(0x01a00000)), + (22, 0xf78f1, e(0x01b00000)), + (22, 0xf78f2, e(0x01c00000)), + (22, 0xf78f3, e(0x01d00000)), + (22, 0xf78f4, e(0x01e00000)), + (22, 0xf78f5, e(0x01f00000)), + (22, 0xf78f6, e(0x02000000)), + (22, 0xf78f7, e(0x02100000)), + (22, 0xf78f8, e(0x02200000)), + (22, 0xf78f9, e(0x02300000)), + (22, 0xf78fa, e(0x02400000)), + (22, 0xf78fb, e(0x02500000)), + (22, 0xf78fc, e(0x02600000)), + (22, 0xf78fd, e(0x02700000)), + (22, 0xf78fe, e(0x02800000)), + (22, 0xf78ff, e(0x02900000)), + (22, 0x3a8300, e(0x02a00000)), + (22, 0x3a8301, e(0x02b00000)), + (22, 0x3a8302, e(0x02c00000)), + (22, 0x3a8303, e(0x02d00000)), + (22, 0x3a8304, e(0x02e00000)), + (22, 0x3a8305, e(0x02f00000)), + (22, 0x3a8306, e(0x03000000)), + (22, 0x3a8307, e(0x03100000)), + (22, 0x3a8308, e(0x03200000)), + (22, 0x3a8309, e(0x03300000)), + (22, 0x3a830a, e(0x03400000)), + (22, 0x3a830b, e(0x03500000)), + (22, 0x3a830c, e(0x03600000)), + (22, 0x3a830d, e(0x03700000)), + (22, 0x3a830e, e(0x03800000)), + (22, 0x3a830f, e(0x03900000)), + (22, 0x3a8310, e(0x03a00000)), + (22, 0x3a8311, e(0x03b00000)), + (22, 0x3a8312, e(0x03c00000)), + (22, 0x3a8313, e(0x03d00000)), + (22, 0x3a8314, e(0x03e00000)), + (22, 0x3a8315, e(0x03f00000)), + (22, 0x3a8316, e(0x04000000)), + (22, 0x3a8317, e(0x04100000)), + (22, 0x3a8318, e(0x04200000)), + (22, 0x3a8319, e(0x04300000)), + (22, 0x3a831a, e(0x04400000)), + (22, 0x3a831b, e(0x04500000)), + (22, 0x3a831c, e(0x04600000)), + (22, 0x3a831d, e(0x04700000)), + (22, 0x3a831e, e(0x04800000)), + (22, 0x3a831f, e(0x04900000)), + (22, 0x3a8320, e(0x04a00000)), + (22, 0x3a8321, e(0x04b00000)), + (22, 0x3a8322, e(0x04c00000)), + (22, 0x3a8323, e(0x04d00000)), + (22, 0x3a8324, e(0x04e00000)), + (22, 0x3a8325, e(0x04f00000)), + (22, 0x3a8326, e(0x05000000)), + (22, 0x3a8327, e(0x05100000)), + (22, 0x3a8328, e(0x05200000)), + (22, 0x3a8329, e(0x05300000)), + (22, 0x3a832a, e(0x05400000)), + (22, 0x3a832b, e(0x05500000)), + (22, 0x3a832c, e(0x05600000)), + (22, 0x3a832d, e(0x05700000)), + (22, 0x3a832e, e(0x05800000)), + (22, 0x3a832f, e(0x05900000)), + (22, 0x3a8330, e(0x05a00000)), + (22, 0x3a8331, e(0x05b00000)), + (22, 0x3a8332, e(0x05c00000)), + (22, 0x3a8333, e(0x05d00000)), + (22, 0x3a8334, e(0x05e00000)), + (22, 0x3a8335, e(0x05f00000)), + (22, 0x3a8336, e(0x06000000)), + (22, 0x3a8337, e(0x06100000)), + (22, 0x3a8338, e(0x06200000)), + (22, 0x3a8339, e(0x06300000)), + (22, 0x3a833a, e(0x06400000)), + (22, 0x3a833b, e(0x06500000)), + (22, 0x3a833c, e(0x06600000)), + (22, 0x3a833d, e(0x06700000)), + (22, 0x3a833e, e(0x06800000)), + (22, 0x3a833f, e(0x06900000)), + (22, 0x3a8340, e(0x06a00000)), + (22, 0x3a8341, e(0x06b00000)), + (22, 0x3a8342, e(0x06c00000)), + (22, 0x3a8343, e(0x06d00000)), + (22, 0x3a8344, e(0x06e00000)), + (22, 0x3a8345, e(0x06f00000)), + (22, 0x3a8346, e(0x07000000)), + (22, 0x3a8347, e(0x07100000)), + (22, 0x3a8348, e(0x07200000)), + (22, 0x3a8349, e(0x07300000)), + (22, 0x3a834a, e(0x07400000)), + (22, 0x3a834b, e(0x07500000)), + (22, 0x3a834c, e(0x07600000)), + (22, 0x3a834d, e(0x07700000)), + (22, 0x3a834e, e(0x07800000)), + (22, 0x3a834f, e(0x07900000)), + (22, 0x3a8350, e(0x07a00000)), + (22, 0x3a8351, e(0x07b00000)), + (22, 0x3a8352, e(0x07c00000)), + (22, 0x3a8353, e(0x07d00000)), + (22, 0x3a8354, e(0x07e00000)), + (22, 0x3a8355, e(0x07f00000)), + (22, 0x3a8356, e(0x08000000)), + (22, 0x3a8357, e(0x08100000)), + (22, 0x3a8358, e(0x08200000)), + (22, 0x3a8359, e(0x08300000)), + (22, 0x3a835a, e(0x08400000)), + (22, 0x3a835b, e(0x08500000)), + (22, 0x3a835c, e(0x08600000)), + (22, 0x3a835d, e(0x08700000)), + (22, 0x3a835e, e(0x08800000)), + (22, 0x3a835f, e(0x08900000)), + (22, 0x3a8360, e(0x08a00000)), + (22, 0x3a8361, e(0x08b00000)), + (22, 0x3a8362, e(0x08c00000)), + (22, 0x3a8363, e(0x08d00000)), + (22, 0x3a8364, e(0x08e00000)), + (22, 0x3a8365, e(0x08f00000)), + (22, 0x3a8366, e(0x09000000)), + (22, 0x3a8367, e(0x09100000)), + (22, 0x3a8368, e(0x09200000)), + (22, 0x3a8369, e(0x09300000)), + (22, 0x3a836a, e(0x09400000)), + (22, 0x3a836b, e(0x09500000)), + (22, 0x3a836c, e(0x09600000)), + (22, 0x3a836d, e(0x09700000)), + (22, 0x3a836e, e(0x09800000)), + (22, 0x3a836f, e(0x09900000)), + (22, 0x3a8370, e(0x09a00000)), + (22, 0x3a8371, e(0x09b00000)), + (22, 0x3a8372, e(0x09c00000)), + (22, 0x3a8373, e(0x09d00000)), + (22, 0x3a8374, e(0x09e00000)), + (22, 0x3a8375, e(0x09f00000)), + (22, 0x3a8376, e(0x0a000000)), + (22, 0x3a8377, e(0x0a100000)), + (22, 0x3a8378, e(0x0a200000)), + (22, 0x3a8379, e(0x0a300000)), + (22, 0x3a837a, e(0x0a400000)), + (22, 0x3a837b, e(0x0a500000)), + (22, 0x3a837c, e(0x0a600000)), + (22, 0x3a837d, e(0x0a700000)), + (22, 0x3a837e, e(0x0a800000)), + (22, 0x3a837f, e(0x0a900000)), + (22, 0x3a8380, e(0x0aa00000)), + (22, 0x3a8381, e(0x0ab00000)), + (22, 0x3a8382, e(0x0ac00000)), + (22, 0x3a8383, e(0x0ad00000)), + (22, 0x3a8384, e(0x0ae00000)), + (22, 0x3a8385, e(0x0af00000)), + (22, 0x3a8386, e(0x0b000000)), + (22, 0x3a8387, e(0x0b100000)), + (22, 0x3a8388, e(0x0b200000)), + (22, 0x3a8389, e(0x0b300000)), + (22, 0x3a838a, e(0x0b400000)), + (22, 0x3a838b, e(0x0b500000)), + (22, 0x3a838c, e(0x0b600000)), + (22, 0x3a838d, e(0x0b700000)), + (22, 0x3a838e, e(0x0b800000)), + (22, 0x3a838f, e(0x0b900000)), + (22, 0x3a8390, e(0x0ba00000)), + (22, 0x3a8391, e(0x0bb00000)), + (22, 0x3a8392, e(0x0bc00000)), + (22, 0x3a8393, e(0x0bd00000)), + (22, 0x3a8394, e(0x0be00000)), + (22, 0x3a8395, e(0x0bf00000)), + (22, 0x3a8396, e(0x0c000000)), + (22, 0x3a8397, e(0x0c100000)), + (22, 0x3a8398, e(0x0c200000)), + (22, 0x3a8399, e(0x0c300000)), + (22, 0x3a839a, e(0x0c400000)), + (22, 0x3a839b, e(0x0c500000)), + (22, 0x3a839c, e(0x0c600000)), + (22, 0x3a839d, e(0x0c700000)), + (22, 0x3a839e, e(0x0c800000)), + (22, 0x3a839f, e(0x0c900000)), + (22, 0x3a83a0, e(0x0ca00000)), + (22, 0x3a83a1, e(0x0cb00000)), + (22, 0x3a83a2, e(0x0cc00000)), + (22, 0x3a83a3, e(0x0cd00000)), + (22, 0x3a83a4, e(0x0ce00000)), + (22, 0x3a83a5, e(0x0cf00000)), + (22, 0x3a83a6, e(0x0d000000)), + (22, 0x3a83a7, e(0x0d100000)), + (22, 0x3a83a8, e(0x0d200000)), + (22, 0x3a83a9, e(0x0d300000)), + (22, 0x3a83aa, e(0x0d400000)), + (22, 0x3a83ab, e(0x0d500000)), + (22, 0x3a83ac, e(0x0d600000)), + (22, 0x3a83ad, e(0x0d700000)), + (22, 0x3a83ae, e(0x0d800000)), + (22, 0x3a83af, e(0x0d900000)), + (22, 0x3a83b0, e(0x0da00000)), + (22, 0x3a83b1, e(0x0db00000)), + (22, 0x3a83b2, e(0x0dc00000)), + (22, 0x3a83b3, e(0x0dd00000)), + (22, 0x3a83b4, e(0x0de00000)), + (22, 0x3a83b5, e(0x0df00000)), + (22, 0x3a83b6, e(0x0e000000)), + (22, 0x3a83b7, e(0x0e100000)), + (22, 0x3a83b8, e(0x0e200000)), + (22, 0x3a83b9, e(0x0e300000)), + (22, 0x3a83ba, e(0x0e400000)), + (22, 0x3a83bb, e(0x0e500000)), + (22, 0x3a83bc, e(0x0e600000)), + (22, 0x3a83bd, e(0x0e700000)), + (22, 0x3a83be, e(0x0e800000)), + (22, 0x3a83bf, e(0x0e900000)), + (22, 0x3a83c0, e(0x0ea00000)), + (22, 0x3a83c1, e(0x0eb00000)), + (22, 0x3a83c2, e(0x0ec00000)), + (22, 0x3a83c3, e(0x0ed00000)), + (22, 0x3a83c4, e(0x0ee00000)), + (22, 0x3a83c5, e(0x0ef00000)), + (22, 0x3a83c6, e(0x0f000000)), + (22, 0x3a83c7, e(0x0f100000)), + (22, 0x3a83c8, e(0x0f200000)), + (22, 0x3a83c9, e(0x0f300000)), + (22, 0x3a83ca, e(0x0f400000)), + (22, 0x3a83cb, e(0x0f500000)), + (22, 0x3a83cc, e(0x0f600000)), + (22, 0x3a83cd, e(0x0f700000)), + (22, 0x3a83ce, e(0x0f800000)), + (22, 0x3a83cf, e(0x0f900000)), + (22, 0x3a83d0, e(0x0fa00000)), + (22, 0x3a83d1, e(0x0fb00000)), + (22, 0x3a83d2, e(0x0fc00000)), + (22, 0x3a83d3, e(0x0fd00000)), + (22, 0x3a83d4, e(0x0fe00000)), + (22, 0x3a83d5, e(0x0ff00000)), + (22, 0x3a83d6, e(0x10000000)), + (22, 0x3a83d7, e(0x10100000)), + (22, 0x3a83d8, e(0x10200000)), + (22, 0x3a83d9, e(0x10300000)), + (22, 0x3a83da, e(0x10400000)), + (22, 0x3a83db, e(0x10500000)), + (22, 0x3a83dc, e(0x10600000)), + (22, 0x3a83dd, e(0x10700000)), + (22, 0x3a83de, e(0x10800000)), + (22, 0x3a83df, e(0x10900000)), + (22, 0x3a83e0, e(0x10a00000)), + (22, 0x3a83e1, e(0x10b00000)), + (22, 0x3a83e2, e(0x10c00000)), + (22, 0x3a83e3, e(0x10d00000)), + (22, 0x3a83e4, e(0x10e00000)), + (22, 0x3a83e5, e(0x10f00000)), + (22, 0x3a83e6, e(0x11000000)), + (22, 0x3a83e7, e(0x11100000)), + (22, 0x3a83e8, e(0x11200000)), + (22, 0x3a83e9, e(0x11300000)), + (22, 0x3a83ea, e(0x11400000)), + (22, 0x3a83eb, e(0x11500000)), + (22, 0x3a83ec, e(0x11600000)), + (22, 0x3a83ed, e(0x11700000)), + (22, 0x3a83ee, e(0x11800000)), + (22, 0x3a83ef, e(0x11900000)), + (22, 0x3a83f0, e(0x11a00000)), + (22, 0x3a83f1, e(0x11b00000)), + (22, 0x3a83f2, e(0x11c00000)), + (22, 0x3a83f3, e(0x11d00000)), + (22, 0x3a83f4, e(0x11e00000)), + (22, 0x3a83f5, e(0x11f00000)), + (22, 0x3a83f6, e(0x12000000)), + (22, 0x3a83f7, e(0x12100000)), + (22, 0x3a83f8, e(0x12200000)), + (22, 0x3a83f9, e(0x12300000)), + (22, 0x3a83fa, e(0x12400000)), + (22, 0x3a83fb, e(0x12500000)), + (22, 0x3a83fc, e(0x12600000)), + (22, 0x3a83fd, e(0x12700000)), + (22, 0x3a83fe, e(0x12800000)), + (22, 0x3a83ff, e(0x12900000)), + (22, 0x3a8400, e(0x12a00000)), + (22, 0x3a8401, e(0x12b00000)), + (22, 0x3a8402, e(0x12c00000)), + (22, 0x3a8403, e(0x12d00000)), + (22, 0x3a8404, e(0x12e00000)), + (22, 0x3a8405, e(0x12f00000)), + (22, 0x3a8406, e(0x13000000)), + (22, 0x3a8407, e(0x13100000)), + (22, 0x3a8408, e(0x13200000)), + (22, 0x3a8409, e(0x13300000)), + (22, 0x3a840a, e(0x13400000)), + (22, 0x3a840b, e(0x13500000)), + (22, 0x3a840c, e(0x13600000)), + (22, 0x3a840d, e(0x13700000)), + (22, 0x3a840e, e(0x13800000)), + (22, 0x3a840f, e(0x13900000)), + (22, 0x3a8410, e(0x13a00000)), + (22, 0x3a8411, e(0x13b00000)), + (22, 0x3a8412, e(0x13c00000)), + (22, 0x3a8413, e(0x13d00000)), + (22, 0x3a8414, e(0x13e00000)), + (22, 0x3a8415, e(0x13f00000)), + (22, 0x3a8416, e(0x14000000)), + (22, 0x3a8417, e(0x14100000)), + (22, 0x3a8418, e(0x14200000)), + (22, 0x3a8419, e(0x14300000)), + (22, 0x3a841a, e(0x14400000)), + (22, 0x3a841b, e(0x14500000)), + (22, 0x3a841c, e(0x14600000)), + (22, 0x3a841d, e(0x14700000)), + (22, 0x3a841e, e(0x14800000)), + (22, 0x3a841f, e(0x14900000)), + (22, 0x3a8420, e(0x14a00000)), + (22, 0x3a8421, e(0x14b00000)), + (22, 0x3a8422, e(0x14c00000)), + (22, 0x3a8423, e(0x14d00000)), + (22, 0x3a8424, e(0x14e00000)), + (22, 0x3a8425, e(0x14f00000)), + (22, 0x3a8426, e(0x15000000)), + (22, 0x3a8427, e(0x15100000)), + (22, 0x3a8428, e(0x15200000)), + (22, 0x3a8429, e(0x15300000)), + (22, 0x3a842a, e(0x15400000)), + (22, 0x3a842b, e(0x15500000)), + (22, 0x3a842c, e(0x15600000)), + (22, 0x3a842d, e(0x15700000)), + (22, 0x3a842e, e(0x15800000)), + (22, 0x3a842f, e(0x15900000)), + (22, 0x3a8430, e(0x15a00000)), + (22, 0x3a8431, e(0x15b00000)), + (22, 0x3a8432, e(0x15c00000)), + (22, 0x3a8433, e(0x15d00000)), + (22, 0x3a8434, e(0x15e00000)), + (22, 0x3a8435, e(0x15f00000)), + (22, 0x3a8436, e(0x16000000)), + (22, 0x3a8437, e(0x16100000)), + (22, 0x3a8438, e(0x16200000)), + (22, 0x3a8439, e(0x16300000)), + (22, 0x3a843a, e(0x16400000)), + (22, 0x3a843b, e(0x16500000)), + (22, 0x3a843c, e(0x16600000)), + (22, 0x3a843d, e(0x16700000)), + (22, 0x3a843e, e(0x16800000)), + (22, 0x3a843f, e(0x16900000)), + (22, 0x3a8440, e(0x16a00000)), + (22, 0x3a8441, e(0x16b00000)), + (22, 0x3a8442, e(0x16c00000)), + (22, 0x3a8443, e(0x16d00000)), + (22, 0x3a8444, e(0x16e00000)), + (22, 0x3a8445, e(0x16f00000)), + (22, 0x3a8446, e(0x17000000)), + (22, 0x3a8447, e(0x17100000)), + (22, 0x3a8448, e(0x17200000)), + (22, 0x3a8449, e(0x17300000)), + (22, 0x3a844a, e(0x17400000)), + (22, 0x3a844b, e(0x17500000)), + (22, 0x3a844c, e(0x17600000)), + (22, 0x3a844d, e(0x17700000)), + (22, 0x3a844e, e(0x17800000)), + (22, 0x3a844f, e(0x17900000)), + (22, 0x3a8450, e(0x17a00000)), + (22, 0x3a8451, e(0x17b00000)), + (22, 0x3a8452, e(0x17c00000)), + (22, 0x3a8453, e(0x17d00000)), + (22, 0x3a8454, e(0x17e00000)), + (22, 0x3a8455, e(0x17f00000)), + (22, 0x3a8456, e(0x18000000)), + (22, 0x3a8457, e(0x18100000)), + (22, 0x3a8458, e(0x18200000)), + (22, 0x3a8459, e(0x18300000)), + (22, 0x3a845a, e(0x18400000)), + (22, 0x3a845b, e(0x18500000)), + (22, 0x3a845c, e(0x18600000)), + (22, 0x3a845d, e(0x18700000)), + (22, 0x3a845e, e(0x18800000)), + (22, 0x3a845f, e(0x18900000)), + (22, 0x3a8460, e(0x18a00000)), + (22, 0x3a8461, e(0x18b00000)), + (22, 0x3a8462, e(0x18c00000)), + (22, 0x3a8463, e(0x18d00000)), + (22, 0x3a8464, e(0x18e00000)), + (22, 0x3a8465, e(0x18f00000)), + (22, 0x3a8466, e(0x19000000)), + (22, 0x3a8467, e(0x19100000)), + (22, 0x3a8468, e(0x19200000)), + (22, 0x3a8469, e(0x19300000)), + (22, 0x3a846a, e(0x19400000)), + (22, 0x3a846b, e(0x19500000)), + (22, 0x3a846c, e(0x19600000)), + (22, 0x3a846d, e(0x19700000)), + (22, 0x3a846e, e(0x19800000)), + (22, 0x3a846f, e(0x19900000)), + (22, 0x3a8470, e(0x19a00000)), + (22, 0x3a8471, e(0x19b00000)), + (22, 0x3a8472, e(0x19c00000)), + (22, 0x3a8473, e(0x19d00000)), + (22, 0x3a8474, e(0x19e00000)), + (22, 0x3a8475, e(0x19f00000)), + (22, 0x3a8476, e(0x1a000000)), + (22, 0x3a8477, e(0x1a100000)), + (22, 0x3a8478, e(0x1a200000)), + (22, 0x3a8479, e(0x1a300000)), + (22, 0x3a847a, e(0x1a400000)), + (22, 0x3a847b, e(0x1a500000)), + (22, 0x3a847c, e(0x1a600000)), + (22, 0x3a847d, e(0x1a700000)), + (22, 0x3a847e, e(0x1a800000)), + (22, 0x3a847f, e(0x1a900000)), + (22, 0x3a8480, e(0x1aa00000)), + (22, 0x3a8481, e(0x1ab00000)), + (22, 0x3a8482, e(0x1ac00000)), + (22, 0x3a8483, e(0x1ad00000)), + (22, 0x3a8484, e(0x1ae00000)), + (22, 0x3a8485, e(0x1af00000)), + (22, 0x3a8486, e(0x1b000000)), + (22, 0x3a8487, e(0x1b100000)), + (22, 0x3a8488, e(0x1b200000)), + (22, 0x3a8489, e(0x1b300000)), + (22, 0x3a848a, e(0x1b400000)), + (22, 0x3a848b, e(0x1b500000)), + (22, 0x3a848c, e(0x1b600000)), + (22, 0x3a848d, e(0x1b700000)), + (22, 0x3a848e, e(0x1b800000)), + (22, 0x3a848f, e(0x1b900000)), + (22, 0x3a8490, e(0x1ba00000)), + (22, 0x3a8491, e(0x1bb00000)), + (22, 0x3a8492, e(0x1bc00000)), + (22, 0x3a8493, e(0x1bd00000)), + (22, 0x3a8494, e(0x1be00000)), + (22, 0x3a8495, e(0x1bf00000)), + (22, 0x3a8496, e(0x1c000000)), + (22, 0x3a8497, e(0x1c100000)), + (22, 0x3a8498, e(0x1c200000)), + (22, 0x3a8499, e(0x1c300000)), + (22, 0x3a849a, e(0x1c400000)), + (22, 0x3a849b, e(0x1c500000)), + (22, 0x3a849c, e(0x1c600000)), + (22, 0x3a849d, e(0x1c700000)), + (22, 0x3a849e, e(0x1c800000)), + (22, 0x3a849f, e(0x1c900000)), + (22, 0x3a84a0, e(0x1ca00000)), + (22, 0x3a84a1, e(0x1cb00000)), + (22, 0x3a84a2, e(0x1cc00000)), + (22, 0x3a84a3, e(0x1cd00000)), + (22, 0x3a84a4, e(0x1ce00000)), + (22, 0x3a84a5, e(0x1cf00000)), + (22, 0x3a84a6, e(0x1d000000)), + (22, 0x3a84a7, e(0x1d100000)), + (22, 0x3a84a8, e(0x1d200000)), + (22, 0x3a84a9, e(0x1d300000)), + (22, 0x3a84aa, e(0x1d400000)), + (22, 0x3a84ab, e(0x1d500000)), + (22, 0x3a84ac, e(0x1d600000)), + (22, 0x3a84ad, e(0x1d700000)), + (22, 0x3a84ae, e(0x1d800000)), + (22, 0x3a84af, e(0x1d900000)), + (22, 0x3a84b0, e(0x1da00000)), + (22, 0x3a84b1, e(0x1db00000)), + (22, 0x3a84b2, e(0x1dc00000)), + (22, 0x3a84b3, e(0x1dd00000)), + (22, 0x3a84b4, e(0x1de00000)), + (22, 0x3a84b5, e(0x1df00000)), + (22, 0x3a84b6, e(0x1e000000)), + (22, 0x3a84b7, e(0x1e100000)), + (22, 0x3a84b8, e(0x1e200000)), + (22, 0x3a84b9, e(0x1e300000)), + (22, 0x3a84ba, e(0x1e400000)), + (22, 0x3a84bb, e(0x1e500000)), + (22, 0x3a84bc, e(0x1e600000)), + (22, 0x3a84bd, e(0x1e700000)), + (22, 0x3a84be, e(0x1e800000)), + (22, 0x3a84bf, e(0x1e900000)), + (22, 0x3a84c0, e(0x1ea00000)), + (22, 0x3a84c1, e(0x1eb00000)), + (22, 0x3a84c2, e(0x1ec00000)), + (22, 0x3a84c3, e(0x1ed00000)), + (22, 0x3a84c4, e(0x1ee00000)), + (22, 0x3a84c5, e(0x1ef00000)), + (22, 0x3a84c6, e(0x1f000000)), + (22, 0x3a84c7, e(0x1f100000)), + (22, 0x3a84c8, e(0x1f200000)), + (22, 0x3a84c9, e(0x1f300000)), + (22, 0x3a84ca, e(0x1f400000)), + (22, 0x3a84cb, e(0x1f500000)), + (22, 0x3a84cc, e(0x1f600000)), + (22, 0x3a84cd, e(0x1f700000)), + (22, 0x3a84ce, e(0x1f800000)), + (22, 0x3a84cf, e(0x1f900000)), + (22, 0x3a84d0, e(0x1fa00000)), + (22, 0x3a84d1, e(0x1fb00000)), + (22, 0x3a84d2, e(0x1fc00000)), + (22, 0x3a84d3, e(0x1fd00000)), + (22, 0x3a84d4, e(0x1fe00000)), + (22, 0x3a84d5, e(0x1ff00000)), + (22, 0x3a84d6, e(0x20000000)), + (22, 0x3a84d7, e(0x20100000)), + (22, 0x3a84d8, e(0x20200000)), + (22, 0x3a84d9, e(0x20300000)), + (22, 0x3a84da, e(0x20400000)), + (22, 0x3a84db, e(0x20500000)), + (22, 0x3a84dc, e(0x20600000)), + (22, 0x3a84dd, e(0x20700000)), + (22, 0x3a84de, e(0x20800000)), + (22, 0x3a84df, e(0x20900000)), + (22, 0x3a84e0, e(0x20a00000)), + (22, 0x3a84e1, e(0x20b00000)), + (22, 0x3a84e2, e(0x20c00000)), + (22, 0x3a84e3, e(0x20d00000)), + (22, 0x3a84e4, e(0x20e00000)), + (22, 0x3a84e5, e(0x20f00000)), + (22, 0x3a84e6, e(0x21000000)), + (22, 0x3a84e7, e(0x21100000)), + (22, 0x3a84e8, e(0x21200000)), + (22, 0x3a84e9, e(0x21300000)), + (22, 0x3a84ea, e(0x21400000)), + (22, 0x3a84eb, e(0x21500000)), + (22, 0x3a84ec, e(0x21600000)), + (22, 0x3a84ed, e(0x21700000)), + (22, 0x3a84ee, e(0x21800000)), + (22, 0x3a84ef, e(0x21900000)), + (22, 0x3a84f0, e(0x21a00000)), + (22, 0x3a84f1, e(0x21b00000)), + (22, 0x3a84f2, e(0x21c00000)), + (22, 0x3a84f3, e(0x21d00000)), + (22, 0x3a84f4, e(0x21e00000)), + (22, 0x3a84f5, e(0x21f00000)), + (22, 0x3a84f6, e(0x22000000)), + (22, 0x3a84f7, e(0x22100000)), + (22, 0x3a84f8, e(0x22200000)), + (22, 0x3a84f9, e(0x22300000)), + (22, 0x3a84fa, e(0x22400000)), + (22, 0x3a84fb, e(0x22500000)), + (22, 0x3a84fc, e(0x22600000)), + (22, 0x3a84fd, e(0x22700000)), + (22, 0x3a84fe, e(0x22800000)), + (22, 0x3a84ff, e(0x22900000)), + (22, 0x3a8500, e(0x22a00000)), + (22, 0x3a8501, e(0x22b00000)), + (22, 0x3a8502, e(0x22c00000)), + (22, 0x3a8503, e(0x22d00000)), + (22, 0x3a8504, e(0x22e00000)), + (22, 0x3a8505, e(0x22f00000)), + (22, 0x3a8506, e(0x23000000)), + (22, 0x3a8507, e(0x23100000)), + (22, 0x3a8508, e(0x23200000)), + (22, 0x3a8509, e(0x23300000)), + (22, 0x3a850a, e(0x23400000)), + (22, 0x3a850b, e(0x23500000)), + (22, 0x3a850c, e(0x23600000)), + (22, 0x3a850d, e(0x23700000)), + (22, 0x3a850e, e(0x23800000)), + (22, 0x3a850f, e(0x23900000)), + (22, 0x3a8510, e(0x23a00000)), + (22, 0x3a8511, e(0x23b00000)), + (22, 0x3a8512, e(0x23c00000)), + (22, 0x3a8513, e(0x23d00000)), + (22, 0x3a8514, e(0x23e00000)), + (22, 0x3a8515, e(0x23f00000)), + (22, 0x3a8516, e(0x24000000)), + (22, 0x3a8517, e(0x24100000)), + (22, 0x3a8518, e(0x24200000)), + (22, 0x3a8519, e(0x24300000)), + (22, 0x3a851a, e(0x24400000)), + (22, 0x3a851b, e(0x24500000)), + (22, 0x3a851c, e(0x24600000)), + (22, 0x3a851d, e(0x24700000)), + (22, 0x3a851e, e(0x24800000)), + (22, 0x3a851f, e(0x24900000)), + (22, 0x3a8520, e(0x24a00000)), + (22, 0x3a8521, e(0x24b00000)), + (22, 0x3a8522, e(0x24c00000)), + (22, 0x3a8523, e(0x24d00000)), + (22, 0x3a8524, e(0x24e00000)), + (22, 0x3a8525, e(0x24f00000)), + (22, 0x3a8526, e(0x25000000)), + (22, 0x3a8527, e(0x25100000)), + (22, 0x3a8528, e(0x25200000)), + (22, 0x3a8529, e(0x25300000)), + (22, 0x3a852a, e(0x25400000)), + (22, 0x3a852b, e(0x25500000)), + (22, 0x3a852c, e(0x25600000)), + (22, 0x3a852d, e(0x25700000)), + (22, 0x3a852e, e(0x25800000)), + (22, 0x3a852f, e(0x25900000)), + (22, 0x3a8530, e(0x25a00000)), + (22, 0x3a8531, e(0x25b00000)), + (22, 0x3a8532, e(0x25c00000)), + (22, 0x3a8533, e(0x25d00000)), + (22, 0x3a8534, e(0x25e00000)), + (22, 0x3a8535, e(0x25f00000)), + (22, 0x3a8536, e(0x26000000)), + (22, 0x3a8537, e(0x26100000)), + (22, 0x3a8538, e(0x26200000)), + (22, 0x3a8539, e(0x26300000)), + (22, 0x3a853a, e(0x26400000)), + (22, 0x3a853b, e(0x26500000)), + (22, 0x3a853c, e(0x26600000)), + (22, 0x3a853d, e(0x26700000)), + (22, 0x3a853e, e(0x26800000)), + (22, 0x3a853f, e(0x26900000)), + (22, 0x3a8540, e(0x26a00000)), + (22, 0x3a8541, e(0x26b00000)), + (22, 0x3a8542, e(0x26c00000)), + (22, 0x3a8543, e(0x26d00000)), + (22, 0x3a8544, e(0x26e00000)), + (22, 0x3a8545, e(0x26f00000)), + (22, 0x3a8546, e(0x27000000)), + (22, 0x3a8547, e(0x27100000)), + (22, 0x3a8548, e(0x27200000)), + (22, 0x3a8549, e(0x27300000)), + (22, 0x3a854a, e(0x27400000)), + (22, 0x3a854b, e(0x27500000)), + (22, 0x3a854c, e(0x27600000)), + (22, 0x3a854d, e(0x27700000)), + (22, 0x3a854e, e(0x27800000)), + (22, 0x3a854f, e(0x27900000)), + (22, 0x3a8550, e(0x27a00000)), + (22, 0x3a8551, e(0x27b00000)), + (22, 0x3a8552, e(0x27c00000)), + (22, 0x3a8553, e(0x27d00000)), + (22, 0x3a8554, e(0x27e00000)), + (22, 0x3a8555, e(0x27f00000)), + (22, 0x3a8556, e(0x28000000)), + (22, 0x3a8557, e(0x28100000)), + (22, 0x3a8558, e(0x28200000)), + (22, 0x3a8559, e(0x28300000)), + (22, 0x3a855a, e(0x28400000)), + (22, 0x3a855b, e(0x28500000)), + (22, 0x3a855c, e(0x28600000)), + (22, 0x3a855d, e(0x28700000)), + (22, 0x3a855e, e(0x28800000)), + (22, 0x3a855f, e(0x28900000)), + (22, 0x3a8560, e(0x28a00000)), + (22, 0x3a8561, e(0x28b00000)), + (22, 0x3a8562, e(0x28c00000)), + (22, 0x3a8563, e(0x28d00000)), + (22, 0x3a8564, e(0x28e00000)), + (22, 0x3a8565, e(0x28f00000)), + (22, 0x3a8566, e(0x29000000)), + (22, 0x3a8567, e(0x29100000)), + (22, 0x3a8568, e(0x29200000)), + (22, 0x3a8569, e(0x29300000)), + (22, 0x3a856a, e(0x29400000)), + (22, 0x3a856b, e(0x29500000)), + (22, 0x3a856c, e(0x29600000)), + (22, 0x3a856d, e(0x29700000)), + (22, 0x3a856e, e(0x29800000)), + (22, 0x3a856f, e(0x29900000)), + (22, 0x3a8570, e(0x29a00000)), + (22, 0x3a8571, e(0x29b00000)), + (22, 0x3a8572, e(0x29c00000)), + (22, 0x3a8573, e(0x29d00000)), + (22, 0x3a8574, e(0x29e00000)), + (22, 0x3a8575, e(0x29f00000)), + (22, 0x3a8576, e(0x2a000000)), + (22, 0x3a8577, e(0x2a100000)), + (22, 0x3a8578, e(0x2a200000)), + (22, 0x3a8579, e(0x2a300000)), + (22, 0x3a857a, e(0x2a400000)), + (22, 0x3a857b, e(0x2a500000)), + (22, 0x3a857c, e(0x2a600000)), + (22, 0x3a857d, e(0x2a700000)), + (22, 0x3a857e, e(0x2a800000)), + (22, 0x3a857f, e(0x2a900000)), + (22, 0x3a8580, e(0x2aa00000)), + (22, 0x3a8581, e(0x2ab00000)), + (22, 0x3a8582, e(0x2ac00000)), + (22, 0x3a8583, e(0x2ad00000)), + (22, 0x3a8584, e(0x2ae00000)), + (22, 0x3a8585, e(0x2af00000)), + (22, 0x3a8586, e(0x2b000000)), + (22, 0x3a8587, e(0x2b100000)), + (22, 0x3a8588, e(0x2b200000)), + (22, 0x3a8589, e(0x2b300000)), + (22, 0x3a858a, e(0x2b400000)), + (22, 0x3a858b, e(0x2b500000)), + (22, 0x3a858c, e(0x2b600000)), + (22, 0x3a858d, e(0x2b700000)), + (22, 0x3a858e, e(0x2b800000)), + (22, 0x3a858f, e(0x2b900000)), + (22, 0x3a8590, e(0x2ba00000)), + (22, 0x3a8591, e(0x2bb00000)), + (22, 0x3a8592, e(0x2bc00000)), + (22, 0x3a8593, e(0x2bd00000)), + (22, 0x3a8594, e(0x2be00000)), + (22, 0x3a8595, e(0x2bf00000)), + (22, 0x3a8596, e(0x2c000000)), + (22, 0x3a8597, e(0x2c100000)), + (22, 0x3a8598, e(0x2c200000)), + (22, 0x3a8599, e(0x2c300000)), + (22, 0x3a859a, e(0x2c400000)), + (22, 0x3a859b, e(0x2c500000)), + (22, 0x3a859c, e(0x2c600000)), + (22, 0x3a859d, e(0x2c700000)), + (22, 0x3a859e, e(0x2c800000)), + (22, 0x3a859f, e(0x2c900000)), + (22, 0x3a85a0, e(0x2ca00000)), + (22, 0x3a85a1, e(0x2cb00000)), + (22, 0x3a85a2, e(0x2cc00000)), + (22, 0x3a85a3, e(0x2cd00000)), + (22, 0x3a85a4, e(0x2ce00000)), + (22, 0x3a85a5, e(0x2cf00000)), + (22, 0x3a85a6, e(0x2d000000)), + (22, 0x3a85a7, e(0x2d100000)), + (22, 0x3a85a8, e(0x2d200000)), + (22, 0x3a85a9, e(0x2d300000)), + (22, 0x3a85aa, e(0x2d400000)), + (22, 0x3a85ab, e(0x2d500000)), + (22, 0x3a85ac, e(0x2d600000)), + (22, 0x3a85ad, e(0x2d700000)), + (22, 0x3a85ae, e(0x2d800000)), + (22, 0x3a85af, e(0x2d900000)), + (22, 0x3a85b0, e(0x2da00000)), + (22, 0x3a85b1, e(0x2db00000)), + (22, 0x3a85b2, e(0x2dc00000)), + (22, 0x3a85b3, e(0x2dd00000)), + (22, 0x3a85b4, e(0x2de00000)), + (22, 0x3a85b5, e(0x2df00000)), + (22, 0x3a85b6, e(0x2e000000)), + (22, 0x3a85b7, e(0x2e100000)), + (22, 0x3a85b8, e(0x2e200000)), + (22, 0x3a85b9, e(0x2e300000)), + (22, 0x3a85ba, e(0x2e400000)), + (22, 0x3a85bb, e(0x2e500000)), + (22, 0x3a85bc, e(0x2e600000)), + (22, 0x3a85bd, e(0x2e700000)), + (22, 0x3a85be, e(0x2e800000)), + (22, 0x3a85bf, e(0x2e900000)), + (22, 0x3a85c0, e(0x2ea00000)), + (22, 0x3a85c1, e(0x2eb00000)), + (22, 0x3a85c2, e(0x2ec00000)), + (22, 0x3a85c3, e(0x2ed00000)), + (22, 0x3a85c4, e(0x2ee00000)), + (22, 0x3a85c5, e(0x2ef00000)), + (22, 0x3a85c6, e(0x2f000000)), + (22, 0x3a85c7, e(0x2f100000)), + (22, 0x3a85c8, e(0x2f200000)), + (22, 0x3a85c9, e(0x2f300000)), + (22, 0x3a85ca, e(0x2f400000)), + (22, 0x3a85cb, e(0x2f500000)), + (22, 0x3a85cc, e(0x2f600000)), + (22, 0x3a85cd, e(0x2f700000)), + (22, 0x3a85ce, e(0x2f800000)), + (22, 0x3a85cf, e(0x2f900000)), + (22, 0x3a85d0, e(0x2fa00000)), + (22, 0x3a85d1, e(0x2fb00000)), + (22, 0x3a85d2, e(0x2fc00000)), + (22, 0x3a85d3, e(0x2fd00000)), + (22, 0x3a85d4, e(0x2fe00000)), + (22, 0x3a85d5, e(0x2ff00000)), + (22, 0x3a85d6, e(0x30000000)), + (22, 0x3a85d7, e(0x30100000)), + (22, 0x3a85d8, e(0x30200000)), + (22, 0x3a85d9, e(0x30300000)), + (22, 0x3a85da, e(0x30400000)), + (22, 0x3a85db, e(0x30500000)), + (22, 0x3a85dc, e(0x30600000)), + (22, 0x3a85dd, e(0x30700000)), + (22, 0x3a85de, e(0x30800000)), + (22, 0x3a85df, e(0x30900000)), + (22, 0x3a85e0, e(0x30a00000)), + (22, 0x3a85e1, e(0x30b00000)), + (22, 0x3a85e2, e(0x30c00000)), + (22, 0x3a85e3, e(0x30d00000)), + (22, 0x3a85e4, e(0x30e00000)), + (22, 0x3a85e5, e(0x30f00000)), + (22, 0x3a85e6, e(0x31000000)), + (22, 0x3a85e7, e(0x31100000)), + (22, 0x3a85e8, e(0x31200000)), + (22, 0x3a85e9, e(0x31300000)), + (22, 0x3a85ea, e(0x31400000)), + (22, 0x3a85eb, e(0x31500000)), + (22, 0x3a85ec, e(0x31600000)), + (22, 0x3a85ed, e(0x31700000)), + (22, 0x3a85ee, e(0x31800000)), + (22, 0x3a85ef, e(0x31900000)), + (22, 0x3a85f0, e(0x31a00000)), + (22, 0x3a85f1, e(0x31b00000)), + (22, 0x3a85f2, e(0x31c00000)), + (22, 0x3a85f3, e(0x31d00000)), + (22, 0x3a85f4, e(0x31e00000)), + (22, 0x3a85f5, e(0x31f00000)), + (22, 0x3a85f6, e(0x32000000)), + (22, 0x3a85f7, e(0x32100000)), + (22, 0x3a85f8, e(0x32200000)), + (22, 0x3a85f9, e(0x32300000)), + (22, 0x3a85fa, e(0x32400000)), + (22, 0x3a85fb, e(0x32500000)), + (22, 0x3a85fc, e(0x32600000)), + (22, 0x3a85fd, e(0x32700000)), + (22, 0x3a85fe, e(0x32800000)), + (22, 0x3a85ff, e(0x32900000)), + (22, 0x3a8600, e(0x32a00000)), + (22, 0x3a8601, e(0x32b00000)), + (22, 0x3a8602, e(0x32c00000)), + (22, 0x3a8603, e(0x32d00000)), + (22, 0x3a8604, e(0x32e00000)), + (22, 0x3a8605, e(0x32f00000)), + (22, 0x3a8606, e(0x33000000)), + (22, 0x3a8607, e(0x33100000)), + (22, 0x3a8608, e(0x33200000)), + (22, 0x3a8609, e(0x33300000)), + (22, 0x3a860a, e(0x33400000)), + (22, 0x3a860b, e(0x33500000)), + (22, 0x3a860c, e(0x33600000)), + (22, 0x3a860d, e(0x33700000)), + (22, 0x3a860e, e(0x33800000)), + (22, 0x3a860f, e(0x33900000)), + (22, 0x3a8610, e(0x33a00000)), + (22, 0x3a8611, e(0x33b00000)), + (22, 0x3a8612, e(0x33c00000)), + (22, 0x3a8613, e(0x33d00000)), + (22, 0x3a8614, e(0x33e00000)), + (22, 0x3a8615, e(0x33f00000)), + (22, 0x3a8616, e(0x34000000)), + (22, 0x3a8617, e(0x34100000)), + (22, 0x3a8618, e(0x34200000)), + (22, 0x3a8619, e(0x34300000)), + (22, 0x3a861a, e(0x34400000)), + (22, 0x3a861b, e(0x34500000)), + (22, 0x3a861c, e(0x34600000)), + (22, 0x3a861d, e(0x34700000)), + (22, 0x3a861e, e(0x34800000)), + (22, 0x3a861f, e(0x34900000)), + (22, 0x3a8620, e(0x34a00000)), + (22, 0x3a8621, e(0x34b00000)), + (22, 0x3a8622, e(0x34c00000)), + (22, 0x3a8623, e(0x34d00000)), + (22, 0x3a8624, e(0x34e00000)), + (22, 0x3a8625, e(0x34f00000)), + (22, 0x3a8626, e(0x35000000)), + (22, 0x3a8627, e(0x35100000)), + (22, 0x3a8628, e(0x35200000)), + (22, 0x3a8629, e(0x35300000)), + (18, 0xf787, e(0x35400000)), + (18, 0xd1d4, e(0x35500000)), + (19, 0x77f9f, e(0x35600000)), + (18, 0x3b8fb, e(0x35700000)), + (19, 0x1ef1a, e(0x35800000)), + (21, 0x1d4315, e(0x35900000)), + (16, 0x3de2, e(0x35a00000)), + (22, 0x3a862c, e(0x35b00000)), + (22, 0x3a862d, e(0x35c00000)), + (22, 0x3a862e, e(0x35d00000)), + (22, 0x3a862f, e(0x35e00000)), + (22, 0x3a8630, e(0x35f00000)), + (22, 0x3a8631, e(0x36000000)), + (22, 0x3a8632, e(0x36100000)), + (22, 0x3a8633, e(0x36200000)), + (22, 0x3a8634, e(0x36300000)), + (22, 0x3a8635, e(0x36400000)), + (22, 0x3a8636, e(0x36500000)), + (22, 0x3a8637, e(0x36600000)), + (22, 0x3a8638, e(0x36700000)), + (21, 0x1d431d, e(0x36800000)), + (22, 0x3a8639, e(0x36900000)), + (22, 0x3a863c, e(0x36a00000)), + (16, 0x3de0, e(0x36b00000)), + (18, 0x3a95e, e(0x36c00000)), + (21, 0x1d431f, e(0x36d00000)), + (22, 0x3a863d, e(0x36e00000)), + (22, 0x3a8640, e(0x36f00000)), + (20, 0x34749, e(0x37000000)), + (20, 0x3474d, e(0x37100000)), + (22, 0x3a8641, e(0x37200000)), + (22, 0x3a8642, e(0x37300000)), + (22, 0x3a8643, e(0x37400000)), + (22, 0x3a8644, e(0x37500000)), + (22, 0x3a8645, e(0x37600000)), + (22, 0x3a8646, e(0x37700000)), + (22, 0x3a8647, e(0x37800000)), + (22, 0x3a8648, e(0x37900000)), + (22, 0x3a8649, e(0x37a00000)), + (22, 0x3a864a, e(0x37b00000)), + (22, 0x3a864b, e(0x37c00000)), + (22, 0x3a864c, e(0x37d00000)), + (22, 0x3a864d, e(0x37e00000)), + (22, 0x3a864e, e(0x37f00000)), + (22, 0x3a864f, e(0x38000000)), + (22, 0x3a8650, e(0x38100000)), + (22, 0x3a8651, e(0x38200000)), + (22, 0x3a8652, e(0x38300000)), + (22, 0x3a8653, e(0x38400000)), + (22, 0x3a8654, e(0x38500000)), + (22, 0x3a8655, e(0x38600000)), + (22, 0x3a8656, e(0x38700000)), + (22, 0x3a8657, e(0x38800000)), + (22, 0x3a8658, e(0x38900000)), + (22, 0x3a8659, e(0x38a00000)), + (22, 0x3a865a, e(0x38b00000)), + (22, 0x3a865b, e(0x38c00000)), + (22, 0x3a865c, e(0x38d00000)), + (22, 0x3a865d, e(0x38e00000)), + (21, 0x1d432f, e(0x38f00000)), + (22, 0x3a8660, e(0x39000000)), + (22, 0x3a8661, e(0x39100000)), + (22, 0x3a8662, e(0x39200000)), + (21, 0x1d4332, e(0x39300000)), + (18, 0xd1d5, e(0x39400000)), + (18, 0x3bfda, e(0x39500000)), + (15, 0x7528, e(0x39600000)), + (15, 0x7529, e(0x39700000)), + (18, 0x3a95f, e(0x39800000)), + (17, 0x1d434, e(0x39900000)), + (19, 0x750cd, e(0x39a00000)), + (18, 0x3a867, e(0x39b00000)), + (19, 0x771fd, e(0x39c00000)), + (15, 0x7764, e(0x39d00000)), + (20, 0xee3f9, e(0x39e00000)), + (18, 0x3bfdb, e(0x39f00000)), + (16, 0xeff7, e(0x3a000000)), + (18, 0xd1d6, e(0x3a100000)), + (22, 0x3a8663, e(0x3a200000)), + (22, 0x3a8666, e(0x3a300000)), + (18, 0x3b8fa, e(0x3a400000)), + (17, 0x1d435, e(0x3a500000)), + (17, 0x1dfe4, e(0x3a600000)), + (19, 0x750d8, e(0x3a700000)), + (18, 0x3a95b, e(0x3a800000)), + (19, 0x77f9e, e(0x3a900000)), + (19, 0x750d9, e(0x3aa00000)), + (18, 0xd1d7, e(0x3ab00000)), + (18, 0x3b8ff, e(0x3ac00000)), + (17, 0x1dc7c, e(0x3ad00000)), + (19, 0x750da, e(0x3ae00000)), + (17, 0x7bc2, e(0x3af00000)), + (18, 0x3bfca, e(0x3b000000)), + (19, 0x1a3a5, e(0x3b100000)), + (17, 0x1d4ac, e(0x3b200000)), + (18, 0x3a86e, e(0x3b300000)), + (17, 0x1d438, e(0x3b400000)), + (18, 0x3bfcb, e(0x3b500000)), + (19, 0x1a3a7, e(0x3b600000)), + (17, 0x1d4ae, e(0x3b700000)), + (18, 0x3bfce, e(0x3b800000)), + (18, 0xf78c, e(0x3b900000)), + (17, 0x1dfec, e(0x3ba00000)), + (17, 0x1d439, e(0x3bb00000)), + (17, 0x68e8, e(0x3bc00000)), + (18, 0xf786, e(0x3bd00000)), + (15, 0x771e, e(0x3be00000)), + (17, 0x1dfe6, e(0x3bf00000)), + (15, 0x77f8, e(0x3c000000)), + (14, 0x3bb3, e(0x3c100000)), + (14, 0x3b8e, e(0x3c200000)), + (14, 0x3a82, e(0x3c300000)), + (14, 0x3b96, e(0x3c400000)), + (13, 0x68f, e(0x3c500000)), + (12, 0x3d4, e(0x3c600000)), + (13, 0x1dca, e(0x3c700000)), + (12, 0x346, e(0x3c800000)), + (12, 0xee7, e(0x3c900000)), + (12, 0xeea, e(0x3ca00000)), + (11, 0x1ed, e(0x3cb00000)), + (12, 0x3df, e(0x3cc00000)), + (11, 0x1a2, e(0x3cd00000)), + (11, 0x56f, e(0x3ce00000)), + (11, 0xb9, e(0x3cf00000)), + (13, 0x7b2, e(0x3d000000)), + (13, 0x1dd8, e(0x3d100000)), + (13, 0x15ba, e(0x3d200000)), + (12, 0xee6, e(0x3d300000)), + (13, 0x15b8, e(0x3d400000)), + (14, 0xf79, e(0x3d500000)), + (14, 0x3a81, e(0x3d600000)), + (14, 0xd1c, e(0x3d700000)), + (15, 0x7765, e(0x3d800000)), + (14, 0xf54, e(0x3d900000)), + (13, 0x15b9, e(0x3da00000)), + (13, 0x7ab, e(0x3db00000)), + (15, 0x7500, e(0x3dc00000)), + (15, 0x1eaa, e(0x3dd00000)), + (15, 0x7501, e(0x3de00000)), + (15, 0x1eab, e(0x3df00000)), + (14, 0x3b97, e(0x3e000000)), + (15, 0x752a, e(0x3e100000)), + (15, 0x77fa, e(0x3e200000)), + (10, 0xf4, d(0x3e35798e, 0xe2308c3a)), + (14, 0x3a93, e(0x3e300000)), + (11, 0x77c, d(0x3e45798e, 0xe2308c3a)), + (13, 0x1dc6, e(0x3e400000)), + (13, 0x7bd, e(0x3e500000)), + (13, 0x1dff, e(0x3e600000)), + (12, 0xefe, e(0x3e700000)), + (8, 0xaf, d(0x3e8ad7f2, 0x9abcaf4a)), + (12, 0xeed, e(0x3e800000)), + (11, 0xb8, e(0x3e900000)), + (12, 0x3d8, e(0x3ea00000)), + (11, 0x1eb, e(0x3eb00000)), + (9, 0x1d2, d(0x3ec0c6f7, 0xa0b5ed8e)), + (13, 0x1d4b, e(0x3ec00000)), + (13, 0x7b3, e(0x3ed00000)), + (10, 0x5d, e(0x3ee00000)), + (12, 0xeeb, e(0x3ef00000)), + (11, 0x1ee, e(0x3f000000)), + (10, 0x5f, e(0x3f100000)), + (10, 0x2b6, e(0x3f200000)), + (9, 0x1de, e(0x3f300000)), + (10, 0xd0, d(0x3f454c98, 0x5f06f694)), + (6, 0x9, d(0x3f4a36e2, 0xeb1c432d)), + (8, 0xe8, e(0x3f400000)), + (4, 0xf, d(0x3f50624d, 0xd2f1a9fc)), + (8, 0xae, e(0x3f500000)), + (5, 0x16, d(0x3f60624d, 0xd2f1a9fc)), + (7, 0x1b, e(0x3f600000)), + (7, 0x76, e(0x3f700000)), + (7, 0xa, e(0x3f800000)), + (6, 0x8, e(0x3f900000)), + (6, 0xe, e(0x3fa00000)), + (11, 0x751, d(0x3fbe69ad, 0x42c3c9ee)), + (6, 0x4, e(0x3fb00000)), + (6, 0xc, e(0x3fc00000)), + (5, 0x3, e(0x3fd00000)), + (11, 0x777, d(0x3fe00000, 0x00000000)), + (9, 0x1d6, d(0x3fefffff, 0xf8000002)), + (4, 0x8, e(0x3fe00000)), + (4, 0x0, d(0x3ff00000, 0x00000000)), + (5, 0x13, e(0x3ff00000)), + (5, 0x1b, e(0x40000000)), + (9, 0x15a, d(0x401921fb, 0x54442d18)), + (5, 0x17, e(0x40100000)), + (5, 0x12, e(0x40200000)), + (11, 0x774, d(0x4035ee14, 0x80000000)), + (5, 0x19, e(0x40300000)), + (9, 0x1d3, d(0x404ca5dc, 0x1a63c1f8)), + (5, 0x1a, e(0x40400000)), + (11, 0x77e, d(0x405bb32f, 0xe0000000)), + (10, 0x5e, d(0x405c332f, 0xe0000000)), + (5, 0x18, e(0x40500000)), + (9, 0x1d7, d(0x40668000, 0x00000000)), + (5, 0x1c, e(0x40600000)), + (9, 0x1d5, d(0x40768000, 0x00000000)), + (5, 0x14, e(0x40700000)), + (11, 0x77d, d(0x408f4000, 0x00000000)), + (5, 0x5, e(0x40800000)), + (10, 0xd2, d(0x409233ff, 0xffffffff)), + (8, 0x3c, d(0x40923400, 0x00000000)), + (11, 0x753, d(0x40923400, 0x00000001)), + (10, 0xd3, d(0x4092abff, 0xffffffff)), + (8, 0x35, d(0x4092ac00, 0x00000000)), + (11, 0x770, d(0x4092ac00, 0x00000001)), + (8, 0x16, e(0x40900000)), + (12, 0xee2, e(0x40a00000)), + (12, 0xee4, e(0x40b00000)), + (7, 0x1f, d(0x40c81c80, 0x00000000)), + (8, 0xac, e(0x40c00000)), + (13, 0x15bb, e(0x40d00000)), + (22, 0x3a8667, e(0x40e00000)), + (22, 0x3a86d8, e(0x40f00000)), + (22, 0x3a86d9, e(0x41000000)), + (22, 0x3a86da, e(0x41100000)), + (17, 0x1dc7e, e(0x41200000)), + (22, 0x3a86db, e(0x41300000)), + (22, 0x3a86dc, e(0x41400000)), + (22, 0x3a86dd, e(0x41500000)), + (22, 0x3a86de, e(0x41600000)), + (22, 0x3a86df, e(0x41700000)), + (22, 0x3a86f0, e(0x41800000)), + (22, 0x3a86f1, e(0x41900000)), + (22, 0x3a86f2, e(0x41a00000)), + (22, 0x3a86f3, e(0x41b00000)), + (6, 0x2a, d(0x41cdcd64, 0xff800000)), + (22, 0x3a86f4, e(0x41c00000)), + (22, 0x3a86f5, e(0x41d00000)), + (22, 0x3a86f6, e(0x41e00000)), + (22, 0x3a86f7, e(0x41f00000)), + (22, 0x3a86f8, e(0x42000000)), + (22, 0x3a86f9, e(0x42100000)), + (22, 0x3a86fa, e(0x42200000)), + (22, 0x3a86fb, e(0x42300000)), + (22, 0x3a86fc, e(0x42400000)), + (22, 0x3a86fd, e(0x42500000)), + (22, 0x3a86fe, e(0x42600000)), + (22, 0x3a86ff, e(0x42700000)), + (22, 0x3a8740, e(0x42800000)), + (22, 0x3a8741, e(0x42900000)), + (22, 0x3a8742, e(0x42a00000)), + (22, 0x3a8743, e(0x42b00000)), + (22, 0x3a8744, e(0x42c00000)), + (22, 0x3a8745, e(0x42d00000)), + (22, 0x3a8746, e(0x42e00000)), + (22, 0x3a8747, e(0x42f00000)), + (22, 0x3a8748, e(0x43000000)), + (22, 0x3a8749, e(0x43100000)), + (22, 0x3a874a, e(0x43200000)), + (22, 0x3a874b, e(0x43300000)), + (22, 0x3a874c, e(0x43400000)), + (22, 0x3a874d, e(0x43500000)), + (22, 0x3a874e, e(0x43600000)), + (22, 0x3a874f, e(0x43700000)), + (22, 0x3a8750, e(0x43800000)), + (22, 0x3a8751, e(0x43900000)), + (22, 0x3a8752, e(0x43a00000)), + (22, 0x3a8753, e(0x43b00000)), + (22, 0x3a8754, e(0x43c00000)), + (22, 0x3a8755, e(0x43d00000)), + (22, 0x3a8756, e(0x43e00000)), + (22, 0x3a8757, e(0x43f00000)), + (22, 0x3a8758, e(0x44000000)), + (15, 0x1a3b, e(0x44100000)), + (22, 0x3a8759, e(0x44200000)), + (22, 0x3a875a, e(0x44300000)), + (22, 0x3a875b, e(0x44400000)), + (22, 0x3a875c, e(0x44500000)), + (22, 0x3a875d, e(0x44600000)), + (22, 0x3a875e, e(0x44700000)), + (22, 0x3a875f, e(0x44800000)), + (22, 0x3a8760, e(0x44900000)), + (22, 0x3a8761, e(0x44a00000)), + (22, 0x3a8762, e(0x44b00000)), + (22, 0x3a8763, e(0x44c00000)), + (22, 0x3a8764, e(0x44d00000)), + (22, 0x3a8765, e(0x44e00000)), + (22, 0x3a8766, e(0x44f00000)), + (22, 0x3a8767, e(0x45000000)), + (22, 0x3a8768, e(0x45100000)), + (22, 0x3a8769, e(0x45200000)), + (22, 0x3a876a, e(0x45300000)), + (22, 0x3a876b, e(0x45400000)), + (22, 0x3a876c, e(0x45500000)), + (22, 0x3a876d, e(0x45600000)), + (22, 0x3a876e, e(0x45700000)), + (22, 0x3a876f, e(0x45800000)), + (22, 0x3a8770, e(0x45900000)), + (22, 0x3a8771, e(0x45a00000)), + (22, 0x3a8772, e(0x45b00000)), + (22, 0x3a8773, e(0x45c00000)), + (22, 0x3a8774, e(0x45d00000)), + (22, 0x3a8775, e(0x45e00000)), + (22, 0x3a8776, e(0x45f00000)), + (22, 0x3a8777, e(0x46000000)), + (22, 0x3a8778, e(0x46100000)), + (22, 0x3a8779, e(0x46200000)), + (22, 0x3a877a, e(0x46300000)), + (22, 0x3a877b, e(0x46400000)), + (22, 0x3a877c, e(0x46500000)), + (22, 0x3a877d, e(0x46600000)), + (22, 0x3a877e, e(0x46700000)), + (22, 0x3a877f, e(0x46800000)), + (22, 0x3a8780, e(0x46900000)), + (22, 0x3a8781, e(0x46a00000)), + (22, 0x3a8782, e(0x46b00000)), + (22, 0x3a8783, e(0x46c00000)), + (22, 0x3a8784, e(0x46d00000)), + (22, 0x3a8785, e(0x46e00000)), + (22, 0x3a8786, e(0x46f00000)), + (22, 0x3a8787, e(0x47000000)), + (22, 0x3a8788, e(0x47100000)), + (22, 0x3a8789, e(0x47200000)), + (22, 0x3a878a, e(0x47300000)), + (22, 0x3a878b, e(0x47400000)), + (22, 0x3a878c, e(0x47500000)), + (22, 0x3a878d, e(0x47600000)), + (22, 0x3a878e, e(0x47700000)), + (22, 0x3a878f, e(0x47800000)), + (22, 0x3a8790, e(0x47900000)), + (22, 0x3a8791, e(0x47a00000)), + (22, 0x3a8792, e(0x47b00000)), + (22, 0x3a8793, e(0x47c00000)), + (22, 0x3a8794, e(0x47d00000)), + (22, 0x3a8795, e(0x47e00000)), + (22, 0x3a8796, e(0x47f00000)), + (22, 0x3a8797, e(0x48000000)), + (22, 0x3a8798, e(0x48100000)), + (22, 0x3a8799, e(0x48200000)), + (22, 0x3a879a, e(0x48300000)), + (22, 0x3a879b, e(0x48400000)), + (22, 0x3a879c, e(0x48500000)), + (22, 0x3a879d, e(0x48600000)), + (22, 0x3a879e, e(0x48700000)), + (22, 0x3a879f, e(0x48800000)), + (22, 0x3a87a0, e(0x48900000)), + (22, 0x3a87a1, e(0x48a00000)), + (22, 0x3a87a2, e(0x48b00000)), + (22, 0x3a87a3, e(0x48c00000)), + (22, 0x3a87a4, e(0x48d00000)), + (22, 0x3a87a5, e(0x48e00000)), + (22, 0x3a87a6, e(0x48f00000)), + (22, 0x3a87a7, e(0x49000000)), + (22, 0x3a87a8, e(0x49100000)), + (22, 0x3a87a9, e(0x49200000)), + (22, 0x3a87aa, e(0x49300000)), + (22, 0x3a87ab, e(0x49400000)), + (22, 0x3a87ac, e(0x49500000)), + (22, 0x3a87ad, e(0x49600000)), + (22, 0x3a87ae, e(0x49700000)), + (22, 0x3a87af, e(0x49800000)), + (22, 0x3a87b0, e(0x49900000)), + (22, 0x3a87b1, e(0x49a00000)), + (22, 0x3a87b2, e(0x49b00000)), + (22, 0x3a87b3, e(0x49c00000)), + (22, 0x3a87b4, e(0x49d00000)), + (22, 0x3a87b5, e(0x49e00000)), + (22, 0x3a87b6, e(0x49f00000)), + (22, 0x3a87b7, e(0x4a000000)), + (22, 0x3a87b8, e(0x4a100000)), + (22, 0x3a87b9, e(0x4a200000)), + (22, 0x3a87ba, e(0x4a300000)), + (22, 0x3a87bb, e(0x4a400000)), + (22, 0x3a87bc, e(0x4a500000)), + (22, 0x3a87bd, e(0x4a600000)), + (22, 0x3a87be, e(0x4a700000)), + (22, 0x3a87bf, e(0x4a800000)), + (22, 0x3a87c0, e(0x4a900000)), + (22, 0x3a87c1, e(0x4aa00000)), + (22, 0x3a87c2, e(0x4ab00000)), + (22, 0x3a87c3, e(0x4ac00000)), + (22, 0x3a87c4, e(0x4ad00000)), + (22, 0x3a87c5, e(0x4ae00000)), + (22, 0x3a87c6, e(0x4af00000)), + (22, 0x3a87c7, e(0x4b000000)), + (22, 0x3a87c8, e(0x4b100000)), + (22, 0x3a87c9, e(0x4b200000)), + (22, 0x3a87ca, e(0x4b300000)), + (22, 0x3a87cb, e(0x4b400000)), + (22, 0x3a87cc, e(0x4b500000)), + (22, 0x3a87cd, e(0x4b600000)), + (22, 0x3a87ce, e(0x4b700000)), + (22, 0x3a87cf, e(0x4b800000)), + (22, 0x3a87d0, e(0x4b900000)), + (22, 0x3a87d1, e(0x4ba00000)), + (22, 0x3a87d2, e(0x4bb00000)), + (22, 0x3a87d3, e(0x4bc00000)), + (22, 0x3a87d4, e(0x4bd00000)), + (22, 0x3a87d5, e(0x4be00000)), + (22, 0x3a87d6, e(0x4bf00000)), + (22, 0x3a87d7, e(0x4c000000)), + (22, 0x3a87d8, e(0x4c100000)), + (22, 0x3a87d9, e(0x4c200000)), + (22, 0x3a87da, e(0x4c300000)), + (22, 0x3a87db, e(0x4c400000)), + (22, 0x3a87dc, e(0x4c500000)), + (22, 0x3a87dd, e(0x4c600000)), + (22, 0x3a87de, e(0x4c700000)), + (22, 0x3a87df, e(0x4c800000)), + (22, 0x3a87e0, e(0x4c900000)), + (22, 0x3a87e1, e(0x4ca00000)), + (22, 0x3a87e2, e(0x4cb00000)), + (22, 0x3a87e3, e(0x4cc00000)), + (22, 0x3a87e4, e(0x4cd00000)), + (22, 0x3a87e5, e(0x4ce00000)), + (22, 0x3a87e6, e(0x4cf00000)), + (22, 0x3a87e7, e(0x4d000000)), + (22, 0x3a87e8, e(0x4d100000)), + (22, 0x3a87e9, e(0x4d200000)), + (22, 0x3a87ea, e(0x4d300000)), + (22, 0x3a87eb, e(0x4d400000)), + (22, 0x3a87ec, e(0x4d500000)), + (22, 0x3a87ed, e(0x4d600000)), + (22, 0x3a87ee, e(0x4d700000)), + (22, 0x3a87ef, e(0x4d800000)), + (22, 0x3a87f0, e(0x4d900000)), + (22, 0x3a87f1, e(0x4da00000)), + (22, 0x3a87f2, e(0x4db00000)), + (22, 0x3a87f3, e(0x4dc00000)), + (22, 0x3a87f4, e(0x4dd00000)), + (22, 0x3a87f5, e(0x4de00000)), + (22, 0x3a87f6, e(0x4df00000)), + (22, 0x3a87f7, e(0x4e000000)), + (22, 0x3a87f8, e(0x4e100000)), + (22, 0x3a87f9, e(0x4e200000)), + (22, 0x3a87fa, e(0x4e300000)), + (22, 0x3a87fb, e(0x4e400000)), + (22, 0x3a87fc, e(0x4e500000)), + (22, 0x3a87fd, e(0x4e600000)), + (22, 0x3a87fe, e(0x4e700000)), + (22, 0x3a87ff, e(0x4e800000)), + (22, 0x3a9000, e(0x4e900000)), + (22, 0x3a9001, e(0x4ea00000)), + (22, 0x3a9002, e(0x4eb00000)), + (22, 0x3a9003, e(0x4ec00000)), + (22, 0x3a9004, e(0x4ed00000)), + (22, 0x3a9005, e(0x4ee00000)), + (22, 0x3a9006, e(0x4ef00000)), + (22, 0x3a9007, e(0x4f000000)), + (22, 0x3a9008, e(0x4f100000)), + (22, 0x3a9009, e(0x4f200000)), + (22, 0x3a900a, e(0x4f300000)), + (22, 0x3a900b, e(0x4f400000)), + (22, 0x3a900c, e(0x4f500000)), + (22, 0x3a900d, e(0x4f600000)), + (22, 0x3a900e, e(0x4f700000)), + (22, 0x3a900f, e(0x4f800000)), + (22, 0x3a9010, e(0x4f900000)), + (22, 0x3a9011, e(0x4fa00000)), + (22, 0x3a9012, e(0x4fb00000)), + (22, 0x3a9013, e(0x4fc00000)), + (22, 0x3a9014, e(0x4fd00000)), + (22, 0x3a9015, e(0x4fe00000)), + (22, 0x3a9016, e(0x4ff00000)), + (22, 0x3a9017, e(0x50000000)), + (22, 0x3a9018, e(0x50100000)), + (22, 0x3a9019, e(0x50200000)), + (22, 0x3a901a, e(0x50300000)), + (22, 0x3a901b, e(0x50400000)), + (22, 0x3a901c, e(0x50500000)), + (22, 0x3a901d, e(0x50600000)), + (22, 0x3a901e, e(0x50700000)), + (22, 0x3a901f, e(0x50800000)), + (22, 0x3a9020, e(0x50900000)), + (22, 0x3a9021, e(0x50a00000)), + (22, 0x3a9022, e(0x50b00000)), + (22, 0x3a9023, e(0x50c00000)), + (22, 0x3a9024, e(0x50d00000)), + (22, 0x3a9025, e(0x50e00000)), + (22, 0x3a9026, e(0x50f00000)), + (22, 0x3a9027, e(0x51000000)), + (22, 0x3a9028, e(0x51100000)), + (22, 0x3a9029, e(0x51200000)), + (22, 0x3a902a, e(0x51300000)), + (22, 0x3a902b, e(0x51400000)), + (22, 0x3a902c, e(0x51500000)), + (22, 0x3a902d, e(0x51600000)), + (22, 0x3a902e, e(0x51700000)), + (22, 0x3a902f, e(0x51800000)), + (22, 0x3a9030, e(0x51900000)), + (22, 0x3a9031, e(0x51a00000)), + (22, 0x3a9032, e(0x51b00000)), + (22, 0x3a9033, e(0x51c00000)), + (22, 0x3a9034, e(0x51d00000)), + (22, 0x3a9035, e(0x51e00000)), + (22, 0x3a9036, e(0x51f00000)), + (22, 0x3a9037, e(0x52000000)), + (22, 0x3a9038, e(0x52100000)), + (22, 0x3a9039, e(0x52200000)), + (22, 0x3a903a, e(0x52300000)), + (22, 0x3a903b, e(0x52400000)), + (22, 0x3a903c, e(0x52500000)), + (22, 0x3a903d, e(0x52600000)), + (22, 0x3a903e, e(0x52700000)), + (22, 0x3a903f, e(0x52800000)), + (22, 0x3a9040, e(0x52900000)), + (22, 0x3a9041, e(0x52a00000)), + (22, 0x3a9042, e(0x52b00000)), + (22, 0x3a9043, e(0x52c00000)), + (22, 0x3a9044, e(0x52d00000)), + (22, 0x3a9045, e(0x52e00000)), + (22, 0x3a9046, e(0x52f00000)), + (22, 0x3a9047, e(0x53000000)), + (22, 0x3a9048, e(0x53100000)), + (22, 0x3a9049, e(0x53200000)), + (22, 0x3a904a, e(0x53300000)), + (22, 0x3a904b, e(0x53400000)), + (22, 0x3a904c, e(0x53500000)), + (22, 0x3a904d, e(0x53600000)), + (22, 0x3a904e, e(0x53700000)), + (22, 0x3a904f, e(0x53800000)), + (22, 0x3a9050, e(0x53900000)), + (22, 0x3a9051, e(0x53a00000)), + (22, 0x3a9052, e(0x53b00000)), + (22, 0x3a9053, e(0x53c00000)), + (22, 0x3a9054, e(0x53d00000)), + (22, 0x3a9055, e(0x53e00000)), + (22, 0x3a9056, e(0x53f00000)), + (22, 0x3a9057, e(0x54000000)), + (22, 0x3a9058, e(0x54100000)), + (22, 0x3a9059, e(0x54200000)), + (22, 0x3a905a, e(0x54300000)), + (22, 0x3a905b, e(0x54400000)), + (22, 0x3a905c, e(0x54500000)), + (22, 0x3a905d, e(0x54600000)), + (22, 0x3a905e, e(0x54700000)), + (22, 0x3a905f, e(0x54800000)), + (22, 0x3a9060, e(0x54900000)), + (22, 0x3a9061, e(0x54a00000)), + (22, 0x3a9062, e(0x54b00000)), + (22, 0x3a9063, e(0x54c00000)), + (22, 0x3a9064, e(0x54d00000)), + (22, 0x3a9065, e(0x54e00000)), + (22, 0x3a9066, e(0x54f00000)), + (22, 0x3a9067, e(0x55000000)), + (22, 0x3a9068, e(0x55100000)), + (22, 0x3a9069, e(0x55200000)), + (22, 0x3a906a, e(0x55300000)), + (22, 0x3a906b, e(0x55400000)), + (22, 0x3a906c, e(0x55500000)), + (22, 0x3a906d, e(0x55600000)), + (22, 0x3a906e, e(0x55700000)), + (22, 0x3a906f, e(0x55800000)), + (22, 0x3a9070, e(0x55900000)), + (22, 0x3a9071, e(0x55a00000)), + (22, 0x3a9072, e(0x55b00000)), + (22, 0x3a9073, e(0x55c00000)), + (22, 0x3a9074, e(0x55d00000)), + (22, 0x3a9075, e(0x55e00000)), + (22, 0x3a9076, e(0x55f00000)), + (22, 0x3a9077, e(0x56000000)), + (22, 0x3a9078, e(0x56100000)), + (22, 0x3a9079, e(0x56200000)), + (22, 0x3a907a, e(0x56300000)), + (22, 0x3a907b, e(0x56400000)), + (22, 0x3a907c, e(0x56500000)), + (22, 0x3a907d, e(0x56600000)), + (22, 0x3a907e, e(0x56700000)), + (22, 0x3a907f, e(0x56800000)), + (22, 0x3a9080, e(0x56900000)), + (22, 0x3a9081, e(0x56a00000)), + (22, 0x3a9082, e(0x56b00000)), + (22, 0x3a9083, e(0x56c00000)), + (22, 0x3a9084, e(0x56d00000)), + (22, 0x3a9085, e(0x56e00000)), + (22, 0x3a9086, e(0x56f00000)), + (22, 0x3a9087, e(0x57000000)), + (22, 0x3a9088, e(0x57100000)), + (22, 0x3a9089, e(0x57200000)), + (22, 0x3a908a, e(0x57300000)), + (22, 0x3a908b, e(0x57400000)), + (22, 0x3a908c, e(0x57500000)), + (22, 0x3a908d, e(0x57600000)), + (22, 0x3a908e, e(0x57700000)), + (22, 0x3a908f, e(0x57800000)), + (22, 0x3a9090, e(0x57900000)), + (22, 0x3a9091, e(0x57a00000)), + (22, 0x3a9092, e(0x57b00000)), + (22, 0x3a9093, e(0x57c00000)), + (22, 0x3a9094, e(0x57d00000)), + (22, 0x3a9095, e(0x57e00000)), + (22, 0x3a9096, e(0x57f00000)), + (22, 0x3a9097, e(0x58000000)), + (22, 0x3a9098, e(0x58100000)), + (22, 0x3a9099, e(0x58200000)), + (22, 0x3a909a, e(0x58300000)), + (22, 0x3a909b, e(0x58400000)), + (22, 0x3a909c, e(0x58500000)), + (22, 0x3a909d, e(0x58600000)), + (22, 0x3a909e, e(0x58700000)), + (22, 0x3a909f, e(0x58800000)), + (22, 0x3a90a0, e(0x58900000)), + (22, 0x3a90a1, e(0x58a00000)), + (22, 0x3a90a2, e(0x58b00000)), + (22, 0x3a90a3, e(0x58c00000)), + (22, 0x3a90a4, e(0x58d00000)), + (22, 0x3a90a5, e(0x58e00000)), + (22, 0x3a90a6, e(0x58f00000)), + (22, 0x3a90a7, e(0x59000000)), + (22, 0x3a90a8, e(0x59100000)), + (22, 0x3a90a9, e(0x59200000)), + (22, 0x3a90aa, e(0x59300000)), + (22, 0x3a90ab, e(0x59400000)), + (22, 0x3a90ac, e(0x59500000)), + (22, 0x3a90ad, e(0x59600000)), + (22, 0x3a90ae, e(0x59700000)), + (22, 0x3a90af, e(0x59800000)), + (22, 0x3a90b0, e(0x59900000)), + (22, 0x3a90b1, e(0x59a00000)), + (22, 0x3a90b2, e(0x59b00000)), + (22, 0x3a90b3, e(0x59c00000)), + (22, 0x3a90b4, e(0x59d00000)), + (22, 0x3a90b5, e(0x59e00000)), + (22, 0x3a90b6, e(0x59f00000)), + (22, 0x3a90b7, e(0x5a000000)), + (22, 0x3a90b8, e(0x5a100000)), + (22, 0x3a90b9, e(0x5a200000)), + (22, 0x3a90ba, e(0x5a300000)), + (22, 0x3a90bb, e(0x5a400000)), + (22, 0x3a90bc, e(0x5a500000)), + (22, 0x3a90bd, e(0x5a600000)), + (22, 0x3a90be, e(0x5a700000)), + (22, 0x3a90bf, e(0x5a800000)), + (22, 0x3a90c0, e(0x5a900000)), + (22, 0x3a90c1, e(0x5aa00000)), + (22, 0x3a90c2, e(0x5ab00000)), + (22, 0x3a90c3, e(0x5ac00000)), + (22, 0x3a90c4, e(0x5ad00000)), + (22, 0x3a90c5, e(0x5ae00000)), + (22, 0x3a90c6, e(0x5af00000)), + (22, 0x3a90c7, e(0x5b000000)), + (22, 0x3a90c8, e(0x5b100000)), + (22, 0x3a90c9, e(0x5b200000)), + (22, 0x3a90ca, e(0x5b300000)), + (22, 0x3a90cb, e(0x5b400000)), + (22, 0x3a90cc, e(0x5b500000)), + (22, 0x3a90cd, e(0x5b600000)), + (22, 0x3a90ce, e(0x5b700000)), + (22, 0x3a90cf, e(0x5b800000)), + (22, 0x3a90d0, e(0x5b900000)), + (22, 0x3a90d1, e(0x5ba00000)), + (22, 0x3a90d2, e(0x5bb00000)), + (22, 0x3a90d3, e(0x5bc00000)), + (22, 0x3a90d4, e(0x5bd00000)), + (22, 0x3a90d5, e(0x5be00000)), + (22, 0x3a90d6, e(0x5bf00000)), + (22, 0x3a90d7, e(0x5c000000)), + (22, 0x3a90d8, e(0x5c100000)), + (22, 0x3a90d9, e(0x5c200000)), + (22, 0x3a90da, e(0x5c300000)), + (22, 0x3a90db, e(0x5c400000)), + (22, 0x3a90dc, e(0x5c500000)), + (22, 0x3a90dd, e(0x5c600000)), + (22, 0x3a90de, e(0x5c700000)), + (22, 0x3a90df, e(0x5c800000)), + (22, 0x3a90e0, e(0x5c900000)), + (22, 0x3a90e1, e(0x5ca00000)), + (22, 0x3a90e2, e(0x5cb00000)), + (22, 0x3a90e3, e(0x5cc00000)), + (22, 0x3a90e4, e(0x5cd00000)), + (22, 0x3a90e5, e(0x5ce00000)), + (22, 0x3a90e6, e(0x5cf00000)), + (22, 0x3a90e7, e(0x5d000000)), + (22, 0x3a90e8, e(0x5d100000)), + (22, 0x3a90e9, e(0x5d200000)), + (22, 0x3a90ea, e(0x5d300000)), + (22, 0x3a90eb, e(0x5d400000)), + (22, 0x3a90ec, e(0x5d500000)), + (22, 0x3a90ed, e(0x5d600000)), + (22, 0x3a90ee, e(0x5d700000)), + (22, 0x3a90ef, e(0x5d800000)), + (22, 0x3a90f0, e(0x5d900000)), + (22, 0x3a90f1, e(0x5da00000)), + (22, 0x3a90f2, e(0x5db00000)), + (22, 0x3a90f3, e(0x5dc00000)), + (22, 0x3a90f4, e(0x5dd00000)), + (22, 0x3a90f5, e(0x5de00000)), + (22, 0x3a90f6, e(0x5df00000)), + (22, 0x3a90f7, e(0x5e000000)), + (22, 0x3a90f8, e(0x5e100000)), + (22, 0x3a90f9, e(0x5e200000)), + (22, 0x3a90fa, e(0x5e300000)), + (22, 0x3a90fb, e(0x5e400000)), + (22, 0x3a90fc, e(0x5e500000)), + (22, 0x3a90fd, e(0x5e600000)), + (22, 0x3a90fe, e(0x5e700000)), + (22, 0x3a90ff, e(0x5e800000)), + (22, 0x3a9100, e(0x5e900000)), + (22, 0x3a9101, e(0x5ea00000)), + (22, 0x3a9102, e(0x5eb00000)), + (22, 0x3a9103, e(0x5ec00000)), + (22, 0x3a9104, e(0x5ed00000)), + (22, 0x3a9105, e(0x5ee00000)), + (22, 0x3a9106, e(0x5ef00000)), + (22, 0x3a9107, e(0x5f000000)), + (22, 0x3a9108, e(0x5f100000)), + (22, 0x3a9109, e(0x5f200000)), + (22, 0x3a910a, e(0x5f300000)), + (22, 0x3a910b, e(0x5f400000)), + (22, 0x3a910c, e(0x5f500000)), + (22, 0x3a910d, e(0x5f600000)), + (22, 0x3a910e, e(0x5f700000)), + (22, 0x3a910f, e(0x5f800000)), + (22, 0x3a9110, e(0x5f900000)), + (22, 0x3a9111, e(0x5fa00000)), + (22, 0x3a9112, e(0x5fb00000)), + (22, 0x3a9113, e(0x5fc00000)), + (22, 0x3a9114, e(0x5fd00000)), + (22, 0x3a9115, e(0x5fe00000)), + (22, 0x3a9116, e(0x5ff00000)), + (22, 0x3a9117, e(0x60000000)), + (22, 0x3a9118, e(0x60100000)), + (22, 0x3a9119, e(0x60200000)), + (22, 0x3a911a, e(0x60300000)), + (22, 0x3a911b, e(0x60400000)), + (22, 0x3a911c, e(0x60500000)), + (22, 0x3a911d, e(0x60600000)), + (22, 0x3a911e, e(0x60700000)), + (22, 0x3a911f, e(0x60800000)), + (22, 0x3a9120, e(0x60900000)), + (22, 0x3a9121, e(0x60a00000)), + (22, 0x3a9122, e(0x60b00000)), + (22, 0x3a9123, e(0x60c00000)), + (22, 0x3a9124, e(0x60d00000)), + (22, 0x3a9125, e(0x60e00000)), + (22, 0x3a9126, e(0x60f00000)), + (22, 0x3a9127, e(0x61000000)), + (22, 0x3a9128, e(0x61100000)), + (22, 0x3a9129, e(0x61200000)), + (22, 0x3a912a, e(0x61300000)), + (22, 0x3a912b, e(0x61400000)), + (22, 0x3a912c, e(0x61500000)), + (22, 0x3a912d, e(0x61600000)), + (22, 0x3a912e, e(0x61700000)), + (22, 0x3a912f, e(0x61800000)), + (22, 0x3a9130, e(0x61900000)), + (22, 0x3a9131, e(0x61a00000)), + (22, 0x3a9132, e(0x61b00000)), + (22, 0x3a9133, e(0x61c00000)), + (22, 0x3a9134, e(0x61d00000)), + (22, 0x3a9135, e(0x61e00000)), + (22, 0x3a9136, e(0x61f00000)), + (22, 0x3a9137, e(0x62000000)), + (22, 0x3a9138, e(0x62100000)), + (22, 0x3a9139, e(0x62200000)), + (22, 0x3a913a, e(0x62300000)), + (22, 0x3a913b, e(0x62400000)), + (22, 0x3a913c, e(0x62500000)), + (22, 0x3a913d, e(0x62600000)), + (22, 0x3a913e, e(0x62700000)), + (22, 0x3a913f, e(0x62800000)), + (22, 0x3a9140, e(0x62900000)), + (22, 0x3a9141, e(0x62a00000)), + (22, 0x3a9142, e(0x62b00000)), + (22, 0x3a9143, e(0x62c00000)), + (22, 0x3a9144, e(0x62d00000)), + (22, 0x3a9145, e(0x62e00000)), + (22, 0x3a9146, e(0x62f00000)), + (22, 0x3a9147, e(0x63000000)), + (22, 0x3a9148, e(0x63100000)), + (22, 0x3a9149, e(0x63200000)), + (22, 0x3a914a, e(0x63300000)), + (22, 0x3a914b, e(0x63400000)), + (22, 0x3a914c, e(0x63500000)), + (22, 0x3a914d, e(0x63600000)), + (22, 0x3a914e, e(0x63700000)), + (22, 0x3a914f, e(0x63800000)), + (22, 0x3a9150, e(0x63900000)), + (22, 0x3a9151, e(0x63a00000)), + (22, 0x3a9152, e(0x63b00000)), + (22, 0x3a9153, e(0x63c00000)), + (22, 0x3a9154, e(0x63d00000)), + (22, 0x3a9155, e(0x63e00000)), + (22, 0x3a9156, e(0x63f00000)), + (22, 0x3a9157, e(0x64000000)), + (22, 0x3a9158, e(0x64100000)), + (22, 0x3a9159, e(0x64200000)), + (22, 0x3a915a, e(0x64300000)), + (22, 0x3a915b, e(0x64400000)), + (22, 0x3a915c, e(0x64500000)), + (22, 0x3a915d, e(0x64600000)), + (22, 0x3a915e, e(0x64700000)), + (22, 0x3a915f, e(0x64800000)), + (22, 0x3a9160, e(0x64900000)), + (22, 0x3a9161, e(0x64a00000)), + (22, 0x3a9162, e(0x64b00000)), + (22, 0x3a9163, e(0x64c00000)), + (22, 0x3a9164, e(0x64d00000)), + (22, 0x3a9165, e(0x64e00000)), + (22, 0x3a9166, e(0x64f00000)), + (22, 0x3a9167, e(0x65000000)), + (22, 0x3a9168, e(0x65100000)), + (22, 0x3a9169, e(0x65200000)), + (22, 0x3a916a, e(0x65300000)), + (22, 0x3a916b, e(0x65400000)), + (22, 0x3a916c, e(0x65500000)), + (22, 0x3a916d, e(0x65600000)), + (22, 0x3a916e, e(0x65700000)), + (22, 0x3a916f, e(0x65800000)), + (22, 0x3a9170, e(0x65900000)), + (22, 0x3a9171, e(0x65a00000)), + (22, 0x3a9172, e(0x65b00000)), + (22, 0x3a9173, e(0x65c00000)), + (22, 0x3a9174, e(0x65d00000)), + (22, 0x3a9175, e(0x65e00000)), + (22, 0x3a9176, e(0x65f00000)), + (22, 0x3a9177, e(0x66000000)), + (22, 0x3a9178, e(0x66100000)), + (22, 0x3a9179, e(0x66200000)), + (22, 0x3a917a, e(0x66300000)), + (22, 0x3a917b, e(0x66400000)), + (22, 0x3a917c, e(0x66500000)), + (22, 0x3a917d, e(0x66600000)), + (22, 0x3a917e, e(0x66700000)), + (22, 0x3a917f, e(0x66800000)), + (22, 0x3a9180, e(0x66900000)), + (22, 0x3a9181, e(0x66a00000)), + (22, 0x3a9182, e(0x66b00000)), + (22, 0x3a9183, e(0x66c00000)), + (22, 0x3a9184, e(0x66d00000)), + (22, 0x3a9185, e(0x66e00000)), + (22, 0x3a9186, e(0x66f00000)), + (22, 0x3a9187, e(0x67000000)), + (22, 0x3a9188, e(0x67100000)), + (22, 0x3a9189, e(0x67200000)), + (22, 0x3a918a, e(0x67300000)), + (22, 0x3a918b, e(0x67400000)), + (22, 0x3a918c, e(0x67500000)), + (22, 0x3a918d, e(0x67600000)), + (22, 0x3a918e, e(0x67700000)), + (22, 0x3a918f, e(0x67800000)), + (22, 0x3a9190, e(0x67900000)), + (22, 0x3a9191, e(0x67a00000)), + (22, 0x3a9192, e(0x67b00000)), + (22, 0x3a9193, e(0x67c00000)), + (22, 0x3a9194, e(0x67d00000)), + (22, 0x3a9195, e(0x67e00000)), + (22, 0x3a9196, e(0x67f00000)), + (22, 0x3a9197, e(0x68000000)), + (22, 0x3a9198, e(0x68100000)), + (22, 0x3a9199, e(0x68200000)), + (22, 0x3a919a, e(0x68300000)), + (22, 0x3a919b, e(0x68400000)), + (22, 0x3a919c, e(0x68500000)), + (22, 0x3a919d, e(0x68600000)), + (22, 0x3a919e, e(0x68700000)), + (22, 0x3a919f, e(0x68800000)), + (22, 0x3a91a0, e(0x68900000)), + (22, 0x3a91a1, e(0x68a00000)), + (22, 0x3a91a2, e(0x68b00000)), + (22, 0x3a91a3, e(0x68c00000)), + (22, 0x3a91a4, e(0x68d00000)), + (22, 0x3a91a5, e(0x68e00000)), + (22, 0x3a91a6, e(0x68f00000)), + (22, 0x3a91a7, e(0x69000000)), + (22, 0x3a91a8, e(0x69100000)), + (22, 0x3a91a9, e(0x69200000)), + (22, 0x3a91aa, e(0x69300000)), + (22, 0x3a91ab, e(0x69400000)), + (22, 0x3a91ac, e(0x69500000)), + (22, 0x3a91ad, e(0x69600000)), + (22, 0x3a91ae, e(0x69700000)), + (22, 0x3a91af, e(0x69800000)), + (22, 0x3a91b0, e(0x69900000)), + (22, 0x3a91b1, e(0x69a00000)), + (22, 0x3a91b2, e(0x69b00000)), + (22, 0x3a91b3, e(0x69c00000)), + (22, 0x3a91b4, e(0x69d00000)), + (22, 0x3a91b5, e(0x69e00000)), + (22, 0x3a91b6, e(0x69f00000)), + (22, 0x3a91b7, e(0x6a000000)), + (22, 0x3a91b8, e(0x6a100000)), + (22, 0x3a91b9, e(0x6a200000)), + (22, 0x3a91ba, e(0x6a300000)), + (22, 0x3a91bb, e(0x6a400000)), + (22, 0x3a91bc, e(0x6a500000)), + (22, 0x3a91bd, e(0x6a600000)), + (22, 0x3a91be, e(0x6a700000)), + (22, 0x3a91bf, e(0x6a800000)), + (22, 0x3a91c0, e(0x6a900000)), + (22, 0x3a91c1, e(0x6aa00000)), + (22, 0x3a91c2, e(0x6ab00000)), + (22, 0x3a91c3, e(0x6ac00000)), + (22, 0x3a91c4, e(0x6ad00000)), + (22, 0x3a91c5, e(0x6ae00000)), + (22, 0x3a91c6, e(0x6af00000)), + (22, 0x3a91c7, e(0x6b000000)), + (22, 0x3a91c8, e(0x6b100000)), + (22, 0x3a91c9, e(0x6b200000)), + (22, 0x3a91ca, e(0x6b300000)), + (22, 0x3a91cb, e(0x6b400000)), + (22, 0x3a91cc, e(0x6b500000)), + (22, 0x3a91cd, e(0x6b600000)), + (22, 0x3a91ce, e(0x6b700000)), + (22, 0x3a91cf, e(0x6b800000)), + (22, 0x3a91d0, e(0x6b900000)), + (22, 0x3a91d1, e(0x6ba00000)), + (22, 0x3a91d2, e(0x6bb00000)), + (22, 0x3a91d3, e(0x6bc00000)), + (22, 0x3a91d4, e(0x6bd00000)), + (22, 0x3a91d5, e(0x6be00000)), + (22, 0x3a91d6, e(0x6bf00000)), + (22, 0x3a91d7, e(0x6c000000)), + (22, 0x3a91d8, e(0x6c100000)), + (22, 0x3a91d9, e(0x6c200000)), + (22, 0x3a91da, e(0x6c300000)), + (22, 0x3a91db, e(0x6c400000)), + (22, 0x3a91dc, e(0x6c500000)), + (22, 0x3a91dd, e(0x6c600000)), + (22, 0x3a91de, e(0x6c700000)), + (22, 0x3a91df, e(0x6c800000)), + (22, 0x3a91e0, e(0x6c900000)), + (22, 0x3a91e1, e(0x6ca00000)), + (22, 0x3a91e2, e(0x6cb00000)), + (22, 0x3a91e3, e(0x6cc00000)), + (22, 0x3a91e4, e(0x6cd00000)), + (22, 0x3a91e5, e(0x6ce00000)), + (22, 0x3a91e6, e(0x6cf00000)), + (22, 0x3a91e7, e(0x6d000000)), + (22, 0x3a91e8, e(0x6d100000)), + (22, 0x3a91e9, e(0x6d200000)), + (22, 0x3a91ea, e(0x6d300000)), + (22, 0x3a91eb, e(0x6d400000)), + (22, 0x3a91ec, e(0x6d500000)), + (22, 0x3a91ed, e(0x6d600000)), + (22, 0x3a91ee, e(0x6d700000)), + (22, 0x3a91ef, e(0x6d800000)), + (22, 0x3a91f0, e(0x6d900000)), + (22, 0x3a91f1, e(0x6da00000)), + (22, 0x3a91f2, e(0x6db00000)), + (22, 0x3a91f3, e(0x6dc00000)), + (22, 0x3a91f4, e(0x6dd00000)), + (22, 0x3a91f5, e(0x6de00000)), + (22, 0x3a91f6, e(0x6df00000)), + (22, 0x3a91f7, e(0x6e000000)), + (22, 0x3a91f8, e(0x6e100000)), + (22, 0x3a91f9, e(0x6e200000)), + (22, 0x3a91fa, e(0x6e300000)), + (22, 0x3a91fb, e(0x6e400000)), + (22, 0x3a91fc, e(0x6e500000)), + (22, 0x3a91fd, e(0x6e600000)), + (22, 0x3a91fe, e(0x6e700000)), + (22, 0x3a91ff, e(0x6e800000)), + (22, 0x3a9200, e(0x6e900000)), + (22, 0x3a9201, e(0x6ea00000)), + (22, 0x3a9202, e(0x6eb00000)), + (22, 0x3a9203, e(0x6ec00000)), + (22, 0x3a9204, e(0x6ed00000)), + (22, 0x3a9205, e(0x6ee00000)), + (22, 0x3a9206, e(0x6ef00000)), + (22, 0x3a9207, e(0x6f000000)), + (22, 0x3a9208, e(0x6f100000)), + (22, 0x3a9209, e(0x6f200000)), + (22, 0x3a920a, e(0x6f300000)), + (22, 0x3a920b, e(0x6f400000)), + (22, 0x3a920c, e(0x6f500000)), + (22, 0x3a920d, e(0x6f600000)), + (22, 0x3a920e, e(0x6f700000)), + (22, 0x3a920f, e(0x6f800000)), + (22, 0x3a9210, e(0x6f900000)), + (22, 0x3a9211, e(0x6fa00000)), + (22, 0x3a9212, e(0x6fb00000)), + (22, 0x3a9213, e(0x6fc00000)), + (22, 0x3a9214, e(0x6fd00000)), + (22, 0x3a9215, e(0x6fe00000)), + (22, 0x3a9216, e(0x6ff00000)), + (22, 0x3a9217, e(0x70000000)), + (22, 0x3a9218, e(0x70100000)), + (22, 0x3a9219, e(0x70200000)), + (22, 0x3a921a, e(0x70300000)), + (22, 0x3a921b, e(0x70400000)), + (22, 0x3a921c, e(0x70500000)), + (22, 0x3a921d, e(0x70600000)), + (22, 0x3a921e, e(0x70700000)), + (22, 0x3a921f, e(0x70800000)), + (22, 0x3a9220, e(0x70900000)), + (22, 0x3a9221, e(0x70a00000)), + (22, 0x3a9222, e(0x70b00000)), + (22, 0x3a9223, e(0x70c00000)), + (22, 0x3a9224, e(0x70d00000)), + (22, 0x3a9225, e(0x70e00000)), + (22, 0x3a9226, e(0x70f00000)), + (22, 0x3a9227, e(0x71000000)), + (22, 0x3a9228, e(0x71100000)), + (22, 0x3a9229, e(0x71200000)), + (22, 0x3a922a, e(0x71300000)), + (22, 0x3a922b, e(0x71400000)), + (22, 0x3a922c, e(0x71500000)), + (22, 0x3a922d, e(0x71600000)), + (22, 0x3a922e, e(0x71700000)), + (22, 0x3a922f, e(0x71800000)), + (22, 0x3a9230, e(0x71900000)), + (22, 0x3a9231, e(0x71a00000)), + (22, 0x3a9232, e(0x71b00000)), + (22, 0x3a9233, e(0x71c00000)), + (22, 0x3a9234, e(0x71d00000)), + (22, 0x3a9235, e(0x71e00000)), + (22, 0x3a9236, e(0x71f00000)), + (22, 0x3a9237, e(0x72000000)), + (22, 0x3a9238, e(0x72100000)), + (22, 0x3a9239, e(0x72200000)), + (22, 0x3a923a, e(0x72300000)), + (22, 0x3a923b, e(0x72400000)), + (22, 0x3a923c, e(0x72500000)), + (22, 0x3a923d, e(0x72600000)), + (22, 0x3a923e, e(0x72700000)), + (22, 0x3a923f, e(0x72800000)), + (22, 0x3a9240, e(0x72900000)), + (22, 0x3a9241, e(0x72a00000)), + (22, 0x3a9242, e(0x72b00000)), + (22, 0x3a9243, e(0x72c00000)), + (22, 0x3a9244, e(0x72d00000)), + (22, 0x3a9245, e(0x72e00000)), + (22, 0x3a9246, e(0x72f00000)), + (22, 0x3a9247, e(0x73000000)), + (22, 0x3a9248, e(0x73100000)), + (22, 0x3a9249, e(0x73200000)), + (22, 0x3a924a, e(0x73300000)), + (22, 0x3a924b, e(0x73400000)), + (22, 0x3a924c, e(0x73500000)), + (22, 0x3a924d, e(0x73600000)), + (22, 0x3a924e, e(0x73700000)), + (22, 0x3a924f, e(0x73800000)), + (22, 0x3a9250, e(0x73900000)), + (22, 0x3a9251, e(0x73a00000)), + (22, 0x3a9252, e(0x73b00000)), + (22, 0x3a9253, e(0x73c00000)), + (22, 0x3a9254, e(0x73d00000)), + (22, 0x3a9255, e(0x73e00000)), + (22, 0x3a9256, e(0x73f00000)), + (22, 0x3a9257, e(0x74000000)), + (22, 0x3a9258, e(0x74100000)), + (22, 0x3a9259, e(0x74200000)), + (22, 0x3a925a, e(0x74300000)), + (22, 0x3a925b, e(0x74400000)), + (22, 0x3a925c, e(0x74500000)), + (22, 0x3a925d, e(0x74600000)), + (22, 0x3a925e, e(0x74700000)), + (22, 0x3a925f, e(0x74800000)), + (22, 0x3a9260, e(0x74900000)), + (22, 0x3a9261, e(0x74a00000)), + (22, 0x3a9262, e(0x74b00000)), + (22, 0x3a9263, e(0x74c00000)), + (22, 0x3a9264, e(0x74d00000)), + (22, 0x3a9265, e(0x74e00000)), + (22, 0x3a9266, e(0x74f00000)), + (22, 0x3a9267, e(0x75000000)), + (22, 0x3a9268, e(0x75100000)), + (22, 0x3a9269, e(0x75200000)), + (22, 0x3a926a, e(0x75300000)), + (22, 0x3a926b, e(0x75400000)), + (22, 0x3a926c, e(0x75500000)), + (22, 0x3a926d, e(0x75600000)), + (22, 0x3a926e, e(0x75700000)), + (22, 0x3a926f, e(0x75800000)), + (22, 0x3a9270, e(0x75900000)), + (22, 0x3a9271, e(0x75a00000)), + (22, 0x3a9272, e(0x75b00000)), + (22, 0x3a9273, e(0x75c00000)), + (22, 0x3a9274, e(0x75d00000)), + (22, 0x3a9275, e(0x75e00000)), + (22, 0x3a9276, e(0x75f00000)), + (22, 0x3a9277, e(0x76000000)), + (22, 0x3a9278, e(0x76100000)), + (22, 0x3a9279, e(0x76200000)), + (22, 0x3a927a, e(0x76300000)), + (22, 0x3a927b, e(0x76400000)), + (22, 0x3a927c, e(0x76500000)), + (22, 0x3a927d, e(0x76600000)), + (22, 0x3a927e, e(0x76700000)), + (22, 0x3a927f, e(0x76800000)), + (22, 0x3a9280, e(0x76900000)), + (22, 0x3a9281, e(0x76a00000)), + (22, 0x3a9282, e(0x76b00000)), + (22, 0x3a9283, e(0x76c00000)), + (22, 0x3a9284, e(0x76d00000)), + (22, 0x3a9285, e(0x76e00000)), + (22, 0x3a9286, e(0x76f00000)), + (22, 0x3a9287, e(0x77000000)), + (22, 0x3a9288, e(0x77100000)), + (22, 0x3a9289, e(0x77200000)), + (22, 0x3a928a, e(0x77300000)), + (22, 0x3a928b, e(0x77400000)), + (22, 0x3a928c, e(0x77500000)), + (22, 0x3a928d, e(0x77600000)), + (22, 0x3a928e, e(0x77700000)), + (22, 0x3a928f, e(0x77800000)), + (22, 0x3a9290, e(0x77900000)), + (22, 0x3a9291, e(0x77a00000)), + (22, 0x3a9292, e(0x77b00000)), + (22, 0x3a9293, e(0x77c00000)), + (22, 0x3a9294, e(0x77d00000)), + (22, 0x3a9295, e(0x77e00000)), + (22, 0x3a9296, e(0x77f00000)), + (22, 0x3a9297, e(0x78000000)), + (22, 0x3a9298, e(0x78100000)), + (22, 0x3a9299, e(0x78200000)), + (22, 0x3a929a, e(0x78300000)), + (22, 0x3a929b, e(0x78400000)), + (22, 0x3a929c, e(0x78500000)), + (22, 0x3a929d, e(0x78600000)), + (22, 0x3a929e, e(0x78700000)), + (22, 0x3a929f, e(0x78800000)), + (22, 0x3a92a0, e(0x78900000)), + (22, 0x3a92a1, e(0x78a00000)), + (22, 0x3a92a2, e(0x78b00000)), + (22, 0x3a92a3, e(0x78c00000)), + (22, 0x3a92a4, e(0x78d00000)), + (22, 0x3a92a5, e(0x78e00000)), + (22, 0x3a92a6, e(0x78f00000)), + (22, 0x3a92a7, e(0x79000000)), + (22, 0x3a92a8, e(0x79100000)), + (22, 0x3a92a9, e(0x79200000)), + (22, 0x3a92aa, e(0x79300000)), + (22, 0x3a92ab, e(0x79400000)), + (22, 0x3a92ac, e(0x79500000)), + (22, 0x3a92ad, e(0x79600000)), + (22, 0x3a92ae, e(0x79700000)), + (22, 0x3a92af, e(0x79800000)), + (22, 0x3a92b0, e(0x79900000)), + (22, 0x3a92b1, e(0x79a00000)), + (22, 0x3a92b2, e(0x79b00000)), + (22, 0x3a92b3, e(0x79c00000)), + (22, 0x3a92b4, e(0x79d00000)), + (22, 0x3a92b5, e(0x79e00000)), + (22, 0x3a92b6, e(0x79f00000)), + (22, 0x3a92b7, e(0x7a000000)), + (22, 0x3a92b8, e(0x7a100000)), + (22, 0x3a92b9, e(0x7a200000)), + (22, 0x3a92ba, e(0x7a300000)), + (22, 0x3a92bb, e(0x7a400000)), + (22, 0x3a92bc, e(0x7a500000)), + (22, 0x3a92bd, e(0x7a600000)), + (22, 0x3a92be, e(0x7a700000)), + (22, 0x3a92bf, e(0x7a800000)), + (22, 0x3a92c0, e(0x7a900000)), + (22, 0x3a92c1, e(0x7aa00000)), + (22, 0x3a92c2, e(0x7ab00000)), + (22, 0x3a92c3, e(0x7ac00000)), + (22, 0x3a92c4, e(0x7ad00000)), + (22, 0x3a92c5, e(0x7ae00000)), + (22, 0x3a92c6, e(0x7af00000)), + (22, 0x3a92c7, e(0x7b000000)), + (22, 0x3a92c8, e(0x7b100000)), + (22, 0x3a92c9, e(0x7b200000)), + (22, 0x3a92ca, e(0x7b300000)), + (22, 0x3a92cb, e(0x7b400000)), + (22, 0x3a92cc, e(0x7b500000)), + (22, 0x3a92cd, e(0x7b600000)), + (22, 0x3a92ce, e(0x7b700000)), + (22, 0x3a92cf, e(0x7b800000)), + (22, 0x3a92d0, e(0x7b900000)), + (22, 0x3a92d1, e(0x7ba00000)), + (22, 0x3a92d2, e(0x7bb00000)), + (22, 0x3a92d3, e(0x7bc00000)), + (22, 0x3a92d4, e(0x7bd00000)), + (22, 0x3a92d5, e(0x7be00000)), + (22, 0x3a92d6, e(0x7bf00000)), + (22, 0x3a92d7, e(0x7c000000)), + (22, 0x3a92d8, e(0x7c100000)), + (22, 0x3a92d9, e(0x7c200000)), + (22, 0x3a92da, e(0x7c300000)), + (22, 0x3a92db, e(0x7c400000)), + (22, 0x3a92dc, e(0x7c500000)), + (22, 0x3a92dd, e(0x7c600000)), + (22, 0x3a92de, e(0x7c700000)), + (22, 0x3a92df, e(0x7c800000)), + (22, 0x3a92e0, e(0x7c900000)), + (22, 0x3a92e1, e(0x7ca00000)), + (22, 0x3a92e2, e(0x7cb00000)), + (22, 0x3a92e3, e(0x7cc00000)), + (22, 0x3a92e4, e(0x7cd00000)), + (22, 0x3a92e5, e(0x7ce00000)), + (22, 0x3a92e6, e(0x7cf00000)), + (22, 0x3a92e7, e(0x7d000000)), + (22, 0x3a92e8, e(0x7d100000)), + (22, 0x3a92e9, e(0x7d200000)), + (22, 0x3a92ea, e(0x7d300000)), + (22, 0x3a92eb, e(0x7d400000)), + (22, 0x3a92ec, e(0x7d500000)), + (22, 0x3a92ed, e(0x7d600000)), + (22, 0x3a92ee, e(0x7d700000)), + (22, 0x3a92ef, e(0x7d800000)), + (22, 0x3a92f0, e(0x7d900000)), + (22, 0x3a92f1, e(0x7da00000)), + (22, 0x3a92f2, e(0x7db00000)), + (22, 0x3a92f3, e(0x7dc00000)), + (22, 0x3a92f4, e(0x7dd00000)), + (22, 0x3a92f5, e(0x7de00000)), + (22, 0x3a92f6, e(0x7df00000)), + (22, 0x3a92f7, e(0x7e000000)), + (22, 0x3a92f8, e(0x7e100000)), + (22, 0x3a92f9, e(0x7e200000)), + (22, 0x3a92fa, e(0x7e300000)), + (22, 0x3a92fb, e(0x7e400000)), + (22, 0x3a92fc, e(0x7e500000)), + (22, 0x3a92fd, e(0x7e600000)), + (22, 0x3a92fe, e(0x7e700000)), + (22, 0x3a92ff, e(0x7e800000)), + (22, 0x3a95a0, e(0x7e900000)), + (22, 0x3a95a1, e(0x7ea00000)), + (22, 0x3a95a2, e(0x7eb00000)), + (22, 0x3a95a3, e(0x7ec00000)), + (22, 0x3a95a4, e(0x7ed00000)), + (22, 0x3a95a5, e(0x7ee00000)), + (22, 0x3a95a6, e(0x7ef00000)), + (22, 0x3a95a7, e(0x7f000000)), + (22, 0x3a95a8, e(0x7f100000)), + (22, 0x3a95a9, e(0x7f200000)), + (22, 0x3a95aa, e(0x7f300000)), + (22, 0x3a95ab, e(0x7f400000)), + (22, 0x3a95ac, e(0x7f500000)), + (22, 0x3a95ad, e(0x7f600000)), + (22, 0x3a95ae, e(0x7f700000)), + (22, 0x3a95af, e(0x7f800000)), + (22, 0x3b8fe0, e(0x7f900000)), + (22, 0x3b8fe1, e(0x7fa00000)), + (22, 0x3b8fe2, e(0x7fb00000)), + (22, 0x3b8fe3, e(0x7fc00000)), + (21, 0x68e90, e(0x7fd00000)), + (21, 0x68e91, e(0x7fe00000)), + (21, 0x68e98, e(0x7ff80000)), +]; diff --git a/crates/prc/src/arrays.rs b/crates/prc/src/arrays.rs new file mode 100644 index 0000000000..de13bde6eb --- /dev/null +++ b/crates/prc/src/arrays.rs @@ -0,0 +1,420 @@ +//! The array primitives of compressed tessellation [WD 11.6-11.9] and the Huffman payload +//! they may carry [WD 12.2]. +//! +//! The WD gives only the Huffman writer. The reader decodes what producers write: each +//! leaf's code length is read in the width the payload's 8-bit field states (the WD writes a +//! fixed 8 bits), stored codes carry a leading root bit, and the payload is read LSB-first +//! although the stream around it is MSB-first. + +use std::collections::HashMap; + +use crate::PrcError; +use crate::bits::BitReader; + +/// Longest Huffman code [WD 12.2]. +const MAX_CODE_BITS: u32 = 32; + +/// Widest Huffman symbol a caller asks for: the arrays hold 2- to 16-bit values. +const MAX_SYMBOL_BITS: u32 = 16; + +fn malformed(what: &str) -> PrcError { + PrcError::Malformed(what.to_owned()) +} + +/// An LSB-first cursor over a Huffman payload. +struct Lsb<'a> { + data: &'a [u8], + pos: usize, + end: usize, +} + +impl Lsb<'_> { + fn bit(&mut self) -> Result { + if self.pos >= self.end { + return Err(PrcError::Truncated("a Huffman payload")); + } + let byte = self.data.get(self.pos / 8).ok_or(PrcError::Truncated("a Huffman payload"))?; + let b = u32::from((byte >> (self.pos % 8)) & 1); + self.pos += 1; + Ok(b) + } + + /// `n` ≤ 32 bits, the first read being the least significant. + fn bits(&mut self, n: u32) -> Result { + let mut v = 0u32; + for i in 0..n.min(32) { + v |= self.bit()? << i; + } + Ok(v) + } + + fn remaining(&self) -> usize { + self.end.saturating_sub(self.pos) + } +} + +/// The low `n` (1..=16) bits of `v` as a symbol, two's complement when `signed`. +fn symbol(v: u32, n: u32, signed: bool) -> i32 { + let v = i64::from(v); + let wide = if signed && n > 0 && v & (1 << (n - 1)) != 0 { v - (1i64 << n) } else { v }; + // n <= 16, so `wide` is within ±2^16. + i32::try_from(wide).unwrap_or(0) +} + +/// The Huffman container of a compressed Character/Short array [WD 11.6, 11.7, 12.2]: +/// symbols are `nbits` wide, sign-extended when `signed`. +fn huffman(r: &mut BitReader<'_>, nbits: u32, signed: bool) -> Result, PrcError> { + if nbits == 0 || nbits > MAX_SYMBOL_BITS { + return Err(malformed("Huffman symbol width")); + } + let words = r.unsigned_integer()? as usize; + if words == 0 { + return Ok(Vec::new()); + } + if words.saturating_mul(32) > r.remaining() { + return Err(PrcError::Truncated("a Huffman payload")); + } + // words * 32 bits remain, so words * 4 bytes cannot overflow. + let blob = (0..words * 4).map(|_| r.character()).collect::, _>>()?; + let last = r.unsigned_integer()?; + if last > 32 { + return Err(malformed("Huffman last-word bit count above 32")); + } + let end = (words - 1) * 32 + last as usize; + let mut b = Lsb { data: &blob, pos: 0, end }; + let codes = huffman_leaves(&mut b, nbits, signed)?; + if codes.is_empty() { + return Ok(Vec::new()); + } + let count = b.bits(32)? as usize; + // Every element is at least one bit. + if count > b.remaining() { + return Err(PrcError::Truncated("a Huffman element count")); + } + r.take_elements(count)?; + let mut out = Vec::with_capacity(count.min(1024)); + for _ in 0..count { + out.push(huffman_symbol(&mut b, &codes)?); + } + Ok(out) +} + +/// The leaf table: `(code length, stored code)` to symbol. +fn huffman_leaves(b: &mut Lsb<'_>, nbits: u32, signed: bool) -> Result, PrcError> { + let leaves = b.bits(nbits + 1)?; + if leaves == 0 { + return Ok(HashMap::new()); + } + // prc-rs: the code-length width is read from the payload, not fixed at 8 bits. + let len_width = b.bits(8)?; + if len_width == 0 || len_width > MAX_CODE_BITS { + return Err(malformed("Huffman code-length width")); + } + // Every leaf is at least nbits + 2 bits, so the payload bounds the table. + if (leaves as usize).saturating_mul((nbits + 2) as usize) > b.remaining() { + return Err(PrcError::Truncated("a Huffman payload")); + } + let mut codes = HashMap::new(); + for _ in 0..leaves { + let s = symbol(b.bits(nbits)?, nbits, signed); + let len = b.bits(len_width)?; + if len == 0 || len > MAX_CODE_BITS { + return Err(malformed("Huffman code length")); + } + let code = b.bits(len)?; + if codes.insert((len, code), s).is_some() { + return Err(malformed("duplicate Huffman code")); + } + } + // A leaf whose code begins another's would end that code early, leaving its unread bits + // in front of the next element. + let shadowed = codes.keys().any(|&(len, code)| (1..len).any(|l| codes.contains_key(&(l, code >> (len - l))))); + if shadowed { + return Err(malformed("a Huffman code that is a prefix of another")); + } + Ok(codes) +} + +/// One element: code bits, most significant first, until a leaf matches. +fn huffman_symbol(b: &mut Lsb<'_>, codes: &HashMap<(u32, u32), i32>) -> Result { + // prc-rs: the stored code's first bit read is its most significant (the root bit). + let (mut code, mut len) = (0u32, 0u32); + while len < MAX_CODE_BITS { + code = (code << 1) | b.bit()?; + len += 1; + if let Some(&s) = codes.get(&(len, code)) { + return Ok(s); + } + } + Err(malformed("a Huffman code matching no leaf")) +} + +/// A `CharacterArray(nbits)` [WD 11.6]. `compressed` is `None` when the flag is written, +/// else the caller's implicit value. Uncompressed elements are whole bytes, sign-extended +/// from 8 bits when `signed`. +pub(crate) fn character_array(r: &mut BitReader<'_>, nbits: u32, compressed: Option, signed: bool) -> Result, PrcError> { + let compressed = match compressed { + Some(c) => c, + None => r.bit()?, + }; + if compressed { + return huffman(r, nbits, signed); + } + let n = r.unsigned_integer()? as usize; + if n.saturating_mul(8) > r.remaining() { + return Err(PrcError::Truncated("a character array")); + } + r.take_elements(n)?; + (0..n) + .map(|_| { + let c = r.character()?; + Ok(if signed { i32::from(c.cast_signed()) } else { i32::from(c) }) + }) + .collect() +} + +/// A `ShortArray(nbits)` [WD 11.7]: uncompressed elements are a low then a high byte, read +/// unsigned. +pub(crate) fn short_array(r: &mut BitReader<'_>, nbits: u32) -> Result, PrcError> { + if r.bit()? { + return huffman(r, nbits, false); + } + let n = r.unsigned_integer()? as usize; + if n.saturating_mul(16) > r.remaining() { + return Err(PrcError::Truncated("a short array")); + } + r.take_elements(n)?; + (0..n) + .map(|_| { + let lo = r.character()?; + let hi = r.character()?; + Ok(i32::from(u16::from_le_bytes([lo, hi]))) + }) + .collect() +} + +/// A `CompressedIntegerArray` [WD 11.8]: a CharacterArray(6) of widths, then one +/// `IntegerWithVariableBitNumber` per width. +pub(crate) fn compressed_integer_array(r: &mut BitReader<'_>) -> Result, PrcError> { + let widths = character_array(r, 6, None, false)?; + widths + .into_iter() + .map(|w| { + let w = u32::try_from(w).map_err(|_| malformed("integer width"))?; + r.integer_vbn(w) + }) + .collect() +} + +/// A `CompressedIndiceArray` [WD 11.9], read as producers write it: the stored signed 6-bit +/// values are width differences summed into each element's width, and each element is a +/// delta from the previous (the WD's own reading desyncs from the third element). Indices +/// are never negative. +pub(crate) fn compressed_indice_array(r: &mut BitReader<'_>, compressed: Option) -> Result, PrcError> { + let diffs = character_array(r, 6, compressed, true)?; + let mut out = Vec::with_capacity(diffs.len()); + let (mut width, mut prev) = (0i64, 0i64); + for d in diffs { + // prc-rs: widths are running sums of the stored differences, values running deltas. + width = width.checked_add(i64::from(d)).ok_or_else(|| malformed("indice width"))?; + let w = u32::try_from(width).map_err(|_| malformed("indice width"))?; + let v = prev.checked_add(r.integer_vbn(w)?).ok_or_else(|| malformed("indice delta"))?; + out.push(u32::try_from(v).map_err(|_| malformed("an index outside 0..2^32"))?); + prev = v; + } + Ok(out) +} + +/// An `UncompressedBoolArray` of `n` raw bits (no length prefix). +pub(crate) fn bool_array(r: &mut BitReader<'_>, n: usize) -> Result, PrcError> { + if n > r.remaining() { + return Err(PrcError::Truncated("a boolean array")); + } + r.take_elements(n)?; + (0..n).map(|_| r.bit()).collect() +} + +#[cfg(test)] +mod tests { + #![allow(clippy::unwrap_used)] + + use super::*; + use crate::testw::W; + + fn read(w: &W, f: impl FnOnce(&mut BitReader<'_>) -> Result) -> T { + let bytes = w.bytes(); + let mut r = BitReader::new(&bytes); + let v = f(&mut r).unwrap(); + assert_eq!(r.position(), w.len(), "bits consumed"); + v + } + + #[test] + fn huffman_decodes_the_wd_example_with_root_bits() { + // WD 12.2's input {1,3,5,7,7,9,5,3,3}; codes carry the root bit. + let leaves = [(3, 0b100, 3), (7, 0b101, 3), (5, 0b110, 3), (1, 0b1110, 4), (9, 0b1111, 4)]; + let values = [1, 3, 5, 7, 7, 9, 5, 3, 3]; + let mut w = W::default(); + w.bit(true).huffman(4, 3, &leaves, &values); + assert_eq!(read(&w, |r| character_array(r, 4, None, false)), values); + } + + #[test] + fn huffman_symbols_sign_extend_only_when_signed() { + let leaves = [(0b111110, 0b10, 2), (1, 0b11, 2)]; + let mut w = W::default(); + w.huffman(6, 2, &leaves, &[62, 1, 62]); + assert_eq!(read(&w, |r| huffman(r, 6, true)), [-2, 1, -2]); + assert_eq!(read(&w, |r| huffman(r, 6, false)), [62, 1, 62]); + } + + #[test] + fn huffman_rejects_codes_that_match_no_leaf_and_empty_payloads_are_empty() { + let mut bad = W::default(); + bad.huffman_raw(2, 2, &[(1, 0b10, 2)], &[(0b11, 2)]); + let bytes = bad.bytes(); + assert!(matches!(huffman(&mut BitReader::new(&bytes), 2, false), Err(PrcError::Malformed(_) | PrcError::Truncated(_)))); + let mut empty = W::default(); + empty.uint(0); + assert!(read(&empty, |r| huffman(r, 2, false)).is_empty()); + } + + #[test] + fn a_huffman_element_count_past_the_payload_is_truncated_not_allocated() { + // One leaf, then an element count of 2^32 - 1 with no elements behind it. + let mut w = W::default(); + w.huffman_count(2, 2, &[(1, 0b10, 2)], u32::MAX); + let bytes = w.bytes(); + assert!(matches!(huffman(&mut BitReader::new(&bytes), 2, false), Err(PrcError::Truncated("a Huffman element count")))); + } + + #[test] + fn a_huffman_code_that_begins_another_is_malformed() { + // `10` would end every `100` one bit early. + let mut w = W::default(); + w.huffman(2, 2, &[(1, 0b10, 2), (2, 0b100, 3)], &[2, 1]); + let bytes = w.bytes(); + let r = huffman(&mut BitReader::new(&bytes), 2, false); + assert!(malformed_by(r, "a Huffman code that is a prefix of another")); + } + + #[test] + fn huffman_widths_out_of_range_are_errors() { + // Well-formed payloads at each width, which only the width check refuses. + for nbits in [0, 17] { + let mut w = W::default(); + w.huffman(nbits, 2, &[(1, 0b10, 2)], &[1]); + let bytes = w.bytes(); + assert!(matches!(huffman(&mut BitReader::new(&bytes), nbits, false), Err(PrcError::Malformed(_))), "{nbits}"); + } + } + + #[test] + fn uncompressed_arrays() { + let mut w = W::default(); + w.bit(false).uint(2).put(0xfe, 8).put(3, 8); + assert_eq!(read(&w, |r| character_array(r, 6, None, true)), [-2, 3]); + let mut s = W::default(); + s.bit(false).uint(1).put(0x34, 8).put(0x12, 8); + assert_eq!(read(&s, |r| short_array(r, 16)), [0x1234]); + } + + #[test] + fn integer_array_reads_one_width_per_value() { + let mut w = W::default(); + w.bit(false).uint(3).put(3, 8).put(0, 8).put(4, 8); + w.bit(true).put(3, 2); // -3 in 3 bits + w.bit(false).put(5, 3); // 5 in 4 bits + assert_eq!(read(&w, compressed_integer_array), [-3, 0, 5]); + } + + #[test] + fn indice_array_sums_widths_and_deltas() { + // Values 5, 6, 4: widths 4, 2, 3 stored as diffs 4, -2, +1. + let mut w = W::default(); + w.bit(false).uint(3).put(4, 8).put(0xfe, 8).put(1, 8); + w.bit(false).put(5, 3); // 5 + w.bit(false).put(1, 1); // +1 + w.bit(true).put(2, 2); // -2 + assert_eq!(read(&w, |r| compressed_indice_array(r, None)), [5, 6, 4]); + // The implicit-flag form reads no flag bit. + let mut i = W::default(); + i.uint(1).put(2, 8).bit(false).put(1, 1); + assert_eq!(read(&i, |r| compressed_indice_array(r, Some(false))), [1]); + // A negative running index is damage. + let mut n = W::default(); + n.bit(false).uint(1).put(2, 8).bit(true).put(1, 1); + let bytes = n.bytes(); + assert!(compressed_indice_array(&mut BitReader::new(&bytes), None).is_err()); + } + + #[test] + fn a_bool_array_longer_than_the_data_is_truncated() { + assert!(matches!(bool_array(&mut BitReader::new(&[0]), 9), Err(PrcError::Truncated(_)))); + assert_eq!(bool_array(&mut BitReader::new(&[0x80]), 2).unwrap(), [true, false]); + } + + #[test] + fn decoded_elements_are_charged_against_one_allowance() { + // One bit can decode to a 4-byte element, so the byte cap alone does not bound memory. + let leaves = [(0b111110, 0b10, 2), (1, 0b11, 2)]; + let mut w = W::default(); + w.huffman(6, 2, &leaves, &[62, 1, 62]); + let bytes = w.bytes(); + let mut r = BitReader::with_elements(&bytes, 2); + assert!(matches!(huffman(&mut r, 6, false), Err(PrcError::TooManyElements { .. }))); + let mut r = BitReader::with_elements(&bytes, 4); + assert_eq!(huffman(&mut r, 6, false).unwrap(), [62, 1, 62]); + assert_eq!(r.elements_left(), 1); + assert!(matches!(bool_array(&mut BitReader::with_elements(&[0], 1), 2), Err(PrcError::TooManyElements { .. }))); + } + + /// Uncompressed character and short arrays are charged against the allowance too, and a + /// character count past the data is refused by name before anything is read. + #[test] + fn uncompressed_arrays_are_charged_and_their_counts_checked() { + let mut w = W::default(); + w.uint(3).put(0, 8).put(1, 8).put(2, 8); + let chars = w.bytes(); + let read = |n| character_array(&mut BitReader::with_elements(&chars, n), 8, Some(false), false); + assert!(matches!(read(2), Err(PrcError::TooManyElements { .. }))); + assert_eq!(read(3).unwrap(), [0, 1, 2]); + let mut w = W::default(); + w.bit(false).uint(2).put(1, 8).put(0, 8).put(2, 8).put(0, 8); + let shorts = w.bytes(); + assert!(matches!(short_array(&mut BitReader::with_elements(&shorts, 1), 8), Err(PrcError::TooManyElements { .. }))); + assert_eq!(short_array(&mut BitReader::with_elements(&shorts, 2), 8).unwrap(), [1, 2]); + let mut w = W::default(); + w.uint(1000).put(7, 8); + let short_of_data = w.bytes(); + let res = character_array(&mut BitReader::new(&short_of_data), 8, Some(false), false); + assert!(matches!(res, Err(PrcError::Truncated("a character array"))), "{res:?}"); + } + + fn malformed_by(res: Result, PrcError>, what: &str) -> bool { + matches!(res, Err(PrcError::Malformed(m)) if m == what) + } + + #[test] + fn a_zero_code_length_width_or_length_is_refused_by_name() { + let mut w = W::default(); + w.huffman_raw(2, 0, &[(1, 0, 0)], &[]); + let bytes = w.bytes(); + assert!(malformed_by(huffman(&mut BitReader::new(&bytes), 2, false), "Huffman code-length width")); + let mut w = W::default(); + w.huffman_raw(2, 2, &[(1, 0, 0)], &[]); + let bytes = w.bytes(); + assert!(malformed_by(huffman(&mut BitReader::new(&bytes), 2, false), "Huffman code length")); + } + + #[test] + fn a_boolean_array_past_the_data_is_refused_before_it_is_read() { + let mut w = W::default(); + w.put(0b1011, 4); + let bytes = w.bytes(); + let mut r = BitReader::new(&bytes); + let n = r.remaining() + 1; + assert_eq!(bool_array(&mut r, n), Err(PrcError::Truncated("a boolean array"))); + assert_eq!(r.position(), 0, "nothing read"); + } +} diff --git a/crates/prc/src/bits.rs b/crates/prc/src/bits.rs new file mode 100644 index 0000000000..837d281f3b --- /dev/null +++ b/crates/prc/src/bits.rs @@ -0,0 +1,491 @@ +//! The bit-level reader for PRC compressed sections [WD 10.3, 11.1-11.17]. +//! +//! Values are consumed MSB-first from each inflated byte and straddle byte boundaries freely +//! [WD 10.3.1, 11.3]. Multi-byte integers are groups of eight bits, low group first. Two sign +//! conventions coexist: `Integer` is two's complement, the `*WithVariableBitNumber` types are +//! sign-magnitude. + +use crate::PrcError; +use crate::acof::{self, Entry}; + +/// The most array elements the readers of one PRC stream decode together (2^24): at most 8 +/// bytes each, so 128 MiB. +pub const MAX_DECODED_ELEMENTS: usize = 1 << 24; + +/// The most schema-program steps (tokens executed, loop iterations included) the readers of +/// one PRC stream run together. +pub const MAX_SCHEMA_STEPS: u64 = 1 << 22; + +/// What the readers of one stream may still spend, carried from each reader to the next. +#[derive(Debug, Clone, Copy, PartialEq, Eq)] +pub(crate) struct Budget { + /// Array elements still to decode. + pub(crate) elements: usize, + /// Schema-program steps still to run. + pub(crate) steps: u64, +} + +impl Budget { + /// A whole stream's allowance. + pub(crate) const FULL: Self = Self { elements: MAX_DECODED_ELEMENTS, steps: MAX_SCHEMA_STEPS }; +} + +/// A cursor over one inflated section. +#[derive(Debug, Clone)] +pub struct BitReader<'a> { + data: &'a [u8], + pos: usize, + budget: Budget, +} + +impl<'a> BitReader<'a> { + /// A reader positioned at the first bit of `data`, allowed [`MAX_DECODED_ELEMENTS`] and + /// [`MAX_SCHEMA_STEPS`]. + pub fn new(data: &'a [u8]) -> Self { + Self::with_budget(data, Budget::FULL) + } + + /// A reader of `data` allowed `budget`: what an earlier reader of the same stream left. + pub(crate) fn with_budget(data: &'a [u8], budget: Budget) -> Self { + Self { data, pos: 0, budget } + } + + /// A reader of `data` allowed `elements` array elements and a full step allowance. + #[cfg(test)] + pub(crate) fn with_elements(data: &'a [u8], elements: usize) -> Self { + Self::with_budget(data, Budget { elements, ..Budget::FULL }) + } + + /// This reader, at its position, allowed `budget` instead: what readers of the same + /// stream left since it stopped. + pub(crate) fn with_remaining(mut self, budget: Budget) -> Self { + self.budget = budget; + self + } + + /// What this reader may still spend. + pub(crate) fn budget(&self) -> Budget { + self.budget + } + + /// Array elements this reader may still decode. + #[cfg(test)] + pub(crate) fn elements_left(&self) -> usize { + self.budget.elements + } + + /// Charge `n` decoded array elements against the allowance. + /// + /// # Errors + /// [`PrcError::TooManyElements`] when fewer than `n` are left. + pub(crate) fn take_elements(&mut self, n: usize) -> Result<(), PrcError> { + self.budget.elements = self.budget.elements.checked_sub(n).ok_or(PrcError::TooManyElements { limit: MAX_DECODED_ELEMENTS })?; + Ok(()) + } + + /// Charge one schema-program step; false when none is left. + pub(crate) fn take_step(&mut self) -> bool { + match self.budget.steps.checked_sub(1) { + Some(left) => { + self.budget.steps = left; + true + } + None => false, + } + } + + /// Bits consumed so far. + pub fn position(&self) -> usize { + self.pos + } + + /// Bits left before the end of the data. + pub fn remaining(&self) -> usize { + self.data.len().saturating_mul(8).saturating_sub(self.pos) + } + + /// Step over `n` bits of opaque data (`UserData` [WD 8.6]). + /// + /// # Errors + /// [`PrcError::Truncated`] when fewer than `n` bits remain; the cursor does not move. + pub fn skip_bits(&mut self, n: usize) -> Result<(), PrcError> { + if n > self.remaining() { + return Err(PrcError::Truncated("user data")); + } + self.pos = self.pos.saturating_add(n); + Ok(()) + } + + /// One bit; a PRC `Boolean` [WD 10.3.3]. + /// + /// # Errors + /// [`PrcError::Truncated`] at the end of the data. + pub fn bit(&mut self) -> Result { + let byte = self.data.get(self.pos / 8).ok_or(PrcError::Truncated("a bit"))?; + let b = (byte >> (7 - (self.pos % 8))) & 1 == 1; + // `pos / 8` indexed the data, so `pos + 1` is at most `8 * len`. + self.pos = self.pos.saturating_add(1); + Ok(b) + } + + /// `n` bits (at most 32), MSB-first; `UnsignedIntegerWithVariableBitNumber` [WD 11.13]. + /// + /// # Errors + /// [`PrcError::Truncated`] at the end of the data; [`PrcError::Malformed`] for `n > 32`. + pub fn bits(&mut self, n: u32) -> Result { + if n > 32 { + return Err(PrcError::Malformed(format!("{n}-bit field"))); + } + if self.remaining() < n as usize { + return Err(PrcError::Truncated("a bit field")); + } + let mut v: u64 = 0; + for _ in 0..n { + v = (v << 1) | u64::from(self.bit()?); + } + // n <= 32, so v fits. + Ok(u32::try_from(v).unwrap_or(u32::MAX)) + } + + fn byte(&mut self) -> Result { + Ok(u8::try_from(self.bits(8)?).unwrap_or(u8::MAX)) + } + + /// A `Character`: one byte [WD 10.3.4]. + /// + /// # Errors + /// [`PrcError::Truncated`]. + pub fn character(&mut self) -> Result { + self.byte() + } + + /// An `UnsignedInteger` [WD 11.10]: while a 1 bit precedes it, one more eight-bit group, + /// low group first. + /// + /// # Errors + /// [`PrcError::Truncated`]; [`PrcError::Malformed`] past four groups. + pub fn unsigned_integer(&mut self) -> Result { + let mut v: u32 = 0; + let mut shift = 0; + while self.bit()? { + if shift >= 32 { + return Err(PrcError::Malformed("UnsignedInteger over 32 bits".into())); + } + v |= u32::from(self.byte()?) << shift; + shift += 8; + } + Ok(v) + } + + /// An `Integer` [WD 11.11]: a 0 bit is zero; otherwise eight-bit groups, low group first, + /// each followed by a continue bit, sign-extended from the last group's top bit. + /// + /// # Errors + /// [`PrcError::Truncated`]; [`PrcError::Malformed`] past four groups. + pub fn integer(&mut self) -> Result { + if !self.bit()? { + return Ok(0); + } + let mut v: u32 = 0; + let mut shift = 0; + loop { + if shift >= 32 { + return Err(PrcError::Malformed("Integer over 32 bits".into())); + } + let group = self.byte()?; + v |= u32::from(group) << shift; + shift += 8; + if !self.bit()? { + if group & 0x80 != 0 && shift < 32 { + v |= u32::MAX << shift; + } + return Ok(v.cast_signed()); + } + } + } + + /// An `IntegerWithVariableBitNumber(n)` [WD 11.12]: a negative flag (1 = negative; the WD + /// prose says the opposite, its sample code and real files agree on this), then the + /// magnitude in `n - 1` bits. + /// + /// # Errors + /// [`PrcError::Truncated`]; [`PrcError::Malformed`] for `n > 33`. + pub fn integer_vbn(&mut self, n: u32) -> Result { + let Some(magnitude_bits) = n.checked_sub(1) else { + return Ok(0); + }; + let negative = self.bit()?; + let m = i64::from(self.bits(magnitude_bits)?); + Ok(if negative { -m } else { m }) + } + + /// A `DoubleWithVariableBitNumber(tolerance, n)` [WD 11.14]: a sign-magnitude quantum + /// count times `tolerance`. + /// + /// # Errors + /// As [`Self::integer_vbn`]. + pub fn double_vbn(&mut self, tolerance: f64, n: u32) -> Result { + Ok(self.integer_vbn(n)? as f64 * tolerance) + } + + /// `NumberOfBitsThenUnsignedInteger` [WD 11.15]: a 5-bit width, then the value in that + /// many bits. + /// + /// # Errors + /// [`PrcError::Truncated`]. + pub fn nbits_then_unsigned(&mut self) -> Result { + let n = self.bits(5)?; + self.bits(n) + } + + /// A `String` [WD 10.3.7, 11.4]: `None` for the null string, else UTF-8 (invalid + /// sequences replaced). + /// + /// # Errors + /// [`PrcError::Truncated`] when the byte count runs past the data. + pub fn string(&mut self) -> Result, PrcError> { + if !self.bit()? { + return Ok(None); + } + let len = self.unsigned_integer()? as usize; + if len.saturating_mul(8) > self.remaining() { + return Err(PrcError::Truncated("a String")); + } + let bytes = (0..len).map(|_| self.byte()).collect::, _>>()?; + Ok(Some(String::from_utf8_lossy(&bytes).into_owned())) + } + + /// `FloatAsBytes` [WD 11.5]: an IEEE binary32, least-significant byte first. + /// + /// # Errors + /// [`PrcError::Truncated`]. + pub fn float_as_bytes(&mut self) -> Result { + let b = [self.byte()?, self.byte()?, self.byte()?, self.byte()?]; + Ok(f32::from_le_bytes(b)) + } + + /// A `Double` [WD 11.17]: a prefix code from the Double table, then a sign bit and, for + /// an exponent code, a byte-wise mantissa with back-references. + /// + /// The two-bit zero code has no sign bit. Back-reference offsets 1 to 5 are all accepted, + /// distance 1 included (PDF Association pdf-issues #769). + /// + /// # Errors + /// [`PrcError::UnknownDoubleCode`], [`PrcError::Truncated`], or [`PrcError::Malformed`] + /// for a back-reference outside the value. + pub fn double(&mut self) -> Result { + // prc-rs: the decoding steps (code lookup, sign, exponent, byte-wise mantissa with + // back-references) follow prc-rs's. + let mut code = 0u32; + let mut entry = None; + for n in 1..=acof::MAX_CODE_BITS { + code = (code << 1) | u32::from(self.bit()?); + if let Some(e) = acof::lookup(n, code) { + if n == 2 && code == 0b01 { + return Ok(0.0); + } + entry = Some(e); + break; + } + } + let entry = entry.ok_or(PrcError::UnknownDoubleCode)?; + let negative = self.bit()?; + let upper = match entry { + Entry::Double(v) => return Ok(if negative { -v } else { v }), + Entry::Exponent(upper) => upper, + }; + let mut b = (u64::from(upper) << 32).to_le_bytes(); + if self.bit()? { + self.mantissa(&mut b)?; + } + let sign = if negative { 1u64 << 63 } else { 0 }; + Ok(f64::from_bits(u64::from_le_bytes(b) | sign)) + } + + /// Mantissa bits 51..0 into `b` (little-endian byte order): a nibble, then bytes 5 down + /// to 0, each a literal or a back-reference to a byte already set. + fn mantissa(&mut self, b: &mut [u8; 8]) -> Result<(), PrcError> { + let nibble = u8::try_from(self.bits(4)?).unwrap_or(0); + let [.., b6, _] = b; + *b6 |= nibble; + let at = |b: &[u8; 8], i: usize| b.get(i).copied().ok_or_else(|| PrcError::Malformed("Double back-reference past the value".into())); + for cbi in (0..=5usize).rev() { + let prev = at(b, cbi + 1)?; + let value = if self.bit()? { + self.byte()? + } else { + match self.bits(3)? { + // Fill this byte and every lower one with the previous byte. + 0 => { + b.iter_mut().take(cbi + 1).for_each(|x| *x = prev); + return Ok(()); + } + // As 0, but the lowest byte is a literal. + 6 => { + b.iter_mut().take(cbi + 1).skip(1).for_each(|x| *x = prev); + let low = self.byte()?; + let [b0, ..] = b; + *b0 = low; + return Ok(()); + } + off @ 1..=5 => at(b, cbi + off as usize)?, + _ => return Err(PrcError::Malformed("Double mantissa offset 7".into())), + } + }; + if let Some(x) = b.get_mut(cbi) { + *x = value; + } + } + Ok(()) + } +} + +#[cfg(test)] +#[allow(clippy::unwrap_used, clippy::expect_used)] +mod tests { + use super::*; + use crate::testw::W; + + /// The code for IEEE exponent `e`. + fn exponent_code(e: u32) -> (u32, u32) { + acof::exponent_code(e).expect("the table has a row per exponent") + } + + fn read_double(w: &W) -> f64 { + BitReader::new(&w.bytes()).double().unwrap() + } + + #[test] + fn zero_is_the_two_bit_code_with_no_sign() { + let mut w = W::default(); + w.put(0b01, 2).put(0b1111_1111, 8); + let bytes = w.bytes(); + let mut r = BitReader::new(&bytes); + assert_eq!(r.double().unwrap(), 0.0); + assert_eq!(r.position(), 2); + } + + #[test] + fn exponent_with_literal_mantissa() { + let v = -1234.5678f64; + let le = v.to_bits().to_le_bytes(); + let e = u32::try_from((v.to_bits() >> 52) & 0x7ff).unwrap(); + let (n, c) = exponent_code(e); + let mut w = W::default(); + w.put(u64::from(c), n).put(1, 1).put(1, 1).put(u64::from(le[6] & 0x0f), 4); + for byte in le[..6].iter().rev() { + w.put(1, 1).put(u64::from(*byte), 8); + } + assert_eq!(read_double(&w), v); + } + + #[test] + fn exponent_without_mantissa_is_a_power_of_two() { + let (n, c) = exponent_code(1023 + 3); + let mut w = W::default(); + w.put(u64::from(c), n).put(0, 1).put(0, 1); + assert_eq!(read_double(&w), 8.0); + } + + #[test] + fn fill_back_reference_and_literal_low_byte() { + // Mantissa bytes 6..0 = 0x05, 0xAA, 0xAA (offset 1), 0xAA (offset 2), fill 0xAA to byte 1, byte 0 = 0x11. + let (n, c) = exponent_code(1023); + let mut w = W::default(); + w.put(u64::from(c), n).put(0, 1).put(1, 1).put(0x5, 4); + w.put(1, 1).put(0xAA, 8); + w.put(0, 1).put(1, 3); + w.put(0, 1).put(2, 3); + w.put(0, 1).put(6, 3).put(0x11, 8); + assert_eq!(read_double(&w), f64::from_le_bytes([0x11, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xF5, 0x3F])); + + // Byte 5 literal, then fill 4..0. + let mut w = W::default(); + w.put(u64::from(c), n).put(0, 1).put(1, 1).put(0x5, 4); + w.put(1, 1).put(0x77, 8).put(0, 1).put(0, 3); + assert_eq!(read_double(&w), f64::from_le_bytes([0x77, 0x77, 0x77, 0x77, 0x77, 0x77, 0xF5, 0x3F])); + } + + #[test] + fn a_back_reference_past_the_value_is_refused() { + let (n, c) = exponent_code(1023); + let mut w = W::default(); + w.put(u64::from(c), n).put(0, 1).put(1, 1).put(0, 4); + w.put(0, 1).put(5, 3).put(0, 16); // byte 5 + offset 5 = byte 10 + assert!(matches!(BitReader::new(&w.bytes()).double(), Err(PrcError::Malformed(_)))); + } + + #[test] + fn a_truncated_double_is_an_error_not_a_panic() { + // No code is 8 bits or shorter and a prefix of 0x17, so one byte ends inside a code. + assert!(matches!(BitReader::new(&[0x17]).double(), Err(PrcError::Truncated(_)))); + assert!(matches!(BitReader::new(&[]).double(), Err(PrcError::Truncated(_)))); + } + + #[test] + fn integers_and_strings() { + let mut w = W::default(); + // UnsignedInteger 300 = groups 0x2C, 0x01. + w.put(1, 1).put(0x2C, 8).put(1, 1).put(0x01, 8).put(0, 1); + // Integer -2 = group 0xFE, stop. + w.put(1, 1).put(0xFE, 8).put(0, 1); + // Integer 200 = 0xC8 then 0x00 (top bit clear), stop. + w.put(1, 1).put(0xC8, 8).put(1, 1).put(0x00, 8).put(0, 1); + // IntegerWithVariableBitNumber(4) = -5. + w.put(1, 1).put(5, 3); + // String "hé". + w.put(1, 1).put(1, 1).put(3, 8).put(0, 1).put(0x68, 8).put(0xC3, 8).put(0xA9, 8); + // Null string. + w.put(0, 1); + let bytes = w.bytes(); + let mut r = BitReader::new(&bytes); + assert_eq!(r.unsigned_integer().unwrap(), 300); + assert_eq!(r.integer().unwrap(), -2); + assert_eq!(r.integer().unwrap(), 200); + assert_eq!(r.integer_vbn(4).unwrap(), -5); + assert_eq!(r.string().unwrap().as_deref(), Some("hé")); + assert_eq!(r.string().unwrap(), None); + } + + #[test] + fn the_writer_round_trips_through_the_reader() { + let mut w = W::default(); + w.uint(0).uint(u32::MAX).int(-129).int(i32::MIN).int(i32::MAX).string(Some("glass")).string(None); + let bytes = w.bytes(); + let mut r = BitReader::new(&bytes); + assert_eq!(r.unsigned_integer().unwrap(), 0); + assert_eq!(r.unsigned_integer().unwrap(), u32::MAX); + assert_eq!(r.integer().unwrap(), -129); + assert_eq!(r.integer().unwrap(), i32::MIN); + assert_eq!(r.integer().unwrap(), i32::MAX); + assert_eq!(r.string().unwrap().as_deref(), Some("glass")); + assert_eq!(r.string().unwrap(), None); + } + + #[test] + fn oversized_groups_and_widths_are_malformed() { + // Five UnsignedInteger groups. + let mut w = W::default(); + (0..5).for_each(|_| { + w.put(1, 1).put(0xFF, 8); + }); + w.put(0, 1); + assert!(matches!(BitReader::new(&w.bytes()).unsigned_integer(), Err(PrcError::Malformed(_)))); + assert!(matches!(BitReader::new(&[0; 8]).bits(33), Err(PrcError::Malformed(_)))); + assert!(matches!(BitReader::new(&[0; 8]).integer_vbn(34), Err(PrcError::Malformed(_)))); + assert_eq!(BitReader::new(&[]).integer_vbn(0), Ok(0)); + } + + #[test] + fn truncation_is_an_error_not_a_panic() { + let mut r = BitReader::new(&[0xFF]); + assert!(r.unsigned_integer().is_err()); + let mut r = BitReader::new(&[0xC0, 0x40]); + assert!(matches!(r.string(), Err(PrcError::Truncated(_)))); + let mut r = BitReader::new(&[0xAB]); + assert!(matches!(r.skip_bits(9), Err(PrcError::Truncated(_)))); + assert_eq!(r.position(), 0, "a refused skip does not move the cursor"); + r.skip_bits(8).unwrap(); + assert_eq!(r.remaining(), 0); + } +} diff --git a/crates/prc/src/container.rs b/crates/prc/src/container.rs new file mode 100644 index 0000000000..3ff18508f1 --- /dev/null +++ b/crates/prc/src/container.rs @@ -0,0 +1,340 @@ +//! The PRC container: file header, file-structure descriptions, section extents and +//! inflation [WD 6.1-6.2, 10.2]. +//! +//! Section extents are not stored: each section runs to the next one's start offset, and the +//! last one to the model file's start. Inflation stops at the zlib stream's end, so padding +//! between sections is tolerated. + +use std::io::Read; + +use crate::PrcError; +use crate::bits::BitReader; +use crate::schema::Schema; + +/// The PRC version this reader implements. +pub const PRC_READER_VERSION: u32 = 8137; + +/// The most file structures [`PrcFile::parse`] accepts. +pub const MAX_FILE_STRUCTURES: usize = 65_536; + +/// Ceiling on the total inflated size of every section in one PRC stream, together with the +/// picture and uncompressed blocks it copies. +pub const MAX_INFLATED_BYTES: usize = 512 * 1024 * 1024; + +/// Sections per file structure [WD 6.1.2]: its header, then the five in [`SectionKind`] +/// order. +const SECTIONS: usize = 6; + +/// A 128-bit PRC identifier [WD 10.2.4]: four little-endian `u32`s. +#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Default)] +pub struct UniqueId(pub [u32; 4]); + +/// The compressed sections of one file structure, in file order [WD 6.1.2, 6.2.1]. +#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)] +pub enum SectionKind { + /// Starts with the file structure's schema, then its globals entity. + Globals, + /// The product-structure tree. + Tree, + /// Tessellated (mesh) data. + Tessellation, + /// Exact (B-rep) geometry. + Geometry, + /// Supplementary geometry. + ExtraGeometry, +} + +impl SectionKind { + /// Every kind, in file order. + pub const ALL: [SectionKind; 5] = + [SectionKind::Globals, SectionKind::Tree, SectionKind::Tessellation, SectionKind::Geometry, SectionKind::ExtraGeometry]; + + /// The section's name, as used in errors. + pub fn name(self) -> &'static str { + match self { + SectionKind::Globals => "globals", + SectionKind::Tree => "tree", + SectionKind::Tessellation => "tessellation", + SectionKind::Geometry => "geometry", + SectionKind::ExtraGeometry => "extra geometry", + } + } + + fn index(self) -> usize { + match self { + SectionKind::Globals => 0, + SectionKind::Tree => 1, + SectionKind::Tessellation => 2, + SectionKind::Geometry => 3, + SectionKind::ExtraGeometry => 4, + } + } +} + +/// The uncompressed file header [WD 6.1.1]. +#[derive(Debug, Clone, PartialEq, Eq)] +#[non_exhaustive] +pub struct PrcHeader { + /// The oldest reader version the producer says can read the file. + pub min_version_for_read: u32, + /// The producer's PRC version. + pub authoring_version: u32, + /// The file's identifier. + pub file_id: UniqueId, + /// The producing application's identifier (all zero = unregistered). + pub application_id: UniqueId, +} + +/// One file structure: its uncompressed header fields and its five inflated sections +/// [WD 6.2]. +#[derive(Debug, Clone, PartialEq, Eq)] +#[non_exhaustive] +pub struct FileStructure { + /// The identifier; equals the file header's description of it. + pub id: UniqueId, + /// Per-structure reader version floor; may differ from the file's. + pub min_version_for_read: u32, + /// Per-structure producer version. + pub authoring_version: u32, + /// Picture payloads (JPEG, PNG, ...) carried uncompressed in the header, referenced by + /// index from picture entities [WD 6.2.2]. + pub pictures: Vec>, + sections: [Vec; 5], +} + +impl FileStructure { + /// The inflated bytes of one section. + pub fn section(&self, kind: SectionKind) -> &[u8] { + self.sections.get(kind.index()).map_or(&[], Vec::as_slice) + } + + /// The globals section's leading schema: the fields producers newer than this reader + /// added to each entity type [WD 6.3]. + /// + /// # Errors + /// As [`Schema::read`]. + pub fn schema(&self) -> Result { + Schema::read(&mut BitReader::new(self.section(SectionKind::Globals))) + } +} + +/// A parsed PRC stream: header, file structures and model-file section. +#[derive(Debug, Clone, PartialEq, Eq)] +#[non_exhaustive] +pub struct PrcFile { + /// The file header. + pub header: PrcHeader, + /// File structures in physical order (referenced before referencing). + pub file_structures: Vec, + /// The inflated model-file section: its schema, then the model file entity. + pub model_file: Vec, + /// The `UncompressedFiles` blocks after the file header [WD 6.1.1], which picture + /// entities may also reference. + pub uncompressed_files: Vec>, +} + +/// A little-endian byte cursor over the uncompressed parts of the stream. +struct Bytes<'a> { + data: &'a [u8], + pos: usize, +} + +impl<'a> Bytes<'a> { + fn take(&mut self, n: usize, what: &'static str) -> Result<&'a [u8], PrcError> { + let end = self.pos.checked_add(n).ok_or(PrcError::Truncated(what))?; + let s = self.data.get(self.pos..end).ok_or(PrcError::Truncated(what))?; + self.pos = end; + Ok(s) + } + + fn u32(&mut self, what: &'static str) -> Result { + let s = self.take(4, what)?; + let b: [u8; 4] = s.try_into().map_err(|_| PrcError::Truncated(what))?; + Ok(u32::from_le_bytes(b)) + } + + fn id(&mut self, what: &'static str) -> Result { + Ok(UniqueId([self.u32(what)?, self.u32(what)?, self.u32(what)?, self.u32(what)?])) + } + + fn magic(&mut self) -> Result<(), PrcError> { + match self.take(3, "magic") { + Ok(b"PRC") => Ok(()), + _ => Err(PrcError::NotPrc), + } + } + + /// `UncompressedBlock`s: a count, then size-prefixed blocks [WD 10.2.2-3], each charged to + /// `budget` before it is copied. + fn blocks(&mut self, what: &'static str, budget: &mut Budget) -> Result>, PrcError> { + let n = offset(self.u32(what)?); + // Each block needs at least its 4-byte size, so `n` is bounded by the data left. + if n > self.data.len().saturating_sub(self.pos) / 4 { + return Err(PrcError::Truncated(what)); + } + (0..n) + .map(|_| { + let len = offset(self.u32(what)?); + let block = self.take(len, what)?; + budget.charge(block.len())?; + Ok(block.to_vec()) + }) + .collect() + } +} + +/// A stored `u32` offset or count as a `usize` (saturating on a 16-bit target, where no data +/// that large could exist). +pub(crate) fn offset(v: u32) -> usize { + usize::try_from(v).unwrap_or(usize::MAX) +} + +/// One file structure description of the file header [WD 6.2]: its id and the offsets of +/// its sections. +fn file_structure_description(r: &mut Bytes) -> Result<(UniqueId, Vec), PrcError> { + let id = r.id("file structure description")?; + if r.u32("file structure description")? != 0 { + return Err(PrcError::Malformed("file structure reserved field is not 0".into())); + } + let count = offset(r.u32("file structure description")?); + if count < SECTIONS { + return Err(PrcError::Malformed(format!("{count} sections in a file structure"))); + } + let mut offs = Vec::with_capacity(SECTIONS); + // Offsets past the six this reader uses are read and dropped; each read is bounded by + // the data, so a huge `count` ends in `Truncated`. + for i in 0..count { + let o = r.u32("section offset")?; + if i < SECTIONS { + offs.push(offset(o)); + } + } + Ok((id, offs)) +} + +/// Where one file structure's parts lie: its section offsets and the end of its last section. +struct Extent<'a> { + offs: &'a [usize], + last: usize, +} + +impl Extent<'_> { + fn bound(&self, i: usize) -> usize { + self.offs.get(i).copied().unwrap_or(self.last) + } +} + +/// The file structure `id` lying at `extent`: its header and its inflated sections, charged +/// to `budget`. +fn file_structure(data: &[u8], id: UniqueId, extent: &Extent, budget: &mut Budget) -> Result { + if extent.offs.windows(2).any(|w| w.first() > w.get(1)) || extent.bound(SECTIONS - 1) > extent.last { + return Err(PrcError::Malformed("file structure section offsets are not ascending".into())); + } + let header = data.get(..extent.bound(1)).ok_or(PrcError::Truncated("file structure header"))?; + let mut h = Bytes { data: header, pos: extent.bound(0) }; + h.magic()?; + let min_version_for_read = h.u32("file structure header")?; + let authoring_version = h.u32("file structure header")?; + if h.id("file structure header")? != id { + return Err(PrcError::Malformed("file structure id differs from its description".into())); + } + let _application = h.id("file structure header")?; + // Headers may overlap, so each structure's copy of its pictures is charged. + let pictures = h.blocks("file structure pictures", budget)?; + let mut sections: [Vec; 5] = Default::default(); + for (i, (slot, kind)) in sections.iter_mut().zip(SectionKind::ALL).enumerate() { + let end = if i + 2 < SECTIONS { extent.bound(i + 2) } else { extent.last }; + *slot = budget.inflate(data, (extent.bound(i + 1), end), kind.name())?; + } + Ok(FileStructure { id, min_version_for_read, authoring_version, pictures, sections }) +} + +/// What is left of the inflation ceiling, and the ceiling itself for the error. +struct Budget { + left: usize, + limit: usize, +} + +impl Budget { + /// The zlib stream in `data[range]`, its size charged to the budget. + fn inflate(&mut self, data: &[u8], range: (usize, usize), section: &'static str) -> Result, PrcError> { + let (start, end) = range; + if start == end { + return Ok(Vec::new()); + } + let src = data.get(start..end).ok_or_else(|| PrcError::Malformed(format!("{section} section {start}..{end} is outside the data")))?; + let mut out = Vec::new(); + // One byte past the budget is enough to know it was exceeded. + let cap = u64::try_from(self.left).unwrap_or(u64::MAX).saturating_add(1); + flate2::read::ZlibDecoder::new(src).take(cap).read_to_end(&mut out).map_err(|e| PrcError::Inflate { section, reason: e.to_string() })?; + self.charge(out.len())?; + Ok(out) + } + + /// Spend `n` bytes of the ceiling. + fn charge(&mut self, n: usize) -> Result<(), PrcError> { + self.left = self.left.checked_sub(n).ok_or(PrcError::TooLarge { limit: self.limit })?; + Ok(()) + } +} + +impl PrcFile { + /// Parse a PRC stream (the decoded data of a `/3D` stream with `/Subtype /PRC`). + /// + /// Newer `min_version_for_read` values are not refused: the fields read here are + /// unchanged across published versions, and the caller sees the versions in + /// [`PrcHeader`]. + /// + /// # Errors + /// [`PrcError::NotPrc`] without the magic; [`PrcError::Truncated`] or + /// [`PrcError::Malformed`] for a damaged header or offset table; [`PrcError::Inflate`] + /// for a section that is not zlib; [`PrcError::TooLarge`] past [`MAX_INFLATED_BYTES`] + /// in total, inflated sections and copied blocks together. + /// + /// # Examples + /// ``` + /// use pdfcraft_prc::{PrcError, PrcFile}; + /// assert_eq!(PrcFile::parse(b"%PDF"), Err(PrcError::NotPrc)); + /// ``` + pub fn parse(data: &[u8]) -> Result { + Self::parse_with_limit(data, MAX_INFLATED_BYTES) + } + + /// [`Self::parse`] with a caller-chosen ceiling on the total inflated size, for callers + /// with less memory to spend than [`MAX_INFLATED_BYTES`]. + /// + /// # Errors + /// As [`Self::parse`]; [`PrcError::TooLarge`] carries `limit`. + pub fn parse_with_limit(data: &[u8], limit: usize) -> Result { + let mut r = Bytes { data, pos: 0 }; + r.magic()?; + let header = PrcHeader { + min_version_for_read: r.u32("file header")?, + authoring_version: r.u32("file header")?, + file_id: r.id("file header")?, + application_id: r.id("file header")?, + }; + let n_fs = offset(r.u32("file structure count")?); + if n_fs == 0 || n_fs > MAX_FILE_STRUCTURES { + return Err(PrcError::Malformed(format!("{n_fs} file structures"))); + } + let descriptions = (0..n_fs).map(|_| file_structure_description(&mut r)).collect::, _>>()?; + let mf_start = offset(r.u32("model file offsets")?); + let mf_end = offset(r.u32("model file offsets")?); + let mut budget = Budget { left: limit, limit }; + let uncompressed_files = r.blocks("uncompressed files", &mut budget)?; + if mf_end < mf_start { + return Err(PrcError::Malformed("model file ends before it starts".into())); + } + if mf_end > data.len() { + return Err(PrcError::Malformed(format!("model file ends at {mf_end}, past the {}-byte stream", data.len()))); + } + + // prc-rs: the last section runs to the model file's start. + let last = mf_start.min(data.len()); + let file_structures = + descriptions.iter().map(|(id, offs)| file_structure(data, *id, &Extent { offs, last }, &mut budget)).collect::, _>>()?; + let model_file = budget.inflate(data, (mf_start, mf_end), "model file")?; + Ok(PrcFile { header, file_structures, model_file, uncompressed_files }) + } +} diff --git a/crates/prc/src/entity.rs b/crates/prc/src/entity.rs new file mode 100644 index 0000000000..9e3c6c2a7b --- /dev/null +++ b/crates/prc/src/entity.rs @@ -0,0 +1,127 @@ +//! What every section reader shares: the cursor, the schema, the authoring version, and the +//! base fields every PRC entity starts with [WD 7.2.3, 7.4.3, 8.6]. + +use crate::PrcError; +use crate::bits::BitReader; +use crate::schema::Schema; +use crate::tree::Graphics; + +const ATTRIBUTE: u32 = 201; + +/// A [`PrcError::Malformed`] naming `what`. +pub(crate) fn malformed(what: String) -> PrcError { + PrcError::Malformed(what) +} + +/// A section reader. +pub(crate) struct Ctx<'a, 's> { + pub(crate) r: BitReader<'a>, + pub(crate) schema: &'s Schema, + /// The file structure's authoring version; gates version-added fields. + pub(crate) version: u32, + /// The current name: what a `same_name` entity reuses; every entity's own name replaces + /// it [WD 7.2.3.4]. Each section starts with none. + pub(crate) name: Option, + /// The PRC unique id of the last `ContentPRCRefBase` read [WD 7.2.3.2]. + pub(crate) uid: u32, + /// The current graphics: what a `same_graphics` entity reuses [WD 7.2.4]. + pub(crate) graphics: Graphics, +} + +impl<'a, 's> Ctx<'a, 's> { + pub(crate) fn new(r: BitReader<'a>, schema: &'s Schema, version: u32) -> Self { + Ctx { r, schema, version, name: None, uid: 0, graphics: Graphics::default() } + } + + /// Reads an entity type code; any value but `want` is malformed. + pub(crate) fn expect_type(&mut self, want: u32) -> Result<(), PrcError> { + let t = self.r.unsigned_integer()?; + if t != want { + return Err(malformed(format!("entity type {t} where {want} belongs"))); + } + Ok(()) + } + + /// A count that must fit in the remaining data at `min_bits` per item, which bounds any + /// allocation sized by it. + pub(crate) fn count(&mut self, min_bits: usize, what: &'static str) -> Result { + let n = self.r.unsigned_integer()? as usize; + if n.saturating_mul(min_bits) > self.r.remaining() { + return Err(PrcError::Truncated(what)); + } + self.r.take_elements(n)?; + Ok(n) + } + + /// A counted array of `UnsignedInteger`s. + pub(crate) fn uints(&mut self, what: &'static str) -> Result, PrcError> { + let n = self.count(1, what)?; + (0..n).map(|_| self.r.unsigned_integer()).collect() + } + + /// A counted array of `Double`s. + pub(crate) fn doubles(&mut self, what: &'static str) -> Result, PrcError> { + let n = self.count(2, what)?; + (0..n).map(|_| self.r.double()).collect() + } + + /// `AttributeEntry` [WD 7.4.3]: a predefined title code or a string. + fn attribute_entry(&mut self) -> Result<(), PrcError> { + if self.r.bit()? { + self.r.unsigned_integer()?; + } else { + self.r.string()?; + } + Ok(()) + } + + /// One attribute value [WD 7.4.3], read past. + fn attribute_value(&mut self) -> Result<(), PrcError> { + match self.r.unsigned_integer()? { + 0 => {} + 1 | 3 => { + self.r.integer()?; + } + 2 => { + self.r.double()?; + } + 4 => { + self.r.string()?; + } + // prc-rs: a 64-bit integer is its high part, then its low part. + 5 => { + self.r.integer()?; + self.r.unsigned_integer()?; + } + k => return Err(malformed(format!("attribute value kind {k}"))), + } + Ok(()) + } + + /// `ContentPRCBase` [WD 7.2.3]: attributes, then the name; returns the entity's name, + /// which is also the current name afterwards. + pub(crate) fn content_prc_base(&mut self) -> Result, PrcError> { + let n = self.count(1, "attributes")?; + for _ in 0..n { + self.expect_type(ATTRIBUTE)?; + self.attribute_entry()?; + let pairs = self.count(2, "attribute pairs")?; + for _ in 0..pairs { + self.attribute_entry()?; + self.attribute_value()?; + } + self.schema.skip_added_fields(ATTRIBUTE, &mut self.r)?; + } + // `same_name` TRUE reuses the current name [WD 7.2.3.4]. + if !self.r.bit()? { + self.name = self.r.string()?; + } + Ok(self.name.clone()) + } + + /// `UserData` [WD 8.6]: a bit count, then that many opaque bits. + pub(crate) fn user_data(&mut self) -> Result<(), PrcError> { + let n = self.r.unsigned_integer()? as usize; + self.r.skip_bits(n) + } +} diff --git a/crates/prc/src/error.rs b/crates/prc/src/error.rs new file mode 100644 index 0000000000..c1d927a224 --- /dev/null +++ b/crates/prc/src/error.rs @@ -0,0 +1,46 @@ +//! The crate's error type. + +/// Why a PRC stream could not be read. +#[derive(Debug, Clone, PartialEq, Eq, thiserror::Error)] +#[non_exhaustive] +pub enum PrcError { + /// The data does not start with the `PRC` magic [WD 6.1.1]. + #[error("not a PRC stream: missing the \"PRC\" magic")] + NotPrc, + /// A value ran past the end of its byte or bit stream. + #[error("PRC data truncated while reading {0}")] + Truncated(&'static str), + /// A structural field holds a value the format forbids. + #[error("malformed PRC: {0}")] + Malformed(String), + /// No `Double` code of up to 22 bits matched [WD 11.17]. + #[error("PRC Double: no code matches the next 22 bits")] + UnknownDoubleCode, + /// A section is not a valid zlib stream. + #[error("PRC {section} section does not inflate: {reason}")] + Inflate { + /// The section's name. + section: &'static str, + /// The decoder's message. + reason: String, + }, + /// The inflated sections and copied blocks together exceed the ceiling + /// ([`crate::MAX_INFLATED_BYTES`] or the caller's limit). + #[error("PRC sections inflate past the {limit}-byte ceiling")] + TooLarge { + /// The ceiling that was hit. + limit: usize, + }, + /// The arrays read from one stream hold more elements than [`crate::MAX_DECODED_ELEMENTS`]. + /// One bit can decode to a 4- or 8-byte element, so the inflation ceiling alone does not + /// bound the memory the arrays take. + #[error("PRC arrays decode past the {limit}-element ceiling")] + TooManyElements { + /// The ceiling that was hit. + limit: usize, + }, + /// A construct this reader does not decode; the part of the file holding it cannot be + /// read past. + #[error("PRC {0} is not supported")] + Unsupported(&'static str), +} diff --git a/crates/prc/src/geometry.rs b/crates/prc/src/geometry.rs new file mode 100644 index 0000000000..c8d7680e05 --- /dev/null +++ b/crates/prc/src/geometry.rs @@ -0,0 +1,276 @@ +//! Model-space vectors and matrices, and their application to meshes. + +use crate::TriangleMesh; + +/// A point or direction in model space. +#[derive(Debug, Clone, Copy, PartialEq, Default)] +pub(crate) struct Vec3(pub(crate) [f64; 3]); + +impl Vec3 { + /// The right-handed cross product `self × b`. + pub(crate) fn cross(self, b: Vec3) -> Vec3 { + let ([ax, ay, az], [bx, by, bz]) = (self.0, b.0); + Vec3([ay * bz - az * by, az * bx - ax * bz, ax * by - ay * bx]) + } +} + +/// A 4×4 matrix, `m.0[row][col]`, acting on column vectors. +#[derive(Debug, Clone, Copy, PartialEq)] +pub struct Matrix(pub [[f64; 4]; 4]); + +impl Default for Matrix { + fn default() -> Self { + Matrix::IDENTITY + } +} + +impl Matrix { + /// The identity matrix. + pub const IDENTITY: Matrix = Matrix([[1.0, 0.0, 0.0, 0.0], [0.0, 1.0, 0.0, 0.0], [0.0, 0.0, 1.0, 0.0], [0.0, 0.0, 0.0, 1.0]]); + + /// The entry at `row`, `col`; 0 outside the matrix. + #[must_use] + pub fn at(&self, row: usize, col: usize) -> f64 { + self.0.get(row).and_then(|r| r.get(col)).copied().unwrap_or(0.0) + } + + /// `self × b`: `b` applied first. + #[must_use] + pub fn multiply(&self, b: &Matrix) -> Matrix { + let mut m = [[0.0; 4]; 4]; + for (i, row) in m.iter_mut().enumerate() { + for (j, cell) in row.iter_mut().enumerate() { + *cell = (0..4).map(|k| self.at(i, k) * b.at(k, j)).sum(); + } + } + Matrix(m) + } + + /// `self × (p, 1)`, divided by the homogeneous coordinate when it is neither 0 nor 1. + #[must_use] + pub fn transform_point(&self, p: [f64; 3]) -> [f64; 3] { + let [x, y, z] = p; + let row = |i: usize| self.at(i, 0) * x + self.at(i, 1) * y + self.at(i, 2) * z + self.at(i, 3); + let w = row(3); + let s = if w == 0.0 || w == 1.0 { 1.0 } else { 1.0 / w }; + [row(0) * s, row(1) * s, row(2) * s] + } + + /// The determinant of the upper-left 3×3 block. + fn det3(&self) -> f64 { + let a = |i, j| self.at(i, j); + a(0, 0) * (a(1, 1) * a(2, 2) - a(1, 2) * a(2, 1)) - a(0, 1) * (a(1, 0) * a(2, 2) - a(1, 2) * a(2, 0)) + + a(0, 2) * (a(1, 0) * a(2, 1) - a(1, 1) * a(2, 0)) + } + + /// The determinant of the whole matrix, by expansion along the last row. + fn det4(&self) -> f64 { + (0..4) + .map(|j| { + let cols: Vec = (0..4).filter(|&c| c != j).collect(); + let a = |i: usize, k: usize| cols.get(k).map_or(0.0, |&c| self.at(i, c)); + let minor = a(0, 0) * (a(1, 1) * a(2, 2) - a(1, 2) * a(2, 1)) - a(0, 1) * (a(1, 0) * a(2, 2) - a(1, 2) * a(2, 0)) + + a(0, 2) * (a(1, 0) * a(2, 1) - a(1, 1) * a(2, 0)); + let sign = if (3 + j) % 2 == 0 { 1.0 } else { -1.0 }; + sign * self.at(3, j) * minor + }) + .sum() + } + + /// Whether mapping points through this matrix turns them inside out. Where the point is + /// divided by `w`, the Jacobian's sign is the sign of the 4×4 determinant; a last row of + /// zeros (no division) falls back to the 3×3 block. Each row is first divided by its + /// largest entry, which keeps both signs while the products stay in range. + fn mirrors(&self) -> bool { + let scaled = Matrix(self.0.map(|row| { + let big = row.iter().fold(0.0_f64, |m, c| m.max(c.abs())); + if big.is_finite() && big > 0.0 { row.map(|c| c / big) } else { row } + })); + let det = scaled.det4(); + if det == 0.0 { scaled.det3() < 0.0 } else { det < 0.0 } + } + + /// Cofactor `(i, j)` of the upper-left 3×3 block (`i`, `j` < 3). + fn cofactor(&self, i: usize, j: usize) -> f64 { + let (r0, r1) = ((i + 1) % 3, (i + 2) % 3); + let (c0, c1) = ((j + 1) % 3, (j + 2) % 3); + self.at(r0, c0) * self.at(r1, c1) - self.at(r0, c1) * self.at(r1, c0) + } +} + +impl TriangleMesh { + /// This mesh with every position mapped through `m` and every stored normal through its + /// inverse transpose (renormalised); a mirroring `m` (negative 4×4 determinant, which a + /// negative `w` can cause too) reverses each triangle so the outside stays + /// counter-clockwise. A projective `m` (a last row other than + /// `[0, 0, 0, w]`) bends the surface per point, so the stored normals are dropped. + /// + /// ``` + /// use pdfcraft_prc::{Matrix, TriangleMesh}; + /// let mut mirror = Matrix::IDENTITY; + /// mirror.0[0][0] = -1.0; + /// let mut mesh = TriangleMesh::default(); + /// mesh.positions = vec![[1.0, 0.0, 0.0], [0.0, 1.0, 0.0], [0.0, 0.0, 1.0]]; + /// mesh.triangles = vec![[0, 1, 2]]; + /// let placed = mesh.transformed(&mirror); + /// assert_eq!(placed.positions, [[-1.0, 0.0, 0.0], [0.0, 1.0, 0.0], [0.0, 0.0, 1.0]]); + /// assert_eq!(placed.triangles, [[0, 2, 1]]); + /// ``` + #[must_use] + pub fn transformed(&self, m: &Matrix) -> Self { + let mut out = self.clone(); + for p in &mut out.positions { + *p = m.transform_point(*p); + } + // The cofactor matrix is det times the inverse transpose, so its sign is corrected + // and the result renormalised. + let mirrors = m.mirrors(); + let sign = if mirrors { -1.0 } else { 1.0 }; + if m.at(3, 0) != 0.0 || m.at(3, 1) != 0.0 || m.at(3, 2) != 0.0 { + out.normals.clear(); + out.triangle_normals.clear(); + } + for n in &mut out.normals { + let [x, y, z] = *n; + let v = [0, 1, 2].map(|i| sign * (m.cofactor(i, 0) * x + m.cofactor(i, 1) * y + m.cofactor(i, 2) * z)); + let len = v.iter().map(|c| c * c).sum::().sqrt(); + *n = if len > 0.0 && len.is_finite() { v.map(|c| c / len) } else { v }; + } + if mirrors { + for t in &mut out.triangles { + t.swap(1, 2); + } + for t in &mut out.triangle_normals { + t.swap(1, 2); + } + for t in out.triangle_uvs.iter_mut().flatten().flatten() { + t.swap(1, 2); + } + } + out + } +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn cross_follows_the_right_hand_rule() { + assert_eq!(Vec3([1.0, 0.0, 0.0]).cross(Vec3([0.0, 1.0, 0.0])), Vec3([0.0, 0.0, 1.0])); + assert_eq!(Vec3([0.0, 1.0, 0.0]).cross(Vec3([1.0, 0.0, 0.0])), Vec3([0.0, 0.0, -1.0])); + } + + #[test] + fn multiply_applies_the_right_operand_first() { + let mut t = Matrix::IDENTITY; + t.0[0][3] = 5.0; + let mut s = Matrix::IDENTITY; + s.0[0][0] = 2.0; + assert_eq!(t.multiply(&s).transform_point([1.0, 0.0, 0.0]), [7.0, 0.0, 0.0]); + assert_eq!(s.multiply(&t).transform_point([1.0, 0.0, 0.0]), [12.0, 0.0, 0.0]); + } + + #[test] + fn a_projective_point_is_divided_by_w() { + let mut m = Matrix::IDENTITY; + m.0[3][3] = 2.0; + assert_eq!(m.transform_point([2.0, 4.0, 6.0]), [1.0, 2.0, 3.0]); + m.0[3][3] = 0.0; + assert_eq!(m.transform_point([2.0, 4.0, 6.0]), [2.0, 4.0, 6.0]); + } + + #[test] + fn normals_take_the_inverse_transpose_and_mirror_with_their_triangle() { + // A shear in x by y turns the plane x = 0 into x = y; its x normal tilts to stay perpendicular. + let mut shear = Matrix::IDENTITY; + shear.0[0][1] = 1.0; + let mesh = TriangleMesh { + positions: vec![[0.0; 3], [1.0, 0.0, 0.0], [0.0, 0.0, 1.0]], + triangles: vec![[0, 1, 2]], + normals: vec![[1.0, 0.0, 0.0]], + triangle_normals: vec![[0, 0, 0]], + ..TriangleMesh::default() + }; + let n = shear.transformed_normal(&mesh); + let s = std::f64::consts::FRAC_1_SQRT_2; + assert!((n[0] - s).abs() < 1e-12 && (n[1] + s).abs() < 1e-12 && n[2] == 0.0, "{n:?}"); + + let mut mirror = Matrix::IDENTITY; + mirror.0[2][2] = -1.0; + let placed = mesh.transformed(&mirror); + assert_eq!(placed.triangles, [[0, 2, 1]]); + assert_eq!(placed.triangle_normals, [[0, 0, 0]]); + assert_eq!(placed.normals, [[1.0, 0.0, 0.0]]); + } + + #[test] + fn a_projective_transform_drops_the_stored_normals() { + let mesh = TriangleMesh { + positions: vec![[0.0; 3], [1.0, 0.0, 0.0], [0.0, 0.0, 1.0]], + triangles: vec![[0, 1, 2]], + normals: vec![[1.0, 0.0, 1.0]], + triangle_normals: vec![[0, 0, 0]], + ..TriangleMesh::default() + }; + let mut projective = Matrix::IDENTITY; + projective.0[3][0] = 1.0; + let placed = mesh.transformed(&projective); + assert!(placed.normals.is_empty() && placed.triangle_normals.is_empty()); + // A uniform w keeps them. + let mut uniform = Matrix::IDENTITY; + uniform.0[3][3] = 2.0; + assert_eq!(mesh.transformed(&uniform).triangle_normals, [[0, 0, 0]]); + } + + /// A negative `w` turns the mapped triangles inside out though the 3×3 block does not + /// mirror, so the winding is reversed by the sign of the whole determinant. + #[test] + fn a_negative_w_reverses_the_winding() { + let mesh = TriangleMesh { + positions: vec![[1.0, 0.0, 0.0], [0.0, 1.0, 0.0], [0.0, 0.0, 1.0]], + triangles: vec![[0, 1, 2]], + normals: vec![[1.0, 0.0, 0.0]], + triangle_normals: vec![[0, 0, 0]], + ..TriangleMesh::default() + }; + let mut negated = Matrix::IDENTITY; + negated.0[3][3] = -1.0; + let placed = mesh.transformed(&negated); + assert_eq!(placed.positions, [[-1.0, 0.0, 0.0], [0.0, -1.0, 0.0], [0.0, 0.0, -1.0]]); + assert_eq!(placed.triangles, [[0, 2, 1]]); + assert_eq!(placed.normals, [[-1.0, 0.0, 0.0]]); + // A projective w that is negative over the mesh: w = z - 3. + let mut projective = Matrix::IDENTITY; + projective.0[3][2] = 1.0; + projective.0[3][3] = -3.0; + assert_eq!(mesh.transformed(&projective).triangles, [[0, 2, 1]]); + // A positive w, uniform or projective, keeps the winding. + let mut positive = Matrix::IDENTITY; + positive.0[3][2] = 1.0; + positive.0[3][3] = 3.0; + assert_eq!(mesh.transformed(&positive).triangles, [[0, 1, 2]]); + } + + #[test] + fn a_mirror_whose_determinant_underflows_still_reverses_the_winding() { + let mesh = TriangleMesh { + positions: vec![[1.0, 0.0, 0.0], [0.0, 1.0, 0.0], [0.0, 0.0, 1.0]], + triangles: vec![[0, 1, 2]], + ..TriangleMesh::default() + }; + let diagonal = |x: f64| { + let mut m = Matrix::IDENTITY; + (m.0[0][0], m.0[1][1], m.0[2][2]) = (x, 1e-200, 1e-200); + m + }; + assert_eq!(mesh.transformed(&diagonal(-1e200)).triangles, [[0, 2, 1]]); + assert_eq!(mesh.transformed(&diagonal(1e200)).triangles, [[0, 1, 2]]); + } + + impl Matrix { + fn transformed_normal(&self, mesh: &TriangleMesh) -> [f64; 3] { + mesh.transformed(self).normals.first().copied().unwrap_or_default() + } + } +} diff --git a/crates/prc/src/lib.rs b/crates/prc/src/lib.rs new file mode 100644 index 0000000000..d594e74065 --- /dev/null +++ b/crates/prc/src/lib.rs @@ -0,0 +1,37 @@ +//! A reader for PRC, the 3D model format a PDF `/3D` stream of `/Subtype /PRC` carries +//! (ISO 32000-2 §13.6.3, ISO 14739-1). +//! +//! Layer L0, standalone: this crate depends on no other PdfCraft crate. It reads the +//! uncompressed container, inflates each section, and decodes the bit-level primitives and +//! the schema a newer producer uses to describe fields this reader does not know; it decodes +//! tessellations and walks the product tree into placed, coloured meshes +//! ([`PrcFile::placements`], [`PrcFile::model_tree`]). +//! +//! Clause numbers in `[WD x.y]` refer to the PRC working draft, ISO/TC 171/SC 2 N570. +//! +//! Every read is bounded: inflation by [`MAX_INFLATED_BYTES`] (or a caller's limit), decoded +//! array elements by [`MAX_DECODED_ELEMENTS`] and schema-program steps by [`MAX_SCHEMA_STEPS`] +//! per reading pass, schema programs by a depth ceiling. Damaged data is an error, never a panic. +#![forbid(unsafe_code)] +#![deny(clippy::unwrap_used, clippy::expect_used, clippy::panic, clippy::unimplemented, clippy::todo, clippy::unreachable)] + +mod acof; +mod arrays; +mod bits; +mod container; +mod entity; +mod error; +mod geometry; +mod schema; +mod tess; +#[cfg(test)] +mod testw; +mod tree; + +pub use bits::{BitReader, MAX_DECODED_ELEMENTS, MAX_SCHEMA_STEPS}; +pub use container::{FileStructure, MAX_FILE_STRUCTURES, MAX_INFLATED_BYTES, PRC_READER_VERSION, PrcFile, PrcHeader, SectionKind, UniqueId}; +pub use error::PrcError; +pub use geometry::Matrix; +pub use schema::Schema; +pub use tess::{Tessellation, TriangleMesh, TriangleStyle}; +pub use tree::{EntityOverrides, MAX_PLACEMENTS, ModelNode, NameSource, Placement, Rgba, StyleAlpha}; diff --git a/crates/prc/src/schema.rs b/crates/prc/src/schema.rs new file mode 100644 index 0000000000..2b505c14ea --- /dev/null +++ b/crates/prc/src/schema.rs @@ -0,0 +1,565 @@ +//! `FileStructureSchema` and its token interpreter [WD 6.3, 9]. +//! +//! A producer newer than the reader describes, per entity type, the fields it appended; the +//! reader runs that program after the fields it knows, to step over the rest [WD 9.1]. Token +//! values follow the WD 9.2 table as corrected by PDF Association pdf-issues #407 +//! (`Block_Start` = 19; 39 and 40 swallow one token). +//! +//! Choices the sources leave open, made here: +//! - `SimpleFor` reads its count as an `UnsignedInteger` [WD 9.3.17 pseudocode; pdf-issues +//! #575]: that read also decodes an `Integer`-written non-negative count, while a signed +//! read does not decode an unsigned one whose top group is 0x80 or higher. +//! - `Block_Version V` runs its block only when `V` is newer than +//! [`crate::PRC_READER_VERSION`]: an older block holds fields this reader already reads +//! natively [WD 9.3.20]. +//! - `Extent_1D` and `Extent_2D` read 2 and 4 `Double`s: `Interval` and `Domain` +//! [WD 8.2.2, 8.2.4]. +//! - Expression values are `i64`; a `Double` read as an operand truncates (NaN reads as 0). +//! - Variables are not block-scoped: a program that reuses an index after its block ends +//! reads the stale value instead of an error. +//! - Pointer tokens (12-14) and `Value_CurveIs3D` (28) need the exact-geometry reader and are +//! refused with [`PrcError::Unsupported`]. + +use std::collections::HashMap; + +use crate::PrcError; +use crate::bits::BitReader; + +/// Nesting depth of blocks, loops, expressions and `Father_Type` calls. +const MAX_DEPTH: u32 = 64; + +/// The per-entity-type programs at the start of a globals or model-file +/// bitstream [WD 6.3.1]. +#[derive(Debug, Clone, Default, PartialEq, Eq)] +pub struct Schema { + programs: HashMap>, +} + +impl Schema { + /// Read a schema: `UnsignedInteger` count, then per entry the entity + /// type, a token count and the tokens [WD 6.3.2]. + /// + /// # Errors + /// [`PrcError::Truncated`] when a count runs past the data; + /// [`PrcError::Malformed`] for an entity type listed twice; [`PrcError::TooManyElements`] + /// past the reader's allowance, one element per entry and per token. + pub fn read(r: &mut BitReader<'_>) -> Result { + let n = r.unsigned_integer()? as usize; + // Every entry takes at least two bits. + if n > r.remaining() / 2 { + return Err(PrcError::Truncated("the schema")); + } + let mut programs = HashMap::new(); + for _ in 0..n { + // The entry itself costs one element, so empty programs are not free. + r.take_elements(1)?; + let entity_type = r.unsigned_integer()?; + let len = r.unsigned_integer()? as usize; + if len > r.remaining() { + return Err(PrcError::Truncated("a schema program")); + } + r.take_elements(len)?; + let tokens = (0..len).map(|_| r.unsigned_integer()).collect::, _>>()?; + if programs.insert(entity_type, tokens).is_some() { + return Err(PrcError::Malformed(format!("schema lists entity type {entity_type} twice"))); + } + } + Ok(Self { programs }) + } + + /// True when no entity type has added fields (the common case: the file + /// was written at the version this reader implements). + pub fn is_empty(&self) -> bool { + self.programs.is_empty() + } + + /// The entity types the schema describes, unordered. + pub fn entity_types(&self) -> impl Iterator + '_ { + self.programs.keys().copied() + } + + /// Consume the fields a newer producer appended to `entity_type`, if the + /// schema describes it; a no-op otherwise. Each step is charged to `r`, so every + /// program run on one stream shares [`crate::MAX_SCHEMA_STEPS`]. + /// + /// # Errors + /// [`PrcError::Unsupported`] for a token that needs exact geometry; + /// [`PrcError::Malformed`] for an ill-formed program, one past the depth ceiling, or + /// one that runs out of the stream's steps; any read error from `r`. + pub fn skip_added_fields(&self, entity_type: u32, r: &mut BitReader<'_>) -> Result<(), PrcError> { + let Some(program) = self.programs.get(&entity_type) else { + return Ok(()); + }; + let mut exec = Exec { schema: self, r, vars: HashMap::new(), loop_counter: 0 }; + exec.program(program, 0) + } +} + +struct Exec<'s, 'r, 'd> { + schema: &'s Schema, + r: &'r mut BitReader<'d>, + vars: HashMap, + loop_counter: i64, +} + +fn malformed(what: &str) -> PrcError { + PrcError::Malformed(format!("schema program: {what}")) +} + +/// The token at `*pos`, advancing past it. +fn next(prog: &[u32], pos: &mut usize) -> Result { + let t = prog.get(*pos).copied().ok_or_else(|| malformed("ends mid-statement"))?; + *pos += 1; + Ok(t) +} + +impl Exec<'_, '_, '_> { + fn tick(&mut self, depth: u32) -> Result<(), PrcError> { + if !self.r.take_step() { + return Err(malformed("step ceiling reached")); + } + if depth > MAX_DEPTH { + return Err(malformed("nesting ceiling reached")); + } + Ok(()) + } + + fn program(&mut self, prog: &[u32], depth: u32) -> Result<(), PrcError> { + let mut pos = 0; + while pos < prog.len() { + self.stmt(prog, &mut pos, true, depth)?; + } + Ok(()) + } + + /// One data read for tokens 0-5 and 7-11, returning its value as an + /// operand. + fn read(&mut self, token: u32) -> Result { + let doubles = match token { + 0 => return Ok(i64::from(self.r.bit()?)), + 1 => return Ok(self.r.double()? as i64), + 2 => return Ok(i64::from(self.r.character()?)), + 3 => return Ok(i64::from(self.r.unsigned_integer()?)), + 4 => return Ok(i64::from(self.r.integer()?)), + 5 => { + self.r.string()?; + return Ok(0); + } + 7 => 2, + 8 => 3, + 9 => 2, + 10 => 4, + 11 => 6, + _ => return Err(malformed("not a data token")), + }; + for _ in 0..doubles { + self.r.double()?; + } + Ok(0) + } + + /// Statements until `Block_End`, which is consumed. + fn block(&mut self, prog: &[u32], pos: &mut usize, run: bool, depth: u32) -> Result<(), PrcError> { + loop { + if prog.get(*pos) == Some(&21) { + *pos += 1; + return Ok(()); + } + self.stmt(prog, pos, run, depth + 1)?; + } + } + + /// Run `body` (one statement at `*pos`) `count` times; `*pos` ends past it. + fn repeat(&mut self, prog: &[u32], pos: &mut usize, count: i64, run: bool, depth: u32) -> Result<(), PrcError> { + let start = *pos; + self.stmt(prog, pos, false, depth + 1)?; + if run { + let saved = self.loop_counter; + for i in 0..count.max(0) { + self.tick(depth)?; + self.loop_counter = i; + let mut p = start; + self.stmt(prog, &mut p, true, depth + 1)?; + } + self.loop_counter = saved; + } + Ok(()) + } + + /// One statement; with `run` false, only step over its tokens. + fn stmt(&mut self, prog: &[u32], pos: &mut usize, run: bool, depth: u32) -> Result<(), PrcError> { + self.tick(depth)?; + match next(prog, pos)? { + t @ (0..=5 | 7..=11) => { + if run { + self.read(t)?; + } + } + 6 => self.parent_program(prog, pos, run, depth)?, + 12 => { + next(prog, pos)?; + if run { + return Err(PrcError::Unsupported("a schema pointer field")); + } + } + 13 | 14 => { + if run { + return Err(PrcError::Unsupported("a schema pointer field")); + } + } + 15 => { + let count = self.expr(prog, pos, run, depth + 1)?; + self.repeat(prog, pos, count, run, depth)?; + } + 16 => { + // prc-rs: reads SimpleFor's count as an Integer; this reader deliberately reads an + // UnsignedInteger (module doc). + let count = if run { i64::from(self.r.unsigned_integer()?) } else { 0 }; + self.repeat(prog, pos, count, run, depth)?; + } + 17 => self.conditional(prog, pos, run, depth)?, + 19 => self.block(prog, pos, run, depth)?, + 20 => { + let version = next(prog, pos)?; + let newer = version > crate::PRC_READER_VERSION; + self.block(prog, pos, run && newer, depth)?; + } + 22 => { + let v = next(prog, pos)?; + if run { + self.vars.insert(v, 0); + } + } + 23 | 24 => { + let v = next(prog, pos)?; + let value = self.expr(prog, pos, run, depth + 1)?; + if run { + self.vars.insert(v, value); + } + } + 39 | 40 => { + next(prog, pos)?; + } + t => return Err(malformed(&format!("token {t} where a statement belongs"))), + } + Ok(()) + } + + /// Token 6: run the named type's own program, the fields it inherits. + fn parent_program(&mut self, prog: &[u32], pos: &mut usize, run: bool, depth: u32) -> Result<(), PrcError> { + let parent = next(prog, pos)?; + if run && let Some(p) = self.schema.programs.get(&parent) { + self.tick(depth + 1)?; + let mut q = 0; + while q < p.len() { + self.stmt(p, &mut q, true, depth + 1)?; + } + } + Ok(()) + } + + /// Token 17: `if` with an optional token-18 `else`. + fn conditional(&mut self, prog: &[u32], pos: &mut usize, run: bool, depth: u32) -> Result<(), PrcError> { + let cond = self.expr(prog, pos, run, depth + 1)? != 0; + self.stmt(prog, pos, run && cond, depth + 1)?; + if prog.get(*pos) == Some(&18) { + *pos += 1; + self.stmt(prog, pos, run && !cond, depth + 1)?; + } + Ok(()) + } + + /// One expression; with `run` false its value is 0 and nothing is read. + fn expr(&mut self, prog: &[u32], pos: &mut usize, run: bool, depth: u32) -> Result { + self.tick(depth)?; + let t = next(prog, pos)?; + match t { + 0..=4 => { + if run { + self.read(t) + } else { + Ok(0) + } + } + 25 => { + let v = next(prog, pos)?; + if !run { + return Ok(0); + } + self.vars.get(&v).copied().ok_or_else(|| malformed(&format!("variable {v} read before it is set"))) + } + 26 => Ok(i64::from(next(prog, pos)?)), + 27 => Ok(self.loop_counter), + 28 => { + if run { + Err(PrcError::Unsupported("Value_CurveIs3D")) + } else { + Ok(0) + } + } + 29..=38 => { + let a = self.expr(prog, pos, run, depth + 1)?; + let b = self.expr(prog, pos, run, depth + 1)?; + if !run { + return Ok(0); + } + let overflow = || malformed("a schema expression overflows"); + Ok(match t { + 29 => a.checked_mul(b).ok_or_else(overflow)?, + 30 => a.checked_div(b).ok_or_else(|| malformed("division by zero"))?, + 31 => a.checked_add(b).ok_or_else(overflow)?, + 32 => a.checked_sub(b).ok_or_else(overflow)?, + 33 => i64::from(a < b), + 34 => i64::from(a <= b), + 35 => i64::from(a > b), + 36 => i64::from(a >= b), + 37 => i64::from(a == b), + _ => i64::from(a != b), + }) + } + t => Err(malformed(&format!("token {t} where a value belongs"))), + } + } +} + +#[cfg(test)] +#[allow(clippy::unwrap_used, clippy::indexing_slicing)] +mod tests { + use super::*; + use crate::testw::W; + + fn schema(entries: &[(u32, &[u32])]) -> W { + let mut w = W::default(); + w.uint(entries.len() as u32); + for (t, toks) in entries { + w.uint(*t).uint(toks.len() as u32); + for k in *toks { + w.uint(*k); + } + } + w + } + + /// Each schema entry and its program's tokens are charged against the reader's element + /// allowance. + #[test] + fn program_tokens_are_charged_against_the_allowance() { + let bytes = schema(&[(172, &[1, 2, 3])]).bytes(); + assert!(matches!(Schema::read(&mut BitReader::with_elements(&bytes, 3)), Err(PrcError::TooManyElements { .. }))); + assert!(Schema::read(&mut BitReader::with_elements(&bytes, 4)).is_ok()); + let empty = schema(&[(172, &[]), (173, &[]), (174, &[])]).bytes(); + assert!(matches!(Schema::read(&mut BitReader::with_elements(&empty, 2)), Err(PrcError::TooManyElements { .. }))); + assert!(Schema::read(&mut BitReader::with_elements(&empty, 3)).is_ok()); + } + + fn run(entries: &[(u32, &[u32])], data: &W) -> (Result<(), PrcError>, usize) { + let mut w = schema(entries); + w.append(data); + let bytes = w.bytes(); + let mut r = BitReader::new(&bytes); + let s = Schema::read(&mut r).unwrap(); + let start = r.position(); + let res = s.skip_added_fields(172, &mut r); + (res, r.position() - start) + } + + #[test] + fn integer_string_and_boolean_tokens_consume_their_fields() { + let mut d = W::default(); + d.bit(true).uint(300).int(-7).string(Some("ab")).put(0x41, 8); + let want = d.len(); + d.uint(99); // not part of the added fields + let (res, used) = run(&[(172, &[0, 3, 4, 5, 2])], &d); + res.unwrap(); + assert_eq!(used, want); + } + + #[test] + fn data_tokens_consume_their_fields() { + let mut d = W::default(); + d.bit(true).double(1.5).uint(300).int(-7).string(Some("ab")).double(2.0).double(3.0); + let want = d.len(); + d.uint(99); // not part of the added fields + let (res, used) = run(&[(172, &[0, 1, 3, 4, 5, 7])], &d); + res.unwrap(); + assert_eq!(used, want); + } + + #[test] + fn vector_extent_and_box_tokens_read_their_doubles() { + // Vec3, Extent1D, Extent2D, BBox: 3 + 2 + 4 + 6 Doubles. + let mut d = W::default(); + for i in 0..15 { + d.double(f64::from(i) + 0.5); + } + let want = d.len(); + d.uint(99); + let (res, used) = run(&[(172, &[8, 9, 10, 11])], &d); + res.unwrap(); + assert_eq!(used, want); + } + + #[test] + fn arithmetic_and_comparison_operators() { + // For (5 - 3) { UInt }, then If (2 < 2) { Bool } Else { Char }. + let mut d = W::default(); + d.uint(7).uint(8).put(0x41, 8); + let want = d.len(); + d.uint(99); + let prog = [15, 32, 26, 5, 26, 3, 3, 17, 33, 26, 2, 26, 2, 0, 18, 2]; + let (res, used) = run(&[(172, &prog)], &d); + res.unwrap(); + assert_eq!(used, want); + } + + #[test] + fn an_absent_type_is_a_no_op() { + let mut w = schema(&[(300, &[3])]); + w.uint(5); + let bytes = w.bytes(); + let mut r = BitReader::new(&bytes); + let s = Schema::read(&mut r).unwrap(); + let at = r.position(); + s.skip_added_fields(172, &mut r).unwrap(); + assert_eq!(r.position(), at); + assert!(!s.is_empty()); + } + + #[test] + fn simple_for_reads_its_count_then_repeats_the_body() { + let mut d = W::default(); + d.int(3).uint(1).uint(2).uint(3); + let want = d.len(); + d.uint(77); + let (res, used) = run(&[(172, &[16, 3])], &d); + res.unwrap(); + assert_eq!(used, want); + + // A block body. + let mut d = W::default(); + d.int(2).bit(true).uint(9).bit(false).uint(8); + let want = d.len(); + let (res, used) = run(&[(172, &[16, 19, 0, 3, 21])], &d); + res.unwrap(); + assert_eq!(used, want); + + // A count whose top byte is 0x80 or higher, written either way. + for signed in [false, true] { + let mut d = W::default(); + if signed { + d.int(200); + } else { + d.uint(200); + } + (0..200).for_each(|_| { + d.bit(true); + }); + let want = d.len(); + d.uint(77); + let (res, used) = run(&[(172, &[16, 0])], &d); + res.unwrap(); + assert_eq!(used, want, "signed={signed}"); + } + } + + #[test] + fn if_else_and_variables_choose_the_branch_by_file_data() { + // DeclareAndSet v0 = ; if v0 > 1 { UInt } else { Character }. + let prog: &[u32] = &[24, 0, 3, 17, 35, 25, 0, 26, 1, 3, 18, 2]; + let mut d = W::default(); + d.uint(5).uint(4250); + let want = d.len(); + let (res, used) = run(&[(172, prog)], &d); + res.unwrap(); + assert_eq!(used, want); + + let mut d = W::default(); + d.uint(0).put(0xAB, 8); + let want = d.len(); + let (res, used) = run(&[(172, prog)], &d); + res.unwrap(); + assert_eq!(used, want); + } + + #[test] + fn for_counts_by_expression_and_exposes_the_counter() { + // For (2 * 2) { if counter == 2 { UInt } } => one UInt read. + let prog: &[u32] = &[15, 29, 26, 2, 26, 2, 17, 37, 27, 26, 2, 3]; + let mut d = W::default(); + d.uint(1234); + let want = d.len(); + let (res, used) = run(&[(172, prog)], &d); + res.unwrap(); + assert_eq!(used, want); + } + + #[test] + fn an_expression_that_overflows_is_malformed() { + // For ((2^31 * 2^31) * 4) { UInt }: 2^64 would wrap to 0 and read nothing. + let prog: &[u32] = &[15, 29, 29, 26, 1 << 31, 26, 1 << 31, 26, 4, 3]; + let (res, _) = run(&[(172, prog)], &W::default()); + assert!(matches!(res, Err(PrcError::Malformed(m)) if m.contains("overflows"))); + } + + #[test] + fn version_blocks_run_only_when_newer_than_the_reader() { + let mut d = W::default(); + d.uint(42); + let want = d.len(); + let (res, used) = run(&[(172, &[20, 8137, 3, 21, 20, 9000, 3, 21])], &d); + res.unwrap(); + assert_eq!(used, want); + } + + #[test] + fn father_type_runs_the_parent_program_and_obsolete_tokens_swallow_one() { + let mut d = W::default(); + d.uint(1).uint(2); + let want = d.len(); + let (res, used) = run(&[(172, &[39, 3, 6, 50, 40, 8]), (50, &[3, 3])], &d); + res.unwrap(); + assert_eq!(used, want); + } + + #[test] + fn hostile_programs_are_refused_not_run_forever() { + let d = W::default(); + // For (1000000 * 1000000) { empty block }. + let prog: &[u32] = &[15, 29, 26, 1_000_000, 26, 1_000_000, 19, 21]; + assert!(matches!(run(&[(172, prog)], &d).0, Err(PrcError::Malformed(_)))); + // Father_Type recursing into itself. + assert!(matches!(run(&[(172, &[6, 172])], &d).0, Err(PrcError::Malformed(_)))); + // A pointer field. + assert!(matches!(run(&[(172, &[13])], &d).0, Err(PrcError::Unsupported(_)))); + // A stray Block_End, an unterminated block, a division by zero. + assert!(matches!(run(&[(172, &[21])], &d).0, Err(PrcError::Malformed(_)))); + assert!(matches!(run(&[(172, &[19, 0])], &d).0, Err(PrcError::Malformed(_)))); + assert!(matches!(run(&[(172, &[23, 0, 30, 26, 1, 26, 0])], &d).0, Err(PrcError::Malformed(_)))); + } + + /// Every program run on one reader draws on one step allowance. + #[test] + fn programs_on_one_stream_share_the_step_allowance() { + // For (2^19) { empty block }: about 2^20 steps, so the 2^22 allowance fits three runs. + let bytes = schema(&[(172, &[15, 26, 1 << 19, 19, 21])]).bytes(); + let mut r = BitReader::new(&bytes); + let s = Schema::read(&mut r).unwrap(); + let runs = (0..5).take_while(|_| s.skip_added_fields(172, &mut r).is_ok()).count(); + assert_eq!(runs, 3); + assert!(matches!(s.skip_added_fields(172, &mut r), Err(PrcError::Malformed(m)) if m.contains("step ceiling"))); + } + + #[test] + fn a_duplicated_entity_type_is_malformed() { + let bytes = schema(&[(172, &[3]), (172, &[3])]).bytes(); + assert!(matches!(Schema::read(&mut BitReader::new(&bytes)), Err(PrcError::Malformed(_)))); + } + + #[test] + fn an_entry_count_past_the_data_is_refused_before_any_entry_is_read() { + let mut w = W::default(); + w.uint(1_000).uint(172).uint(0); + let bytes = w.bytes(); + assert_eq!(Schema::read(&mut BitReader::new(&bytes)).unwrap_err(), PrcError::Truncated("the schema")); + } +} diff --git a/crates/prc/src/tess.rs b/crates/prc/src/tess.rs new file mode 100644 index 0000000000..c9bffb30f4 --- /dev/null +++ b/crates/prc/src/tess.rs @@ -0,0 +1,440 @@ +//! The tessellation section: `FileStructureTessellation` (305) and its entities, `TESS_3D` +//! (172) with its `TESS_Face`s (174), `TESS_3D_Wire` (175), `TESS_Markup` (176) and +//! `TESS_3D_Compressed` (173) [WD 7.3.7, 7.8]. +//! +//! `TESS_3D_Compressed` is read field by field to its end [WD 7.8.9.7] so the entities after +//! it decode; its triangles are not rebuilt. +//! +//! The schema runs after the fields of each concrete type read here (201, 172, 173, 174, 175, +//! 176, 305), not for the abstract levels (`PRCBase`, `TESS`). + +mod triangulate; + +use crate::arrays; +use crate::bits::{BitReader, Budget}; +use crate::entity::{Ctx, malformed}; +use crate::{FileStructure, PrcError, SectionKind}; +use triangulate::{Corners, Face, triangle_elements}; + +const FILE_STRUCTURE_TESSELLATION: u32 = 305; +const TESS_3D: u32 = 172; +const TESS_3D_COMPRESSED: u32 = 173; +const TESS_FACE: u32 = 174; +const TESS_3D_WIRE: u32 = 175; +const TESS_MARKUP: u32 = 176; + +/// prc-rs: `origin` exists in a compressed tessellation from this version (pdf-issues #705). +const ORIGIN_FROM: u32 = 7031; +/// prc-rs: `has_loops` exists from this file-structure authoring version. +const HAS_LOOPS_FROM: u32 = 7039; +/// prc-rs: `must_recalculate_normals` exists from this version. +const RECALCULATE_FROM: u32 = 7047; + +/// A wire header's count bits; the top four are flags [WD 7.8.7.5]. +const WIRE_COUNT_MASK: u32 = 0x0FFF_FFFF; +const WIRE_IS_CLOSING: u32 = 0x1000_0000; +const WIRE_IS_CONTINUOUS: u32 = 0x2000_0000; + +/// Normal angles wider than this are not a `ShortArray` [WD 7.8.9.7]. +const MAX_ANGLE_BITS: u32 = 16; + +/// One entry of the tessellation array; a representation item's `index_tessellation` selects +/// one by position. +#[derive(Debug, Clone, PartialEq)] +#[non_exhaustive] +pub enum Tessellation { + /// A `TESS_3D` triangle mesh. + Mesh(TriangleMesh), + /// A `TESS_3D_Wire`: polylines. + Wire(Vec>), + /// A `TESS_Markup` (PMI drawing), read past but not decoded. + Markup, + /// A `TESS_3D_Compressed` mesh, read past but not rebuilt. + #[non_exhaustive] + Compressed { + /// The entity's triangle count. + triangles: usize, + }, +} + +/// A `TESS_3D` decoded to indexed triangles, in the entity's own frame (the representation +/// item's placement is not applied). +#[derive(Debug, Clone, Default, PartialEq)] +#[non_exhaustive] +pub struct TriangleMesh { + /// The coordinate array as points. + pub positions: Vec<[f64; 3]>, + /// Counter-clockwise triangles seen from the outside, as indices into + /// [`Self::positions`] [WD 7.8.5.2]. + pub triangles: Vec<[u32; 3]>, + /// `triangles` split by `TESS_Face`: face `i` is `triangles[faces[i].clone()]`. + pub faces: Vec>, + /// The producer asks the reader to compute normals (none are stored). + pub normals_recalculated: bool, + /// The stored vertex normals [WD 7.8.5.1], as written (not renormalised); empty when the + /// mesh stores none or any of its normal indices is unusable. + pub normals: Vec<[f64; 3]>, + /// Per triangle, the index into [`Self::normals`] of each corner's normal, in the corner + /// order of [`Self::triangles`]; empty exactly when [`Self::normals`] is. + pub triangle_normals: Vec<[u32; 3]>, + /// Per triangle, the style its face's line attributes give it [WD 7.8.6]; empty when no + /// face carries any. + pub triangle_styles: Vec, + /// The stored texture coordinates as (u, v) pairs [WD 7.8.5.1]; empty when the mesh + /// stores none. + pub uvs: Vec<[f64; 2]>, + /// Per texture-coordinate set, per triangle, the index into [`Self::uvs`] of each + /// corner's coordinate, in the corner order of [`Self::triangles`]; each set is as long + /// as [`Self::triangles`], `None` for a triangle whose face stores no coordinate for the + /// set. Empty when [`Self::uvs`] is. + pub triangle_uvs: Vec>>, +} + +/// The graphics a face's `line_attributes` give one triangle [WD 7.8.6]. +#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)] +#[non_exhaustive] +pub struct TriangleStyle { + /// An index into the globals' style array plus one; 0 keeps the owner's graphics. + pub style: u32, + /// The face's behaviour bits [WD 5.4]. + pub behaviour: u16, +} + +impl FileStructure { + /// The tessellation section's entities, in the order a representation item's + /// `index_tessellation` counts them [WD 7.3.7]. + /// + /// # Errors + /// As [`crate::Schema::read`] for the globals' schema; [`PrcError::Truncated`] or + /// [`PrcError::Malformed`] for a damaged section; [`PrcError::Unsupported`] for + /// optimised vertex colours. + pub fn tessellations(&self) -> Result, PrcError> { + self.tessellations_within(Budget::FULL) + } + + /// As [`Self::tessellations`], the globals' schema and the section sharing `budget`. + pub(crate) fn tessellations_within(&self, budget: Budget) -> Result, PrcError> { + let mut globals = BitReader::with_budget(self.section(SectionKind::Globals), budget); + let schema = crate::Schema::read(&mut globals)?; + let r = BitReader::with_budget(self.section(SectionKind::Tessellation), globals.budget()); + Ctx::new(r, &schema, self.authoring_version).file_structure_tessellation() + } +} + +impl Ctx<'_, '_> { + /// `FileStructureTessellation` (305) [WD 7.3.7]. + pub(crate) fn file_structure_tessellation(&mut self) -> Result, PrcError> { + self.expect_type(FILE_STRUCTURE_TESSELLATION)?; + self.content_prc_base()?; + let n = self.count(1, "tessellations")?; + let mut out = Vec::with_capacity(n.min(1024)); + for _ in 0..n { + out.push(match self.r.unsigned_integer()? { + TESS_3D => Tessellation::Mesh(self.tess_3d()?), + TESS_3D_WIRE => Tessellation::Wire(self.wire()?), + TESS_MARKUP => { + self.markup()?; + Tessellation::Markup + } + TESS_3D_COMPRESSED => Tessellation::Compressed { triangles: self.tess_3d_compressed()? }, + t => return Err(malformed(format!("entity type {t} in the tessellation array"))), + }); + } + self.schema.skip_added_fields(FILE_STRUCTURE_TESSELLATION, &mut self.r)?; + self.user_data()?; + Ok(out) + } + + /// `ContentBaseTessData` [WD 7.8.4]: `is_calculated`, then coordinates. + fn base_tess_data(&mut self) -> Result, PrcError> { + self.r.bit()?; + self.doubles("tessellation coordinates") + } + + /// `VertexColors` [WD 7.8.7.2] for `count` points, read past. + fn vertex_colors(&mut self, count: usize, in_wire: bool) -> Result<(), PrcError> { + let rgba = self.r.bit()?; + // prc-rs: `is_segment_color` exists only inside a wire. + let per_segment = in_wire && self.r.bit()?; + if self.r.bit()? { + return Err(PrcError::Unsupported("optimised vertex colours")); + } + let count = if per_segment { count / 2 } else { count }; + let bytes = if rgba { 4 } else { 3 }; + // The first colour is `8 * bytes` bits; every later one is at least its reuse bit. + if count > 0 && (count - 1).saturating_add(8 * bytes) > self.r.remaining() { + return Err(PrcError::Truncated("vertex colours")); + } + for i in 0..count { + if i == 0 || !self.r.bit()? { + for _ in 0..bytes { + self.r.character()?; + } + } + } + Ok(()) + } + + /// `TESS_3D` (172) after its type code [WD 7.8.5]. + fn tess_3d(&mut self) -> Result { + let coords = self.base_tess_data()?; + let _has_faces = self.r.bit()?; + if self.version >= HAS_LOOPS_FROM { + let _has_loops = self.r.bit()?; + } + let recalc = self.version >= RECALCULATE_FROM && self.r.bit()?; + if recalc { + let _flags = self.r.character()?; + let _crease_angle = self.r.double()?; + } + let normal_coords = self.doubles("normal coordinates")?; + let _wire_indices = self.uints("wire indices")?; + let indices = self.uints("triangulated indices")?; + let n_faces = self.count(1, "faces")?; + let mut faces = Vec::with_capacity(n_faces.min(1024)); + for _ in 0..n_faces { + self.expect_type(TESS_FACE)?; + faces.push(self.face()?); + } + let texture = self.doubles("texture coordinates")?; + self.schema.skip_added_fields(TESS_3D, &mut self.r)?; + + let mut mesh = TriangleMesh { positions: points(&coords)?, normals_recalculated: recalc, ..TriangleMesh::default() }; + let styled = faces.iter().any(|f| !f.styles.is_empty()); + let mut corners = Corners::default(); + // Every triangle carries a corner triple for each texture set any face uses. + let sets = faces.iter().map(Face::texture_sets).max().unwrap_or(0); + let per_triangle = triangle_elements(!recalc, sets, styled); + for f in &faces { + // Faces may reuse one index range, so each face's triangles, with what they + // carry, are charged before they are made. + self.r.take_elements(f.triangle_count()?.saturating_mul(per_triangle))?; + let start = mesh.triangles.len(); + let entities = f.triangulate(&indices, !recalc, mesh.positions.len(), &mut mesh.triangles, &mut corners)?; + mesh.faces.push(start..mesh.triangles.len()); + if styled { + f.styles_into(&entities, &mut mesh.triangle_styles); + } + } + (mesh.normals, mesh.triangle_normals) = stored_normals(&normal_coords, &corners.normals, mesh.triangles.len()); + (mesh.uvs, mesh.triangle_uvs) = stored_uvs(&texture, corners.textures); + Ok(mesh) + } + + /// `TESS_Face` (174) after its type code [WD 7.8.6]. + fn face(&mut self) -> Result { + let styles = self.uints("face line attributes")?; + let _start_of_wire = self.r.unsigned_integer()?; + let _sizes_wire = self.uints("face wire sizes")?; + let flags = self.r.unsigned_integer()?; + let start = self.r.unsigned_integer()? as usize; + let data = self.uints("triangulated data")?; + let textures = self.r.unsigned_integer()? as usize; + let mut face = Face { flags, start, data, textures, styles, behaviour: 0 }; + if self.r.bit()? { + let n = face.point_count()?; + self.vertex_colors(n, false)?; + } + if !face.styles.is_empty() { + let bits = self.r.unsigned_integer()?; + face.behaviour = u16::try_from(bits).map_err(|_| malformed(format!("face behaviour {bits} is wider than 16 bits")))?; + } + self.schema.skip_added_fields(TESS_FACE, &mut self.r)?; + Ok(face) + } + + /// `TESS_3D_Wire` (175) after its type code [WD 7.8.7]. + fn wire(&mut self) -> Result>, PrcError> { + let pts = points(&self.base_tess_data()?)?; + let words = self.uints("wire indices")?; + let mut wires: Vec> = Vec::new(); + let mut colours = 0usize; + if words.is_empty() { + colours = pts.len(); + if !pts.is_empty() { + wires.push(pts.clone()); + } + } + let mut it = words.iter().copied(); + while let Some(header) = it.next() { + let n = (header & WIRE_COUNT_MASK) as usize; + let mut wire = Vec::with_capacity(n.min(words.len())); + for _ in 0..n { + let idx = it.next().ok_or_else(|| malformed("wire runs past its indices".into()))?; + wire.push(point_at(&pts, idx)?); + } + colours = colours.saturating_add(n); + if header & WIRE_IS_CLOSING != 0 { + colours = colours.saturating_add(1); + if let Some(&first) = wire.first() { + wire.push(first); + } + } + match wires.last_mut() { + Some(prev) if header & WIRE_IS_CONTINUOUS != 0 => prev.extend(wire), + _ => wires.push(wire), + } + } + if self.r.bit()? { + self.vertex_colors(colours, true)?; + } + self.schema.skip_added_fields(TESS_3D_WIRE, &mut self.r)?; + Ok(wires) + } + + /// `TESS_Markup` (176) after its type code [WD 7.8.8], read past. + fn markup(&mut self) -> Result<(), PrcError> { + self.base_tess_data()?; + self.uints("markup codes")?; + let n = self.count(1, "markup strings")?; + for _ in 0..n { + self.r.string()?; + } + self.r.string()?; + self.r.character()?; + self.schema.skip_added_fields(TESS_MARKUP, &mut self.r) + } + + /// `TESS_3D_Compressed` (173) after its type code [WD 7.8.9.7], read past; returns its + /// triangle count. prc-rs: the field order and version gates follow its schema. + /// + /// Sizes the WD leaves unstated: the face count is the largest face index + 1, + /// `normal_is_reversed` is one bit per triangle, and the point-reference array's + /// compressed flag is implicit (more than three references). + fn tess_3d_compressed(&mut self) -> Result { + let r = &mut self.r; + r.bit()?; // is_calculated + r.bit()?; // has_faces + r.double()?; // tolerance + if self.version >= ORIGIN_FROM { + for _ in 0..3 { + r.float_as_bytes()?; + } + } + arrays::compressed_integer_array(r)?; // point_array + let edge_status = arrays::character_array(r, 2, None, false)?; + let face_of = arrays::compressed_indice_array(r, None)?; + let t = face_of.len(); + if edge_status.len() != t && edge_status.len() != t.saturating_mul(3) { + return Err(malformed(format!("{} edge statuses for {t} triangles", edge_status.len()))); + } + let faces = face_of.iter().max().map_or(0, |&m| (m as usize).saturating_add(1)); + let n = r.unsigned_integer()? as usize; + let references = arrays::bool_array(r, n)?.iter().filter(|&&b| b).count(); + let refs = arrays::compressed_indice_array(r, Some(references > 3))?; + if refs.len() != references { + return Err(malformed(format!("{} point references for {references} flagged points", refs.len()))); + } + self.compressed_normals(t, faces)?; + let r = &mut self.r; + if r.bit()? { + arrays::bool_array(r, faces)?; // is_point_color_on_face + arrays::character_array(r, 8, None, false)?; // point_color_array + } + if r.bit()? { + arrays::bool_array(r, faces)?; // is_multiple_line_attribute_on_face + } + arrays::short_array(r, 16)?; // line_attributes + if !r.bit()? { + self.compressed_texture_parameter()?; + if !self.r.bit()? { + arrays::bool_array(&mut self.r, faces)?; // face_has_texture + } + } + if self.r.bit()? { + arrays::character_array(&mut self.r, 8, None, false)?; // behaviours + } + self.schema.skip_added_fields(TESS_3D_COMPRESSED, &mut self.r)?; + Ok(t) + } + + /// A compressed tessellation's normals, from `must_recalculate_normals` on, read past. + fn compressed_normals(&mut self, triangles: usize, faces: usize) -> Result<(), PrcError> { + let r = &mut self.r; + if r.bit()? { + arrays::bool_array(r, triangles)?; // normal_is_reversed + r.double()?; // crease_angle + r.character()?; // normal_recalculation_flags + return Ok(()); + } + let angle_bits = u32::from(r.character()?); + if angle_bits > MAX_ANGLE_BITS { + return Err(malformed(format!("{angle_bits}-bit normal angles"))); + } + let n = r.unsigned_integer()? as usize; + arrays::bool_array(r, n)?; // is_binary_normal + arrays::short_array(r, angle_bits)?; // normal angles + arrays::bool_array(r, faces)?; // is_face_planar + Ok(()) + } + + /// `CompressedTextureParameter` [WD 7.8.9.8-7.8.9.9; pdf-issues #729, #749], read past. + fn compressed_texture_parameter(&mut self) -> Result<(), PrcError> { + let words = self.count(32, "texture data")?; + // count() bounded words * 32 by the remaining bits. + self.r.skip_bits(words.saturating_mul(32))?; + self.r.unsigned_integer()?; // last_integer_used_bit_number + let n = self.count(5, "texture references")?; + for _ in 0..n { + self.r.nbits_then_unsigned()?; + } + self.r.double()?; // tolerance + let n = self.count(32, "texture parameters")?; + for _ in 0..n { + self.r.float_as_bytes()?; + } + Ok(()) + } +} + +/// The normal array and each triangle's corner normals, or both empty when any slot fails to +/// name a normal: a mesh shades from its stored normals wholly or not at all. +fn stored_normals(coords: &[f64], slots: &[[u32; 3]], triangles: usize) -> (Vec<[f64; 3]>, Vec<[u32; 3]>) { + // Like a slot naming no normal, a partial last normal drops the stored normals. + let Ok(normals) = points(coords) else { + return (Vec::new(), Vec::new()); + }; + let named = |k: u32| k.is_multiple_of(3) && (k as usize / 3) < normals.len(); + if normals.is_empty() || slots.len() != triangles || !slots.iter().flatten().all(|&k| named(k)) { + return (Vec::new(), Vec::new()); + } + let corners = slots.iter().map(|t| t.map(|k| k / 3)).collect(); + (normals, corners) +} + +/// Per texture-coordinate set, each triangle's corner pair indices. +type UvSets = Vec>>; + +/// The texture coordinates as (u, v) pairs, and per set the corner slots turned into pair +/// indices (a slot is a `Double` offset, so even); a triangle with a slot naming no pair is +/// `None`. +fn stored_uvs(coords: &[f64], sets: UvSets) -> (Vec<[f64; 2]>, UvSets) { + let uvs: Vec<[f64; 2]> = coords.as_chunks::<2>().0.to_vec(); + if uvs.is_empty() { + return (Vec::new(), Vec::new()); + } + let pair = |k: u32| (k.is_multiple_of(2) && (k as usize / 2) < uvs.len()).then_some(k / 2); + let sets = + sets.into_iter().map(|set| set.into_iter().map(|tri| tri.and_then(|[a, b, c]| Some([pair(a)?, pair(b)?, pair(c)?]))).collect()).collect(); + (uvs, sets) +} + +/// `coords` as points; an array that does not divide into whole points is malformed. +fn points(coords: &[f64]) -> Result, PrcError> { + let (whole, rest) = coords.as_chunks::<3>(); + if !rest.is_empty() { + return Err(malformed(format!("{} coordinates are not whole points", coords.len()))); + } + Ok(whole.to_vec()) +} + +/// The point a `Double`-offset index names [WD 7.8.5.2: multiples of 3]. +fn point_at(pts: &[[f64; 3]], idx: u32) -> Result<[f64; 3], PrcError> { + if !idx.is_multiple_of(3) { + return Err(malformed(format!("point index {idx} is not a multiple of 3"))); + } + pts.get(idx as usize / 3).copied().ok_or_else(|| malformed(format!("point index {idx} past the coordinates"))) +} + +#[cfg(test)] +pub(crate) mod tests; diff --git a/crates/prc/src/tess/tests.rs b/crates/prc/src/tess/tests.rs new file mode 100644 index 0000000000..26219f2865 --- /dev/null +++ b/crates/prc/src/tess/tests.rs @@ -0,0 +1,738 @@ +#![allow(clippy::unwrap_used, clippy::expect_used, clippy::panic)] + +use super::triangulate::MAX_TEXTURE_SETS; +use super::*; +use crate::Schema; +use crate::testw::W; + +const SQUARE: [f64; 12] = [0., 0., 0., 1., 0., 0., 1., 1., 0., 0., 1., 0.]; +const ATTRIBUTE: u32 = 201; +const NORMAL_SINGLE: u32 = 0x4000_0000; + +/// A 305 section holding `body` (already-written entities, `n` of them). +fn section(n: u32, body: &W) -> W { + let mut w = W::default(); + w.uint(305).uint(0).bit(true).uint(n).append(body).uint(0); + w +} + +fn uints(w: &mut W, v: &[u32]) { + w.uint(u32::try_from(v.len()).unwrap()); + for &x in v { + w.uint(x); + } +} + +fn square(w: &mut W, entity: u32) { + coordinates(w, entity, &SQUARE); +} + +fn coordinates(w: &mut W, entity: u32, coords: &[f64]) { + w.uint(entity).bit(false).uint(u32::try_from(coords.len()).unwrap()); + for &c in coords { + w.double(c); + } +} + +#[derive(Default)] +struct Mesh<'a> { + recalc: bool, + indices: &'a [u32], + faces: &'a [(u32, u32, &'a [u32])], + textures: u32, + uvs: &'a [f64], + /// The coordinates; [`SQUARE`] when `None`. + coords: Option<&'a [f64]>, +} + +/// A TESS_3D over [`SQUARE`] at authoring version `v`. +fn tess_3d(w: &mut W, v: u32, m: &Mesh<'_>) { + coordinates(w, 172, m.coords.unwrap_or(&SQUARE)); + w.bit(true); + if v >= 7039 { + w.bit(false); + } + if v >= 7047 { + w.bit(m.recalc); + if m.recalc { + w.put(0, 8).double(0.5); + } + } + if m.recalc { + w.uint(0); + } else { + // Two normals: slot 0 is +z, slot 3 is -z. + w.uint(6); + for c in [0., 0., 1., 0., 0., -1.] { + w.double(c); + } + } + w.uint(0); + uints(w, m.indices); + w.uint(u32::try_from(m.faces.len()).unwrap()); + for &(flags, start, data) in m.faces { + w.uint(174).uint(0).uint(0).uint(0).uint(flags).uint(start); + uints(w, data); + w.uint(m.textures).bit(false); + } + w.uint(u32::try_from(m.uvs.len()).unwrap()); + for &c in m.uvs { + w.double(c); + } +} + +fn decode(w: &W, schema: &Schema, version: u32) -> Result, PrcError> { + let bytes = w.bytes(); + Ctx::new(BitReader::new(&bytes), schema, version).file_structure_tessellation() +} + +fn mesh(t: &Tessellation) -> &TriangleMesh { + match t { + Tessellation::Mesh(m) => m, + other => panic!("not a mesh: {other:?}"), + } +} + +fn one(indices: &[u32], faces: &[(u32, u32, &[u32])], textures: u32) -> TriangleMesh { + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { indices, faces, textures, ..Mesh::default() }); + mesh(&decode(§ion(1, &b), &Schema::default(), 8137).unwrap()[0]).clone() +} + +#[test] +fn triangle_and_fan_with_normals() { + // One triangle (n,p x3), then a fan of 4 vertices (n,p each). + let m = one(&[3, 0, 0, 3, 3, 6, 0, 0, 3, 3, 0, 6, 3, 9], &[(0x2 | 0x4, 0, &[1, 1, 4])], 0); + assert_eq!(m.positions.len(), 4); + assert_eq!(m.positions[2], [1., 1., 0.]); + assert_eq!(m.triangles, [[0, 1, 2], [0, 1, 2], [0, 2, 3]]); + assert_eq!(m.faces, vec![0..3]); + assert!(!m.normals_recalculated); + assert_eq!(m.normals, [[0., 0., 1.], [0., 0., -1.]]); + // Each corner keeps the normal stored beside its point. + assert_eq!(m.triangle_normals, [[1, 0, 1], [0, 1, 0], [0, 0, 1]]); +} + +#[test] +fn strip_normals_follow_the_alternating_winding() { + let m = one(&[3, 0, 0, 3, 3, 6, 0, 9], &[(0x8, 0, &[1, 4])], 0); + assert_eq!(m.triangles, [[0, 1, 2], [2, 1, 3]]); + assert_eq!(m.triangle_normals, [[1, 0, 1], [1, 0, 0]]); +} + +#[test] +fn a_normal_index_past_the_array_drops_every_normal_but_no_triangle() { + let m = one(&[0, 0, 6, 3, 0, 6], &[(0x2, 0, &[1])], 0); + assert_eq!(m.triangles, [[0, 1, 2]]); + assert!(m.normals.is_empty()); + assert!(m.triangle_normals.is_empty()); +} + +#[test] +fn recalculated_normals_drop_the_normal_slots_and_strips_alternate() { + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { recalc: true, indices: &[0, 3, 6, 9], faces: &[(0x8, 0, &[1, 4])], ..Mesh::default() }); + let t = decode(§ion(1, &b), &Schema::default(), 8137).unwrap(); + let m = mesh(&t[0]); + assert!(m.normals_recalculated); + assert_eq!(m.triangles, [[0, 1, 2], [2, 1, 3]]); + assert!(m.normals.is_empty() && m.triangle_normals.is_empty()); +} + +#[test] +fn version_gates_the_loop_and_recalculation_flags() { + for v in [7038, 7039, 7046, 7047, 8137] { + let mut b = W::default(); + tess_3d(&mut b, v, &Mesh { indices: &[0, 0, 0, 3, 0, 6], faces: &[(0x2, 0, &[1])], ..Mesh::default() }); + // A trailing wire: a misread flag desyncs it. + wire(&mut b, false); + let t = decode(§ion(2, &b), &Schema::default(), v).unwrap(); + assert_eq!(mesh(&t[0]).triangles, [[0, 1, 2]], "version {v}"); + assert!(matches!(t[1], Tessellation::Wire(_)), "version {v}"); + } +} + +#[test] +fn one_normal_blocks_and_textures() { + // 0x20: n,p,p,p. 0x40 with NORMAL_Single: n then p per vertex. + // 0x2000 with one texture set: n,t,p,t,p,t,p. + let idx = [ + 0, 0, 3, 6, // 0x20 + 3, 0, 6, 9, // 0x40 single, 3 vertices + 0, 0, 9, 2, 0, 4, 3, // 0x2000 + ]; + let m = one(&idx, &[(0x20 | 0x40 | 0x2000, 0, &[1, 1, 3 | NORMAL_SINGLE, 1])], 1); + assert_eq!(m.triangles, [[0, 1, 2], [0, 2, 3], [3, 0, 1]]); + // A one-normal entity gives every corner its single normal. + assert_eq!(m.triangle_normals, [[0, 0, 0], [1, 1, 1], [0, 0, 0]]); +} + +#[test] +fn one_normal_fan_without_single_keeps_a_normal_per_vertex() { + let m = one(&[0, 0, 0, 3, 0, 6, 0, 9], &[(0x40, 0, &[1, 4])], 0); + assert_eq!(m.triangles, [[0, 1, 2], [0, 2, 3]]); +} + +#[test] +fn textured_fan_and_strip_carry_texture_slots_per_vertex() { + // 0x400 then 0x800, two texture sets: n,t,t,p per vertex. + let mut idx = Vec::new(); + for p in [0, 3, 6, 0, 3, 6, 9] { + idx.extend([0, 0, 2, p]); + } + let m = one(&idx, &[(0x400 | 0x800, 0, &[1, 3, 1, 4])], 2); + assert_eq!(m.triangles, [[0, 1, 2], [0, 1, 2], [2, 1, 3]]); +} + +/// The unit square as two textured triangles (one set; per vertex a normal, a texture slot, +/// a point), its corners at the picture's corners. +fn textured_square() -> W { + let mut body = W::default(); + let indices = [0, 0, 0, 0, 2, 3, 0, 4, 6, 0, 0, 0, 0, 4, 6, 0, 6, 9]; + tess_3d( + &mut body, + 8137, + &Mesh { indices: &indices, faces: &[(0x200, 0, &[2])], textures: 1, uvs: &[0., 0., 1., 0., 1., 1., 0., 1.], ..Mesh::default() }, + ); + section(1, &body) +} + +#[test] +fn a_textured_mesh_keeps_each_corners_coordinates() { + let t = decode(&textured_square(), &Schema::default(), 8137).unwrap(); + let m = mesh(&t[0]); + assert_eq!(m.triangles, [[0, 1, 2], [0, 2, 3]]); + assert_eq!(m.uvs, [[0.0, 0.0], [1.0, 0.0], [1.0, 1.0], [0.0, 1.0]]); + assert_eq!(m.triangle_uvs, [vec![Some([0, 1, 2]), Some([0, 2, 3])]]); + // An odd slot names no pair: that triangle has no coordinates. + let mut b = W::default(); + tess_3d( + &mut b, + 8137, + &Mesh { indices: &[0, 1, 0, 0, 2, 3, 0, 4, 6], faces: &[(0x200, 0, &[1])], textures: 1, uvs: &[0.0; 6], ..Mesh::default() }, + ); + let t = decode(§ion(1, &b), &Schema::default(), 8137).unwrap(); + let m = mesh(&t[0]); + assert_eq!(m.triangles, [[0, 1, 2]]); + assert_eq!(m.triangle_uvs, [vec![None]]); +} + +#[test] +fn too_many_texture_sets_are_refused_before_anything_is_allocated() { + // An empty fan on a face claiming 2^32 - 1 texture sets: no slot bounds the count. + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { indices: &[0], faces: &[(0x400, 0, &[1, 0])], textures: u32::MAX, ..Mesh::default() }); + assert!( + matches!(decode(§ion(1, &b), &Schema::default(), 8137), Err(PrcError::Malformed(m)) if m == format!("{} texture-coordinate sets on a face", u32::MAX)) + ); + let sets = u32::try_from(MAX_TEXTURE_SETS).unwrap(); + let mut ok = W::default(); + tess_3d(&mut ok, 8137, &Mesh { indices: &[0], faces: &[(0x400, 0, &[1, 0])], textures: sets, ..Mesh::default() }); + assert!(decode(§ion(1, &ok), &Schema::default(), 8137).is_ok()); +} + +/// A tessellation section holding one TESS_3D: [`SQUARE`] as two faces of one triangle each. +pub(crate) fn two_face_square() -> W { + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { indices: &[0, 0, 0, 3, 0, 6, 0, 0, 0, 6, 0, 9], faces: &[(0x2, 0, &[1]), (0x2, 6, &[1])], ..Mesh::default() }); + section(1, &b) +} + +#[test] +fn two_faces_partition_the_triangles() { + let m = one(&[0, 0, 0, 3, 0, 6, 0, 0, 0, 6, 0, 9], &[(0x2, 0, &[1]), (0x2, 6, &[1])], 0); + assert_eq!(m.triangles, [[0, 1, 2], [0, 2, 3]]); + assert_eq!(m.faces, [0..1, 1..2]); +} + +/// A wire over [`SQUARE`]: an open two-point wire continued by a closing two-point wire, +/// then optionally per-segment RGB colours. +fn wire(w: &mut W, colours: bool) { + square(w, 175); + uints(w, &[2, 0, 3, WIRE_IS_CONTINUOUS | WIRE_IS_CLOSING | 2, 6, 9]); + w.bit(colours); + if colours { + // 5 points (4 + the implicit closing one) -> 2 segment colours. + w.bit(false).bit(true).bit(false).put(0xff0000, 24); + w.bit(true); + } +} + +#[test] +fn wires_close_and_continue_and_colours_are_counted() { + for colours in [false, true] { + let mut b = W::default(); + wire(&mut b, colours); + // A second entity: a wrong colour count would desync it. + wire(&mut b, false); + let t = decode(§ion(2, &b), &Schema::default(), 8137).unwrap(); + for e in &t { + let Tessellation::Wire(w) = e else { panic!("not a wire") }; + assert_eq!(w, &[vec![[0., 0., 0.], [1., 0., 0.], [1., 1., 0.], [0., 1., 0.], [1., 1., 0.]]]); + } + } +} + +#[test] +fn a_closing_wire_colours_its_implicit_point() { + // Per-point RGB: 4 indexed points + 1 implicit = 5 colours. + let mut b = W::default(); + square(&mut b, 175); + uints(&mut b, &[2, 0, 3, WIRE_IS_CONTINUOUS | WIRE_IS_CLOSING | 2, 6, 9]); + b.bit(true).bit(false).bit(false).bit(false).put(1, 24); + for _ in 1..5 { + b.bit(true); + } + wire(&mut b, false); + let t = decode(§ion(2, &b), &Schema::default(), 8137).unwrap(); + assert!(matches!(t[1], Tessellation::Wire(_))); + assert_eq!(t[0], t[1]); +} + +#[test] +fn a_wire_index_must_name_a_point() { + let mut b = W::default(); + square(&mut b, 175); + uints(&mut b, &[2, 0, 12]); + b.bit(false); + assert!(matches!(decode(§ion(1, &b), &Schema::default(), 8137), Err(PrcError::Malformed(_)))); +} + +#[test] +fn a_wire_without_indices_is_the_coordinate_polyline() { + let mut b = W::default(); + square(&mut b, 175); + b.uint(0).bit(true); + // 4 implicit points -> 4 RGB colours, all "same" after the first. + b.bit(false).bit(false).bit(false).put(1, 24); + for _ in 1..4 { + b.bit(true); + } + wire(&mut b, false); + let t = decode(§ion(2, &b), &Schema::default(), 8137).unwrap(); + let Tessellation::Wire(w) = &t[0] else { panic!("not a wire") }; + assert_eq!(w.len(), 1); + assert_eq!(w[0].len(), 4); + assert!(matches!(t[1], Tessellation::Wire(_))); +} + +/// Reused colours cost one bit each, so a long run of them may end the section. +#[test] +fn reused_vertex_colours_may_be_the_last_data() { + let n = 40; + let mut b = W::default(); + b.uint(175).bit(false).uint(3 * n); + for i in 0..3 * n { + b.double(f64::from(i)); + } + // No indices: the polyline of all `n` points, one RGB colour then `n - 1` reuses. + b.uint(0).bit(true).bit(false).bit(false).bit(false).put(1, 24); + for _ in 1..n { + b.bit(true); + } + let t = decode(§ion(1, &b), &Schema::default(), 8137).unwrap(); + let Tessellation::Wire(w) = &t[0] else { panic!("not a wire") }; + assert_eq!(w[0].len(), n as usize); +} + +/// A square as one triangle plus one 4-vertex fan (two entities, three triangles), the face +/// carrying `attrs` and, when it has any, behaviour 0x30; then `colours` per-point colours +/// and a trailing wire. +fn styled_face(attrs: &[u32], colours: usize) -> Vec { + decode(&styled_face_section(attrs, colours, 0x30), &Schema::default(), 8137).unwrap() +} + +/// [`styled_face`]'s section, the face's behaviour `behaviour`. +fn styled_face_section(attrs: &[u32], colours: usize, behaviour: u32) -> W { + let mut b = W::default(); + square(&mut b, 172); + b.bit(true).bit(false).bit(false).uint(0).uint(0); + uints(&mut b, &[0, 0, 0, 3, 0, 6, 0, 0, 0, 3, 0, 6, 0, 9]); + b.uint(1).uint(174); + uints(&mut b, attrs); + b.uint(0).uint(0).uint(0x2 | 0x4).uint(0); + uints(&mut b, &[1, 1, 4]); + b.uint(0).bit(colours > 0); + if colours > 0 { + b.bit(false).bit(false).put(0x10_2030, 24); + for _ in 1..colours { + b.bit(true); + } + } + if !attrs.is_empty() { + b.uint(behaviour); + } + b.uint(0); + wire(&mut b, false); + section(2, &b) +} + +#[test] +fn face_vertex_colours_count_points_not_data() { + // A triangle plus a fan of 4 vertices = 7 points -> 7 colours. + let t = styled_face(&[], 7); + assert_eq!(mesh(&t[0]).triangles.len(), 3); + assert!(matches!(t[1], Tessellation::Wire(_))); +} + +fn styles(attrs: &[u32]) -> Vec<(u32, u16)> { + let t = styled_face(attrs, 0); + assert!(matches!(t[1], Tessellation::Wire(_))); + mesh(&t[0]).triangle_styles.iter().map(|g| (g.style, g.behaviour)).collect() +} + +#[test] +fn face_line_attributes_style_its_triangles() { + assert!(styles(&[]).is_empty()); + assert_eq!(styles(&[5]), [(5, 0x30); 3]); + // One per entity: the triangle, then both triangles of the fan. + assert_eq!(styles(&[4, 7]), [(4, 0x30), (7, 0x30), (7, 0x30)]); + // An entity styled 0 inherits, as do entities past the list. + assert_eq!(styles(&[4, 0, 9]), [(4, 0x30), (0, 0x30), (0, 0x30)]); +} + +#[test] +fn markup_is_read_past() { + let mut b = W::default(); + b.uint(176).bit(false).uint(3); + for c in [1., 2., 3.] { + b.double(c); + } + uints(&mut b, &[6, 0]); + b.uint(1).string(Some("R1")).string(Some("label")).put(7, 8); + wire(&mut b, false); + let t = decode(§ion(2, &b), &Schema::default(), 8137).unwrap(); + assert_eq!(t[0], Tessellation::Markup); + assert!(matches!(t[1], Tessellation::Wire(_))); +} + +#[test] +fn schema_fields_are_skipped_after_each_entity() { + // A newer producer added one UnsignedInteger to TESS_3D and one to TESS_Face. + let mut s = W::default(); + s.uint(2).uint(172).uint(1).uint(3).uint(174).uint(1).uint(3); + let sb = s.bytes(); + let schema = Schema::read(&mut BitReader::new(&sb)).unwrap(); + + let mut b = W::default(); + square(&mut b, 172); + b.bit(true).bit(false).bit(false).uint(0).uint(0); + uints(&mut b, &[0, 0, 0, 3, 0, 6]); + b.uint(1).uint(174).uint(0).uint(0).uint(0).uint(0x2).uint(0); + uints(&mut b, &[1]); + b.uint(0).bit(false).uint(99); // the face's added field + b.uint(0).uint(77); // texture count, then TESS_3D's added field + wire(&mut b, false); + let t = decode(§ion(2, &b), &schema, 8137).unwrap(); + assert_eq!(mesh(&t[0]).triangles, [[0, 1, 2]]); + assert!(matches!(t[1], Tessellation::Wire(_))); + // Without the schema the same bytes do not decode to the same thing. + assert_ne!(decode(§ion(2, &b), &Schema::default(), 8137).ok(), Some(t)); +} + +#[test] +fn attributes_names_and_user_data_are_read_past() { + let mut w = W::default(); + w.uint(305).uint(1).uint(ATTRIBUTE); + w.bit(true).uint(2); // predefined title + w.uint(2); + w.bit(false).string(Some("k")).uint(4).string(Some("v")); + w.bit(true).uint(1).uint(5).int(-1).uint(9); + w.uint(42); // the attribute's schema-added field + w.bit(false).string(Some("tess")).uint(1); + wire(&mut w, false); + w.uint(5).put(0b10110, 5); + let mut sw = W::default(); + sw.uint(1).uint(ATTRIBUTE).uint(1).uint(3); + let sb = sw.bytes(); + let schema = Schema::read(&mut BitReader::new(&sb)).unwrap(); + let bytes = w.bytes(); + let mut ctx = Ctx::new(BitReader::new(&bytes), &schema, 8137); + let t = ctx.file_structure_tessellation().unwrap(); + assert!(matches!(t[..], [Tessellation::Wire(_)])); + assert_eq!(ctx.name.as_deref(), Some("tess")); + assert_eq!(ctx.r.position(), w.len()); +} + +/// A one-triangle TESS_3D_Compressed. `full` takes every optional branch: stored normals, +/// colours, line attributes, texture, behaviours. Otherwise `three_t` picks the 3T edge form. +fn compressed(w: &mut W, v: u32, full: bool, three_t: bool) { + compressed_with_angles(w, v, full, three_t, 10); +} + +/// [`compressed`] whose `full` form stores `angle_bits`-bit normal angles. +fn compressed_with_angles(w: &mut W, v: u32, full: bool, three_t: bool, angle_bits: u8) { + w.uint(173).bit(false).bit(true).double(0.001); + if v >= 7031 { + w.put(0, 32).put(0, 32).put(0, 32); + } + // point_array: nine 2-bit values. + w.bit(false).uint(9); + for _ in 0..9 { + w.put(2, 8); + } + for _ in 0..9 { + w.bit(false).bit(true); + } + // edge_status_array: the 3T form (Huffman) or the T form. + if full { + w.bit(true).huffman(2, 2, &[(0, 0b10, 2)], &[0, 0, 0]); + } else if three_t { + w.bit(false).uint(3).put(0, 8).put(0, 8).put(0, 8); + } else { + w.bit(false).uint(1).put(0, 8); + } + // triangle_face_array [0]. `full`: four references, so the flagless reference array is + // Huffman; else none. + w.bit(false).uint(1).put(1, 8).bit(false); + if full { + w.uint(4).put(0b1111, 4); + w.huffman(6, 2, &[(1, 0b10, 2), (0, 0b11, 2)], &[1, 0, 0, 0]); + w.put(0, 4); + } else { + w.uint(3).put(0, 3).uint(0); + } + if full { + w.bit(false).put(u64::from(angle_bits), 8).uint(3).put(0b101, 3); + w.bit(true).huffman(10, 2, &[(5, 0b10, 2)], &[5, 5]); + w.bit(true); // is_face_planar + w.bit(true).bit(true).bit(false).uint(5); + for c in [1, 255, 0, 0, 128] { + w.put(c, 8); + } + w.bit(true).bit(false); + } else { + w.bit(true).bit(false).double(0.5).put(0, 8); + w.bit(false).bit(false); + } + w.bit(false).uint(1).put(1, 8).put(0, 8); // line_attribute_array + if full { + w.bit(false).uint(1).put(0, 32).uint(32); + w.uint(1).put(1, 5).bit(true).double(0.5).uint(2); + w.put(0, 32).put(0, 32).bit(false).bit(true); + w.bit(true).bit(false).uint(1).put(1, 8); + } else { + w.bit(true).bit(false); + } +} + +#[test] +fn a_compressed_mesh_is_read_to_its_end() { + for (full, three_t) in [(false, false), (false, true), (true, true)] { + for v in [7030, 7031, 8137] { + let mut b = W::default(); + compressed(&mut b, v, full, three_t); + let indices = [0, 0, 0, 3, 0, 6, 0, 0, 0, 3, 0, 6, 0, 9]; + tess_3d(&mut b, v, &Mesh { indices: &indices, faces: &[(0x2 | 0x4, 0, &[1, 1, 4])], ..Mesh::default() }); + let w = section(2, &b); + let bytes = w.bytes(); + let schema = Schema::default(); + let mut ctx = Ctx::new(BitReader::new(&bytes), &schema, v); + let t = ctx.file_structure_tessellation().unwrap(); + assert!(matches!(t[..], [Tessellation::Compressed { triangles: 1, .. }, Tessellation::Mesh(_)]), "full={full} v={v}: {t:?}"); + assert_eq!(ctx.r.position(), w.len(), "full={full} v={v}"); + } + } +} + +#[test] +fn refusals_and_damage_are_errors() { + // Edge statuses neither T nor 3T long. + let mut c = W::default(); + c.uint(173).bit(false).bit(true).double(0.001); + c.put(0, 32).put(0, 32).put(0, 32); + c.bit(false).uint(0).bit(false).uint(2).put(0, 16); + c.bit(false).uint(1).put(1, 8).bit(false); + assert!(matches!(decode(§ion(1, &c), &Schema::default(), 8137), Err(PrcError::Malformed(_)))); + // Optimised vertex colours. + let mut o = W::default(); + square(&mut o, 175); + o.uint(0).bit(true).bit(false).bit(false).bit(true); + assert!(matches!(decode(§ion(1, &o), &Schema::default(), 8137), Err(PrcError::Unsupported(_)))); + // A point index that is not a multiple of 3, and one past the end. + for bad in [4, 12] { + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { indices: &[0, 0, 0, bad, 0, 6], faces: &[(0x2, 0, &[1])], ..Mesh::default() }); + assert!(matches!(decode(§ion(1, &b), &Schema::default(), 8137), Err(PrcError::Malformed(_)))); + } + // A fan whose count claims a billion points. + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { indices: &[0, 0], faces: &[(0x4, 0, &[1, 0x3FFF_FFFF])], ..Mesh::default() }); + assert!(decode(§ion(1, &b), &Schema::default(), 8137).is_err()); + // A tessellation count far past the data. + let mut w = W::default(); + w.uint(305).uint(0).bit(true).uint(u32::MAX); + assert!(matches!(decode(&w, &Schema::default(), 8137), Err(PrcError::Truncated(_)))); + // A wrong entity type where the section's own belongs. + let mut w = W::default(); + w.uint(306); + assert!(matches!(decode(&w, &Schema::default(), 8137), Err(PrcError::Malformed(_)))); +} + +#[test] +fn every_truncation_of_a_section_is_an_error_not_a_panic() { + let mut b = W::default(); + compressed(&mut b, 8137, true, true); + wire(&mut b, true); + let bytes = section(2, &b).bytes(); + for end in 0..bytes.len() { + let cut = bytes.get(..end).unwrap(); + // The zero padding of the last byte may still decode; anything shorter must not. + let _ = Ctx::new(BitReader::new(cut), &Schema::default(), 8137).file_structure_tessellation(); + } + let cut = bytes.get(..bytes.len() / 2).unwrap(); + assert!(Ctx::new(BitReader::new(cut), &Schema::default(), 8137).file_structure_tessellation().is_err()); +} + +/// A one-file-structure PRC stream whose tessellation section is `tess`, with an empty +/// globals section schema. +fn prc_file(tess: &[u8]) -> Vec { + let mut globals = W::default(); + globals.uint(0); + prc_file_with(&globals.bytes(), tess) +} + +/// As [`prc_file`] with the globals section `globals`. +fn prc_file_with(globals: &[u8], tess: &[u8]) -> Vec { + use std::io::Write as _; + const HEADER_LEN: usize = 107; + let zlib = |data: &[u8]| { + let mut e = flate2::write::ZlibEncoder::new(Vec::new(), flate2::Compression::default()); + e.write_all(data).unwrap(); + e.finish().unwrap() + }; + let u32s = |out: &mut Vec, vs: &[u32]| vs.iter().for_each(|v| out.extend_from_slice(&v.to_le_bytes())); + let sections: [&[u8]; 5] = [globals, b"", tess, b"", b""]; + let mut body = b"PRC".to_vec(); + u32s(&mut body, &[8137, 8137, 5, 6, 7, 8, 0, 0, 0, 0, 0]); + let mut offs = vec![HEADER_LEN]; + for s in sections { + offs.push(HEADER_LEN + body.len()); + body.extend(zlib(s)); + } + let mf_start = HEADER_LEN + body.len(); + body.extend(zlib(b"")); + let mf_end = HEADER_LEN + body.len(); + let mut out = b"PRC".to_vec(); + u32s(&mut out, &[8137, 8137, 1, 2, 3, 4, 0, 0, 0, 0, 1]); + u32s(&mut out, &[5, 6, 7, 8, 0, 6]); + u32s(&mut out, &offs.iter().map(|&o| u32::try_from(o).unwrap()).collect::>()); + u32s(&mut out, &[u32::try_from(mf_start).unwrap(), u32::try_from(mf_end).unwrap(), 0]); + out.extend(body); + out +} + +#[test] +fn a_prc_file_yields_its_tessellations() { + let f = crate::PrcFile::parse(&prc_file(&textured_square().bytes())).unwrap(); + let t = f.file_structures[0].tessellations().unwrap(); + assert_eq!(mesh(&t[0]).triangles, [[0, 1, 2], [0, 2, 3]]); +} + +#[test] +fn normal_angles_wider_than_16_bits_are_refused_by_name() { + let mut b = W::default(); + compressed_with_angles(&mut b, 8137, true, true, 17); + assert!(matches!(decode(§ion(1, &b), &Schema::default(), 8137), Err(PrcError::Malformed(m)) if m == "17-bit normal angles")); +} + +/// An entity's counted array is charged against the reader's element allowance. +#[test] +fn counted_entity_arrays_are_charged_against_the_allowance() { + let mut w = W::default(); + w.uint(3).uint(1).uint(2).uint(3); + let bytes = w.bytes(); + let schema = Schema::default(); + let read = |n| Ctx::new(BitReader::with_elements(&bytes, n), &schema, 8137).uints("indices"); + assert!(matches!(read(2), Err(PrcError::TooManyElements { .. }))); + assert_eq!(read(3).unwrap(), [1, 2, 3]); +} + +#[test] +fn a_face_behaviour_wider_than_16_bits_is_malformed() { + let r = decode(&styled_face_section(&[1], 0, 0x1_0030), &Schema::default(), 8137); + assert!(matches!(r, Err(PrcError::Malformed(m)) if m == "face behaviour 65584 is wider than 16 bits")); +} + +#[test] +fn coordinates_that_are_not_whole_points_are_malformed() { + let mut partial = SQUARE.to_vec(); + partial.push(7.0); + let mut b = W::default(); + let faces: &[(u32, u32, &[u32])] = &[(0x2, 0, &[1])]; + tess_3d(&mut b, 8137, &Mesh { indices: &[0, 0, 0, 3, 0, 6], faces, coords: Some(&partial), ..Mesh::default() }); + let r = decode(§ion(1, &b), &Schema::default(), 8137); + assert!(matches!(r, Err(PrcError::Malformed(m)) if m == "13 coordinates are not whole points")); + // A partial last normal drops the stored normals, as a slot naming none does. + assert_eq!(stored_normals(&[0.0, 0.0, 1.0, 0.0], &[[0, 0, 0]], 1), (Vec::new(), Vec::new())); +} + +/// Elements a tessellation section spends from a full allowance. +fn spent(section: &W) -> usize { + let bytes = section.bytes(); + let schema = Schema::default(); + let mut ctx = Ctx::new(BitReader::new(&bytes), &schema, 8137); + ctx.file_structure_tessellation().unwrap(); + crate::MAX_DECODED_ELEMENTS - ctx.r.elements_left() +} + +#[test] +fn faces_reusing_one_index_range_are_charged_for_each_triangle_they_make() { + const INDICES: [u32; 12] = [0, 0, 0, 3, 0, 6, 0, 0, 0, 6, 0, 9]; + let square = |faces: &[(u32, u32, &[u32])]| { + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { indices: &INDICES, faces, ..Mesh::default() }); + spent(§ion(1, &b)) + }; + let one = square(&[(0x2, 0, &[2])]); + let three = square(&[(0x2, 0, &[2]), (0x2, 0, &[2]), (0x2, 0, &[2])]); + // Per extra face: its count entry, its one data element and its two triangles, each + // 12 bytes of corners and 12 of normal slots. + assert_eq!(three - one, 2 * (2 + 2 * 4)); +} + +#[test] +fn reused_textured_faces_are_charged_for_every_corner_array() { + // The textured square, then one untextured triangle's normal and point slots. + const INDICES: [u32; 24] = [0, 0, 0, 0, 2, 3, 0, 4, 6, 0, 0, 0, 0, 4, 6, 0, 6, 9, 0, 0, 0, 3, 0, 6]; + let square = |faces: &[(u32, u32, &[u32])], sets: u32| { + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { indices: &INDICES, faces, textures: sets, uvs: &[0.0; 8], ..Mesh::default() }); + spent(§ion(1, &b)) + }; + let face: (u32, u32, &[u32]) = (0x200, 0, &[2]); + let untextured: (u32, u32, &[u32]) = (0x2, 18, &[1]); + let one = square(&[face], 1); + let three = square(&[face, face, face], 1); + // Per extra face: its count, its data element and two triangles of corners, normal slots + // and one texture set, 16 bytes a corner triple. + assert_eq!(three - one, 2 * (2 + 2 * 6)); + // One textured face gives every triangle of the mesh a texture triple. + let mixed = square(&[face, untextured], 1); + assert_eq!(mixed - one, 2 + 6); +} + +#[test] +fn a_face_is_charged_before_it_is_triangulated() { + // One face claiming a million triangles over a 12-entry index array. + const INDICES: [u32; 12] = [0, 0, 0, 3, 0, 6, 0, 0, 0, 6, 0, 9]; + let mut b = W::default(); + tess_3d(&mut b, 8137, &Mesh { indices: &INDICES, faces: &[(0x2, 0, &[1_000_000])], ..Mesh::default() }); + let bytes = section(1, &b).bytes(); + let schema = Schema::default(); + let r = Ctx::new(BitReader::with_elements(&bytes, 1000), &schema, 8137).file_structure_tessellation(); + assert!(matches!(r, Err(PrcError::TooManyElements { .. })), "{r:?}"); +} + +#[test] +fn the_globals_schema_and_the_tessellations_share_one_allowance() { + // A schema of one entry with three tokens: four elements. + let mut globals = W::default(); + globals.uint(1).uint(999).uint(3).uint(0).uint(0).uint(0); + let tess = textured_square(); + let f = crate::PrcFile::parse(&prc_file_with(&globals.bytes(), &tess.bytes())).unwrap(); + let need = spent(&tess) + 4; + let within = |elements| f.file_structures[0].tessellations_within(crate::bits::Budget { elements, ..crate::bits::Budget::FULL }); + assert!(within(need).is_ok()); + assert!(matches!(within(need - 1), Err(PrcError::TooManyElements { .. }))); +} diff --git a/crates/prc/src/tess/triangulate.rs b/crates/prc/src/tess/triangulate.rs new file mode 100644 index 0000000000..6611e4abf7 --- /dev/null +++ b/crates/prc/src/tess/triangulate.rs @@ -0,0 +1,282 @@ +//! A `TESS_Face`'s triangulated data turned into triangles [WD 7.8.5.5, 7.8.6]. + +use super::TriangleStyle; +use crate::PrcError; +use crate::entity::malformed; + +/// A fan/strip count's point-count bits; bit 30 is `NORMAL_Single` [WD 7.8.6.2]. +const COUNT_MASK: u32 = 0x3FFF_FFFF; +const NORMAL_SINGLE: u32 = 0x4000_0000; + +/// Texture-coordinate sets a face may carry. Each set costs a slot per point and a corner +/// triple per triangle of the whole mesh, so the count is capped. +pub(super) const MAX_TEXTURE_SETS: usize = 8; + +pub(super) struct Face { + pub(super) flags: u32, + pub(super) start: usize, + pub(super) data: Vec, + pub(super) textures: usize, + /// `line_attributes`, each a style index + 1: none = the owner's graphics, one = the + /// whole face's, more = one per triangulation entity [WD 7.8.6]. + pub(super) styles: Vec, + /// The behaviour bits read when `styles` is not empty. + pub(super) behaviour: u16, +} + +/// A `used_entities_flag` block's shape [WD 7.8.5.5]. +#[derive(Clone, Copy)] +enum Shape { + Triangles, + Fan, + Strip, +} + +/// One `used_entities_flag` block. +#[derive(Clone, Copy)] +struct Block { + bit: u32, + shape: Shape, + /// One normal per entity rather than per point. + one_normal: bool, + textured: bool, +} + +const fn block(bit: u32, shape: Shape, one_normal: bool, textured: bool) -> Block { + Block { bit, shape, one_normal, textured } +} + +/// The blocks in stream order, low bit first [WD 7.8.5.5]. +const BLOCKS: [Block; 12] = [ + block(0x0002, Shape::Triangles, false, false), + block(0x0004, Shape::Fan, false, false), + block(0x0008, Shape::Strip, false, false), + block(0x0020, Shape::Triangles, true, false), + block(0x0040, Shape::Fan, true, false), + block(0x0080, Shape::Strip, true, false), + block(0x0200, Shape::Triangles, false, true), + block(0x0400, Shape::Fan, false, true), + block(0x0800, Shape::Strip, false, true), + block(0x2000, Shape::Triangles, true, true), + block(0x4000, Shape::Fan, true, true), + block(0x8000, Shape::Strip, true, true), +]; + +/// One triangulation entity's index slots: per point its position, normal slot (when the +/// array stores normals) and texture slots. +#[derive(Default)] +struct Slots { + verts: Vec, + normals: Vec, + /// Per texture set, each point's slot. + textures: Vec>, +} + +/// What a mesh's triangles carry besides positions, built in step with them: corner normal +/// slots and, per texture set, corner texture slots (`None` for a triangle whose face stores +/// fewer sets). +#[derive(Default)] +pub(super) struct Corners { + pub(super) normals: Vec<[u32; 3]>, + pub(super) textures: Vec>>, +} + +impl Face { + /// Entity sizes per set block, from `TriangulatedData` [WD 7.8.6]. A triangle block is + /// one size word (its triangle count); fans and strips list a vertex count each. + fn entities(&self) -> Result)>, PrcError> { + let mut d = self.data.iter().copied(); + let mut next = || d.next().ok_or_else(|| malformed("triangulated data ends early".into())); + let mut out = Vec::new(); + for b in BLOCKS { + if self.flags & b.bit == 0 { + continue; + } + let sizes = match b.shape { + Shape::Triangles => vec![next()?], + Shape::Fan | Shape::Strip => { + let n = next()? as usize; + if n > self.data.len() { + return Err(malformed("triangulated data ends early".into())); + } + (0..n).map(|_| next()).collect::>()? + } + }; + out.push((b, sizes)); + } + Ok(out) + } + + /// Points the face's index slots name: the vertex-colour count (pdf-issues #820). + pub(super) fn point_count(&self) -> Result { + let mut n = 0usize; + for (b, sizes) in self.entities()? { + for s in sizes { + let s = (s & COUNT_MASK) as usize; + n = n.saturating_add(if matches!(b.shape, Shape::Triangles) { s.saturating_mul(3) } else { s }); + } + } + Ok(n) + } + + /// Triangles [`Self::triangulate`] makes: one per triangle-block count, two fewer than the + /// points of each fan and strip. + pub(super) fn triangle_count(&self) -> Result { + let mut n = 0usize; + for (b, sizes) in self.entities()? { + for s in sizes { + let s = (s & COUNT_MASK) as usize; + n = n.saturating_add(if matches!(b.shape, Shape::Triangles) { s } else { s.saturating_sub(2) }); + } + } + Ok(n) + } + + /// Texture sets the face's triangles carry: its count when one of its blocks is textured, + /// at most [`MAX_TEXTURE_SETS`] (a face past it is refused when triangulated). + pub(super) fn texture_sets(&self) -> usize { + if BLOCKS.iter().any(|b| b.textured && self.flags & b.bit != 0) { self.textures.min(MAX_TEXTURE_SETS) } else { 0 } + } + + /// Emit the face's triangles into `out`; `normals` = the index array stores normal slots, + /// in which case each triangle's corner normal slots go to `corners`, as do its corner + /// texture slots. Returns the triangles each triangulation entity emitted, in order: one + /// entity per triangle of a triangle block, per fan, per strip [WD 7.8.6]. + pub(super) fn triangulate( + &self, + indices: &[u32], + normals: bool, + n_points: usize, + out: &mut Vec<[u32; 3]>, + corners: &mut Corners, + ) -> Result, PrcError> { + if self.textures > MAX_TEXTURE_SETS { + return Err(malformed(format!("{} texture-coordinate sets on a face", self.textures))); + } + let mut entities = Vec::new(); + let mut slots = indices.get(self.start..).unwrap_or(&[]).iter().copied(); + for (b, sizes) in self.entities()? { + let t = if b.textured { self.textures } else { 0 }; + for word in sizes { + let count = (word & COUNT_MASK) as usize; + // Points in this entity, and whether each carries its own normal slot. + let (points, per_point_normal) = match b.shape { + Shape::Triangles => (count.saturating_mul(3), !b.one_normal), + Shape::Fan | Shape::Strip => (count, !b.one_normal || word & NORMAL_SINGLE == 0), + }; + if points.saturating_mul(t.saturating_add(1)) > indices.len() { + return Err(malformed("face runs past the triangulated index array".into())); + } + let s = read_slots(&mut slots, b.shape, points, t, normals.then_some(per_point_normal), n_points)?; + let made = emit(b.shape, &s, out, corners); + match b.shape { + Shape::Triangles => entities.extend(std::iter::repeat_n(1, made)), + Shape::Fan | Shape::Strip => entities.push(made), + } + } + } + Ok(entities) + } + + /// Append one [`TriangleStyle`] per triangle of the face, whose entities emitted + /// `entities[i]` triangles each. Entities past the face's styles inherit (style 0). + pub(super) fn styles_into(&self, entities: &[usize], out: &mut Vec) { + let whole = match self.styles.as_slice() { + [] => Some(0), + [one] => Some(*one), + _ => None, + }; + for (i, &n) in entities.iter().enumerate() { + let style = whole.unwrap_or_else(|| self.styles.get(i).copied().unwrap_or(0)); + out.extend(std::iter::repeat_n(TriangleStyle { style, behaviour: self.behaviour }, n)); + } + } +} + +/// The 8-byte elements of [`crate::MAX_DECODED_ELEMENTS`] one triangle holds: its corners +/// (12 bytes), its normal-slot corners when the index array stores them (12), a corner triple +/// per texture set (16 each) and its style when the mesh is styled (8). +pub(super) fn triangle_elements(normals: bool, sets: usize, styled: bool) -> usize { + let normals = if normals { 2 } else { 0 }; + 2usize.saturating_add(normals).saturating_add(sets.saturating_mul(2)).saturating_add(usize::from(styled)) +} + +/// One entity's `points` index slots. `normals` is `None` when the array stores no normal +/// slots, else whether every point has its own (rather than one per entity). +fn read_slots( + slots: &mut impl Iterator, + shape: Shape, + points: usize, + texture_sets: usize, + normals: Option, + n_points: usize, +) -> Result { + let mut take = || slots.next().ok_or_else(|| malformed("face runs past the triangulated index array".into())); + // The caller bounded points * (texture_sets + 1) by the index array's length. + let mut s = Slots { textures: vec![Vec::with_capacity(points); texture_sets], ..Slots::default() }; + let mut current = 0; + for k in 0..points { + let new_entity = match shape { + Shape::Triangles => k % 3 == 0, + Shape::Fan | Shape::Strip => k == 0, + }; + if let Some(per_point) = normals { + if per_point || new_entity { + current = take()?; + } + s.normals.push(current); + } + for set in &mut s.textures { + set.push(take()?); + } + let p = take()?; + if !p.is_multiple_of(3) || p as usize / 3 >= n_points { + return Err(malformed(format!("point index {p} is not a point"))); + } + s.verts.push(p / 3); + } + Ok(s) +} + +/// Append the triangles of one entity's `slots` to `out` and `corners`, keeping every +/// texture set as long as `out`. Returns how many it made. +fn emit(shape: Shape, slots: &Slots, out: &mut Vec<[u32; 3]>, corners: &mut Corners) -> usize { + let before = out.len(); + assemble(shape, &slots.verts, out); + assemble(shape, &slots.normals, &mut corners.normals); + let made = out.len().saturating_sub(before); + while corners.textures.len() < slots.textures.len() { + corners.textures.push(vec![None; before]); + } + for (k, set) in corners.textures.iter_mut().enumerate() { + match slots.textures.get(k) { + Some(own) => { + let mut tris = Vec::with_capacity(made); + assemble(shape, own, &mut tris); + set.extend(tris.into_iter().map(Some)); + } + None => set.extend(std::iter::repeat_n(None, made)), + } + } + made +} + +/// Append the triangles `shape` makes of `verts`. +fn assemble(shape: Shape, verts: &[u32], out: &mut Vec<[u32; 3]>) { + match shape { + Shape::Triangles => out.extend(verts.as_chunks::<3>().0.iter().copied()), + Shape::Fan => { + if let Some((&c, rim)) = verts.split_first() { + out.extend(rim.windows(2).filter_map(|w| match w { + [a, b] => Some([c, *a, *b]), + _ => None, + })); + } + } + Shape::Strip => out.extend(verts.windows(3).enumerate().filter_map(|(k, w)| match w { + [a, b, c] if k % 2 == 0 => Some([*a, *b, *c]), + [a, b, c] => Some([*b, *a, *c]), + _ => None, + })), + } +} diff --git a/crates/prc/src/testw.rs b/crates/prc/src/testw.rs new file mode 100644 index 0000000000..37e90606b6 --- /dev/null +++ b/crates/prc/src/testw.rs @@ -0,0 +1,166 @@ +//! Test-only PRC bit writer: the inverse of [`crate::BitReader`]. + +#![allow(clippy::expect_used)] + +use crate::acof; + +/// An MSB-first bit sink for building test streams. +#[derive(Debug, Default, Clone)] +pub(crate) struct W { + bits: Vec, +} + +impl W { + /// The low `n` bits of `v` (`n` ≤ 64), most significant first. + pub(crate) fn put(&mut self, v: u64, n: u32) -> &mut Self { + for i in (0..n).rev() { + self.bits.push((v >> i) & 1 == 1); + } + self + } + + /// One bit; a PRC `Boolean`. + pub(crate) fn bit(&mut self, b: bool) -> &mut Self { + self.bits.push(b); + self + } + + /// Bits written so far. + pub(crate) fn len(&self) -> usize { + self.bits.len() + } + + /// Every bit of `other`, unaligned. + pub(crate) fn append(&mut self, other: &Self) -> &mut Self { + self.bits.extend_from_slice(&other.bits); + self + } + + /// An `UnsignedInteger` [WD 11.10]. + pub(crate) fn uint(&mut self, mut v: u32) -> &mut Self { + while v != 0 { + self.bit(true).put(u64::from(v & 0xff), 8); + v >>= 8; + } + self.bit(false) + } + + /// An `Integer` [WD 11.11]: two's-complement groups, low first, each followed by a + /// continue bit. + pub(crate) fn int(&mut self, v: i32) -> &mut Self { + if v == 0 { + return self.bit(false); + } + self.bit(true); + let mut v = i64::from(v); + loop { + let group = v & 0xff; + self.put(group.cast_unsigned(), 8); + v >>= 8; + let done = (v == 0 && group & 0x80 == 0) || (v == -1 && group & 0x80 != 0); + self.bit(!done); + if done { + return self; + } + } + } + + /// A `String` [WD 11.4]: presence bit, byte count, raw UTF-8 bytes. + pub(crate) fn string(&mut self, s: Option<&str>) -> &mut Self { + let Some(s) = s else { + return self.bit(false); + }; + self.bit(true).uint(u32::try_from(s.len()).expect("a short test string")); + for b in s.bytes() { + self.put(u64::from(b), 8); + } + self + } + + /// A finite `Double` [WD 11.17]: zero as the two-bit zero code; anything else as its + /// exponent row, the sign, then the mantissa as literal bytes. + pub(crate) fn double(&mut self, v: f64) -> &mut Self { + if v == 0.0 { + return self.put(0b01, 2); + } + let bits = v.to_bits(); + let exponent = u32::try_from((bits >> 52) & 0x7ff).expect("11 bits"); + let (n, c) = acof::exponent_code(exponent).expect("the Double table has a row per exponent"); + let le = bits.to_le_bytes(); + self.put(u64::from(c), n).bit(v < 0.0).bit(true).put(u64::from(le[6] & 0x0f), 4); + for b in le[..6].iter().rev() { + self.bit(true).put(u64::from(*b), 8); + } + self + } + + /// A Huffman container [WD 12.2]: `leaves` are `(symbol, stored code, length)`, each value + /// is emitted as its leaf's code. + pub(crate) fn huffman(&mut self, nbits: u32, len_width: u32, leaves: &[(u32, u32, u32)], values: &[i32]) -> &mut Self { + let mask = (1u32 << nbits) - 1; + let codes: Vec<(u32, u32)> = values + .iter() + .map(|&v| { + let l = leaves.iter().find(|l| l.0 & mask == v.cast_unsigned() & mask).expect("a leaf per value"); + (l.1, l.2) + }) + .collect(); + self.huffman_raw(nbits, len_width, leaves, &codes) + } + + /// As [`Self::huffman`] with each element's `(code, length)` given. + pub(crate) fn huffman_raw(&mut self, nbits: u32, len_width: u32, leaves: &[(u32, u32, u32)], codes: &[(u32, u32)]) -> &mut Self { + let mut blob = Self::huffman_head(nbits, len_width, leaves); + lsb(&mut blob, u32::try_from(codes.len()).expect("a short test array"), 32); + for &(c, l) in codes { + for i in (0..l).rev() { + blob.push((c >> i) & 1 == 1); + } + } + self.huffman_words(blob) + } + + /// A Huffman container whose element count is `count` with no elements written. + pub(crate) fn huffman_count(&mut self, nbits: u32, len_width: u32, leaves: &[(u32, u32, u32)], count: u32) -> &mut Self { + let mut blob = Self::huffman_head(nbits, len_width, leaves); + lsb(&mut blob, count, 32); + self.huffman_words(blob) + } + + fn huffman_head(nbits: u32, len_width: u32, leaves: &[(u32, u32, u32)]) -> Vec { + let mut blob = Vec::new(); + lsb(&mut blob, u32::try_from(leaves.len()).expect("a short test table"), nbits + 1); + lsb(&mut blob, len_width, 8); + for &(s, c, l) in leaves { + lsb(&mut blob, s, nbits); + lsb(&mut blob, l, len_width); + lsb(&mut blob, c, l); + } + blob + } + + /// The payload as whole 32-bit words of LSB-first bytes, then the last word's used bits. + fn huffman_words(&mut self, mut blob: Vec) -> &mut Self { + let bits = blob.len(); + let words = bits.div_ceil(32); + blob.resize(words * 32, false); + self.uint(u32::try_from(words).expect("a short test payload")); + for byte in blob.chunks(8) { + let b = byte.iter().enumerate().fold(0u64, |a, (i, &x)| a | (u64::from(x) << i)); + self.put(b, 8); + } + self.uint(u32::try_from(bits - (words - 1) * 32).expect("at most 32")) + } + + /// The bits packed MSB-first, the last byte zero-padded. + pub(crate) fn bytes(&self) -> Vec { + self.bits.chunks(8).map(|c| c.iter().enumerate().fold(0u8, |a, (i, &b)| a | (u8::from(b) << (7 - i)))).collect() + } +} + +/// The low `n` bits of `v`, least significant first. +fn lsb(blob: &mut Vec, v: u32, n: u32) { + for i in 0..n { + blob.push((v >> i) & 1 == 1); + } +} diff --git a/crates/prc/src/tree/markup.rs b/crates/prc/src/tree/markup.rs new file mode 100644 index 0000000000..193c4fa219 --- /dev/null +++ b/crates/prc/src/tree/markup.rs @@ -0,0 +1,199 @@ +//! What a tree carries beside its geometry, read past and not modelled: markup data (PMI) +//! and views [WD 7.3.10.4, 7.7], and a scene's lights and clipping planes [WD 7.5.15-7.5.18, +//! 7.11.13]. The tree section has no length to skip by, so each is read field by field. + +use super::{ + AMBIENT_LIGHT, ANNOTATION_ITEM, ANNOTATION_REFERENCE, ANNOTATION_SET, DIRECTIONAL_LIGHT, LIGHT_INTENSITY_FROM, MARKUP_LINKED_ITEM, MAX_DEPTH, + MKP_LEADER, MKP_MARKUP, MKP_VIEW, POINT_LIGHT, SPOT_LIGHT, SURF_PLANE, VIEW_FLAGS_FROM, +}; +use crate::PrcError; +use crate::entity::{Ctx, malformed}; + +impl Ctx<'_, '_> { + /// `MarkupData` [WD 7.3.10.4]: linked items, leaders, markups, then annotation entities. + pub(super) fn markup_data(&mut self) -> Result<(), PrcError> { + let n = self.count(1, "markup linked items")?; + for _ in 0..n { + self.markup_linked_item()?; + } + let n = self.count(1, "leaders")?; + for _ in 0..n { + self.leader()?; + } + let n = self.count(1, "markups")?; + for _ in 0..n { + self.mkp_markup()?; + } + let n = self.count(1, "annotation entities")?; + for _ in 0..n { + self.annotation_entity(0)?; + } + Ok(()) + } + + /// A counted list of `MKP_View`s. + pub(super) fn views(&mut self) -> Result<(), PrcError> { + let n = self.count(1, "views")?; + for _ in 0..n { + self.view()?; + } + Ok(()) + } + + /// A counted list of `ReferenceOnPRCBase`s. + fn references(&mut self, what: &'static str) -> Result<(), PrcError> { + let n = self.count(1, what)?; + for _ in 0..n { + self.reference_on_prc_base()?; + } + Ok(()) + } + + /// `MISC_MarkupLinkedItem` (204) [WD 7.4.5]. + // prc-rs: a `has_reference_data` flag and its reference data follow the content entity + // reference (pdf-issues #722). + fn markup_linked_item(&mut self) -> Result<(), PrcError> { + self.expect_type(MARKUP_LINKED_ITEM)?; + self.base_with_graphics()?; + self.r.unsigned_integer()?; // local coordinate system + 1 + if self.r.bit()? { + self.reference_data()?; + } + if self.r.bit()? { + self.reference_data()?; + } + for _ in 0..4 { + self.r.bit()?; // show and delete, markup and leader + } + self.schema.skip_added_fields(MARKUP_LINKED_ITEM, &mut self.r)?; + self.user_data() + } + + /// `MKP_Leader` (503) [WD 7.7.5; pdf-issues #698]. + // prc-rs: the leader's field order. + fn leader(&mut self) -> Result<(), PrcError> { + self.expect_type(MKP_LEADER)?; + self.base_with_graphics()?; + for _ in 0..2 { + if self.r.bit()? { + self.reference_on_prc_base()?; // first, second linked item + } + } + self.r.unsigned_integer()?; // tessellation + 1 + self.schema.skip_added_fields(MKP_LEADER, &mut self.r)?; + self.user_data() + } + + /// `MKP_Markup` (502) [WD 7.7.4]. + // prc-rs: the markup's field order. + fn mkp_markup(&mut self) -> Result<(), PrcError> { + self.expect_type(MKP_MARKUP)?; + self.base_with_graphics()?; + self.r.unsigned_integer()?; // type + self.r.unsigned_integer()?; // subtype + self.references("markup linked items")?; + self.references("markup leaders")?; + self.r.unsigned_integer()?; // tessellation + 1 + self.schema.skip_added_fields(MKP_MARKUP, &mut self.r)?; + self.user_data() + } + + /// An annotation item (504), set (505) or reference (506) [WD 7.7.6-7.7.8]; sets nest. + // prc-rs: the annotation entities' field orders. + fn annotation_entity(&mut self, depth: usize) -> Result<(), PrcError> { + if depth > MAX_DEPTH { + return Err(malformed("annotation sets nest too deep".into())); + } + let t = self.r.unsigned_integer()?; + if !matches!(t, ANNOTATION_ITEM | ANNOTATION_SET | ANNOTATION_REFERENCE) { + return Err(malformed(format!("entity type {t} as an annotation"))); + } + self.base_with_graphics()?; + match t { + ANNOTATION_ITEM => self.reference_on_prc_base()?, + ANNOTATION_SET => { + let n = self.count(1, "annotation set")?; + for _ in 0..n { + self.annotation_entity(depth + 1)?; + } + } + _ => self.references("annotation reference")?, + } + self.schema.skip_added_fields(t, &mut self.r)?; + self.user_data() + } + + /// `MKP_View` (501) [WD 7.7.3]. + // prc-rs: the view's field order, and its flags, linked items and filters from version + // 8137. + fn view(&mut self) -> Result<(), PrcError> { + self.expect_type(MKP_VIEW)?; + self.base_with_graphics()?; + self.references("view annotations")?; + self.plane()?; + if self.r.bit()? { + self.scene_display_parameters()?; + } + if self.version >= VIEW_FLAGS_FROM { + for _ in 0..3 { + self.r.bit()?; // annotation view, default view, direction + } + self.references("view linked items")?; + let n = self.count(1, "view filters")?; + for _ in 0..n { + self.filter()?; + } + } + self.schema.skip_added_fields(MKP_VIEW, &mut self.r)?; + self.user_data() + } + + /// `SURF_Plane` (86) [WD 7.11.13; pdf-issues #658]: `ContentSurface` [WD 7.11.3.2], a + /// `Transformation3D` with no presence flag, the `Domain` [WD 8.2.4] and the + /// parameterisation's four coefficients. + // prc-rs: the plane's field order. + pub(super) fn plane(&mut self) -> Result<(), PrcError> { + self.expect_type(SURF_PLANE)?; + if self.r.bit()? { + self.content_prc_base()?; // base geometry: attributes and name + self.r.unsigned_integer()?; // identifier + } + self.r.unsigned_integer()?; // extend type + self.transformation_3d()?; + for _ in 0..8 { + self.r.double()?; // domain (4), coefficients (4) + } + self.schema.skip_added_fields(SURF_PLANE, &mut self.r) + } + + /// A light (731-734) [WD 7.5.15-7.5.18]. + // prc-rs: the lights' field orders, and a directional light's intensity from version 8030 + // (pdf-issues #745). + pub(super) fn light(&mut self) -> Result<(), PrcError> { + let t = self.r.unsigned_integer()?; + if !(AMBIENT_LIGHT..=SPOT_LIGHT).contains(&t) { + return Err(malformed(format!("entity type {t} as a light"))); + } + self.content_prc_ref_base()?; + for _ in 0..3 { + self.r.unsigned_integer()?; // ambient, diffuse, specular colour + 1 + } + if matches!(t, POINT_LIGHT | SPOT_LIGHT) { + self.vector3()?; // location + for _ in 0..3 { + self.r.double()?; // attenuation + } + } + if matches!(t, DIRECTIONAL_LIGHT | SPOT_LIGHT) { + self.vector3()?; // direction + } + if t == DIRECTIONAL_LIGHT && self.version >= LIGHT_INTENSITY_FROM { + self.r.double()?; + } + if t == SPOT_LIGHT { + self.r.double()?; // fall-off angle + self.r.double()?; // fall-off exponent + } + self.schema.skip_added_fields(t, &mut self.r) + } +} diff --git a/crates/prc/src/tree/markup_tests.rs b/crates/prc/src/tree/markup_tests.rs new file mode 100644 index 0000000000..dcd658bb62 --- /dev/null +++ b/crates/prc/src/tree/markup_tests.rs @@ -0,0 +1,280 @@ +#![allow(clippy::unwrap_used, clippy::expect_used, clippy::panic)] + +use super::tests::{Occ, PLAIN, S_RED, base, graphics, part_with, plate, ref_base, single}; +use crate::PrcError; +use crate::bits::BitReader; +use crate::entity::Ctx; +use crate::testw::W; +use crate::{Schema, Tessellation}; + +/// Read after each entity, to show it was read to its last bit and no further. +const SENTINEL: u32 = 0x5A5A; + +/// Runs `read` over `w` then the sentinel, at authoring version `version`. +fn read_at(w: &W, version: u32, read: impl FnOnce(&mut Ctx<'_, '_>) -> Result<(), PrcError>) -> Result<(), PrcError> { + let mut all = W::default(); + all.append(w).uint(SENTINEL); + let bytes = all.bytes(); + let schema = Schema::default(); + let mut ctx = Ctx::new(BitReader::new(&bytes), &schema, version); + read(&mut ctx)?; + assert_eq!(ctx.r.unsigned_integer()?, SENTINEL, "read past the wrong number of bits"); + Ok(()) +} + +fn read(w: &W, read: impl FnOnce(&mut Ctx<'_, '_>) -> Result<(), PrcError>) -> Result<(), PrcError> { + read_at(w, 8137, read) +} + +/// Each truncation of `w` is an error, never a panic. +fn truncations_fail(w: &W, read: impl Fn(&mut Ctx<'_, '_>) -> Result<(), PrcError>) { + let bytes = w.bytes(); + let schema = Schema::default(); + for end in 0..bytes.len().saturating_sub(1) { + let mut ctx = Ctx::new(BitReader::new(&bytes[..end]), &schema, 8137); + assert!(read(&mut ctx).is_err(), "cut at {end}"); + } +} + +fn reference(w: &mut W, uid: u32) { + w.uint(205).uint(237).bit(true).uint(uid); +} + +fn markup_data(linked: &[W], leaders: &[W], markups: &[W], annotations: &[W]) -> W { + let mut w = W::default(); + for list in [linked, leaders, markups, annotations] { + w.uint(u32::try_from(list.len()).unwrap()); + for e in list { + w.append(e); + } + } + w +} + +fn linked_item() -> W { + let mut w = W::default(); + w.uint(204); + graphics(&mut w, None, 1, PLAIN); + w.uint(0).bit(true); + reference(&mut w, 7); + w.bit(true).uint(206).uint(149).bit(true).bit(true).uint(1).uint(1).uint(1).uint(0); + w.bit(true).bit(false).bit(true).bit(false).uint(0); + w +} + +fn leader() -> W { + let mut w = W::default(); + w.uint(503); + graphics(&mut w, None, 2, PLAIN); + w.bit(true); + reference(&mut w, 3); + w.bit(false).uint(4).uint(0); + w +} + +fn markup() -> W { + let mut w = W::default(); + w.uint(502); + graphics(&mut w, Some("Note"), 3, PLAIN); + w.uint(1).uint(2).uint(1); + reference(&mut w, 7); + w.uint(2); + reference(&mut w, 2); + reference(&mut w, 2); + w.uint(3).uint(0); + w +} + +fn annotation(t: u32, inner: impl FnOnce(&mut W)) -> W { + let mut w = W::default(); + w.uint(t); + graphics(&mut w, None, 4, PLAIN); + inner(&mut w); + w.uint(0); + w +} + +fn annotation_item() -> W { + annotation(504, |w| reference(w, 3)) +} + +fn annotation_set(members: &[W]) -> W { + annotation(505, |w| { + w.uint(u32::try_from(members.len()).unwrap()); + for m in members { + w.append(m); + } + }) +} + +fn annotation_reference() -> W { + annotation(506, |w| { + w.uint(1); + reference(w, 3); + }) +} + +fn plane(w: &mut W) { + w.uint(86).bit(true); + base(w, Some("Front")); + w.uint(9).uint(0).put(0x01, 8); + for v in [1., 2., 3., 0., 0., 1., 1., 1., 1., 0., 0.] { + w.double(v); + } +} + +fn light(t: u32) -> W { + let mut w = W::default(); + w.uint(t); + ref_base(&mut w, None, 5); + w.uint(1).uint(2).uint(3); + let doubles = match t { + 732 => 6, + 733 => 4, + 734 => 11, + _ => 0, + }; + for _ in 0..doubles { + w.double(0.5); + } + w +} + +/// A scene with `lights` and `planes` clipping planes; `is_absolute` is written from 8137. +fn scene(lights: &[W], planes: usize, version: u32) -> W { + let mut w = W::default(); + w.uint(741); + ref_base(&mut w, None, 6); + w.bit(true).uint(u32::try_from(lights.len()).unwrap()); + for l in lights { + w.append(l); + } + w.bit(false).bit(false).uint(u32::try_from(planes).unwrap()); + for _ in 0..planes { + plane(&mut w); + } + w.uint(0).uint(0).uint(0); + if version >= 8137 { + w.bit(true); + } + w +} + +fn filter(w: &mut W) { + w.uint(320); + ref_base(w, None, 8); + w.bit(true).bit(true).uint(0).bit(true).uint(0).uint(0); +} + +fn view(version: u32) -> W { + let mut w = W::default(); + w.uint(501); + graphics(&mut w, Some("Top"), 7, PLAIN); + w.uint(1); + reference(&mut w, 3); + plane(&mut w); + w.bit(true).append(&scene(&[light(731)], 1, version)); + if version >= 8137 { + w.bit(true).bit(false).bit(true).uint(1); + reference(&mut w, 7); + w.uint(1); + filter(&mut w); + } + w.uint(0); + w +} + +fn views(vs: &[W]) -> W { + let mut w = W::default(); + w.uint(u32::try_from(vs.len()).unwrap()); + for v in vs { + w.append(v); + } + w +} + +#[test] +fn a_markup_linked_item_is_read_past() { + let w = markup_data(&[linked_item()], &[], &[], &[]); + read(&w, |c| c.markup_data()).unwrap(); + truncations_fail(&w, |c| c.markup_data()); +} + +#[test] +fn a_leader_is_read_past() { + let w = markup_data(&[], &[leader()], &[], &[]); + read(&w, |c| c.markup_data()).unwrap(); + truncations_fail(&w, |c| c.markup_data()); +} + +#[test] +fn a_markup_is_read_past() { + let w = markup_data(&[], &[], &[markup()], &[]); + read(&w, |c| c.markup_data()).unwrap(); + truncations_fail(&w, |c| c.markup_data()); +} + +#[test] +fn annotation_items_sets_and_references_are_read_past() { + let set = annotation_set(&[annotation_item(), annotation_set(&[annotation_reference()])]); + let w = markup_data(&[], &[], &[], &[annotation_item(), set, annotation_reference()]); + read(&w, |c| c.markup_data()).unwrap(); + truncations_fail(&w, |c| c.markup_data()); +} + +#[test] +fn annotation_sets_nested_past_the_ceiling_are_an_error() { + let mut e = annotation_item(); + for _ in 0..=super::MAX_DEPTH { + e = annotation_set(&[e]); + } + let w = markup_data(&[], &[], &[], &[e]); + assert!(matches!(read(&w, |c| c.markup_data()), Err(PrcError::Malformed(_)))); +} + +#[test] +fn an_annotation_of_another_type_is_an_error() { + let w = markup_data(&[], &[], &[], &[annotation(502, |_| {})]); + assert!(matches!(read(&w, |c| c.markup_data()), Err(PrcError::Malformed(_)))); +} + +#[test] +fn a_view_is_read_past_with_its_flags_only_from_8137() { + let w = views(&[view(8137)]); + read(&w, |c| c.views()).unwrap(); + truncations_fail(&w, |c| c.views()); + read_at(&views(&[view(8030)]), 8030, |c| c.views()).unwrap(); +} + +#[test] +fn scene_lights_of_each_kind_are_read_past() { + let w = scene(&[light(731), light(732), light(733), light(734)], 0, 8137); + read(&w, |c| c.scene_display_parameters()).unwrap(); + truncations_fail(&w, |c| c.scene_display_parameters()); + let mut old = W::default(); + old.uint(733); + ref_base(&mut old, None, 5); + old.uint(1).uint(2).uint(3).double(0.).double(0.).double(1.); + read_at(&old, 8020, |c| c.light()).unwrap(); +} + +#[test] +fn scene_clipping_planes_are_read_past() { + let w = scene(&[], 2, 8137); + read(&w, |c| c.scene_display_parameters()).unwrap(); + truncations_fail(&w, |c| c.scene_display_parameters()); +} + +#[test] +fn a_tree_with_markups_and_views_still_places_its_parts() { + let mut tail = markup_data(&[linked_item()], &[leader()], &[markup()], &[annotation_set(&[annotation_item()])]); + tail.append(&views(&[view(8137)])); + let part = part_with(Some("Plate"), PLAIN, 1, |w| super::tests::poly_item(w, 7, 0, 1, (S_RED, super::SHOW)), &tail); + let occs = [Occ { name: Some("Asm"), sons: vec![1], markups: Some(tail.clone()), ..Occ::default() }, Occ { part: 1, ..Occ::default() }]; + let f = single(&[part, plate(S_RED)], &occs); + let nodes = f.model_tree().unwrap(); + assert_eq!(nodes.len(), 2); + let p = f.placements().unwrap(); + assert_eq!(p.len(), 1); + assert!(matches!(f.file_structures[0].tessellations().unwrap().first(), Some(Tessellation::Mesh(_)))); +} diff --git a/crates/prc/src/tree/mod.rs b/crates/prc/src/tree/mod.rs new file mode 100644 index 0000000000..3ed9d8eb20 --- /dev/null +++ b/crates/prc/src/tree/mod.rs @@ -0,0 +1,329 @@ +//! The product tree: the tree section (`FileStructureTree`, 304), the globals' styles and +//! coordinate systems (303) and the model file (301), walked into placed tessellations +//! [WD 7.3.3-7.3.11, 7.5, 7.6]. +//! +//! Markups (PMI), views, scene lights and clipping planes are read past and not modelled. + +// prc-rs: where the working draft leaves an entity's field order open, the order here is +// prc-rs's format description's; each such entity's reader carries its own marker. + +use std::sync::Arc; + +use crate::bits::{BitReader, Budget}; +use crate::container::offset; +use crate::entity::{Ctx, malformed}; +use crate::geometry::{Matrix, Vec3}; +use crate::{PrcError, PrcFile, Schema, SectionKind, UniqueId}; + +mod markup; +mod node; +mod overrides; +mod read; +mod style; +mod walk; + +pub use node::{ModelNode, NameSource}; +pub use overrides::EntityOverrides; +use overrides::{Override, Target}; +pub(crate) use style::Graphics; +use style::{Globals, Paint, resolve_style}; +pub use style::{Rgba, StyleAlpha}; +use walk::{At, Walk}; + +const BASE_WITH_GRAPHICS: u32 = 2; +const MODEL_FILE: u32 = 301; +const FILE_STRUCTURE: u32 = 302; +const FILE_STRUCTURE_GLOBALS: u32 = 303; +const FILE_STRUCTURE_TREE: u32 = 304; +const PRODUCT_OCCURRENCE: u32 = 310; +const PART_DEFINITION: u32 = 311; +const FILTER: u32 = 320; +const SURF_PLANE: u32 = 86; +const CARTESIAN_TRANSFORMATION: u32 = 202; +const ENTITY_REFERENCE: u32 = 203; +const MARKUP_LINKED_ITEM: u32 = 204; +const REFERENCE_ON_PRC_BASE: u32 = 205; +const REFERENCE_ON_TOPOLOGY: u32 = 206; +const GENERAL_TRANSFORMATION: u32 = 207; +const RI_BREP_MODEL: u32 = 232; +const RI_CURVE: u32 = 233; +const RI_DIRECTION: u32 = 234; +const RI_PLANE: u32 = 235; +const RI_POINT_SET: u32 = 236; +const RI_POLY_BREP_MODEL: u32 = 237; +const RI_POLY_WIRE: u32 = 238; +const RI_SET: u32 = 239; +const RI_COORDINATE_SYSTEM: u32 = 240; +const MKP_VIEW: u32 = 501; +const MKP_MARKUP: u32 = 502; +const MKP_LEADER: u32 = 503; +const ANNOTATION_ITEM: u32 = 504; +const ANNOTATION_SET: u32 = 505; +const ANNOTATION_REFERENCE: u32 = 506; +const STYLE: u32 = 701; +const MATERIAL: u32 = 702; +const PICTURE: u32 = 703; +const TEXTURE_APPLICATION: u32 = 711; +const TEXTURE_DEFINITION: u32 = 712; +const TEXTURE_TRANSFORMATION: u32 = 713; +const LINE_PATTERN: u32 = 721; +const AMBIENT_LIGHT: u32 = 731; +const POINT_LIGHT: u32 = 732; +const DIRECTIONAL_LIGHT: u32 = 733; +const SPOT_LIGHT: u32 = 734; +const SCENE_DISPLAY_PARAMETERS: u32 = 741; +const CAMERA: u32 = 742; +/// prc-rs: a scene's `is_absolute` exists from this authoring version. +const ABSOLUTE_SCENE_FROM: u32 = 8137; +/// prc-rs: a view's flags, linked items and filters exist from this authoring version. +const VIEW_FLAGS_FROM: u32 = 8137; +/// prc-rs: a directional light's intensity exists from this authoring version. +const LIGHT_INTENSITY_FROM: u32 = 8030; +/// `TextureDefinition` mapping type "operator" and function "blend", 0-based +/// (pdf-issues #485). +const TEXTURE_MAPPING_OPERATOR: i32 = 3; +const TEXTURE_FUNCTION_BLEND: i32 = 3; +/// Texture application mode: alpha test [WD 7.5.7]. +const TEXTURE_ALPHA_TEST: u8 = 0x02; + +/// Ceiling on `RI_Set` nesting, the occurrence walk's depth and a prototype chain's length. +const MAX_DEPTH: usize = 64; +/// Ceiling on occurrences visited in one walk: prototypes share subtrees, so a small file can +/// describe an exponential walk. +const MAX_VISITS: usize = 1 << 20; +/// Ceiling on placements one walk yields. +pub const MAX_PLACEMENTS: usize = 1 << 20; + +/// `GraphicsContent` behaviour bits [WD 7.2.4.2]. +const SHOW: u16 = 0x0001; +const SON_HERIT_COLOR: u16 = 0x0008; +const FATHER_HERIT_COLOR: u16 = 0x0010; +const REMOVED: u16 = 0x2000; +/// `product_behavior` SUPPRESSED [WD 7.3.10]. +const SUPPRESSED: u8 = 0x01; + +/// Per file structure, entry `b` is the paint of biased style `b`. +type Palettes = Arc<[Arc<[Option]>]>; + +/// A tessellation drawn at a place in the model: world = [`Self::matrix`] × the +/// tessellation's own coordinates. +#[derive(Debug, Clone, PartialEq)] +#[non_exhaustive] +pub struct Placement { + /// Index into [`PrcFile::file_structures`]. + pub file_structure: usize, + /// Index into that file structure's [`tessellations`](crate::FileStructure::tessellations), + /// as the tree stores it: [`PrcFile::placements`] does not decode that section, so an + /// index past its end is the caller's to refuse. + pub tessellation: usize, + /// The occurrence path's locations, then the representation item's coordinate systems. + pub matrix: Matrix, + /// The colour of the style the tree resolves for this item; `None` when no style + /// reaches it or the style names no colour this reader resolves (a textured material + /// without a base). Faces with their own style are coloured by + /// [`Self::triangle_colours`]. + pub colour: Option, + /// The graphics from the root to this item, outermost first. + chain: Vec, + palettes: Palettes, + /// An occurrence's entity reference replacing the item's style. + item_override: Option, + /// Per face index, sorted, an entity reference replacing that face's style. + face_overrides: Vec<(u32, Graphics)>, +} + +/// The paint of the style `won` names, in its file structure's globals. +fn paint_of(palettes: &Palettes, won: Graphics) -> Option { + palettes.get(won.fs)?.get(offset(won.style)).copied().flatten() +} + +impl Placement { + /// Per triangle of `mesh` (this placement's tessellation), its colour as + /// [`Self::colour`] describes, with each face's own style taking part in the inheritance + /// [WD 7.2.4, 7.8.6]; `None` when no face of `mesh` carries a style or an override, so + /// every triangle is [`Self::colour`]. + #[must_use] + pub fn triangle_colours(&self, mesh: &crate::TriangleMesh) -> Option>> { + if mesh.triangle_styles.is_empty() && self.face_overrides.is_empty() { + return None; + } + let faces = overrides::triangle_faces(mesh); + let by_face = |k: usize| { + let f = faces.get(k)?; + let i = self.face_overrides.binary_search_by_key(f, |(i, _)| *i).ok()?; + self.face_overrides.get(i).map(|(_, g)| *g) + }; + let colour = |won| paint_of(&self.palettes, won).map(|p| p.rgba); + if mesh.triangle_styles.is_empty() { + let whole = self.item_override.unwrap_or_else(|| resolve_style(&self.chain)); + return Some((0..mesh.triangles.len()).map(|k| colour(by_face(k).unwrap_or(whole))).collect()); + } + let mut chain = self.chain.clone(); + chain.push(Graphics::default()); + let looks = mesh.triangle_styles.iter().enumerate().map(|(k, s)| { + if let Some(won) = by_face(k).or(self.item_override) { + return colour(won); + } + if let Some(last) = chain.last_mut() { + *last = Graphics { style: s.style, bits: s.behaviour, fs: self.file_structure }; + } + colour(resolve_style(&chain)) + }); + Some(looks.collect()) + } +} + +/// A representation item that can be drawn, flattened out of any `RI_Set`. +#[derive(Debug, Clone, PartialEq)] +struct Item { + /// `index_local_coordinate_system + 1` of each enclosing set and the item, outermost + /// first; 0 entries dropped. + local: Vec, + /// `index_tessellation + 1`; never 0 here. + tessellation: u32, + /// The part's graphics, each enclosing set's, then the item's own. + graphics: Vec, + /// The PRC unique id of each enclosing set, then the item's own. + uids: Vec, + /// A B-rep model's stored `(context, body)` ids [WD 7.6.2]. + brep: Option<(u32, u32)>, +} + +/// A `ProductOccurrence` (310) as far as placing and naming. +#[derive(Debug, Clone, PartialEq, Default)] +struct Product { + part: u32, + prototype: u32, + prototype_fs: Option, + external: u32, + external_fs: Option, + sons: Vec, + /// Its own graphics hide it (Show clear or Removed set). + hidden: bool, + suppressed: bool, + graphics: Graphics, + location: Option, + name: Option, + /// Its entity references that name a target. + overrides: Vec, +} + +/// One file structure's tree section. +#[derive(Debug, Clone, PartialEq, Default)] +struct Tree { + parts: Vec>, + /// Each part definition's name, parallel to `parts`. + part_names: Vec>, + products: Vec, +} + +/// The model file's roots: file structure id and biased occurrence index. +#[derive(Debug, Clone, PartialEq)] +struct ModelFile { + roots: Vec<(UniqueId, u32)>, +} + +impl PrcFile { + /// Where each drawable tessellation sits in the model: the product-occurrence tree walked + /// from the model file's roots, each occurrence's location composed onto its father's, + /// and each representation item's local coordinate systems applied last [WD 7.3.10.1, + /// 7.6.3.2], coloured with [`StyleAlpha::ZeroUnset`] and entity references applied by + /// [`EntityOverrides::Subtree`]. Suppressed occurrences and hidden entities are left + /// out; a tessellation several occurrences draw appears once per occurrence. Empty when + /// the file stores no tree ([`Self::stores_tree`]). + /// + /// # Errors + /// [`PrcError::Unsupported`] when an item or an entity reference names a reference + /// coordinate system and the globals carry fonts or fill patterns; [`PrcError::Malformed`] for a tree index naming no entity, + /// a cycle of occurrences or prototypes, or a tree past [`MAX_PLACEMENTS`] or the depth + /// ceiling; read errors as [`crate::FileStructure::tessellations`]. + pub fn placements(&self) -> Result, PrcError> { + self.placements_with(StyleAlpha::default(), EntityOverrides::default()) + } + + /// [`Self::placements`], combining style transparency with material alpha by `rule` and + /// applying occurrences' entity references by `scope`. + /// + /// # Errors + /// As [`Self::placements`]. + pub fn placements_with(&self, rule: StyleAlpha, scope: EntityOverrides) -> Result, PrcError> { + Ok(self.walk(rule, scope)?.0) + } + + /// The assembly tree as a model-tree panel lists it: every product occurrence the model + /// file's roots reach, depth first, with its display name and stored visibility. Hidden + /// and suppressed subtrees are listed (with `drawn` false) though [`Self::placements`] + /// leaves them out; each node's `placements` range indexes that list, so attributes each + /// placed mesh to its part. Empty when the file stores no tree. + /// + /// # Errors + /// As [`Self::placements`]. + pub fn model_tree(&self) -> Result, PrcError> { + Ok(self.walk(StyleAlpha::default(), EntityOverrides::default())?.1) + } + + /// Whether the file stores a tree: its model file holds an entity. ISO 14739-1 requires + /// one, but a tessellation-only file still draws (unplaced), so its absence reads as an + /// empty tree, not an error. + #[must_use] + pub fn stores_tree(&self) -> bool { + self.model_file.iter().any(|&b| b != 0) + } + + fn walk(&self, rule: StyleAlpha, scope: EntityOverrides) -> Result<(Vec, Vec), PrcError> { + self.walk_within(rule, scope, Budget::FULL) + } + + /// [`Self::walk`] with `budget` shared by all its readers. + pub(crate) fn walk_within(&self, rule: StyleAlpha, scope: EntityOverrides, budget: Budget) -> Result<(Vec, Vec), PrcError> { + if !self.stores_tree() { + return Ok((Vec::new(), Vec::new())); + } + // One allowance for the whole walk, so many sections cannot each spend a full one. + let mut left = budget; + let mut trees = Vec::with_capacity(self.file_structures.len()); + for fs in &self.file_structures { + // The globals open with the schema every reader of this structure uses: read it + // once, charged to the walk, and go on from there for the globals themselves. + let mut globals_reader = BitReader::with_budget(fs.section(SectionKind::Globals), left); + let schema = Schema::read(&mut globals_reader)?; + left = globals_reader.budget(); + let mut ctx = Ctx::new(BitReader::with_budget(fs.section(SectionKind::Tree), left), &schema, fs.authoring_version); + let tree = ctx.file_structure_tree()?; + left = ctx.r.budget(); + // The globals carry the coordinate systems and the colours; colours alone are not + // worth refusing the placements over. + let mut ctx = Ctx::new(globals_reader.with_remaining(left), &schema, fs.authoring_version); + let globals = ctx.globals(); + left = ctx.r.budget(); + let globals = match globals { + Ok(g) => g, + Err(e) if places_by_system(&tree) => return Err(e), + Err(_) => Globals::default(), + }; + trees.push((fs.id, tree, globals)); + } + let mut r = BitReader::with_budget(&self.model_file, left); + let schema = Schema::read(&mut r)?; + let model = Ctx::new(r, &schema, self.header.authoring_version).model_file()?; + let mut walk = Walk::new(trees.iter().map(|(id, t, g)| (*id, t, g)).collect(), rule, scope); + for (id, root) in model.roots { + let fs = trees.iter().position(|t| t.0 == id).ok_or_else(|| malformed("a root names an unknown file structure".into()))?; + if let Some(index) = root.checked_sub(1) { + walk.occurrence(fs, offset(index), &Matrix::IDENTITY, &[], At::ROOT)?; + } + } + Ok((walk.out, walk.nodes)) + } +} + +/// Whether an item of `tree`, or an entity reference one of its occurrences holds, names a +/// coordinate system: then the globals that store them must have been read. +fn places_by_system(tree: &Tree) -> bool { + tree.parts.iter().flatten().any(|i| !i.local.is_empty()) || tree.products.iter().flat_map(|p| &p.overrides).any(|o| o.local != 0) +} + +#[cfg(test)] +mod markup_tests; +#[cfg(test)] +mod tests; diff --git a/crates/prc/src/tree/node.rs b/crates/prc/src/tree/node.rs new file mode 100644 index 0000000000..19a8e04e5c --- /dev/null +++ b/crates/prc/src/tree/node.rs @@ -0,0 +1,68 @@ +//! The assembly tree as a list: what a model-tree panel shows. + +use std::ops::Range; + +/// Where a [`ModelNode`]'s name came from. PRC does not say which name a +/// viewer should show; this reader takes the first present of these, in order. +#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)] +#[non_exhaustive] +pub enum NameSource { + /// The occurrence's own name. + Occurrence, + /// The name of an occurrence in its prototype chain. + Prototype, + /// The name of the part definition it draws. + Part, + /// No name anywhere: [`ModelNode::name`] is `None`. + Unnamed, +} + +/// One product occurrence of a PRC model's assembly tree, as +/// [`PrcFile::model_tree`](crate::PrcFile::model_tree) lists it: depth +/// first, a parent before its children, children in file order. +#[derive(Debug, Clone, PartialEq, Eq)] +#[non_exhaustive] +pub struct ModelNode { + /// The name to show, by [`Self::name_from`]; `None` when unnamed. + pub name: Option, + /// Where [`Self::name`] came from. + pub name_from: NameSource, + /// Index of the parent node in the list; `None` for a root. + pub parent: Option, + /// 0 for a root. + pub depth: usize, + /// Index into [`crate::PrcFile::file_structures`] of the structure + /// holding this occurrence. + pub file_structure: usize, + /// The occurrence's 0-based index in that structure's tree. + pub occurrence: usize, + /// The occurrence's own graphics hide it (Show clear, or Removed) + /// [WD 7.2.4.2]: the visibility the file stores, which a viewer opens on. + pub hidden: bool, + /// The occurrence is suppressed (`product_behavior`) [WD 7.3.10]. + pub suppressed: bool, + /// Whether anything of this node is drawn: false when it or an + /// ancestor is hidden or suppressed. + pub drawn: bool, + /// Whether the occurrence (or its prototype) carries a part definition, + /// i.e. geometry of its own rather than only children. + pub has_part: bool, + /// The placements this node and its descendants draw: indices into + /// [`PrcFile::placements`](crate::PrcFile::placements) (or + /// [`placements_with`](crate::PrcFile::placements_with), which yields the same order): + /// the meshes attributed to this occurrence's part and its descendants' parts. Empty when + /// nothing under it is drawn. + pub placements: Range, +} + +impl ModelNode { + /// The label the node is listed under: its name, or `occurrence F:I` + /// (file structure, occurrence) when it has none. + #[must_use] + pub fn label(&self) -> String { + match &self.name { + Some(name) => name.clone(), + None => format!("occurrence {}:{}", self.file_structure, self.occurrence), + } + } +} diff --git a/crates/prc/src/tree/overrides.rs b/crates/prc/src/tree/overrides.rs new file mode 100644 index 0000000000..a85a83ba09 --- /dev/null +++ b/crates/prc/src/tree/overrides.rs @@ -0,0 +1,128 @@ +//! Product-occurrence entity references: the colour, visibility and placement an occurrence +//! imposes on an entity it reaches [WD 7.3.10.1, 7.4.4]. + +use super::{Graphics, Item, RI_BREP_MODEL, RI_COORDINATE_SYSTEM, UniqueId}; + +/// `PRC_TYPE_TOPO_Face` [WD 6.2]. +const TOPO_FACE: u32 = 149; + +/// Which entities an occurrence's entity references reach. ISO 14739-1 says a reference +/// overrides the referenced entity's properties, and not where. +#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Default)] +#[non_exhaustive] +pub enum EntityOverrides { + /// Only where the occurrence's own subtree places the target, so one instance of a + /// shared part can be coloured alone. + #[default] + Subtree, + /// Every placement of the target anywhere in the model. + Everywhere, + /// Read past them: every entity keeps its own and inherited graphics. + Ignore, +} + +/// One `MISC_EntityReference` held by a product occurrence. +#[derive(Debug, Clone, PartialEq)] +pub(super) struct Override { + /// The reference's graphics; style 0 leaves the target's colour alone. + pub(super) graphics: Graphics, + /// The graphics hide the target (Show clear or Removed set). + pub(super) hidden: bool, + /// `index_local_coordinate_system + 1` replacing the target's; 0 none. + pub(super) local: u32, + pub(super) target: Target, +} + +/// What an entity reference names. +#[derive(Debug, Clone, PartialEq)] +pub(super) enum Target { + /// `ReferenceOnPRCBase`: the entity of type `kind` whose PRC unique id is `uid` in + /// structure `fs` (`None`: the occurrence's own). + Entity { kind: u32, fs: Option, uid: u32 }, + /// `ReferenceOnTopology` into the B-rep stored as `(context, body)`: its topological + /// items of type `kind` at `indices` (0-based). With `fs` `None` it matches that B-rep in + /// any structure the subtree reaches: producers write "same structure" for an assembly's + /// reference into a part's own structure. + Topology { kind: u32, fs: Option, brep: (u32, u32), indices: Vec }, +} + +impl Target { + /// The structure the reference names, if it names one. + pub(super) fn fs(&self) -> Option { + match self { + Target::Entity { fs, .. } | Target::Topology { fs, .. } => *fs, + } + } +} + +/// An entity reference in force: the structure of the occurrence holding it, the structure +/// it names (`None`: any), and the reference. +pub(super) type Active<'t> = (usize, Option, &'t Override); + +/// The overrides that reach one representation item. +#[derive(Debug, Default)] +pub(super) struct Reached { + /// Graphics replacing the item's resolved style, in its owner's structure. + pub(super) item: Option, + pub(super) hidden: bool, + /// The owner's structure and its local coordinate system. + pub(super) local: Option<(usize, u32)>, + /// Per face index, sorted, graphics replacing that face's style. + pub(super) faces: Vec<(u32, Graphics)>, +} + +impl Reached { + /// Applies `active` innermost first, so an outer occurrence's override wins over an + /// inner one's, as an assembly's appearance beats its sub-assembly's. + pub(super) fn collect(active: &[Active<'_>], fs: usize, item: &Item) -> Self { + let mut r = Reached::default(); + // A map, not a list searched per index: one reference may name millions of faces. + let mut faces = std::collections::BTreeMap::new(); + for &(owner, named, ov) in active.iter().rev() { + let in_fs = named.is_none_or(|n| n == fs); + let g = Graphics { fs: owner, ..ov.graphics }; + match &ov.target { + Target::Entity { kind, uid, .. } => { + if !in_fs || !is_item(*kind) || !item.uids.contains(uid) { + continue; + } + if ov.graphics.style != 0 { + r.item = Some(g); + } + r.hidden = ov.hidden; + if ov.local != 0 { + r.local = Some((owner, ov.local)); + } + } + Target::Topology { kind, brep, indices, .. } => { + if !in_fs || *kind != TOPO_FACE || item.brep != Some(*brep) || ov.graphics.style == 0 { + continue; + } + for &f in indices { + faces.insert(f, g); + } + } + } + } + r.faces = faces.into_iter().collect(); + r + } +} + +/// Representation-item types a `ReferenceOnPRCBase` can colour. +fn is_item(kind: u32) -> bool { + (RI_BREP_MODEL..=RI_COORDINATE_SYSTEM).contains(&kind) +} + +/// Per triangle of `mesh`, the index of the `TESS_Face` it was read from (`u32::MAX` for +/// none). +pub(super) fn triangle_faces(mesh: &crate::TriangleMesh) -> Vec { + let mut out = vec![u32::MAX; mesh.triangles.len()]; + for (i, range) in mesh.faces.iter().enumerate() { + let face = u32::try_from(i).unwrap_or(u32::MAX); + for f in out.get_mut(range.clone()).into_iter().flatten() { + *f = face; + } + } + out +} diff --git a/crates/prc/src/tree/read.rs b/crates/prc/src/tree/read.rs new file mode 100644 index 0000000000..ff3affaeee --- /dev/null +++ b/crates/prc/src/tree/read.rs @@ -0,0 +1,452 @@ +//! The tree section's and the model file's entities: part definitions, representation items, +//! product occurrences and their transformations and entity references [WD 7.3, 7.4, 7.6]. + +use super::{ + ABSOLUTE_SCENE_FROM, BASE_WITH_GRAPHICS, CAMERA, CARTESIAN_TRANSFORMATION, ENTITY_REFERENCE, FILE_STRUCTURE, FILE_STRUCTURE_TREE, FILTER, + GENERAL_TRANSFORMATION, Graphics, Item, MAX_DEPTH, MODEL_FILE, Matrix, ModelFile, Override, PART_DEFINITION, PRODUCT_OCCURRENCE, Product, + REFERENCE_ON_PRC_BASE, REFERENCE_ON_TOPOLOGY, REMOVED, RI_BREP_MODEL, RI_COORDINATE_SYSTEM, RI_CURVE, RI_DIRECTION, RI_PLANE, RI_POINT_SET, + RI_POLY_BREP_MODEL, RI_POLY_WIRE, RI_SET, SCENE_DISPLAY_PARAMETERS, SHOW, SUPPRESSED, Target, Tree, UniqueId, Vec3, +}; +use crate::PrcError; +use crate::entity::{Ctx, malformed}; + +impl Ctx<'_, '_> { + pub(super) fn vector3(&mut self) -> Result { + Ok(Vec3([self.r.double()?, self.r.double()?, self.r.double()?])) + } + + pub(super) fn unique_id(&mut self) -> Result { + let mut id = [0; 4]; + for v in &mut id { + *v = self.r.unsigned_integer()?; + } + Ok(UniqueId(id)) + } + + /// `ContentPRCRefBase` [WD 7.2.3.2; pdf-issues #697]: the base, then the CAD ids and the + /// PRC unique id, which becomes [`Ctx::uid`]. + pub(super) fn content_prc_ref_base(&mut self) -> Result<(), PrcError> { + self.content_prc_base()?; + self.r.unsigned_integer()?; // CAD identifier + self.r.unsigned_integer()?; // CAD persistent identifier + self.uid = self.r.unsigned_integer()?; + Ok(()) + } + + /// `PRCBaseWithGraphics` [WD 7.2.4; pdf-issues #405]: whether the entity's own graphics + /// hide it (Show clear or Removed set), and its graphics. Reused graphics + /// (`same_graphics`) are the current graphics and never hide. + pub(super) fn base_with_graphics(&mut self) -> Result<(bool, Graphics), PrcError> { + self.content_prc_ref_base()?; + let hidden = if self.r.bit()? { + false + } else { + self.r.unsigned_integer()?; // layer + 1 + let style = self.r.unsigned_integer()?; + let lo = u16::from(self.r.character()?); + let hi = u16::from(self.r.character()?); + let bits = lo | hi << 8; + self.graphics = Graphics { style, bits, fs: 0 }; + bits & SHOW == 0 || bits & REMOVED != 0 + }; + // prc-rs: schema additions to the base itself follow the graphics. + self.schema.skip_added_fields(BASE_WITH_GRAPHICS, &mut self.r)?; + Ok((hidden, self.graphics)) + } + + /// `CartesianTransformation` (202) or `GeneralTransformation` (207), type code included + /// [WD 7.4.8, 7.4.9, 7.4.11; pdf-issues #430]. + pub(super) fn transformation(&mut self) -> Result { + let t = self.r.unsigned_integer()?; + let m = match t { + GENERAL_TRANSFORMATION => { + // Column-major [WD 7.4.9]. + let mut m = Matrix::IDENTITY; + for col in 0..4 { + for row in &mut m.0 { + if let Some(c) = row.get_mut(col) { + *c = self.r.double()?; + } + } + } + m + } + CARTESIAN_TRANSFORMATION => self.transformation_3d()?, + t => return Err(malformed(format!("entity type {t} as a transformation"))), + }; + self.schema.skip_added_fields(t, &mut self.r)?; + Ok(m) + } + + /// The embedded `Transformation3D` [WD 7.4.11.1]: a behaviour byte, then each field its + /// bits name. + pub(super) fn transformation_3d(&mut self) -> Result { + let b = self.r.character()?; + let mut m = Matrix::IDENTITY; + let set_col = |m: &mut Matrix, col: usize, v: Vec3| { + for (row, x) in m.0.iter_mut().zip(v.0) { + if let Some(c) = row.get_mut(col) { + *c = x; + } + } + }; + if b & 0x01 != 0 { + let t = self.vector3()?; + set_col(&mut m, 3, t); + } + if b & 0x20 != 0 { + for col in 0..3 { + let v = self.vector3()?; + set_col(&mut m, col, v); + } + } else if b & 0x02 != 0 { + let x = self.vector3()?; + let y = self.vector3()?; + // Mirrored (0x04): the third axis is y × x. + let z = if b & 0x04 != 0 { y.cross(x) } else { x.cross(y) }; + set_col(&mut m, 0, x); + set_col(&mut m, 1, y); + set_col(&mut m, 2, z); + } + let mut scale = [1.0; 3]; + if b & 0x10 != 0 { + scale = self.vector3()?.0; + } + // prc-rs: with both scale bits set, both are read and multiply. + if b & 0x08 != 0 { + let s = self.r.double()?; + scale = scale.map(|v| v * s); + } + for row in m.0.iter_mut().take(3) { + for (c, s) in row.iter_mut().zip(scale) { + *c *= s; + } + } + if b & 0x40 != 0 { + let h = [self.r.double()?, self.r.double()?, self.r.double()?, self.r.double()?]; + if let Some(r) = m.0.get_mut(3) { + *r = h; + } + } + Ok(m) + } + + /// A polymorphic representation item [WD 7.6]; drawable ones are pushed onto `out` with + /// `local` (the enclosing sets' coordinate systems) prefixed. + // prc-rs: the item's own fields follow its local coordinate system and tessellation. + fn representation_item(&mut self, at: &Item, out: &mut Vec, depth: usize) -> Result<(), PrcError> { + if depth > MAX_DEPTH { + return Err(malformed("representation sets nest too deep".into())); + } + let t = self.r.unsigned_integer()?; + let (hidden, own) = self.base_with_graphics()?; + let mut item = at.clone(); + item.uids.push(self.uid); + item.graphics.push(own); + let cs = self.r.unsigned_integer()?; + let tess = self.r.unsigned_integer()?; + if cs != 0 { + item.local.push(cs); + } + let (drawable, brep) = if t == RI_SET { + let n = self.count(1, "representation set")?; + let mut members = Vec::new(); + for _ in 0..n { + self.representation_item(&item, &mut members, depth + 1)?; + } + if !hidden { + out.extend(members); + } + (false, None) + } else { + self.leaf_item_fields(t)? + }; + self.schema.skip_added_fields(t, &mut self.r)?; + self.user_data()?; + if drawable && !hidden && tess != 0 { + out.push(Item { tessellation: tess, brep, ..item }); + } + Ok(()) + } + + /// The type-specific fields of a representation item other than `RI_Set` [WD 7.6]; + /// whether the item can be drawn, and a B-rep model's `(context, body)`. + fn leaf_item_fields(&mut self, t: u32) -> Result<(bool, Option<(u32, u32)>), PrcError> { + match t { + RI_BREP_MODEL => { + let brep = self.exact_tolerance()?; + self.r.bit()?; // is_closed + return Ok((true, brep)); + } + RI_CURVE | RI_PLANE => { + self.exact_tolerance()?; + } + RI_DIRECTION => { + if self.r.bit()? { + self.vector3()?; // origin + } + self.vector3()?; + return Ok((false, None)); + } + RI_POINT_SET => { + let n = self.count(6, "point set")?; + for _ in 0..n { + self.vector3()?; + } + } + RI_POLY_BREP_MODEL => { + self.r.bit()?; // is_closed + } + RI_POLY_WIRE => {} + RI_COORDINATE_SYSTEM => { + self.transformation()?; + return Ok((false, None)); + } + t => return Err(malformed(format!("entity type {t} as a representation item"))), + } + Ok((true, None)) + } + + /// A B-rep, curve or plane item's optional exact-geometry `(context, body)` ids + /// [WD 7.6.2]. + fn exact_tolerance(&mut self) -> Result, PrcError> { + if self.r.bit()? { + return Ok(Some((self.r.unsigned_integer()?, self.r.unsigned_integer()?))); + } + Ok(None) + } + + /// `SceneDisplayParameters` (741) [WD 7.5.19]: read past, lights and clipping planes + /// included. + // prc-rs: the scene's field order, and `is_absolute` from version 8137. + pub(super) fn scene_display_parameters(&mut self) -> Result<(), PrcError> { + self.expect_type(SCENE_DISPLAY_PARAMETERS)?; + self.content_prc_ref_base()?; + self.r.bit()?; // is_active + let n = self.count(1, "scene lights")?; + for _ in 0..n { + self.light()?; + } + if self.r.bit()? { + self.camera()?; + } + if self.r.bit()? { + self.vector3()?; // rotation centre + } + let n = self.count(1, "scene clipping planes")?; + for _ in 0..n { + self.plane()?; + } + self.r.unsigned_integer()?; // background style + 1 + self.r.unsigned_integer()?; // default style + 1 + let n = self.count(2, "scene default styles")?; + for _ in 0..n { + self.r.unsigned_integer()?; // entity type + self.r.unsigned_integer()?; // style + 1 + } + if self.version >= ABSOLUTE_SCENE_FROM { + self.r.bit()?; // is_absolute + } + self.schema.skip_added_fields(SCENE_DISPLAY_PARAMETERS, &mut self.r) + } + + /// `Camera` (742) [WD 7.5.20]: read past. + // prc-rs: the camera's field order. + fn camera(&mut self) -> Result<(), PrcError> { + self.expect_type(CAMERA)?; + self.content_prc_ref_base()?; + self.r.bit()?; // is_orthographic + for _ in 0..3 { + self.vector3()?; // position, look, up + } + for _ in 0..6 { + self.r.double()?; // x, y, ratio, near, far, zoom + } + self.schema.skip_added_fields(CAMERA, &mut self.r) + } + + /// `MISC_EntityReference` (203) [WD 7.4.4, 7.4.10]; `None` when it names no target. + // prc-rs: the local coordinate system precedes the optional reference data. + fn entity_reference(&mut self) -> Result, PrcError> { + self.expect_type(ENTITY_REFERENCE)?; + let (hidden, graphics) = self.base_with_graphics()?; + let local = self.r.unsigned_integer()?; + let target = if self.r.bit()? { self.reference_data()? } else { None }; + self.schema.skip_added_fields(ENTITY_REFERENCE, &mut self.r)?; + self.user_data()?; + Ok(target.map(|target| Override { graphics, hidden, local, target })) + } + + /// `ReferenceData` [WD 7.4.10.2]: a `ReferenceOnPRCBase` (205) or a + /// `ReferenceOnTopology` (206), type code included; `None` when it names no target. + pub(super) fn reference_data(&mut self) -> Result, PrcError> { + let t = self.r.unsigned_integer()?; + let target = match t { + REFERENCE_ON_PRC_BASE => Some(self.reference_on_prc_base_body()?), + REFERENCE_ON_TOPOLOGY => self.reference_on_topology()?, + t => return Err(malformed(format!("entity type {t} as reference data"))), + }; + self.schema.skip_added_fields(t, &mut self.r)?; + Ok(target) + } + + /// `ReferenceOnPRCBase` (205) [WD 7.4.6] after its type code. + fn reference_on_prc_base_body(&mut self) -> Result { + let kind = self.r.unsigned_integer()?; + let fs = self.file_identifier()?; + let uid = self.r.unsigned_integer()?; + Ok(Target::Entity { kind, fs, uid }) + } + + /// `ReferenceOnPRCBase` (205) [WD 7.4.6], type code included: read past. + pub(super) fn reference_on_prc_base(&mut self) -> Result<(), PrcError> { + self.expect_type(REFERENCE_ON_PRC_BASE)?; + self.reference_on_prc_base_body()?; + self.schema.skip_added_fields(REFERENCE_ON_PRC_BASE, &mut self.r) + } + + /// `ReferenceOnTopology` (206) after its type code; `None` without a body. + fn reference_on_topology(&mut self) -> Result, PrcError> { + let kind = self.r.unsigned_integer()?; + if !self.r.bit()? { + return Ok(None); + } + let fs = self.file_identifier()?; + let brep = (self.r.unsigned_integer()?, self.r.unsigned_integer()?); + let indices = self.uints("topology reference indices")?; + Ok(Some(Target::Topology { kind, fs, brep, indices })) + } + + /// `ASM_Filter` (320) [WD 7.3.12]: read past. + // prc-rs: the filter's field order. + pub(super) fn filter(&mut self) -> Result<(), PrcError> { + self.expect_type(FILTER)?; + self.content_prc_ref_base()?; + self.r.bit()?; // is_active + self.r.bit()?; // layers inclusive + self.uints("filter layers")?; + self.r.bit()?; // entities inclusive + let n = self.count(1, "filter entities")?; + for _ in 0..n { + self.entity_reference()?; + } + self.schema.skip_added_fields(FILTER, &mut self.r)?; + self.user_data() + } + + /// `PartDefinition` (311): its drawable items and its name. + fn part_definition(&mut self) -> Result<(Vec, Option), PrcError> { + self.expect_type(PART_DEFINITION)?; + let (hidden, own) = self.base_with_graphics()?; + let name = self.name.clone(); + self.vector3()?; // bounding box + self.vector3()?; + let n = self.count(1, "representation items")?; + let mut items = Vec::new(); + let at = Item { local: Vec::new(), tessellation: 0, graphics: vec![own], uids: Vec::new(), brep: None }; + for _ in 0..n { + self.representation_item(&at, &mut items, 0)?; + } + self.markup_data()?; + self.views()?; + self.schema.skip_added_fields(PART_DEFINITION, &mut self.r)?; + self.user_data()?; + if hidden { + items.clear(); + } + Ok((items, name)) + } + + /// `FileIdentifier` [WD 7.3.10.2.2]: `None` for the same structure. + fn file_identifier(&mut self) -> Result, PrcError> { + if self.r.bit()? { Ok(None) } else { self.unique_id().map(Some) } + } + + /// `ProductOccurrence` (310) [WD 7.3.10]. + fn product_occurrence(&mut self) -> Result { + self.expect_type(PRODUCT_OCCURRENCE)?; + let (hidden, graphics) = self.base_with_graphics()?; + let mut p = Product { hidden, graphics, name: self.name.clone(), part: self.r.unsigned_integer()?, ..Product::default() }; + p.prototype = self.r.unsigned_integer()?; + if p.prototype != 0 { + p.prototype_fs = self.file_identifier()?; + } + p.external = self.r.unsigned_integer()?; + if p.external != 0 { + p.external_fs = self.file_identifier()?; + } + p.sons = self.uints("son occurrences")?; + p.suppressed = self.r.character()? & SUPPRESSED != 0; + // ProductInformation [WD 7.3.10.3]: unit flag, unit, load status, information. + self.r.bit()?; + self.r.double()?; + self.r.character()?; + self.r.integer()?; + if self.r.bit()? { + p.location = Some(self.transformation()?); + } + let n = self.count(1, "entity references")?; + for _ in 0..n { + p.overrides.extend(self.entity_reference()?); + } + self.markup_data()?; + self.views()?; + if self.r.bit()? { + self.filter()?; + } + let n = self.count(1, "display filters")?; + for _ in 0..n { + self.filter()?; + } + let n = self.count(1, "scene display parameters")?; + for _ in 0..n { + self.scene_display_parameters()?; + } + self.schema.skip_added_fields(PRODUCT_OCCURRENCE, &mut self.r)?; + self.user_data()?; + Ok(p) + } + + /// `FileStructureTree` (304) [WD 7.3.6], with its `FileStructure` (302) inside. + pub(super) fn file_structure_tree(&mut self) -> Result { + self.expect_type(FILE_STRUCTURE_TREE)?; + self.content_prc_base()?; + let n = self.count(1, "part definitions")?; + let mut tree = Tree::default(); + for _ in 0..n { + let (items, name) = self.part_definition()?; + tree.parts.push(items); + tree.part_names.push(name); + } + let n = self.count(1, "product occurrences")?; + for _ in 0..n { + tree.products.push(self.product_occurrence()?); + } + self.expect_type(FILE_STRUCTURE)?; + self.content_prc_base()?; + self.r.unsigned_integer()?; // next_available_index + self.r.unsigned_integer()?; // product_occurrence_index + 1 (the structure's root) + self.schema.skip_added_fields(FILE_STRUCTURE, &mut self.r)?; + self.schema.skip_added_fields(FILE_STRUCTURE_TREE, &mut self.r)?; + self.user_data()?; + Ok(tree) + } + + /// `ModelFile` (301) after its schema [WD 7.3.3; pdf-issues #724, #740]: `units_in_mm` is + /// always present. + pub(super) fn model_file(&mut self) -> Result { + self.expect_type(MODEL_FILE)?; + self.content_prc_base()?; + self.r.bit()?; // units_from_cad_file + self.r.double()?; // units_in_mm + let n = self.count(6, "root occurrences")?; + let mut roots = Vec::with_capacity(n); + for _ in 0..n { + let id = self.unique_id()?; + let index = self.r.unsigned_integer()?; + self.r.bit()?; // is_active + roots.push((id, index)); + } + Ok(ModelFile { roots }) + } +} diff --git a/crates/prc/src/tree/style.rs b/crates/prc/src/tree/style.rs new file mode 100644 index 0000000000..d313fb2f64 --- /dev/null +++ b/crates/prc/src/tree/style.rs @@ -0,0 +1,348 @@ +//! Graphics, styles, materials and the globals that hold them [WD 7.2.4, 7.3.5, 7.5]. + +use super::{ + FATHER_HERIT_COLOR, FILE_STRUCTURE_GLOBALS, LINE_PATTERN, MATERIAL, Matrix, PICTURE, RI_COORDINATE_SYSTEM, SON_HERIT_COLOR, STYLE, + TEXTURE_ALPHA_TEST, TEXTURE_APPLICATION, TEXTURE_DEFINITION, TEXTURE_FUNCTION_BLEND, TEXTURE_MAPPING_OPERATOR, TEXTURE_TRANSFORMATION, Vec3, +}; +use crate::PrcError; +use crate::container::offset; +use crate::entity::{Ctx, malformed}; + +/// One entity's `GraphicsContent` [WD 7.2.4]: `style` is `line_style_index + 1` (0 none), +/// `bits` the behaviour bits. `fs` is the file structure whose globals `style` indexes: the +/// one whose tree holds the entity. The walk sets it; a reader leaves it 0. +#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)] +pub(crate) struct Graphics { + pub(crate) style: u32, + pub(crate) bits: u16, + pub(crate) fs: usize, +} + +/// A straight (not premultiplied) colour, each channel 0-1; `a` is the opacity. +#[derive(Debug, Clone, Copy, PartialEq, Default)] +#[non_exhaustive] +pub struct Rgba { + /// Red. + pub r: f64, + /// Green. + pub g: f64, + /// Blue. + pub b: f64, + /// Opacity: 0 transparent, 1 opaque. + pub a: f64, +} + +/// How a style's `transparency` combines with its material's diffuse alpha. ISO 14739-1 +/// defines both and not their combination, and gives a diffuse alpha of 0.0 no "unset" +/// meaning. +#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Default)] +#[non_exhaustive] +pub enum StyleAlpha { + /// As [`Self::StyleWins`], except that a material diffuse alpha of exactly 0.0 under a + /// style with no transparency reads as unset, so opaque. Producers write 0.0 on the + /// materials of models meant to be seen; read literally, every part is invisible. + #[default] + ZeroUnset, + /// The style's transparency, when it has one, is the opacity; the material alpha, read + /// literally (0.0 invisible), applies only to a style without one. + StyleWins, + /// The opacity is material alpha × style transparency. + Multiply, +} + +/// A style's resolved colour, and whether its opacity comes from reading a zero material +/// alpha as unset. +#[derive(Debug, Clone, Copy, PartialEq)] +pub(crate) struct Paint { + pub(crate) rgba: Rgba, + pub(crate) alpha_unset: bool, +} + +/// A `Style` (701) as far as colour: `index` is `colour_or_material + 1`. +#[derive(Debug, Clone, Copy, PartialEq, Eq)] +pub(crate) struct Style { + pub(crate) is_material: bool, + pub(crate) index: u32, + pub(crate) transparency: Option, +} + +/// A `materials` entry as far as colour. +#[derive(Debug, Clone, Copy, PartialEq)] +pub(crate) enum Material { + /// `Material` (702): `diffuse + 1` (a colour index) and the diffuse alpha. + Plain { diffuse: u32, alpha: f64 }, + /// `TextureApplication` (711) [WD 7.5.6]: `material_generic_index + 1`, the material + /// under the texture. + Textured { base: u32 }, +} + +/// What `FileStructureGlobals` contributes to placing and colouring. +#[derive(Debug, Clone, PartialEq, Default)] +pub(crate) struct Globals { + /// Entry `i` places an item with biased local coordinate system `i + 1`. + pub(crate) systems: Vec, + /// RGB, `a` 1. + pub(crate) colours: Vec, + pub(crate) materials: Vec, + pub(crate) styles: Vec