M2: layers open in the state the document's view auto-state sets (ISO 32000-2 8.11.4.4)

The default configuration's /AS array (usage application dictionaries) was
never read, so a group whose /Usage /View /ViewState is OFF and that a
/View usage application manages still drew on screen.

- hayro-interpret patch (67): OcgState::from_catalog applies the View
  event's auto state after /BaseState, /ON and /OFF and before
  ocg_overrides. Each application with /Event /View and /View in
  /Category recommends each listed group's /ViewState; a group is ON only
  if every recommendation is ON. /Zoom, /User and /Language give no
  recommendation. At most 256 applications, 4096 /OCGs entries and 16
  categories each are read. InterpreterSettings::view_auto_state (on by
  default) switches it.
- render: RenderConfig::for_print turns it off; print as image sets it,
  since printing is not the View event.
- inspect: a layer's initial and visible state apply the same auto state,
  so the Layers panel and doc_info match the page.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0175FWj4xFhMhSH3UpMU8DSC

crates/engine/src/print_raster.rs

@@ -22,6 +22,9 @@ pub fn rasterize(doc: &Document, pages: &[usize], raster: &Raster, content: Cont
     let mut config = src.config.clone();
     // The copy below holds only what prints, so the viewer's own comment hiding plays no part.
     config.hide_comments = false;
+    // Printing is not the View event: a layer the auto state turns off on screen keeps its
+    // configured state on paper, as in a print that goes out as vectors.
+    config.for_print = true;
     let printed = guard(|| -> Result<Vec<u8>, String> {
         let mut cos = pdfcraft_cos::Document::open_with_password(src.bytes.clone(), config.password.as_deref()).map_err(|e| e.to_string())?;
         pdfcraft_print::keep_printed(&mut cos, content);

crates/render/src/inspect/resources.rs

@@ -3,7 +3,7 @@ use super::core::inspect;
 use super::core::{Inspector, load_options};
 use super::model::{Attachment, AttachmentSource, FontInfo, Layer, PageTransition, TransitionStyle};
 use super::navigation::NameTreeSeen;
-use super::{AssertUnwindSafe, Dictionary, DocInfo, Document, HashSet, Object, ObjectId, catch_unwind};
+use super::{AssertUnwindSafe, Dictionary, DocInfo, Document, HashMap, HashSet, Object, ObjectId, catch_unwind};
 
 /// Decoded bytes read from an XFA form's configuration packet (or a whole-XDP `/XFA` stream) to
 /// find `dynamicRender`. Designer's packets are a few kilobytes; a whole XDP can hold the template.
@@ -12,6 +12,13 @@ const MAX_XFA_CONFIG_BYTES: usize = 16 << 20;
 /// At most this many layers are read from all of `/RBGroups` together.
 const MAX_LAYER_GROUP_ENTRIES: usize = 4096;
 
+/// The configuration's auto state (`/AS`): at most this many usage application dictionaries,
+/// `/OCGs` entries across them, and `/Category` names in each are read. The renderer's
+/// (vendored hayro-interpret patch (67)) have the same values.
+const MAX_AUTO_STATE_APPLICATIONS: usize = 256;
+const MAX_AUTO_STATE_GROUPS: usize = 4096;
+const MAX_AUTO_STATE_CATEGORIES: usize = 16;
+
 impl<'a> Inspector<'a> {
     // ── object helpers ──────────────────────────────────────────────────────────────────────
 
@@ -100,11 +107,14 @@ impl<'a> Inspector<'a> {
             .and_then(|o| self.resolve(o).as_array().ok())
             .map(|a| a.iter().filter_map(|x| x.as_reference().ok()).collect())
             .unwrap_or_default();
+        let auto = config.map(|c| self.view_auto_state(c)).unwrap_or_default();
         let Ok(Object::Array(ocgs)) = props.get(b"OCGs").map(|o| self.resolve(o)) else { return };
         for g in ocgs {
             let Some(d) = self.dict(g) else { continue };
             let id = g.as_reference().ok();
+            // The View auto state, when it manages the group, then the configuration.
             let visible = match id {
+                Some(id) if let Some(&state) = auto.get(&id) => state,
                 Some(id) if off.contains(&id) => false,
                 Some(id) if on.contains(&id) => true,
                 _ => !base_off,
@@ -114,6 +124,54 @@ impl<'a> Inspector<'a> {
         }
     }
 
+    /// The state the configuration's auto state (`/AS`) gives each group it manages for the View
+    /// event (ISO 32000-2 §8.11.4.4, Table 101), which overrides `/BaseState`, `/ON` and `/OFF`
+    /// as the document opens. Keep in step with the renderer's (vendored hayro-interpret patch
+    /// (67), `apply_view_auto_state`): each usage application with `/Event /View` and `/View` in
+    /// its `/Category` recommends the `/ViewState` of each group in its `/OCGs` whose `/Usage`
+    /// gives one; a group is on only if all its recommendations are. Groups with none are absent.
+    /// Other categories (`/Zoom`, `/User`, `/Language`) give no recommendation here.
+    fn view_auto_state(&self, config: &Dictionary) -> HashMap<ObjectId, bool> {
+        let mut out: HashMap<ObjectId, bool> = HashMap::new();
+        let Some(applications) = config.get(b"AS").ok().and_then(|o| self.resolve(o).as_array().ok()) else { return out };
+        let mut groups_left = MAX_AUTO_STATE_GROUPS;
+        for application in applications.iter().take(MAX_AUTO_STATE_APPLICATIONS) {
+            let Some(application) = self.dict(application) else { continue };
+            if self.name(application, b"Event").as_deref() != Some("View") {
+                continue;
+            }
+            let views = application.get(b"Category").ok().and_then(|o| self.resolve(o).as_array().ok()).is_some_and(|categories| {
+                categories.iter().take(MAX_AUTO_STATE_CATEGORIES).any(|c| matches!(c, Object::Name(n) if n.as_slice() == b"View"))
+            });
+            if !views {
+                continue;
+            }
+            let Some(groups) = application.get(b"OCGs").ok().and_then(|o| self.resolve(o).as_array().ok()) else { continue };
+            for g in groups {
+                if groups_left == 0 {
+                    return out;
+                }
+                groups_left -= 1;
+                let Ok(id) = g.as_reference() else { continue };
+                let state = self
+                    .dict(g)
+                    .and_then(|d| d.get(b"Usage").ok())
+                    .and_then(|o| self.dict(o))
+                    .and_then(|u| u.get(b"View").ok())
+                    .and_then(|o| self.dict(o))
+                    .and_then(|v| self.name(v, b"ViewState"));
+                let on = match state.as_deref() {
+                    Some("ON") => true,
+                    Some("OFF") => false,
+                    _ => continue,
+                };
+                let entry = out.entry(id).or_insert(true);
+                *entry &= on;
+            }
+        }
+        out
+    }
+
     /// The default configuration's radio-button groups (`/RBGroups`) of two or more layers.
     pub(super) fn layer_groups(&self, catalog: &Dictionary) -> Vec<Vec<ObjectId>> {
         let groups = catalog

crates/render/src/inspect/tests.rs

@@ -769,6 +769,40 @@ trailer << /Root 1 0 R >>
     assert_eq!(layers, [("Red", true), ("Green", false), ("Blue", false)]);
 }
 
+/// ISO 32000-2 §8.11.4.4: the View auto state (`/AS`) decides the initial state of the groups
+/// it manages, over `/ON` and `/OFF`; the others keep the configuration's. Malformed entries
+/// are ignored.
+#[test]
+fn initial_layer_visibility_applies_the_view_auto_state() {
+    let initial = |config: &str| {
+        let pdf = format!(
+            "%PDF-1.7
+1 0 obj << /Type /Catalog /Pages 2 0 R /OCProperties << /OCGs [5 0 R 6 0 R 7 0 R] /D {config} >> >> endobj
+2 0 obj << /Type /Pages /Kids [3 0 R] /Count 1 >> endobj
+3 0 obj << /Type /Page /Parent 2 0 R /MediaBox [0 0 200 200] >> endobj
+5 0 obj << /Type /OCG /Name (ViewOff) /Usage << /View << /ViewState /OFF >> >> >> endobj
+6 0 obj << /Type /OCG /Name (ViewOn) /Usage 8 0 R >> endobj
+7 0 obj << /Type /OCG /Name (Plain) >> endobj
+8 0 obj << /View << /ViewState /ON >> >> endobj
+9 0 obj 9 0 R endobj
+trailer << /Root 1 0 R >>
+%%EOF"
+        );
+        let info = inspect(Arc::new(pdf.into_bytes()), None).expect("opens");
+        // Opened, so each is shown as it starts.
+        info.layers.iter().inspect(|l| assert_eq!(l.initial, l.visible, "{}", l.name)).map(|l| l.initial).collect::<Vec<_>>()
+    };
+    let managed = "/AS [<< /Event /View /Category [/View] /OCGs [5 0 R 6 0 R 7 0 R] >>]";
+    assert_eq!(initial(&format!("<< /OFF [6 0 R 7 0 R] {managed} >>")), [false, true, false]);
+    assert_eq!(initial("<< /OFF [6 0 R 7 0 R] >>"), [true, false, false], "without /AS");
+    assert_eq!(initial("<< /OFF [6 0 R] /AS [<< /Event /Print /Category [/View] /OCGs [5 0 R 6 0 R] >>] >>"), [true, false, true], "Print event");
+    assert_eq!(initial("<< /AS [<< /Event /View /Category [/Zoom] /OCGs [5 0 R] >>] >>"), [true, true, true], "Zoom only");
+    for bad in ["/AS 9 0 R", "/AS [9 0 R 1 null]", "/AS [<< /Event /View /OCGs [5 0 R] >>]", "/AS [<< /Event /View /Category [/View] /OCGs 9 0 R >>]"]
+    {
+        assert_eq!(initial(&format!("<< {bad} >>")), [true, true, true], "{bad}");
+    }
+}
+
 const ORPHAN_FIELDS: &[u8] = b"%PDF-1.7
 1 0 obj << /Type /Catalog /Pages 2 0 R /AcroForm 4 0 R >> endobj
 2 0 obj << /Type /Pages /Kids [3 0 R] /Count 1 >> endobj

crates/render/src/raster.rs

@@ -83,6 +83,9 @@ pub struct RenderConfig {
     /// viewer and preview rasters: exports, printing and OCR render without it, so nothing drawn with an
     /// installed font is saved or printed (AGENTS.md §1.2, §1.4).
     pub system_fonts: bool,
+    /// The render is output for printing (print as image): the configuration's auto state for
+    /// the View event (ISO 32000-2 §8.11.4.4) is not applied, since printing is not viewing.
+    pub for_print: bool,
 }
 
 impl RenderConfig {
@@ -104,6 +107,7 @@ impl RenderConfig {
         let system_fonts = self.system_fonts;
         InterpreterSettings {
             ocg_overrides: self.layers.clone(),
+            view_auto_state: !self.for_print,
             hide_comments: self.hide_comments,
             font_resolver: Arc::new(move |query| {
                 japanese_fallback(query).or_else(|| system_fonts.then(|| system_fallback(query)).flatten()).or_else(|| standard(query))

crates/render/src/raster/tests.rs

@@ -2303,6 +2303,112 @@ trailer << /Root 1 0 R >>
     assert_eq!(px(off), vec![255, 255, 255, 255], "toggled-off layer must not draw");
 }
 
+/// A page whose red square is in group 5 0 R (`ocg` its dictionary), under the default
+/// configuration `config`; `extra` adds objects from number 8.
+fn auto_state_pdf(ocg: &str, config: &str, extra: &str) -> Vec<u8> {
+    format!(
+        "%PDF-1.7
+1 0 obj << /Type /Catalog /Pages 2 0 R /OCProperties << /OCGs [5 0 R] /D {config} >> >> endobj
+2 0 obj << /Type /Pages /Kids [3 0 R] /Count 1 >> endobj
+3 0 obj << /Type /Page /Parent 2 0 R /MediaBox [0 0 100 100] /Contents 4 0 R /Resources << /Properties << /L1 5 0 R >> >> >> endobj
+4 0 obj << /Length 44 >> stream
+/OC /L1 BDC 1 0 0 rg 0 0 100 100 re f EMC
+endstream endobj
+5 0 obj {ocg} endobj
+{extra}
+trailer << /Root 1 0 R >>
+%%EOF"
+    )
+    .into_bytes()
+}
+
+/// Whether the red square in group 5 0 R draws.
+fn auto_state_red(pdf: Vec<u8>, config: RenderConfig) -> bool {
+    let mut r = PageRenderer::new(Arc::new(pdf), config);
+    let p = r.render(RenderRequest { page: 0, kind: RequestKind::Pixels, tile: None, scale: 0.5, tag: 0, ..Default::default() });
+    p.rgba.get(..4) == Some(&[255, 0, 0, 255][..])
+}
+
+const VIEW_OFF: &str = "<< /Type /OCG /Name (A) /Usage << /View << /ViewState /OFF >> >> >>";
+const VIEW_ON: &str = "<< /Type /OCG /Name (A) /Usage << /View << /ViewState /ON >> >> >>";
+const AS_VIEW: &str = "/AS [<< /Event /View /Category [/View] /OCGs [5 0 R] >>]";
+
+/// ISO 32000-2 §8.11.4.4: the configuration's auto state (`/AS`) sets a group's state for
+/// the View event from its usage, over `/BaseState`, `/ON` and `/OFF`. It was ignored, so a
+/// group whose `/ViewState` is OFF still drew on screen.
+#[test]
+fn the_view_auto_state_sets_a_groups_state_on_screen() {
+    let shown = |ocg: &str, config: &str| auto_state_red(auto_state_pdf(ocg, config, ""), RenderConfig::default());
+    assert!(!shown(VIEW_OFF, &format!("<< {AS_VIEW} >>")), "ViewState OFF, managed: hidden");
+    assert!(shown(VIEW_OFF, "<< >>"), "no /AS: the usage alone changes nothing");
+    assert!(shown(VIEW_OFF, "<< /AS [<< /Event /View /Category [/View] /OCGs [] >>] >>"), "not listed in /OCGs");
+    assert!(shown(VIEW_ON, &format!("<< /OFF [5 0 R] {AS_VIEW} >>")), "ViewState ON wins over /OFF");
+    assert!(shown(VIEW_ON, &format!("<< /BaseState /OFF {AS_VIEW} >>")), "and over /BaseState /OFF");
+    assert!(!shown(VIEW_ON, "<< /OFF [5 0 R] >>"), "without /AS, /OFF decides");
+    // The View event only, and the View category only.
+    assert!(shown(VIEW_OFF, "<< /AS [<< /Event /Print /Category [/View] /OCGs [5 0 R] >>] >>"), "Print event: not on screen");
+    assert!(shown(VIEW_OFF, "<< /AS [<< /Event /Export /Category [/View] /OCGs [5 0 R] >>] >>"), "Export event: not on screen");
+    assert!(shown(VIEW_OFF, "<< /AS [<< /Event /View /Category [/Print] /OCGs [5 0 R] >>] >>"), "a category without a View entry");
+    // A category it cannot evaluate gives no recommendation; the View one still applies.
+    assert!(!shown(VIEW_OFF, "<< /AS [<< /Event /View /Category [/Zoom /View] /OCGs [5 0 R] >>] >>"), "Zoom alongside View");
+    let zoomed = "<< /Type /OCG /Name (A) /Usage << /Zoom << /min 4 >> >> >>";
+    assert!(shown(zoomed, "<< /AS [<< /Event /View /Category [/Zoom] /OCGs [5 0 R] >>] >>"), "Zoom alone: unchanged");
+    // Indirect applications, arrays and usage dictionaries are followed.
+    let indirect = "<< /Type /OCG /Name (A) /Usage 10 0 R >>";
+    let extra = "8 0 obj [9 0 R] endobj\n9 0 obj << /Event /View /Category [/View] /OCGs 11 0 R >> endobj\n10 0 obj << /View << /ViewState /OFF >> >> endobj\n11 0 obj [5 0 R] endobj";
+    assert!(!auto_state_red(auto_state_pdf(indirect, "<< /AS 8 0 R >>", extra), RenderConfig::default()), "indirect");
+}
+
+/// The Layers panel's toggles still decide over the auto state, and a render for printing
+/// does not apply it (printing is not the View event).
+#[test]
+fn layer_toggles_and_print_renders_are_not_the_view_auto_state() {
+    let pdf = auto_state_pdf(VIEW_OFF, &format!("<< {AS_VIEW} >>"), "");
+    let on = RenderConfig { layers: Arc::new(vec![(5, 0, true)]), ..Default::default() };
+    assert!(auto_state_red(pdf.clone(), on), "toggled on in the Layers panel");
+    assert!(auto_state_red(pdf.clone(), RenderConfig { for_print: true, ..Default::default() }), "printed: its configured state");
+    assert!(!auto_state_red(pdf, RenderConfig::default()), "on screen: hidden");
+}
+
+/// Malformed auto state is ignored without a panic, and the walk is bounded.
+#[test]
+fn malformed_view_auto_state_is_ignored() {
+    let shown = |config: &str, extra: &str| auto_state_red(auto_state_pdf(VIEW_OFF, config, extra), RenderConfig::default());
+    for config in [
+        "<< /AS 5 >>",
+        "<< /AS << /Event /View /Category [/View] /OCGs [5 0 R] >> >>",
+        "<< /AS [1 null /x (s) [5 0 R]] >>",
+        "<< /AS [<< /Event /View /Category [/View] >>] >>",
+        "<< /AS [<< /Event /View /OCGs [5 0 R] >>] >>",
+        "<< /AS [<< /Category [/View] /OCGs [5 0 R] >>] >>",
+        "<< /AS [<< /Event (View) /Category [/View] /OCGs [5 0 R] >>] >>",
+        "<< /AS [<< /Event /View /Category /View /OCGs [5 0 R] >>] >>",
+        "<< /AS [<< /Event /View /Category [/View] /OCGs 5 0 R >>] >>",
+        "<< /AS [<< /Event /View /Category [/View] /OCGs [(5) 5 /OCG] >>] >>",
+    ] {
+        assert!(shown(config, ""), "{config}: ignored");
+    }
+    let bad_state = |s: &str| {
+        let ocg = format!("<< /Type /OCG /Name (A) /Usage << /View << /ViewState {s} >> >> >>");
+        auto_state_red(auto_state_pdf(&ocg, &format!("<< {AS_VIEW} >>"), ""), RenderConfig::default())
+    };
+    for s in ["/Off", "(OFF)", "0", "null", "[/OFF]"] {
+        assert!(bad_state(s), "ViewState {s}: no recommendation");
+    }
+    // Cyclic references: an /AS, an application and a usage that refer to themselves.
+    assert!(shown("<< /AS 8 0 R >>", "8 0 obj 8 0 R endobj"), "self-referencing /AS");
+    assert!(shown("<< /AS [8 0 R] >>", "8 0 obj 8 0 R endobj"), "self-referencing application");
+    let cyclic = "<< /Type /OCG /Name (A) /Usage 8 0 R >>";
+    assert!(auto_state_red(auto_state_pdf(cyclic, &format!("<< {AS_VIEW} >>"), "8 0 obj 8 0 R endobj"), RenderConfig::default()));
+    // Huge arrays: the group listed after 4096 other entries is past the cap.
+    let many = format!("<< /AS [<< /Event /View /Category [/View] /OCGs [{}5 0 R] >>] >>", "9 0 R ".repeat(4096));
+    assert!(shown(&many, "9 0 obj << /Type /OCG /Name (B) >> endobj"), "past the cap: unchanged");
+    let within = format!("<< /AS [<< /Event /View /Category [/View] /OCGs [{}5 0 R] >>] >>", "9 0 R ".repeat(4095));
+    assert!(!shown(&within, "9 0 obj << /Type /OCG /Name (B) >> endobj"), "within the cap: applied");
+    let apps = format!("<< /AS [{}<< /Event /View /Category [/View] /OCGs [5 0 R] >>] >>", "<< /Event /Print >> ".repeat(256));
+    assert!(shown(&apps, ""), "past the application cap: unchanged");
+}
+
 /// #2986: an optional content membership dictionary with a visibility expression (`/VE`,
 /// ISO 32000-2 §8.11.2.2) shows its content only when the expression is true, and the
 /// expression takes precedence over `/OCGs` and `/P`. It was ignored, so content under it

parity/acrobat-features.toml

@@ -1334,8 +1334,8 @@ milestone = "M3"
 status = "shipped"
 commands = []
 tools = ["doc_info", "layer_set"]
-evidence = ["crates/render/src/raster/tests.rs::layer_override_hides_content", "crates/engine/src/tests.rs::layer_choices_survive_edits", "crates/automation/tests/it/automation.rs::layers_and_hidden_comments_change_what_page_render_draws"]
-notes = "layer_set shows or hides a layer by name (every group with that name, as CAD exporters repeat names) or by position, and page_render draws with the document's layer visibility. Still missing: Reset to initial visibility (view.layers-reset), layer properties, and /OC on annotations (#2176)."
+evidence = ["crates/render/src/raster/tests.rs::layer_override_hides_content", "crates/engine/src/tests.rs::layer_choices_survive_edits", "crates/automation/tests/it/automation.rs::layers_and_hidden_comments_change_what_page_render_draws", "crates/render/src/raster/tests.rs::the_view_auto_state_sets_a_groups_state_on_screen", "crates/render/src/raster/tests.rs::layer_toggles_and_print_renders_are_not_the_view_auto_state", "crates/render/src/raster/tests.rs::malformed_view_auto_state_is_ignored"]
+notes = "layer_set shows or hides a layer by name (every group with that name, as CAD exporters repeat names) or by position, and page_render draws with the document's layer visibility. Still missing: Reset to initial visibility (view.layers-reset), layer properties, and /OC on annotations (#2176). The default configuration's auto state for the View event (/AS, ISO 32000-2 8.11.4.4) sets the groups it manages from their /Usage /View /ViewState as the document opens, on screen and in the panel, and toggles still override it; renders for printing don't apply it. Not yet: the /Zoom, /User and /Language categories, and the Print event's auto state when printing."
 
 [[feature]]
 id = "view.layers-reset"
@@ -1346,7 +1346,7 @@ milestone = "M12"
 status = "partial"
 commands = []
 tools = ["doc_info"]
-evidence = ["crates/engine/src/tests.rs::reset_layers_returns_to_the_files_visibility", "crates/render/src/inspect/tests.rs::initial_layer_visibility_is_the_default_configuration"]
+evidence = ["crates/engine/src/tests.rs::reset_layers_returns_to_the_files_visibility", "crates/render/src/inspect/tests.rs::initial_layer_visibility_is_the_default_configuration", "crates/render/src/inspect/tests.rs::initial_layer_visibility_applies_the_view_auto_state"]
 notes = "Layers panel ▸ Layer options ▸ Reset to Initial Visibility shows each layer as the default configuration (/OCProperties /D) has it; doc_info reports it as `initial`. Not yet: listing the layers of all pages or the current page only."
 
 [[feature]]

vendor/README.md

@@ -39,7 +39,7 @@ Other resource categories retain their existing parent-chain lookup.
 
 | Crate | Version | Licence | Patches | Test |
 |---|---|---|---|---|
-| hayro-interpret | 0.7.0 | Apache-2.0 OR MIT | (1) select `/AP /N` appearance states via `/AS` (checkboxes, radios, stateful icons) and honour the NoView flag; (2) Type3 Unicode fallback to `/Encoding` glyph names, then printable ASCII; (3) `Type3Glyph::advance_width`; (4) `InterpreterSettings::ocg_overrides` for viewer layer toggles; (5) soft masks whose group lacks `/CS` were dropped (Chrome gradient text); (6) paint nesting cap `MAX_PAINT_NESTING` for tiling patterns and Type 3 glyphs (self-referencing content overflowed the stack); (7) CID width ranges in `/W` and `/W2` clamped to `MAX_CID` (a `0 4294967295 w` range hung); (8) `InterpreterSettings::hide_comments` skips markup annotations but keeps widgets and links (View ▸ Hide all comments); (9) `image_size_ok`: images and masks over 2^28 pixels are not decoded (a fuzzed `/W 4294967295` stencil mask allocated 4 GiB); (10) `MAX_NESTED_PAINTS`: nested paints (forms, Type 3 glyphs and tiling patterns inside one another) share a per-page budget of 50,000 (a glyph showing eight glyphs of its own font, sixteen deep, hung); (11) depends on skrifa 0.47 instead of 0.42 (upstream read-fonts fixed a Type 1 tokenizer that recursed through runs of numbers: a fuzzed font hung or overflowed the stack), with the one API change (`Charset::glyph_id` returns `Option`); (12) annotation appearances whose `/BBox` or `/Matrix` collapses to a line or a point are skipped (mapping them onto `/Rect` divided by zero and handed the device infinite or NaN matrices); (13) a Highlight annotation without a usable normal appearance is drawn from its `/QuadPoints`, `/C` and `/CA` with Multiply blending, as pdfcraft-annot draws highlights (it was not drawn at all); (14) simple-font `/Widths` (`read_widths`): entries only for codes 0–255 (one entry per code below `/FirstChar` was allocated: `4294967295` asked for 34 GB), an entry that isn't a number leaves its code missing instead of ending the array, and a `/LastChar` below `/FirstChar` gives no widths instead of discarding the font for a standard one; (15) the appearance box is mapped onto `/Rect` corner to corner: the translation now accounts for the scale (an appearance whose box didn't start at the origin and differed in size from `/Rect` was drawn shifted); (16) a clip set by `W`/`W*` takes effect after whichever path-painting operator ends the path (`S`, `f`, `B`, … as well as `n`; ISO 32000-2 §8.5.4): it was applied only on `n`, so `re W* S` clipped nothing and the clip stayed pending for a later, unrelated `n`; (17) `InterpreterSettings::cancelled`: once set, interpretation stops before the next content operator (a render pool sets it for every worker when dropped, after an edit or on closing, and for one worker when the watchdog gives up on its render, so renders whose answers nobody can receive stop instead of running to the end); (18) `MAX_PAGE_CONTENT`: the content a page holds at once (its contents, and the forms, tiling patterns and Type 3 glyph procedures being painted) stays within 512 MiB (`decode_content`; once it runs out, `InterpreterWarning::ContentTruncated` is reported (once per page by pdfcraft-render) so a truncated page is not a clean render; each level gives its bytes back when it finishes): forms nest fifty deep and each level kept its content while the next ran, so a form painting itself around a stream inflating to 256 MiB would have held over 12 GB; and Type 0 sampled functions take at most `MAX_SAMPLE_ENTRIES` (2^21) table points and `MAX_SAMPLE_VALUES` (2^23) samples, read no further than `/Size` asks (256 MiB of 1-bit samples asked for over 8 GB); (19) mesh shadings (types 4–7) are sampled only inside the area being drawn (`ShadingPattern::encode_within`, which the vendored hayro renderer calls with that area plus a pixel), within 8 visits per pixel of it plus 4 per triangle (`TriangleSampler`), and patches are triangulated one at a time: points far off the page were sampled over their whole bounding box, up to 65536 × 65536 entries (a fuzzed 147-byte patch stream; a synthetic page passed 6 GB in six seconds), ten thousand overlapping triangles ran for minutes, and every patch became 722 triangles before any limit; meshes hold at most `MAX_MESH_PATCHES` (2^16) patches and `MAX_MESH_TRIANGLES` (2^20) triangles, a lattice needs two to 2^19 vertices per row (`/VerticesPerRow 0` pushed empty rows for ever; one huge row read the whole stream), and a page samples at most `MAX_PAGE_MESH_TRIANGLES` (2^25) mesh triangles in all (a mesh painted a thousand times was triangulated a thousand times); each cut logs a warning; (20) PDF function construction is limited to 64 levels and 10,000 total nodes with cycle detection; calculator parsing and evaluation share procedure depth/work budgets, integer division/bitshift/rotation/index are panic-free, and fixed-size stack errors propagate; (21) a Type 0 font drawn with a substitute that isn't CID-keyed (it isn't embedded, or its program doesn't load) draws `.nLine truncated
+| hayro-interpret | 0.7.0 | Apache-2.0 OR MIT | (1) select `/AP /N` appearance states via `/AS` (checkboxes, radios, stateful icons) and honour the NoView flag; (2) Type3 Unicode fallback to `/Encoding` glyph names, then printable ASCII; (3) `Type3Glyph::advance_width`; (4) `InterpreterSettings::ocg_overrides` for viewer layer toggles; (5) soft masks whose group lacks `/CS` were dropped (Chrome gradient text); (6) paint nesting cap `MAX_PAINT_NESTING` for tiling patterns and Type 3 glyphs (self-referencing content overflowed the stack); (7) CID width ranges in `/W` and `/W2` clamped to `MAX_CID` (a `0 4294967295 w` range hung); (8) `InterpreterSettings::hide_comments` skips markup annotations but keeps widgets and links (View ▸ Hide all comments); (9) `image_size_ok`: images and masks over 2^28 pixels are not decoded (a fuzzed `/W 4294967295` stencil mask allocated 4 GiB); (10) `MAX_NESTED_PAINTS`: nested paints (forms, Type 3 glyphs and tiling patterns inside one another) share a per-page budget of 50,000 (a glyph showing eight glyphs of its own font, sixteen deep, hung); (11) depends on skrifa 0.47 instead of 0.42 (upstream read-fonts fixed a Type 1 tokenizer that recursed through runs of numbers: a fuzzed font hung or overflowed the stack), with the one API change (`Charset::glyph_id` returns `Option`); (12) annotation appearances whose `/BBox` or `/Matrix` collapses to a line or a point are skipped (mapping them onto `/Rect` divided by zero and handed the device infinite or NaN matrices); (13) a Highlight annotation without a usable normal appearance is drawn from its `/QuadPoints`, `/C` and `/CA` with Multiply blending, as pdfcraft-annot draws highlights (it was not drawn at all); (14) simple-font `/Widths` (`read_widths`): entries only for codes 0–255 (one entry per code below `/FirstChar` was allocated: `4294967295` asked for 34 GB), an entry that isn't a number leaves its code missing instead of ending the array, and a `/LastChar` below `/FirstChar` gives no widths instead of discarding the font for a standard one; (15) the appearance box is mapped onto `/Rect` corner to corner: the translation now accounts for the scale (an appearance whose box didn't start at the origin and differed in size from `/Rect` was drawn shifted); (16) a clip set by `W`/`W*` takes effect after whichever path-painting operator ends the path (`S`, `f`, `B`, … as well as `n`; ISO 32000-2 §8.5.4): it was applied only on `n`, so `re W* S` clipped nothing and the clip stayed pending for a later, unrelated `n`; (17) `InterpreterSettings::cancelled`: once set, interpretation stops before the next content operator (a render pool sets it for every worker when dropped, after an edit or on closing, and for one worker when the watchdog gives up on its render, so renders whose answers nobody can receive stop instead of running to the end); (18) `MAX_PAGE_CONTENT`: the content a page holds at once (its contents, and the forms, tiling patterns and Type 3 glyph procedures being painted) stays within 512 MiB (`decode_content`; once it runs out, `InterpreterWarning::ContentTruncated` is reported (once per page by pdfcraft-render) so a truncated page is not a clean render; each level gives its bytes back when it finishes): forms nest fifty deep and each level kept its content while the next ran, so a form painting itself around a stream inflating to 256 MiB would have held over 12 GB; and Type 0 sampled functions take at most `MAX_SAMPLE_ENTRIES` (2^21) table points and `MAX_SAMPLE_VALUES` (2^23) samples, read no further than `/Size` asks (256 MiB of 1-bit samples asked for over 8 GB); (19) mesh shadings (types 4–7) are sampled only inside the area being drawn (`ShadingPattern::encode_within`, which the vendored hayro renderer calls with that area plus a pixel), within 8 visits per pixel of it plus 4 per triangle (`TriangleSampler`), and patches are triangulated one at a time: points far off the page were sampled over their whole bounding box, up to 65536 × 65536 entries (a fuzzed 147-byte patch stream; a synthetic page passed 6 GB in six seconds), ten thousand overlapping triangles ran for minutes, and every patch became 722 triangles before any limit; meshes hold at most `MAX_MESH_PATCHES` (2^16) patches and `MAX_MESH_TRIANGLES` (2^20) triangles, a lattice needs two to 2^19 vertices per row (`/VerticesPerRow 0` pushed empty rows for ever; one huge row read the whole stream), and a page samples at most `MAX_PAGE_MESH_TRIANGLES` (2^25) mesh triangles in all (a mesh painted a thousand times was triangulated a thousand times); each cut logs a warning; (20) PDF function construction is limited to 64 levels and 10,000 total nodes with cycle detection; calculator parsing and evaluation share procedure depth/work budgets, integer division/bitshift/rotation/index are panic-free, and fixed-size stack errors propagate; (21) a Type 0 font drawn with a substitute that isn't CID-keyed (it isn't embedded, or its program doesn't load) draws `.nLine truncated
 | hayro | 0.7.1 | Apache-2.0 OR MIT | (1) `RenderSettings::{x_offset, y_offset}` to render one tile of a large page; (2) images over `MAX_IMAGE_PIXELS` (2^28) are skipped (a fuzzed `/W 4294967295` hung in resampling); (3) `tiling_cell_scale` keeps a tiling pattern's cell pixmap within 3000 px a side whatever its `/XStep`/`/YStep` (a fuzzed step far beyond `/BBox` allocated 2 GB); (4) stroke widths clamped to 4 × (canvas width + height) device pixels (a fuzzed `/LW 9223372036854775807` allocated 10 GB in stroke expansion); (5) `usable_dash_pattern`: a stroke whose dash pattern could cut it into more than `MAX_DASHES_PER_STROKE` (250,000) pieces, counting kurbo's restart of the pattern at every subpath and `ClosePath`, is drawn solid (a fuzzed `/D [[11] 0]` on a line from x = 92234775807 allocated over 5 GB, `[0.000001] 0 d` on a 20 pt line over 10 GB); (6) a dash pattern with a negative or non-finite entry, or a period that isn't positive, is drawn solid (`[-1 -1] 0 d` looped for ever in kurbo); (7) `render_size` and `render_into`: render into a buffer the caller allocates (PdfCraft renders into pixel-aligned memory that the GUI takes over as texture data); `render` is `render_into` on a new pixmap; (8) checked image source/target/RGBA, alpha-premultiplication scratch and intermediate sizes before resampling, finite transforms and exact source/mask buffers; sources wider than u16 may shrink before backend limits, with a 2^20 source-axis cap and conservative 128 MiB filter-planning estimate; an over-limit or failed resize keeps the original image resolution and geometry when the source fits the backend, otherwise skips it, with a logged warning; staging, resampling output/scratch, RGBA conversion and padding buffers use fallible reserves; (9) `may_paint_canvas`: a fill or stroke whose control-point box, grown by how far its paint reaches past the path (half the stroke width times the larger of the miter limit and √2) plus 0.5 user units and 2 pixels, misses the canvas is not drawn, so a tile no longer strokes and fills every path of its page (item 5 of #623); `set_skip_offscreen_paths` turns this off for the regression test; (10) `/Interpolate` governs the pre-resize too (#624): a minified image that does not request interpolation is pre-resized with nearest-neighbour, planned without the Catmull-Rom filter limits (`image_resampling_size` takes the flag); a Type 3 glyph image is sampled bicubically only when interpolated; a minified image with a matching soft mask is nearest-sampled straight into premultiplied RGBA; (11) `may_paint_rect`: an image whose stream-size box, mapped to the device, misses the canvas is skipped before decoding (decoded data can only shrink against that box), so a tile of an image-heavy page no longer decodes, resamples and rasterizes every image on it; (12) preview renders (`InterpreterSettings::preview`: backdrops, thumbnails, grid cells, signature-drag backgrounds) skip images under `PREVIEW_MIN_IMAGE_AREA` device pixels, while final renders always draw them; (13) reading colours (#897): `RenderSettings::color_replacement` takes a `ColorReplacement` (text colour, background, `only_black`, `line_art`) that remaps solid text and line-art paint while rendering, glyph outlines counting as text and stencil-mask ink as line art; images, image alpha masks, soft masks and pattern cells keep their colours; `render_pdf` gains a `color_replacement: Option<ColorReplacement>` argument (pass `None` for true colours); the upstream-only `[dev-dependencies.log]` entry is removed from `Cargo.toml` (log is a normal dependency since the logging patch), so `cargo test -p hayro` runs this crate's unit tests; (14) a shrunk, interpolated, axis-aligned image is snapped to whole device pixels: shrunk once to its rounded device rectangle and drawn 1:1 at the rounded corner, instead of a ceil-sized raster resampled again at a fractional scale and offset (`snap_minified_image`; scans looked soft next to Acrobat); (15) a soft mask of another size than its image is filtered when shrunk, whatever its `/Interpolate` (#1649): such a mask carries the shape on its own (Word saves text as a tiny black image with a glyph-bitmap mask), and nearest-neighbour dropped every stroke thinner than the shrink factor; (16) under a soft mask or a blend mode, axial and radial shadings are sampled instead of drawn as native gradients (#2128): vello_cpu 0.0.8 copies an opaque gradient over fully covered tiles without the mask or blend mode, so PowerPoint's faded gradient fills came out solid; (17) the renderer keeps annotation paints out of the reading-colours mapping between `begin_annotations` and `end_annotations` (#1548); (18) the page renders as one transparency group; under reading colours its background is painted inside it as the initial backdrop, so a Multiply highlight blends with the dark background instead of staying an opaque block, and otherwise the background stays outside, so a blend mode over an empty part of thLine truncated
 | hayro-jbig2 | 0.3.0 | Apache-2.0 OR MIT | (1) at most 2^28 pixels per bitmap, page bitmap, halftone grid and grayscale image (each side could be 65535: 4.3 billion pixels decoded one by one); (2) at most 2^26 symbol pixels per symbol dictionary (a fuzzed dictionary decoded for minutes); (3) a symbol dictionary's height classes are capped at one per new symbol and its export runs at `2 × symbols + 1`, each plus `EXTRA_EMPTY_STEPS` (1024): an empty height class or a zero-length export run doesn't advance, and past the end of its data the arithmetic decoder can decode them for ever (one changed byte hung the renderer); (4) a segment header's long-form referred-to segment count is capped at `u16::MAX` and a larger count is refused before its retention or referred-to bytes are read: the 29-bit count went straight to `Vec::with_capacity`, a reservation of about 2 GiB whose failure aborts the process (from storytold/craft-libs#176) | (1) `raster::tests::jbig2_regions_of_billions_of_pixels_are_refused`; (2) the nightly `cargo xtask fuzz` finding, which now finishes in about a second; (3) `raster::tests::jbig2_symbol_dictionaries_that_never_advance_terminate`; (4) `segment::tests::test_referred_to_count_is_capped` (`cargo test --manifest-path vendor/hayro-jbig2/Cargo.toml`) |
 | lopdf | 0.45.0 | MIT | (1) PNG predictor rows longer than the data are refused before their buffers are allocated (a fuzzed `/Columns 4294967295` allocated two 4 GiB rows), and empty frames return without row buffers; (2) predictor row widths use saturating arithmetic or reject overflow; (3) sub-byte TIFF Predictor 2 scratch is bounded by the samples present in the input instead of the declared `/Colors`, aligned with [lopdf#589](https://github.com/J-F-Liu/lopdf/pull/589); (4) the object lookup keeps its cycle history through compressed objects, requires an ordinary in-file container, and caps the chain of indirect lookups at 16 (a chain of new objects, each a `/Length` compressed in a container whose own `/Length` names the next, or a stream whose `/Length` names the next stream, overflowed the stack: a dev build's 2 MiB thread within 800 hops); (5) object-stream members are parsed within their own extent (up to the next distinct offset), an entry whose offset repeats an earlier entry's is skipped (so a repeated object number can resolve to its earlier entry), and offsets are processed in index order; (6) `find_latest_trailer`: a rebuild takes a `/Type /XRef` stream dictionary as the trailer when the file has no `trailer` keyword (the latest in the file wins, as a later `trailer` would), so a concatenated or truncated file keeps its /Info and /ID instead of failing to load; a candidate whose /Root is the catalog beats one whose /Root is another dictionary, and an in-file /Root must match its generation (#2365) | (1) `inspect::tests::png_predictor_rows_longer_than_the_data_are_refused_before_allocating`, `inspect::tests::empty_png_predictor_frames_do_not_allocate_rows`; (2) `inspect::tests::tiff_subbyte_predictor_row_width_overflow_is_refused`; (3) `inspect::tests::tiff_subbyte_predictor_scratch_is_bounded_by_available_samples`, `inspect::tests::tiff_subbyte_predictors_keep_components_rows_and_padding`; (4) `inspect::tests::compressed_object_self_container_does_not_recurse`, `inspect::tests::compressed_object_mutual_containers_do_not_recurse`, `inspect::tests::object_stream_length_cycle_does_not_recurse`, `inspect::tests::compressed_object_container_chain_does_not_recurse`, `inspect::tests::compressed_length_chain_does_not_recurse`, `inspect::tests::plain_stream_length_chain_does_not_recurse`, `inspect::tests::compressed_objects_and_indirect_stream_length_still_load`, `inspect::tests::compressed_length_in_a_container_with_an_indirect_length_still_loads`; (5) `inspect::tests::object_stream_member_extents_bound_owned_strings`, `inspect::tests::object_stream_member_extents_preserve_index_order_and_whitespace`; (6) `inspect::tests::rebuilt_file_keeps_the_info_of_its_cross_reference_stream`, `inspect::tests::rebuilt_file_prefers_a_trailer_whose_root_is_the_catalog` |

vendor/hayro-interpret/src/context.rs

@@ -237,7 +237,7 @@ impl<'a> Context<'a> {
         let mut ocg_state = {
             let root_ref = xref.root_id();
             xref.get::<Dict<'_>>(root_ref)
-                .map(|catalog| OcgState::from_catalog(&catalog))
+                .map(|catalog| OcgState::from_catalog(&catalog, xref, settings.view_auto_state))
                 .unwrap_or_default()
         };
         // PdfCraft patch: apply viewer layer toggles.

vendor/hayro-interpret/src/interpret/mod.rs

@@ -111,6 +111,11 @@ pub struct InterpreterSettings {
     /// PdfCraft patch: viewer overrides for optional content groups (object number, generation,
     /// visible), applied on top of the document's default configuration (Layers panel toggles).
     pub ocg_overrides: Arc<Vec<(i32, i32, bool)>>,
+    /// PdfCraft patch (67): apply the default configuration's auto state for the View event
+    /// (`/AS` usage applications with `/Event /View`, ISO 32000-2 §8.11.4.4) before
+    /// `ocg_overrides`. On by default: that is the state for viewing. Off for output made for
+    /// printing, which the View event is not.
+    pub view_auto_state: bool,
     /// PdfCraft patch: once this is `true`, interpretation stops before the next content
     /// operator. What was drawn is then incomplete, so only set it when the output will be
     /// thrown away (a viewer's render whose result nobody can receive any more).
@@ -140,6 +145,7 @@ impl Default for InterpreterSettings {
             render_annotations: true,
             hide_comments: false,
             ocg_overrides: Arc::new(Vec::new()),
+            view_auto_state: true,
             cancelled: None,
             preview: false,
         }

vendor/hayro-interpret/src/ocg.rs

@@ -1,6 +1,9 @@
-use hayro_syntax::object::dict::keys::{BASE_STATE, D, OCGS, OCMD, OCPROPERTIES, OFF, ON, P, TYPE, VE};
+use hayro_syntax::object::dict::keys::{
+    AS, BASE_STATE, D, OCGS, OCMD, OCPROPERTIES, OFF, ON, P, TYPE, USAGE, VE, VIEW, VIEW_STATE,
+};
 use hayro_syntax::object::{Array, Dict, MaybeRef, Name, Object, ObjectIdentifier};
-use std::collections::HashSet;
+use hayro_syntax::xref::XRef;
+use std::collections::{HashMap, HashSet};
 
 pub(crate) struct OcgState {
     inactive_ocgs: HashSet<ObjectIdentifier>,
@@ -30,7 +33,10 @@ impl OcgState {
         }
     }
 
-    pub(crate) fn from_catalog(catalog: &Dict<'_>) -> Self {
+    /// The groups' states as the document opens: the default configuration (`/BaseState`, then
+    /// `/ON` and `/OFF`), then, when `view_auto_state` is set (PdfCraft patch (67)), the auto
+    /// state its `/AS` array gives for the View event ([`apply_view_auto_state`]).
+    pub(crate) fn from_catalog(catalog: &Dict<'_>, xref: &XRef, view_auto_state: bool) -> Self {
         let Some(oc_properties) = catalog.get::<Dict<'_>>(OCPROPERTIES) else {
             return Self::dummy();
         };
@@ -83,6 +89,10 @@ impl OcgState {
         read_ocg_array(ON, true);
         read_ocg_array(OFF, false);
 
+        if view_auto_state {
+            apply_view_auto_state(&config, xref, &mut inactive);
+        }
+
         Self {
             inactive_ocgs: inactive,
             known_ocgs: known,
@@ -213,6 +223,87 @@ impl Default for OcgState {
     }
 }
 
+/// PdfCraft patch (67): at most this many usage application dictionaries of an `/AS` array are
+/// read; the rest are ignored.
+const MAX_AUTO_STATE_APPLICATIONS: usize = 256;
+/// PdfCraft patch (67): at most this many `/OCGs` entries are read across all of them.
+const MAX_AUTO_STATE_GROUPS: usize = 4096;
+/// PdfCraft patch (67): at most this many `/Category` names are read per application.
+const MAX_AUTO_STATE_CATEGORIES: usize = 16;
+
+/// PdfCraft patch (67): apply the configuration's auto state for the View event (ISO 32000-2
+/// §8.11.4.4, Tables 99 and 101). For each usage application dictionary in `/AS` whose `/Event`
+/// is `/View`, each group in its `/OCGs` gets a recommended state from each category in its
+/// `/Category` that the group's `/Usage` gives one for. A group is ON only if every
+/// recommendation it receives, across all such dictionaries, is ON, and OFF otherwise; a group
+/// that receives none keeps its configured state.
+///
+/// Only the `/View` category (`/ViewState`) is evaluated. `/Zoom`, `/User` and `/Language` need
+/// the magnification, user and locale, which the interpreter doesn't know, so they give no
+/// recommendation. Malformed entries (not a dictionary, no `/Category`, a `/ViewState` other
+/// than ON or OFF) are skipped, and the walk is bounded by the caps above.
+fn apply_view_auto_state(config: &Dict<'_>, xref: &XRef, inactive: &mut HashSet<ObjectIdentifier>) {
+    let Some(applications) = config.get::<Array<'_>>(AS) else {
+        return;
+    };
+    let mut recommended: HashMap<ObjectIdentifier, bool> = HashMap::new();
+    let mut groups_left = MAX_AUTO_STATE_GROUPS;
+    for application in applications.raw_iter().take(MAX_AUTO_STATE_APPLICATIONS) {
+        let application = match application {
+            MaybeRef::Ref(r) => xref.get::<Dict<'_>>(r.into()),
+            MaybeRef::NotRef(Object::Dict(d)) => Some(d),
+            MaybeRef::NotRef(_) => None,
+        };
+        let Some(application) = application else {
+            continue;
+        };
+        if application.get::<Name<'_>>(b"Event").as_deref() != Some(VIEW) {
+            continue;
+        }
+        let views = application.get::<Array<'_>>(b"Category").is_some_and(|categories| {
+            categories
+                .raw_iter()
+                .take(MAX_AUTO_STATE_CATEGORIES)
+                .any(|c| matches!(c, MaybeRef::NotRef(Object::Name(n)) if n.as_ref() == VIEW))
+        });
+        if !views {
+            continue;
+        }
+        let Some(groups) = application.get::<Array<'_>>(OCGS) else {
+            continue;
+        };
+        for group in groups.raw_iter() {
+            if groups_left == 0 {
+                break;
+            }
+            groups_left -= 1;
+            let Some(r) = group.as_obj_ref() else {
+                continue;
+            };
+            let id: ObjectIdentifier = r.into();
+            let state = xref
+                .get::<Dict<'_>>(id)
+                .and_then(|g| g.get::<Dict<'_>>(USAGE))
+                .and_then(|u| u.get::<Dict<'_>>(VIEW))
+                .and_then(|v| v.get::<Name<'_>>(VIEW_STATE));
+            let on = match state.as_deref() {
+                Some(ON) => true,
+                Some(OFF) => false,
+                _ => continue,
+            };
+            let entry = recommended.entry(id).or_insert(true);
+            *entry &= on;
+        }
+    }
+    for (id, on) in recommended {
+        if on {
+            inactive.remove(&id);
+        } else {
+            inactive.insert(id);
+        }
+    }
+}
+
 /// PdfCraft patch (40): how deep visibility expressions may nest; deeper ones are malformed.
 const MAX_VISIBILITY_EXPRESSION_DEPTH: usize = 32;