* CI: the shared gate workflow; nightly fuzzing in fuzz.yml
* CI: run the gate now that the shared workflow is on craft-common-actions main
* CI: adding full-ci to an open PR runs the gate; fuzz checkout keeps no token
* CI speed: line tables for workspace crates, no debug info for dependencies
* CI speed: xtask ci runs nextest and doctests, one wasm check, the CLI without default features
The page counts the spacing after every character, the last of a line
included; epaint only puts it between characters, so a line could fit in
the editor and wrap in the saved text.
* Allow OFL typefaces whoever designed them (AGENTS.md §1.1)
Noto CJK and Source Han from their open-source releases are allowed like any
other OFL font. The fonts build no longer skips them, the asset check accepts
an OFL font credited to Adobe, and the docs describe the Chinese faces that
result. The ban on Adobe icons, artwork, UI graphics and the fonts Adobe
products install stays.
* Leave out whole CJK faces other than the Japanese ones for size
A Chinese or Korean face is embedded whole only under 4 MiB, so the 16.4 MB
Noto Sans CJK SC in craft-fonts stays out of the binary and the embedded
total against the pinned craft-fonts is unchanged (25,204,828 bytes).
* docs: replacement boxes for Chinese without craft-fonts only on the web or without an installed Chinese face
* xtask assets: the OFL font exemption needs at least one licence ID
* fonts build: count a CJK script tag with a region (Hant-HK) as CJK, shared with the embedding test
* AGENTS.md: the asset gate exempts OFL fonts from the Adobe-author check
* fonts build: count a Hangul-only face (Hang) as CJK for the size rule
* fonts: the CJK embedding test checks the size rule, sharing its constants with build.rs
* docs: the missing Hans face is a property of the pinned craft-fonts, not of every build
* fix(links): use resolve_absolute for remote document paths
* fix(ui): refuse output file names that end in a dot or a space
* ci: check that git dependencies come from storytold
* ci: release only with git dependencies from storytold
* ci: check git dependency sources on a locked, parsed Cargo.lock
The git-sources step in ci.yml and release.yml fails when Cargo.lock is
missing, when .cargo/config.toml replaces or patches a source, when
Cargo.lock is out of date with the manifests (cargo metadata --locked),
and when any package source parsed from Cargo.lock as TOML is a git
source outside github.com/storytold/. A grep on Cargo's usual spelling
missed single-quoted strings and stale lockfiles.
* fix(links): a Windows path on Unix is refused, not reported as needing a source
resolve_remote_document_path reported MissingSourcePath for any path that
resolved but not to an absolute one. With an absolute source, such a path
is one this platform cannot open (C:\x.pdf on Unix): it is now
NotLocal(Invalid). A relative target from a missing or relative source
still gives MissingSourcePath.
* ci: check the legacy .cargo/config for source replacement too
Cargo reads .cargo/config as well as .cargo/config.toml, so the
git-sources step greps both; grep -s keeps a missing file from
failing the step.
* ci: parse cargo config for source replacement, patches and path overrides
* Packaging: pin the Windows signing client and check its hash
* Packaging: pin appimagetool and check its hash
* Packaging: fail clearly when no pinned appimagetool exists for the architecture
riscv64 has no pinned appimagetool, so building its AppImage without APPIMAGETOOL or a tool
on PATH stopped on an unbound APPIMAGETOOL_SHA256 after a download attempt.
* Packaging: pin the AppImage runtime and check its hash
appimagetool otherwise downloads the type2-runtime continuous build and embeds it unchecked.
The script fetches release 20251108 by version, checks its SHA-256 on download and on every
run, and passes it with --runtime-file. An architecture with no pinned runtime exits with a
clear message.
The paragraph box, and the added-text box, each keep one Undo history under
one egui id, shared by every tab. A box cleared it when it opened on new
text, but a box in another tab on the same text (two copies of a document)
kept it: typing taken back with Undo in one tab came back with Redo in the
other, from the key or from the menu. The document that last used the
history is now kept beside it, and the history is cleared whenever another
document's box takes it over.
Tests: the paragraph and added-text cases, each with two copies of a
document open; the paragraph case tries the menu's Redo too.
The paragraph box, and the added-text box, each keep one Undo history under
one egui id, shared by every tab. A box cleared it when it opened on new
text, but a box in another tab on the same text (two copies of a document)
kept it: typing taken back with Undo in one tab came back with Redo in the
other. The document that last used the history is now kept beside it, and
the history is cleared whenever another document's box takes it over.
Tests: the paragraph and added-text cases, each with two copies of a
document open.
`/Annots` may hold an annotation dictionary inline instead of a reference
(the standard asks for references, but files do it and PdfCraft accepts
them). Deleting a comment or clearing redaction marks only matched
referenced entries, so an inline reply (`/IRT`) or pop-up (`/Parent`) of
what went stayed behind, pointing at nothing. The pop-up an inline reply
or an inline comment names (`/Popup`) stayed too. This predates the shared
walk.
`goes_with` matches an entry against what is going, inline entries
included; `delete_annotation` and `clear_marks` filter with it. The walk
follows the pop-up an inline reply names, and deleting an inline comment
takes the pop-up it names.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An incremental save wrote every object added during the session, whether
anything still referred to it or not. A redaction mark added and then
cleared (or a comment added and deleted) was therefore written into the
file as an unreferenced object, with its QuadPoints, overlay text and
author.
`SaveOptions::drop_unlinked` writes such additions as free entries instead:
objects whose numbers the file never used and that nothing the update
writes refers to, walked from the trailer and the file's changed objects
through additions only. `/Size` and the numbering stay as before, so undo
past the save can bring an object back under its number. The option is off
by default because the standalone document timestamp is found by scanning,
not by reference; `Session::save_bytes` (Save, and the automation save)
turns it on. Full saves already left these objects out.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Clear all wrote `/Annots []` on a page that had no annotations and replaced
an indirect `/Annots` array with an inline copy. It also left replies to a
mark in place, so their `/IRT` kept the cleared mark in the document (and in
the comment list, as a reply to nothing).
`clear_marks` now writes the list through the annot crate's `set_annots`
(made public), as deleting a comment does, and removes each mark's replies
and pop-ups with it. The "what goes with a deleted comment" walk moves from
`delete_annotation` into `with_replies_and_popups`, which both use; it visits
each annotation once instead of sweeping the list until nothing changes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- A row aligned on its own lost its alignment when the paragraph was then
given a colour, size or font: the rewrite left the rows' alignments out
once they had been applied. They now go with every rewrite.
- egui drops a box's selection while the box is unfocused, so after one
alignment or format a second choice went to the whole paragraph or box.
The selection is given back when formatting returns the focus.
- Redo from the menu or toolbar applied the box's typing first, a new edit
that dropped the document's Redo. It now redoes in the box, as the key does.
alignment_frame found the column by the left edge most of the page's other
rows share. On a page of two columns, a title over the column with fewer rows
took the other column's edge, failed to fit in it, and was aligned in its own
box. Only rows reaching across the title now count.
Tests: the two-column case; a two-colour title (the styled-region path) is
framed too; a justified paragraph sent with its rows split by line breaks, as
the editor sends it, still fills every row but the last.
Finishing a box trimmed its text before comparing it with the item, so
an unchanged box whose text was saved with trailing spaces finished as
an edit. Before an Undo or Redo step that edit was applied, clearing
the document's Redo. A box matching its item is now no edit.
The editor of a box narrower than its 60-point minimum was anchored by
the box's own alignment. A box whose lines are all right-aligned (or
centred) on their own while the box stays Left wasn't anchored, so its
text was placed against the wider editor's edge, past the box. The
lines are now compared with each other and the editor keeps the edge
they share.
Without new text, text_edit rewrote a paragraph from its text joined
with spaces, so its rows were rewrapped into one and only the first
entry of `aligns`, which counts rows, had any effect. With `aligns`
given, the paragraph's own rows (its editor text) are kept, as in the
app.
The editor is at least 60 points wide on screen, so a short box can be
clicked into. With the box's alignment applied, right-aligned or centred
text in a box narrower than that was placed against the wider editor's
edge, past the box's right edge. The editor now reaches past a narrow
box away from the edge its text keeps, so the text stays where the page
draws it.
With typing undone in a paragraph or added text box, its text matches
the page again, so Undo and Redo went to the document. Redo then closed
the box and redid a document change (or nothing), instead of putting the
typing back as the box's own Redo did before. And Shift+Cmd+Z is also
Cmd+Z to the Undo shortcut, so even a Redo the box kept was taken as a
document Undo.
Redo stays with the box while its own history has typing to put back
(the test egui makes when it handles the key), and Shift+Cmd+Z is never
a document Undo.
Centring one highlighted line of a paragraph moves it out of the
paragraph's rows, so the next alignment of it saw a one-row paragraph
and aligned it in its own box: Left left it where it was, and since a
single row grows to the page's margin, each Centre moved it further
right. An alignment for a one-row paragraph is now laid out across the
column of text it sits in (the left edge the page's other text uses, to
the furthest right those rows reach), or its own box when no column
holds it.
Format text's alignment always re-aligned the whole paragraph or added
text box, and clicking it closed an added text box, losing the
selection, so one line couldn't be aligned. With text selected,
alignment now applies to the lines (text between line breaks) the
selection touches; with only the caret, to the whole box as before.
- Added text stores each line's alignment in its /PCAdded parameters
(/Aligns, null where a line follows /Align). A justified line's last
row is not stretched.
- An edited paragraph sends its rows' alignments with each rewrite
while it is open (BlockStyle::aligns), in both the single-style and
the styled-region rewrite.
- Both editors draw each line by its own alignment, and the alignment
buttons show the selected line's. An added text box stays open while
Format text is used, its style applied at once; clicks are judged by
where they were pressed, so a touch or pen tap on the panel doesn't
close either editor.
- Automation: `aligns` on page_add_text, content_update and text_edit.
Ctrl+Z is left to a focused text field, so in the paragraph editor it
went to the box's own Undo history. That history still held the
paragraph's text from before the last formatting change, so Ctrl+Z put
it back into a box laid out for the page after the change, a layout
the page never had, and the document kept the change. The box shared
its id across paragraphs too, so it could put back another paragraph's
text.
The history of the paragraph editor and the added-text box now starts
at what the page has. Undo and Redo keep the keys only while the box
holds typing the page doesn't have; then they reach the document. An
open box is applied (as clicking away applies it) and closed before an
Undo or Redo step, so it never shows a state the document has left;
text it can't apply keeps it open and stops the step.
Opening a centred or right-aligned Add text box drew its text at the left
edge, a layout the box never had, and Ctrl+Z inside the box (which goes to
the text, not the document) left it showing that way. The editor now lays
the text out with the box's alignment. Justified text is still shown
left-aligned while editing (TextEdit can't justify).
CodeRabbit: a recognition run reads a snapshot of the document and its result
names a page by index. If the document changed while it ran (a page deleted or
moved), Session::apply_ocr would put the words on whatever page now has that
index. An Edit-started run now remembers the page and the document's edit
generation (AutoOcr); when its result arrives after a change it is dropped and
the page's "tried" mark is cleared so the page in view is judged again. A run
the user asked for is unchanged. The unit test uses a made-up result (no OCR
models) and fails without the check.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Rebasing onto main met its annotation listing fixes (c82dc91a, dfde9078):
inspect now lists each annotation once, at the first page and position
that lists it, told apart by the object its reference chain ends at,
and reads a page-shared /Annots array under its first page only. The
rebase collects reply slots in that same pass and matches /IRT by the
object it resolves to, so a reply finds its comment wherever the
listing keeps it.
The test fails with the slots keyed by the raw /Annots reference (as
before the rebase) and with /IRT matched by its raw reference.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`link_replies` makes an annotation whose /IRT leads nowhere a comment of
its own. For a status or checkmark that meant a card in the Comments
panel (and an invisible hit area on the page, and a root in
comment_list) that the header's count left out; before threading by
reference, most of these were hidden. The realistic case is a comment
deleted from another page than its status replies.
`Annotation::is_comment` (not a reply, not a status or checkmark) is now
the one test for "listed as a comment": the panel's cards, page badges,
filter menus and count, the page's hit-testing, comment_list's roots and
the summary all use it. A Replace Text caret still shows in its
strikeout's card: it holds the replacement text, and the count already
treats the pair as one comment.
From CodeRabbit review on #4061.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
From review:
- `link_replies` only finds parents among the annotations both readers
list, and neither lists links, form widgets, pop-ups or other
non-markup annotations (`is_comment_subtype`), so an /IRT to one of
those leaves its annotation a comment of its own. Its doc comment now
says so; the UI edge-case test adds an /IRT to a link and to a form
field, and the engine test an /IRT to a link (after an edit too).
Both fail if a reader lists links as comments.
- The Comments panel's grouping comment now says that a status or
checkmark set on a reply, at any depth, is that reply's: the card's
badge and checkmark leave it out, and comment_list reports it on the
reply.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The previous commit filed a status or checkmark under the annotation it
was set on, but comment_list only reported states for comments, so one
set on a reply dropped out of the listing. Each reply in `replies` now
carries its own `status` and `marked`, read the same way as a
comment's, and the tool description says so.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A status or checkmark is a state reply whose /IRT names the annotation
it applies to (ISO 32000-2 §12.5.6.2). comment_list gathered state
replies at any depth of a thread (since #839 listed nested replies), so
a checkmark or status set on a reply was reported as the comment's own,
while the Comments panel, comment properties and the page's context
menu showed the comment unmarked. comment_list now files them under the
annotation they were set on, as the panel does.
A UI regression pins the card and the page's menu (it also fails if the
menu counts checkmarks at any depth), and an automation one pins
comment_list.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The comment tools resolved `id` (the comment's /NM, as comment_list
reports it) to the first annotation with that name. /NM need not be
unique: in a PyMuPDF file every page's first comment is `fitz-A0`, so
comment_reply, comment_set_status and the rest acted on page 1's
comment whatever page the caller meant. A shared id is now refused
with a message pointing at `page` + `index`, which always name one
comment. A new end-to-end test also checks that comment_list threads
such a file by reference.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Follow-up to threading by /IRT reference, from review:
- An /IRT that reaches no listed comment (the annotation itself, a
loop, a pop-up, an object no page lists) left the annotation a reply
shown nowhere, though still counted. `link_replies` now settles the
links in both readers (inspect on open, the engine's list after an
edit): a reply keeps its parent only when the chain reaches a comment
within 64 steps, and anything else stays a comment of its own.
- A reply to a reply was never shown in the Comments panel (only
comment_list followed the chain, #806), but the new header line
counted it. The panel now files every text reply under the comment
its thread starts from (`thread_tops`); a status or checkmark stays
with the annotation it was set on.
- A Replace Text caret (/RT /Group) is grouped with its strikeout, not
a reply to it, so the header no longer counts it (new `grouped`
field on both readers).
- The engine summary and comment_list gather replies once instead of
rescanning every annotation per reply (quadratic on large reviews).
- Inspect takes the first listing in page order instead of a min over
an unordered map.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Eric Kaden
Marco Arnulfo
aiongg
srt8hellcat2-ui
Combiler
midasmatterhorn
RiceDee
Osvaldo DuranandClaude Fable 5.1
alehelandClaude Opus 5.5
PhilipandClaude Sonnet 5.5
Osvaldo Duran MelendezandClaude Opus 5.5