From 32ca91a31f1ee222d48c377e9ae24db6482c724e Mon Sep 17 00:00:00 2001 From: Guillem Hermida Date: Fri, 16 Oct 2026 19:00:02 +0200 Subject: [PATCH 1/3] M5: sticky notes stay upright where they are placed on rotated pages (#2583) Notes and file attachments are saved with the NoRotate flag (/F bit 5), but nothing honoured it: on a page with /Rotate the icon was fitted to /Rect and turned with the page, and sat away from where it was placed. The vendored renderer now honours NoRotate (hayro-interpret patch 37): the appearance fitted to /Rect is turned against the page's rotation about the rectangle's upper-left corner, which stays put (ISO 32000-2 12.5.3). Form fields and links stay in /Rect, where their click targets come from. A counterrotated appearance, as stamps use, would be turned twice by viewers that honour the flag. Annotation::shown_rect gives the box the note is drawn in. The UI uses it for hit-testing, outlines, moving, the hover pop-up, the Comments panel flash and the edit composer; a resize maps it back to /Rect. comment_add anchors an icon's displayed top-left at `at`, and comment_list, doc_info and comment_edit use the drawn box. This also covers the sticky-note part of #1869. --- ROADMAP.md | 1 + crates/annot/README.md | 5 + crates/annot/src/lib.rs | 14 +++ crates/automation/src/comments.rs | 23 ++-- crates/automation/src/lib.rs | 3 +- crates/engine/src/lib.rs | 1 + crates/engine/src/tests.rs | 41 +++++++ crates/render/src/inspect.rs | 115 ++++++++++++++++++++ crates/render/src/raster.rs | 85 +++++++++++++++ crates/ui-egui/src/canvas.rs | 3 +- crates/ui-egui/src/comments.rs | 32 ++++-- crates/ui-egui/src/comments_panel.rs | 7 +- crates/ui-egui/tests/comments.rs | 32 ++++++ parity/acrobat-features.toml | 4 +- vendor/README.md | 2 +- vendor/hayro-interpret/src/interpret/mod.rs | 34 +++++- 16 files changed, 373 insertions(+), 29 deletions(-) diff --git a/ROADMAP.md b/ROADMAP.md index 259dbfec0..6bb2c0833 100644 --- a/ROADMAP.md +++ b/ROADMAP.md @@ -157,6 +157,7 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md | 8 | 1.0 polish: performance, right-to-left, missing languages, keyboard-only (M14) | 55–95 | ## Progress log +- **2026-10-16 (M5, #2583, #1869):** A sticky note on a page with `/Rotate` is drawn upright, with its top-left where it was placed. Notes are saved with the NoRotate flag, and the renderer now honours it: the appearance is turned against the page's rotation about the upper-left corner of `/Rect`, which stays put (ISO 32000-2 §12.5.3; vendored hayro-interpret patch 37). Attachment icons follow the same rule. Selection, hover, the edit composer, the Comments panel's flash and the `comment_list` and `doc_info` rectangles use the drawn box, and `comment_add` anchors an icon's displayed top-left at `at`. Estimates unchanged. - **2026-10-15 (M2, gap 12):** `cargo xtask text-oracle` scores reading order as well as word F1: the longest common subsequence of the two word sequences over the longer one. F1 is a bag of words and cannot see order at all -- measuring #2431, a change that moves text and nothing else, it called 26 of the 29 files whose output changed identical, where the order score separated 18 improvements from 6 regressions. Reports, does not gate; a disagreement with Poppler, not a verdict. Estimates unchanged. - **2026-10-15 (M2, #2305 follow-up):** WinAnsiEncoding 0xA0 and 0xAD and MacRomanEncoding 0xCA extract as the ordinary space and hyphen Annex D.2 names them, in the vendored hayro tables (patch 30) as well as `crates/fonts`, so search, copy and redaction see characters that were hidden behind U+00A0 and U+00AD. Across the pdf.js corpus the soft hyphens we extract fall 117 to 13 and the non-breaking spaces 30 to 26; the rest come from `/ToUnicode`. An explicit `/Differences` name keeps its glyph list value, as pdftotext does. 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. diff --git a/crates/annot/README.md b/crates/annot/README.md index 8ba3ca8c4..55df6d1a7 100644 --- a/crates/annot/README.md +++ b/crates/annot/README.md @@ -51,6 +51,11 @@ deterministic; the engine supplies them. signature) and PdfCraft's rubber stamps counterrotate their appearance on rotated pages, so the picture or label reads upright as the page is displayed. Restyling a picture stamp keeps that appearance matrix. The rotation is the page's `/Rotate` at placement. +- Sticky notes and file attachments are not counterrotated: they carry the NoRotate flag and + their icon is drawn in an unturned 20 × 20 box. A viewer keeps them upright on a turned page, + about the upper-left corner of `/Rect` (ISO 32000-2 §12.5.3). PdfCraft's renderer does so + (vendored hayro-interpret patch 37), and `Summary::no_rotate` tells the UI where the icon is + drawn. ## Not yet diff --git a/crates/annot/src/lib.rs b/crates/annot/src/lib.rs index f4973aa02..590848cb2 100644 --- a/crates/annot/src/lib.rs +++ b/crates/annot/src/lib.rs @@ -751,6 +751,17 @@ fn annotation_flags(doc: &Document, d: &Dict) -> i64 { d.get(b"F").and_then(|f| doc.resolve(f).as_int()).unwrap_or(0) } +/// Whether the renderer keeps the annotation upright on a turned page: the NoRotate flag, with +/// `/F` read as the renderer reads it (a real truncated, a value that isn't a `u32` no flags). +fn drawn_upright(doc: &Document, d: &Dict) -> bool { + let f = match d.get(b"F").map(|f| doc.resolve(f)).as_deref() { + Some(Object::Int(i)) => *i, + Some(Object::Real(r)) => *r as i64, + _ => 0, + }; + u32::try_from(f).is_ok_and(|f| i64::from(f) & FLAG_NO_ROTATE != 0) +} + /// Whether any page has a Fill & Sign annotation that flatten would bake in (not hidden). pub fn has_visible_fill_sign(doc: &Document) -> bool { let Ok(pages) = page_refs(doc) else { return false }; @@ -2082,6 +2093,8 @@ pub struct Summary { pub intent: Option, /// A Fill & Sign mark ([`is_fill_sign`]). pub fill_sign: bool, + /// The NoRotate flag (`/F` bit 5), as `pdfcraft_render::Annotation::no_rotate`. + pub no_rotate: bool, } fn text_value(doc: &Document, d: &Dict, key: &[u8]) -> Option { @@ -2155,6 +2168,7 @@ pub fn summaries(doc: &Document) -> Vec { locked: annotation_flags(doc, d) & FLAG_LOCKED != 0, intent: d.get(b"IT").and_then(|o| doc.resolve(o).as_name().map(|n| String::from_utf8_lossy(n).into_owned())), fill_sign: is_fill_sign(doc, d), + no_rotate: drawn_upright(doc, d), }); } } diff --git a/crates/automation/src/comments.rs b/crates/automation/src/comments.rs index 302688f72..109e4acf3 100644 --- a/crates/automation/src/comments.rs +++ b/crates/automation/src/comments.rs @@ -3,9 +3,7 @@ //! Geometry follows the automation convention: points from the top-left of the displayed page, //! y down. It is converted to PDF user space (crop box, `/Rotate`) here. -use pdfcraft_engine::{ - Edit, LineEnding, Markup, NOTE_SIZE, NewAnnotation, NoteIcon, ReviewState, Rgb, Shape, StampGroup, StampKind, Style, SummarySort, -}; +use pdfcraft_engine::{Edit, LineEnding, Markup, NewAnnotation, NoteIcon, ReviewState, Rgb, Shape, StampGroup, StampKind, Style, SummarySort}; use pdfcraft_render::{Annotation, PageInfo}; use serde_json::{Value, json}; @@ -79,13 +77,11 @@ fn rect_to_user(p: &PageInfo, r: [f64; 4]) -> [f64; 4] { } /// The engine's `at` for a note or attachment icon whose top-left corner *as displayed* is the -/// view point (x, y): the top-left (`[x0, y1]`) of the icon's square in user space. Converting -/// the point alone is only right on unrotated pages — under `/Rotate` that corner of the -/// displayed square is another corner of the user-space one, and the icon lands one icon-width -/// away from where it was asked for. +/// view point (x, y): the upper-left corner (`[x0, y1]`) of its `/Rect`. Notes and attachments +/// are NoRotate, so on a turned page the icon is drawn upright with its top-left at that corner +/// (ISO 32000-2 §12.5.3), and `comment_list` reports that square (`Annotation::shown_rect`). fn icon_anchor(p: &PageInfo, x: f64, y: f64) -> [f64; 2] { - let r = rect_to_user(p, [x, y, x + NOTE_SIZE, y + NOTE_SIZE]); - [r[0], r[3]] + to_user(p, x, y) } /// A user-space rectangle as the displayed-page rectangle every tool reports and accepts: @@ -182,7 +178,7 @@ impl Automation { "author": c.author, "contents": c.contents, "modified": c.modified, - "rect": rect_to_view(p, c.rect), + "rect": rect_to_view(p, c.shown_rect(p.rotation)), "color": c.color.map(hex), }) }; @@ -626,7 +622,12 @@ impl Automation { edits.push(Edit::FontSizeAnnotation { page, index, size }); } if let Some(r) = a.nums::<4>("rect")? { - edits.push(Edit::ResizeAnnotation { page, index, rect: rect_to_user(&info, r) }); + // The rectangle as comment_list reports it: where the comment is drawn. + let doc = self.doc(a)?; + let rect = rect_to_user(&info, r); + let rect = + doc.info.annotations.iter().find(|c| c.page == page && c.index == index).map_or(rect, |c| c.rect_for_shown(rect, info.rotation)); + edits.push(Edit::ResizeAnnotation { page, index, rect }); } if let Some([dx, dy]) = a.nums::<2>("move")? { let (o, d) = (to_user(&info, 0.0, 0.0), to_user(&info, dx, dy)); diff --git a/crates/automation/src/lib.rs b/crates/automation/src/lib.rs index 88856305b..e1abeaa6c 100644 --- a/crates/automation/src/lib.rs +++ b/crates/automation/src/lib.rs @@ -2417,7 +2417,8 @@ fn info(d: &Document) -> Value { // geometry-taking tools (#129). "annotations": i.annotations.iter().map(|a| json!({ "page": page1(a.page), "type": a.subtype, "author": a.author, "contents": a.contents, - "modified": a.modified, "name": a.name, "in_reply_to": a.in_reply_to, "rect": view_rect(i, a.page, a.rect), + "modified": a.modified, "name": a.name, "in_reply_to": a.in_reply_to, + "rect": view_rect(i, a.page, a.shown_rect(i.pages.get(a.page).map_or(0, |p| p.rotation))), })).collect::>(), "fields": i.fields.iter().map(|f| json!({ "name": f.name, "kind": format!("{:?}", f.kind), "value": f.value, "page": f.page.map(page1), diff --git a/crates/engine/src/lib.rs b/crates/engine/src/lib.rs index 3432cd868..fd8c66cac 100644 --- a/crates/engine/src/lib.rs +++ b/crates/engine/src/lib.rs @@ -2710,6 +2710,7 @@ fn comment_list(doc: &pdfcraft_cos::Document) -> Vec Vec<[u8; 4]> { + (80..120).flat_map(|y| (200..240).map(move |x| (x, y))).map(|(x, y)| stamp_pixel(rendered, x, y)).collect() + }; + let mut upright = None; + for degrees in [0, 90, 180, 270] { + let (mut session, id) = rotated_blank(degrees); + let info = session.get(id).unwrap().info.pages[0].clone(); + let [x, y] = info.view_to_user(100.0, 40.0); + let shape = Shape::Note { at: [f64::from(x), f64::from(y)], icon: NoteIcon::Comment }; + let style = Style::default_for(&shape); + session.apply(id, Edit::AddAnnotation(NewAnnotation { page: 0, style, shape, contents: "Turned".into(), author: "Test".into() })).unwrap(); + let mut check = |session: &Session, id: DocId| { + let doc = session.get(id).unwrap(); + let info = &doc.info.pages[0]; + let note = &doc.info.annotations[0]; + assert!(note.no_rotate, "rotation {degrees}: notes are saved NoRotate"); + let shown = note.shown_rect(info.rotation); + let (a, b) = (info.user_to_view(shown[0], shown[1]), info.user_to_view(shown[2], shown[3])); + let shown = [a[0].min(b[0]), a[1].min(b[1]), a[0].max(b[0]), a[1].max(b[1])]; + assert!(shown.iter().zip([100.0, 40.0, 120.0, 60.0]).all(|(v, w)| (v - w).abs() < 0.01), "rotation {degrees}: shown at {shown:?}"); + let icon = crop(&render_page(session, id)); + let reference = upright.get_or_insert_with(|| icon.clone()); + assert!(reference.iter().filter(|c| c.iter().any(|v| *v < 200)).count() > 100, "the note's icon is drawn"); + let off = icon.iter().zip(reference.iter()).filter(|(a, b)| a.iter().zip(b.iter()).any(|(a, b)| a.abs_diff(*b) > 48)).count(); + assert!(off <= 8, "rotation {degrees}: {off} of the icon's {} pixels differ from the unturned page's", icon.len()); + }; + check(&session, id); + let saved = session.save_bytes(id).unwrap(); + let mut reopened = Session::new().with_clock(|| 1_700_000_000); + let again = reopened.open("drawing.pdf", None, saved, None).unwrap(); + check(&reopened, again); + } +} + /// Fill & Sign text and text boxes read across as displayed on turned pages (scanned forms are /// often stored sideways with `/Rotate`), including after save. #[test] diff --git a/crates/render/src/inspect.rs b/crates/render/src/inspect.rs index 3c3988b7b..fb2b5a9c7 100644 --- a/crates/render/src/inspect.rs +++ b/crates/render/src/inspect.rs @@ -383,6 +383,10 @@ pub struct Annotation { /// A Fill & Sign mark: typed text, a check, cross, dot or line, or a typed, drawn or picture /// signature or initials (the rules of `pdfcraft_annot::is_fill_sign`). pub fill_sign: bool, + /// The NoRotate flag (`/F` bit 5), set on sticky notes and attachments: on a turned page the + /// appearance stays upright, turned about the upper-left corner of `rect` + /// ([`Self::shown_rect`]). + pub no_rotate: bool, } impl Annotation { @@ -390,6 +394,58 @@ impl Annotation { pub fn is_mark(&self) -> bool { matches!(self.state.as_deref(), Some("Marked" | "Unmarked")) } + + /// Where the comment is drawn on a page turned by `rotation` (`PageInfo::rotation`), in user + /// space: `rect`, or for a NoRotate comment `rect` turned upright about its upper-left corner + /// (see [`Self::no_rotate`]). Hit-testing, selection and anything placed beside the comment + /// use this, so they match the drawing. + pub fn shown_rect(&self, rotation: u16) -> [f32; 4] { + self.shown(self.rect.map(f64::from), rotation).map(|v| v as f32) + } + + /// [`Self::shown_rect`] for this comment with its `/Rect` at `rect` (one being moved or + /// resized). + pub fn shown(&self, rect: [f64; 4], rotation: u16) -> [f64; 4] { + if self.no_rotate { upright_rect(rect, rotation) } else { rect } + } + + /// The `/Rect` that shows this comment in the user-space box `shown`: the inverse of + /// [`Self::shown`], for a box resized as drawn. + pub fn rect_for_shown(&self, shown: [f64; 4], rotation: u16) -> [f64; 4] { + if self.no_rotate { rect_from_upright(shown, rotation) } else { shown } + } +} + +/// The user-space box a NoRotate annotation with `/Rect` `rect` is drawn in on a page turned +/// clockwise by `rotation` (ISO 32000-2 §12.5.3): `rect` turned as far counterclockwise about its +/// upper-left corner `(x0, y1)`, which stays where it is on the page. As displayed, the box's +/// top-left is that corner, the rectangle's width runs across and its height down, as on an +/// unturned page. `rect` itself for 0 or anything that isn't a quarter turn. The vendored +/// renderer draws the appearance there (hayro-interpret patch 37). +fn upright_rect(rect: [f64; 4], rotation: u16) -> [f64; 4] { + let [x0, y0, x1, y1] = [rect[0].min(rect[2]), rect[1].min(rect[3]), rect[0].max(rect[2]), rect[1].max(rect[3])]; + let (w, h) = (x1 - x0, y1 - y0); + match rotation { + 90 => [x0, y1, x0 + h, y1 + w], + 180 => [x0 - w, y1, x0, y1 + h], + 270 => [x0 - h, y1 - w, x0, y1], + _ => [x0, y0, x1, y1], + } +} + +/// The `/Rect` whose [`upright_rect`] is `shown`: `shown` turned back about the corner that +/// stays put. +fn rect_from_upright(shown: [f64; 4], rotation: u16) -> [f64; 4] { + let [s0, s1, s2, s3] = [shown[0].min(shown[2]), shown[1].min(shown[3]), shown[0].max(shown[2]), shown[1].max(shown[3])]; + let (sw, sh) = (s2 - s0, s3 - s1); + // The upper-left corner of `/Rect` and the rectangle's width and height. + let ((x0, y1), w, h) = match rotation { + 90 => ((s0, s1), sh, sw), + 180 => ((s2, s1), sw, sh), + 270 => ((s2, s3), sh, sw), + _ => return [s0, s1, s2, s3], + }; + [x0, y1 - h, x0 + w, y1] } #[derive(Clone, Copy, Debug, PartialEq, Eq)] @@ -1141,6 +1197,7 @@ impl<'a> Inspector<'a> { locked: d.get(b"F").ok().and_then(|f| self.resolve(f).as_i64().ok()).unwrap_or(0) & 128 != 0, intent: self.name(d, b"IT"), fill_sign, + no_rotate: d.get(b"F").is_ok_and(|f| draw_flags(self.resolve(f)) & 16 != 0), }); } } @@ -1554,6 +1611,17 @@ fn close_outline(open: &mut Vec<(usize, OutlineItem)>, roots: &mut Vec u32 { + let f = match o { + Object::Integer(i) => *i, + Object::Real(r) => *r as i64, + _ => 0, + }; + u32::try_from(f).unwrap_or(0) +} + fn rect4(o: &Object) -> [f32; 4] { match o { Object::Array(a) if a.len() == 4 => { @@ -1900,6 +1968,53 @@ mod tests { assert_eq!(p.view_rect_to_quad([0.0, 0.0, 10.0, 5.0]), [10.0, 320.0, 20.0, 320.0, 10.0, 315.0, 20.0, 315.0]); } + /// #2583: a NoRotate comment is drawn upright with its top-left as displayed at the upper-left + /// corner of `/Rect`, its width across and its height down; a box resized as drawn maps back to + /// the `/Rect` that shows it. Other comments, and pages that aren't quarter-turned, keep `/Rect`. + #[test] + fn no_rotate_comments_are_shown_upright_from_their_corner_and_back() { + let mut note = super::Annotation { + page: 0, + subtype: "Text".into(), + author: None, + contents: None, + modified: None, + name: None, + in_reply_to: None, + rect: [40.0, 50.0, 70.0, 70.0], + color: None, + index: 0, + state: None, + quads: Vec::new(), + locked: false, + intent: None, + fill_sign: false, + no_rotate: true, + }; + let rect = note.rect.map(f64::from); + for rotation in [0u16, 90, 180, 270, 45] { + let (w, h) = if rotation % 180 == 90 { (140.0, 200.0) } else { (200.0, 140.0) }; + let p = super::PageInfo { width: w, height: h, label: String::new(), crop: [0.0, 0.0, 200.0, 140.0], rotation }; + let shown = note.shown_rect(rotation); + let (a, b) = (p.user_to_view(shown[0], shown[1]), p.user_to_view(shown[2], shown[3])); + let [cx, cy] = p.user_to_view(40.0, 70.0); + let want = [cx, cy, cx + 30.0, cy + 20.0]; + let got = [a[0].min(b[0]), a[1].min(b[1]), a[0].max(b[0]), a[1].max(b[1])]; + assert!(got.iter().zip(want).all(|(g, w)| (g - w).abs() < 1e-3), "{rotation}: shown {got:?}, want {want:?}"); + assert_eq!(note.rect_for_shown(note.shown(rect, rotation), rotation), rect, "{rotation}"); + // Moved by (5, -7) and resized to 50 × 10 as drawn, back to a /Rect. + let resized = [shown[0] as f64 + 5.0, shown[1] as f64 - 7.0, 0.0, 0.0]; + let across = if rotation % 180 == 90 { [10.0, 50.0] } else { [50.0, 10.0] }; + let resized = [resized[0], resized[1], resized[0] + across[0], resized[1] + across[1]]; + let back = note.rect_for_shown(resized, rotation); + assert_eq!((back[2] - back[0], back[3] - back[1]), (50.0, 10.0), "{rotation}: {back:?}"); + assert_eq!(note.shown(back, rotation), resized, "{rotation}"); + } + note.no_rotate = false; + assert_eq!(note.shown_rect(90), note.rect); + assert_eq!(note.rect_for_shown([1.0, 2.0, 3.0, 4.0], 90), [1.0, 2.0, 3.0, 4.0]); + } + use super::*; #[test] diff --git a/crates/render/src/raster.rs b/crates/render/src/raster.rs index 8d5686dd8..04f2f46b6 100644 --- a/crates/render/src/raster.rs +++ b/crates/render/src/raster.rs @@ -3034,6 +3034,91 @@ trailer << /Root 1 0 R >> } } + /// #2583, #1869: a sticky note carries the NoRotate flag (`/F` bit 5), so on a page with + /// `/Rotate` its icon stays upright as displayed, turned about the upper-left corner of its + /// `/Rect`, which stays where it is on the page (ISO 32000-2 §12.5.3). It was drawn turned with + /// the page, in `/Rect` as displayed (vendored hayro-interpret patch 37). A note without the + /// flag still turns with the page. `Annotation::shown_rect` is where the UI hit-tests the + /// note, and must be where it is drawn. + #[test] + fn no_rotate_notes_stay_upright_at_their_corner_on_rotated_pages() { + // A 30 × 20 appearance: red in its top-left quarter, blue elsewhere. + let content = "0 0 1 rg 0 0 30 20 re f 1 0 0 rg 0 10 15 10 re f"; + for rotation in [0u16, 90, 180, 270] { + let objects = [ + "<< /Type /Catalog /Pages 2 0 R >>".to_string(), + "<< /Type /Pages /Kids [3 0 R] /Count 1 >>".to_string(), + format!("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 200 140] /Rotate {rotation} /Annots [4 0 R 5 0 R] >>"), + // NoRotate (and Print), as PdfCraft and Acrobat save notes. + "<< /Type /Annot /Subtype /Text /F 20 /Rect [40 50 70 70] /AP << /N 6 0 R >> >>".to_string(), + // The same appearance without the flag. + "<< /Type /Annot /Subtype /Text /F 4 /Rect [120 50 150 70] /AP << /N 6 0 R >> >>".to_string(), + format!("<< /Type /XObject /Subtype /Form /BBox [0 0 30 20] /Length {} >>\nstream\n{content}\nendstream", content.len()), + ]; + let mut pdf = b"%PDF-1.7\n".to_vec(); + let mut offsets = Vec::new(); + for (i, object) in objects.iter().enumerate() { + offsets.push(pdf.len()); + pdf.extend_from_slice(format!("{} 0 obj\n{object}\nendobj\n", i + 1).as_bytes()); + } + let xref = pdf.len(); + pdf.extend_from_slice(format!("xref\n0 {}\n0000000000 65535 f \n", objects.len() + 1).as_bytes()); + for offset in offsets { + pdf.extend_from_slice(format!("{offset:010} 00000 n \n").as_bytes()); + } + pdf.extend_from_slice(format!("trailer << /Size {} /Root 1 0 R >>\nstartxref\n{xref}\n%%EOF\n", objects.len() + 1).as_bytes()); + let pdf = Arc::new(pdf); + let info = crate::inspect(pdf.clone(), None).expect("opens"); + let page = &info.pages[0]; + assert_eq!(page.rotation, rotation); + let mut r = PageRenderer::new(pdf, RenderConfig::default()); + let p = r.render(RenderRequest { page: 0, scale: 1.0, ..Default::default() }); + assert!(p.error.is_none(), "{:?}", p.error); + // Displayed-page points are device pixels at scale 1. + let px = |x: f32, y: f32| -> [u8; 4] { + let (x, y) = (x.clamp(0.0, p.width as f32 - 1.0) as u32, y.clamp(0.0, p.height as f32 - 1.0) as u32); + p.rgba[((y * p.width + x) * 4) as usize..][..4].try_into().unwrap() + }; + let red: fn([u8; 4]) -> bool = |c| c[0] > 200 && c[1] < 60 && c[2] < 60; + let blue: fn([u8; 4]) -> bool = |c| c[2] > 200 && c[0] < 60 && c[1] < 60; + let paper: fn([u8; 4]) -> bool = |c| c.iter().all(|v| *v > 240); + + // The NoRotate note: its upper-left corner (40, 70) is the top-left of an upright + // 30 × 20 icon as displayed, red quarter top-left. + let [cx, cy] = page.user_to_view(40.0, 70.0); + for ((dx, dy), want, what) in [ + ((7.5, 5.0), red, "red top-left quarter"), + ((22.5, 5.0), blue, "blue top-right"), + ((7.5, 15.0), blue, "blue bottom-left"), + ((22.5, 15.0), blue, "blue bottom-right"), + ((-3.0, 10.0), paper, "paper left of the icon"), + ((10.0, -3.0), paper, "paper above the icon"), + ((33.0, 10.0), paper, "paper right of its 30 pt width"), + ((10.0, 23.0), paper, "paper below its 20 pt height"), + ] { + let c = px(cx + dx, cy + dy); + assert!(want(c), "rotation {rotation}: NoRotate note at corner + ({dx}, {dy}) is {c:?}, want {what}"); + } + let note = info.annotations.iter().find(|a| a.no_rotate).expect("the NoRotate note is listed"); + let shown = note.shown_rect(rotation); + let (a, b) = (page.user_to_view(shown[0], shown[1]), page.user_to_view(shown[2], shown[3])); + let box_ = [a[0].min(b[0]), a[1].min(b[1]), a[0].max(b[0]), a[1].max(b[1])]; + let want = [cx, cy, cx + 30.0, cy + 20.0]; + assert!(box_.iter().zip(want).all(|(v, w)| (v - w).abs() < 0.01), "rotation {rotation}: shown {box_:?}, drawn {want:?}"); + + // The note without the flag turns with the page: its red quarter stays at the + // user-space top-left of /Rect, at the displayed top-left only when unturned. + let plain = info.annotations.iter().find(|a| !a.no_rotate).expect("the plain note is listed"); + assert_eq!(plain.shown_rect(rotation), plain.rect); + let [rx, ry] = page.user_to_view(127.5, 65.0); + let [bx, by] = page.user_to_view(142.5, 55.0); + assert!(red(px(rx, ry)) && blue(px(bx, by)), "rotation {rotation}: the plain note's appearance turns with the page"); + let (a, b) = (page.user_to_view(120.0, 50.0), page.user_to_view(150.0, 70.0)); + let corner = px(a[0].min(b[0]) + 4.0, a[1].min(b[1]) + 4.0); + assert_eq!(red(corner), rotation == 0, "rotation {rotation}: displayed top-left of the plain note is {corner:?}"); + } + } + /// #1968: the highlight appearance other editors write, a transparency group whose content is /// filled with `/BM /Multiply`, drew as an opaque block over the text. Every `/Group` form /// got a layer of its own, so the Multiply blended with that layer's transparent backdrop diff --git a/crates/ui-egui/src/canvas.rs b/crates/ui-egui/src/canvas.rs index 0acab64a0..3704a98ea 100644 --- a/crates/ui-egui/src/canvas.rs +++ b/crates/ui-egui/src/canvas.rs @@ -3354,7 +3354,7 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) { let freehand = matches!(tool, QuickTool::Comment(comments::CommentTool::Ink | comments::CommentTool::Eraser)); let gesturing = view.comments.gesture.is_some() || freehand; for a in info.annotations.iter().filter(|a| a.page == i && !gesturing) { - let sr = xf.user_rect(info, i, a.rect); + let sr = xf.user_rect(info, i, a.shown_rect(info.pages.get(i).map_or(0, |p| p.rotation))); let hidden = prefs.hide_empty_highlight_hover && a.subtype == "Highlight" && a.contents.as_deref().is_none_or(str::is_empty); if sr.contains(p) && hover_text.is_none() @@ -5141,6 +5141,7 @@ mod tests { locked: false, intent: None, fill_sign, + no_rotate: false, } } diff --git a/crates/ui-egui/src/comments.rs b/crates/ui-egui/src/comments.rs index 5a708c14f..b677431fe 100644 --- a/crates/ui-egui/src/comments.rs +++ b/crates/ui-egui/src/comments.rs @@ -553,8 +553,14 @@ impl PageCx<'_> { self.xf.norm_to_screen(v[0] / self.xf.pw, v[1] / self.xf.ph) } + /// The page's clockwise `/Rotate`. + fn rotation(&self) -> u16 { + self.info.pages.get(self.page).map_or(0, |p| p.rotation) + } + + /// Where a comment is drawn on screen (a sticky note stays upright on a turned page). fn screen_rect(&self, a: &Annotation) -> Rect { - self.xf.user_rect(self.info, self.page, a.rect) + self.xf.user_rect(self.info, self.page, a.shown_rect(self.rotation())) } /// Top-level comments on this page, in paint order. @@ -625,16 +631,13 @@ impl PageCx<'_> { if let Some(Edit::ResizeAnnotation { page, index, rect }) = pending && (*page, *index) == (self.page, a.index) { - return self.xf.user_rect(self.info, self.page, rect.map(|v| v as f32)); + return self.xf.user_rect(self.info, self.page, a.shown(*rect, self.rotation()).map(|v| v as f32)); } if let Some(Edit::MoveAnnotation { page, index, dx, dy }) = pending && (*page, *index) == (self.page, a.index) { - return self.xf.user_rect( - self.info, - self.page, - [a.rect[0] + *dx as f32, a.rect[1] + *dy as f32, a.rect[2] + *dx as f32, a.rect[3] + *dy as f32], - ); + let r = a.shown_rect(self.rotation()); + return self.xf.user_rect(self.info, self.page, [r[0] + *dx as f32, r[1] + *dy as f32, r[2] + *dx as f32, r[3] + *dy as f32]); } match (gesture, pointer) { (Some(Gesture::Move { page, index, from }), Some(p)) if (*page, *index) == (self.page, a.index) => r.translate(p - *from), @@ -1091,7 +1094,7 @@ fn select_input( let from_rect = cx.screen_rect(a); let shift = ui.input(|i| i.modifiers.shift); let r = resized(from_rect, handle, p - from, corner_ratio(from_rect, aspect_ratio, shift)); - let rect = cx.rect_to_user(r); + let rect = a.rect_for_shown(cx.rect_to_user(r), cx.rotation()); if r.width() >= 4.0 && r.height() >= 4.0 { view.pending_edit = Some(Edit::ResizeAnnotation { page, index, rect }); } @@ -1131,7 +1134,7 @@ fn select_input( cv.selected = Some((cx.page, a.index)); cv.composer = Some(Composer { page: cx.page, - at: [a.rect[2] as f64, a.rect[3] as f64], + at: edit_anchor(cx.info, a), kind: ComposerKind::Edit(a.index), text: a.contents.clone().unwrap_or_default(), focus: true, @@ -1332,6 +1335,15 @@ fn erase(cx: &PageCx<'_>, path: &[[f64; 2]]) -> Option { Some(Edit::Batch { label: "Erase".into(), edits }) } +/// Where the composer for editing `a` opens: the top-right corner of the comment as drawn on the +/// displayed page. +fn edit_anchor(info: &DocInfo, a: &Annotation) -> [f64; 2] { + let Some(p) = info.pages.get(a.page) else { return [f64::from(a.rect[2]), f64::from(a.rect[3])] }; + let r = a.shown_rect(p.rotation); + let (u, v) = (p.user_to_view(r[0], r[1]), p.user_to_view(r[2], r[3])); + p.view_to_user(u[0].max(v[0]), u[1].min(v[1])).map(f64::from) +} + fn new_comment(cx: &PageCx<'_>, tool: CommentTool, shape: Shape, contents: String) -> Edit { Edit::AddAnnotation(NewAnnotation { page: cx.page, shape, style: cx.prefs.style(tool), contents, author: cx.prefs.author.clone() }) } @@ -1588,7 +1600,7 @@ pub(crate) fn context_menu(ui: &mut egui::Ui, view: &mut DocView, info: &DocInfo if ui.add_enabled(allowed, egui::Button::new(tl!("Edit text…"))).clicked() { view.comments.composer = Some(Composer { page, - at: [a.rect[2] as f64, a.rect[3] as f64], + at: edit_anchor(info, a), kind: ComposerKind::Edit(index), text: a.contents.clone().unwrap_or_default(), focus: true, diff --git a/crates/ui-egui/src/comments_panel.rs b/crates/ui-egui/src/comments_panel.rs index c5fbbd0ce..482d40e08 100644 --- a/crates/ui-egui/src/comments_panel.rs +++ b/crates/ui-egui/src/comments_panel.rs @@ -255,7 +255,8 @@ pub(crate) fn show( continue; } let thread = replies_of(a); - if let Some(e) = card(ui, t, a, &thread, view, prefs, allowed, nav) { + let rotation = info.pages.get(a.page).map_or(0, |p| p.rotation); + if let Some(e) = card(ui, t, a, rotation, &thread, view, prefs, allowed, nav) { edit = Some(e); } } @@ -267,6 +268,8 @@ fn card( ui: &mut egui::Ui, t: &Tokens, a: &Annotation, + // The clockwise `/Rotate` of the comment's page. + rotation: u16, thread: &[&Annotation], view: &mut DocView, prefs: &CommentPrefs, @@ -430,7 +433,7 @@ fn card( if click.clicked() { view.comments.selected = Some(key); view.comments.reply.clear(); - *nav = Some(Nav::Flash(a.page, a.rect)); + *nav = Some(Nav::Flash(a.page, a.shown_rect(rotation))); } if click.double_clicked() && allowed { view.comments.editing = Some((a.page, a.index, a.contents.clone().unwrap_or_default())); diff --git a/crates/ui-egui/tests/comments.rs b/crates/ui-egui/tests/comments.rs index 914bc281b..f479aebe4 100644 --- a/crates/ui-egui/tests/comments.rs +++ b/crates/ui-egui/tests/comments.rs @@ -488,6 +488,38 @@ fn sticky_note_composer_posts_and_returns_to_select() { assert_eq!(comments(&h).len(), 1); } +/// #2583: a sticky note is saved NoRotate, so on a turned page its icon is drawn upright from +/// where it was placed (right and down), not where its `/Rect` lies as displayed. The Select tool +/// picks it where it is drawn. +#[test] +fn a_sticky_note_on_a_turned_page_is_picked_where_it_is_drawn() { + let mut h = harness(|app| { + app.apply_edit(pdfcraft_engine::Edit::RotatePages { pages: vec![0], degrees: 90 }); + app.set_option("quick", "note").unwrap(); + }); + settle(&mut h); + let page = h.state().views[0].page_screen_rect(0).expect("page 1 on screen"); + // Screen pixels per point of the displayed page (300 × 200 pt turned to 200 × 300). + let k = page.width() / 200.0; + let placed = page.left_top() + egui::vec2(100.0 * k, 60.0 * k); + click(&mut h, placed); + field(&h, "Add a comment").type_text("Turned"); + h.run_steps(2); + h.get_by_label("Post").click(); + h.run_steps(3); + settle(&mut h); + let c = comments(&h); + assert_eq!(c.len(), 1); + assert_eq!((c[0].subtype.as_str(), c[0].no_rotate), ("Text", true)); + assert_eq!(h.state().quick_tool, QuickTool::Select); + // Where the note's /Rect lies as displayed, left of where it was placed: nothing. + click(&mut h, placed + egui::vec2(-10.0 * k, 10.0 * k)); + assert_eq!(h.state().views[0].comments.selected, None); + // On the icon, drawn right and down from where it was placed. + click(&mut h, placed + egui::vec2(10.0 * k, 10.0 * k)); + assert_eq!(h.state().views[0].comments.selected, Some((0, 0))); +} + #[test] fn comments_are_selected_moved_and_deleted_with_the_select_tool() { let mut h = harness(|app| app.set_option("quick", "square").unwrap()); diff --git a/parity/acrobat-features.toml b/parity/acrobat-features.toml index c3d319f94..72ee77962 100644 --- a/parity/acrobat-features.toml +++ b/parity/acrobat-features.toml @@ -3572,8 +3572,8 @@ milestone = "M5" status = "shipped" commands = ["comment.note"] tools = ["comment_add"] -evidence = ["crates/ui-egui/tests/comments.rs::sticky_note_composer_posts_and_returns_to_select", "crates/ui-egui/tests/comments.rs::the_comment_composer_shortcut_only_posts_the_focused_editor", "crates/annot/src/tests.rs::every_tool_creates_a_drawable_comment"] -notes = "our own Comment/Note icon drawings, with a pop-up; Ctrl+Enter (also Command+Enter) posts from the focused editor, Shift+Enter adds a newline" +evidence = ["crates/ui-egui/tests/comments.rs::sticky_note_composer_posts_and_returns_to_select", "crates/ui-egui/tests/comments.rs::the_comment_composer_shortcut_only_posts_the_focused_editor", "crates/annot/src/tests.rs::every_tool_creates_a_drawable_comment", "crates/engine/src/tests.rs::sticky_notes_stay_upright_where_they_are_placed_on_rotated_pages", "crates/render/src/raster.rs::no_rotate_notes_stay_upright_at_their_corner_on_rotated_pages", "crates/ui-egui/tests/comments.rs::a_sticky_note_on_a_turned_page_is_picked_where_it_is_drawn"] +notes = "our own Comment/Note icon drawings, with a pop-up; Ctrl+Enter (also Command+Enter) posts from the focused editor, Shift+Enter adds a newline. Notes are saved with the NoRotate flag, and it is honoured: on a page with /Rotate the icon stays upright with its top-left where it was placed, and is selected and hovered there (#2583). Missing: NoZoom (the icon scales with the zoom); printing and flattening still place the icon in its /Rect, turned with the page" [[feature]] id = "comment.highlight" diff --git a/vendor/README.md b/vendor/README.md index 62be4cad9..6a815d22b 100644 --- a/vendor/README.md +++ b/vendor/README.md @@ -9,7 +9,7 @@ A PdfCraft patch in `hayro-syntax::Page::base_dimensions` expresses the existing | 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 `.notdef` for a CID of an Adobe character collection (Japan1, GB1, CNS1, Korea1) that it can't find by Unicode: hayro took the CID for a glyph index of the substitute, which numbers its glyphs its own way, so the character was drawn as an unrelated one (𠮷 as "Ｑ" in BIZ UDPGothic); (22) a Type 0 font drawn with a substitute finds a character given with a variation selector (`/ToUnicode`, or Adobe-Japan1-UCS2, which maps the JIS X 0208 forms of the kanji JIS X 0213:2004 redrew that way: CID 1247 is 噂 U+E0100) through the substitute's variation sequences, else as the character alone: such a CID found no glyph and was drawn as an unrelated one; (23) backport of hayro 0.8's `gradient` module: `EncodedShadingPattern::as_svg_gradient` turns an axial or radial shading into stops of an sRGB gradient (within 0.01 per channel, a stop at every stitching-function bound via `Function::stitching_bounds`) when a native gradient draws it exactly (extended past the box drawn in; a radial one with no background colour); (24) `DecodedImageCache`: decoded image pixels shared across renders through the interpreter cache within a 64 MiB budget (an image-heavy page decoded its thousands of images again per tile and per scale), keyed by every decode input (`DecodedImageKey`: the stream's place and dictionary bytes, its object, mask/stencil/raster output, interpolation, decode target size, preview or final); images under a transfer function and inline images are not cached, an image over the budget is never kept, and a poisoned lock decodes uncached; (25) `Device::actual_text` hands the device a marked content sequence's `/ActualText` (inline or named in `/Properties`), so text extraction keeps the Thai, Indic and Arabic clusters Chrome marks this way (their shaped tone marks and vowels have no Unicode of their own); (26) non-embedded simple fonts drawn with a standard-font stand-in build the Latin Extended-A letters the stand-in lacks (`/Gbreve`, `/Idotaccent`, `/Scedilla`, `/aogonek`, `/ccaron`…) from the base letter and accent it has (`StandardKind::composite_glyph`, `ACCENTED_LETTERS`): PDFium's Foxit stand-ins hold only the Standard/WinAnsi/Mac Roman glyphs, so iText letters with Times New Roman unembedded showed blanks for Ğ, İ, Ş, ğ and ş; (27) an embedded TrueType font without a base encoding reads its codes as WinAnsi for text extraction (`EmbeddedKind::code_to_unicode_name`), matching pdftotext's output, instead of Standard: OCR text layers of scans write WinAnsi codes with no `/Encoding`, and Standard has no ü, ö or ß and reads é as Ø (glyph selection unchanged); (28) an image whose pixels hold at most 16 bits of samples (an 8-bit Separation or DeviceN image, a 1-, 2- or 4-bit gray or indexed one) is converted once per possible pixel value through a lookup table instead of once per pixel (`rgb_via_lookup`): a spot-colour tint transform is often a PostScript calculator program, run for every pixel, so a fillable character sheet's 3304 × 4105 DeviceN background took four seconds and 430 MB to draw a page; each entry is converted on its own (`convert_each`), so a tint transform giving the wrong number of components for a value the image never uses cannot shift the colours of the values it does; (29) `Device::begin_annotations`/`end_annotations` bracket the page's annotations, so a device can tell comment paints from document paints (reading colours recolour the document, never a comment, #1548); (30) the Annex D.2 alias codes read as the ink they draw: WinAnsiEncoding 0xA0 and 0xAD, and MacRomanEncoding 0xCA, carry the glyph names `space` and `hyphen` -- the names those tables also give 0x20 and 0x2D -- not `nbspace` and `sfthyphen`, whose glyph list values U+00A0 and U+00AD are invisible formatting characters. Extraction hid drawn ink behind them, so search and redaction matched neither (`XiaoBiaoSong.pdf` in the pdf.js corpus drew 104 soft hyphens; across the 989 files the soft hyphens we extract fall from 117 to 13 and the non-breaking spaces from 30 to 26, what remains coming from `/ToUnicode` maps, which say so explicitly). The base table only, in the text direction only: a `/Differences` name keeps its glyph list value and glyph lookup is untouched, both as pdftotext does; (31) a Type 0 sampled function keeps its samples in one flat array indexed by strides (`Type0::entry`) instead of a hash map keyed by each grid point: the map was built again, one hashed insert per sample, every time a shading using the function was painted (a 2,097,152-sample ramp painted 30 times took 38 s; `bug1721218` in the pdf.js corpus spent most of a 1.5 s page on it); the interpolation is unchanged, so results are bit for bit the same, a sample missing from a short stream still gives no value, and the maximum sample value is computed without overflowing at 32 bits per sample; (32) the decoded-image cache (24) keys a decode by the size the image is drawn at only when a JPX stream decodes it or its soft mask or mask (`decode_reads_target`): JPX is the only filter that decodes a reduced resolution for a size, so a JPEG or Flate scan was decoded again at every zoom step and kept once per zoom level; one decode now serves every size; `InterpreterCache::decoded_image_decodes` counts the cache's decodes; (33) Coons and tensor-product patches copy their vertices' colours less: each point of a patch's 20 × 20 triangle grid is interpolated once (`generate_patch_triangles`; each cell interpolated its four corners itself, 1,444 interpolations a patch for 400 points), a cell's second triangle takes the vertices its first copied, and the mesh sampler moves a patch's triangles into device space instead of copying each vertex (`TriangleSampler::sample_owned`); the same values, 10–16% faster on mesh-gradient pages | (1) `raster::tests::widget_appearance_states`; (2, 3) `xtask text-oracle` on corpus/pdfjs issue918; (4) `raster::tests::layer_override_hides_content`; (5) `raster::tests::soft_mask_without_group_colour_space_is_applied`; (6) `raster::tests::self_referencing_tiling_pattern_terminates`, `raster::tests::self_referencing_type3_glyph_terminates`; (7) `raster::tests::huge_cid_width_ranges_terminate`; (8) `raster::tests::hiding_comments_keeps_fields`; (9) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (10) `raster::tests::type3_glyphs_that_fan_out_into_themselves_finish`; (11) `raster::tests::type1_font_programs_with_long_runs_of_numbers_load_quickly`; (12) `raster::tests::degenerate_annotation_appearances_are_skipped`; (13) `raster::tests::highlights_without_an_appearance_are_drawn`; (14) `raster::tests::malformed_simple_font_widths_keep_later_widths_and_the_font`, `raster::tests::huge_simple_font_first_char_terminates`; (15) `raster::tests::annotation_appearance_box_maps_onto_rect`; (16) `raster::tests::clipping_path_applies_after_any_painting_operator`; (17) `raster::tests::dropping_the_pool_cancels_its_renders`, `raster::tests::the_watchdog_stops_the_render_it_gives_up_on`; (18) `raster::tests::page_contents_and_nested_forms_share_one_budget`, `raster::tests::content_painted_again_and_again_keeps_drawing`, `raster::tests::sampled_functions_with_billions_of_samples_are_refused`; (19) `raster::tests::mesh_shadings_far_off_the_page_render`, `raster::tests::many_overlapping_mesh_triangles_finish`, `raster::tests::mesh_shadings_with_many_patches_or_no_rows_finish`; (20) `raster::tests::function_limits_calculator_arithmetic_and_depth`, `raster::tests::function_limits_public_construction_work_and_depth`, `raster::tests::function_limits_stitching_cycles_are_refused_and_shared_children_work`, `raster::tests::function_limits_invalid_calculator_keeps_the_page_renderable`; (21) `raster::tests::cids_a_substitute_lacks_are_not_drawn_as_other_glyphs`; (26) `raster::tests::accented_letters_the_standard_stand_ins_lack_are_built_from_their_parts`; (22) `raster::tests::variation_sequences_draw_their_kanji_in_a_substitute`; (23) `raster::tests::axial_and_radial_shadings_draw_as_native_gradients`; (24) `raster::tests::tiles_match_full_render`, `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel` (both decode through the cache), `raster::tests::decoded_images_with_identical_dictionaries_keep_their_own_pixels`, `raster::tests::decoded_images_under_a_transfer_function_are_not_shared`; (25) `text::tests::actual_text_replaces_the_glyphs_it_marks_in_every_writing_direction`, `text::tests::actual_text_follows_the_pdf_rules_for_marked_content`; (27) `text::tests::truetype_without_an_encoding_reads_its_codes_as_win_ansi`; (28) `raster::tests::spot_colour_images_run_their_tint_transform_once_per_sample_value`, `raster::tests::images_converted_through_a_table_keep_every_samples_colour`, `raster::tests::a_malformed_tint_for_an_unused_sample_keeps_the_used_colours`; (29) `raster::tests::reading_colors_keep_highlight_annotations_readable`; (30) `text::tests::the_annex_d2_alias_codes_read_as_an_ordinary_hyphen_and_space`; (31) `raster::tests::shadings_with_large_sampled_functions_paint_quickly`, `raster::tests::sampled_functions_interpolate_their_grid_as_before`; (32) `raster::tests::decoded_images_serve_every_size_they_are_drawn_at`; (33) `raster::tests::patch_meshes_keep_their_corner_colours` | +| 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 `.notdef` for a CID of an Adobe character collection (Japan1, GB1, CNS1, Korea1) that it can't find by Unicode: hayro took the CID for a glyph index of the substitute, which numbers its glyphs its own way, so the character was drawn as an unrelated one (𠮷 as "Ｑ" in BIZ UDPGothic); (22) a Type 0 font drawn with a substitute finds a character given with a variation selector (`/ToUnicode`, or Adobe-Japan1-UCS2, which maps the JIS X 0208 forms of the kanji JIS X 0213:2004 redrew that way: CID 1247 is 噂 U+E0100) through the substitute's variation sequences, else as the character alone: such a CID found no glyph and was drawn as an unrelated one; (23) backport of hayro 0.8's `gradient` module: `EncodedShadingPattern::as_svg_gradient` turns an axial or radial shading into stops of an sRGB gradient (within 0.01 per channel, a stop at every stitching-function bound via `Function::stitching_bounds`) when a native gradient draws it exactly (extended past the box drawn in; a radial one with no background colour); (24) `DecodedImageCache`: decoded image pixels shared across renders through the interpreter cache within a 64 MiB budget (an image-heavy page decoded its thousands of images again per tile and per scale), keyed by every decode input (`DecodedImageKey`: the stream's place and dictionary bytes, its object, mask/stencil/raster output, interpolation, decode target size, preview or final); images under a transfer function and inline images are not cached, an image over the budget is never kept, and a poisoned lock decodes uncached; (25) `Device::actual_text` hands the device a marked content sequence's `/ActualText` (inline or named in `/Properties`), so text extraction keeps the Thai, Indic and Arabic clusters Chrome marks this way (their shaped tone marks and vowels have no Unicode of their own); (26) non-embedded simple fonts drawn with a standard-font stand-in build the Latin Extended-A letters the stand-in lacks (`/Gbreve`, `/Idotaccent`, `/Scedilla`, `/aogonek`, `/ccaron`…) from the base letter and accent it has (`StandardKind::composite_glyph`, `ACCENTED_LETTERS`): PDFium's Foxit stand-ins hold only the Standard/WinAnsi/Mac Roman glyphs, so iText letters with Times New Roman unembedded showed blanks for Ğ, İ, Ş, ğ and ş; (27) an embedded TrueType font without a base encoding reads its codes as WinAnsi for text extraction (`EmbeddedKind::code_to_unicode_name`), matching pdftotext's output, instead of Standard: OCR text layers of scans write WinAnsi codes with no `/Encoding`, and Standard has no ü, ö or ß and reads é as Ø (glyph selection unchanged); (28) an image whose pixels hold at most 16 bits of samples (an 8-bit Separation or DeviceN image, a 1-, 2- or 4-bit gray or indexed one) is converted once per possible pixel value through a lookup table instead of once per pixel (`rgb_via_lookup`): a spot-colour tint transform is often a PostScript calculator program, run for every pixel, so a fillable character sheet's 3304 × 4105 DeviceN background took four seconds and 430 MB to draw a page; each entry is converted on its own (`convert_each`), so a tint transform giving the wrong number of components for a value the image never uses cannot shift the colours of the values it does; (29) `Device::begin_annotations`/`end_annotations` bracket the page's annotations, so a device can tell comment paints from document paints (reading colours recolour the document, never a comment, #1548); (30) the Annex D.2 alias codes read as the ink they draw: WinAnsiEncoding 0xA0 and 0xAD, and MacRomanEncoding 0xCA, carry the glyph names `space` and `hyphen` -- the names those tables also give 0x20 and 0x2D -- not `nbspace` and `sfthyphen`, whose glyph list values U+00A0 and U+00AD are invisible formatting characters. Extraction hid drawn ink behind them, so search and redaction matched neither (`XiaoBiaoSong.pdf` in the pdf.js corpus drew 104 soft hyphens; across the 989 files the soft hyphens we extract fall from 117 to 13 and the non-breaking spaces from 30 to 26, what remains coming from `/ToUnicode` maps, which say so explicitly). The base table only, in the text direction only: a `/Differences` name keeps its glyph list value and glyph lookup is untouched, both as pdftotext does; (31) a Type 0 sampled function keeps its samples in one flat array indexed by strides (`Type0::entry`) instead of a hash map keyed by each grid point: the map was built again, one hashed insert per sample, every time a shading using the function was painted (a 2,097,152-sample ramp painted 30 times took 38 s; `bug1721218` in the pdf.js corpus spent most of a 1.5 s page on it); the interpolation is unchanged, so results are bit for bit the same, a sample missing from a short stream still gives no value, and the maximum sample value is computed without overflowing at 32 bits per sample; (32) the decoded-image cache (24) keys a decode by the size the image is drawn at only when a JPX stream decodes it or its soft mask or mask (`decode_reads_target`): JPX is the only filter that decodes a reduced resolution for a size, so a JPEG or Flate scan was decoded again at every zoom step and kept once per zoom level; one decode now serves every size; `InterpreterCache::decoded_image_decodes` counts the cache's decodes; (33) Coons and tensor-product patches copy their vertices' colours less: each point of a patch's 20 × 20 triangle grid is interpolated once (`generate_patch_triangles`; each cell interpolated its four corners itself, 1,444 interpolations a patch for 400 points), a cell's second triangle takes the vertices its first copied, and the mesh sampler moves a patch's triangles into device space instead of copying each vertex (`TriangleSampler::sample_owned`); the same values, 10–16% faster on mesh-gradient pages; (37) the NoRotate flag (`/F` bit 5) is honoured: the appearance, fitted to `/Rect`, is turned against the page's `/Rotate` about the rectangle's upper-left corner, which stays where it is, so it reads upright as the page is displayed (ISO 32000-2 §12.5.3). Sticky notes, which are saved NoRotate, were drawn turned with the page (#2583). Form fields and links stay in `/Rect`, where PdfCraft takes their click targets from | (1) `raster::tests::widget_appearance_states`; (2, 3) `xtask text-oracle` on corpus/pdfjs issue918; (4) `raster::tests::layer_override_hides_content`; (5) `raster::tests::soft_mask_without_group_colour_space_is_applied`; (6) `raster::tests::self_referencing_tiling_pattern_terminates`, `raster::tests::self_referencing_type3_glyph_terminates`; (7) `raster::tests::huge_cid_width_ranges_terminate`; (8) `raster::tests::hiding_comments_keeps_fields`; (9) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (10) `raster::tests::type3_glyphs_that_fan_out_into_themselves_finish`; (11) `raster::tests::type1_font_programs_with_long_runs_of_numbers_load_quickly`; (12) `raster::tests::degenerate_annotation_appearances_are_skipped`; (13) `raster::tests::highlights_without_an_appearance_are_drawn`; (14) `raster::tests::malformed_simple_font_widths_keep_later_widths_and_the_font`, `raster::tests::huge_simple_font_first_char_terminates`; (15) `raster::tests::annotation_appearance_box_maps_onto_rect`; (16) `raster::tests::clipping_path_applies_after_any_painting_operator`; (17) `raster::tests::dropping_the_pool_cancels_its_renders`, `raster::tests::the_watchdog_stops_the_render_it_gives_up_on`; (18) `raster::tests::page_contents_and_nested_forms_share_one_budget`, `raster::tests::content_painted_again_and_again_keeps_drawing`, `raster::tests::sampled_functions_with_billions_of_samples_are_refused`; (19) `raster::tests::mesh_shadings_far_off_the_page_render`, `raster::tests::many_overlapping_mesh_triangles_finish`, `raster::tests::mesh_shadings_with_many_patches_or_no_rows_finish`; (20) `raster::tests::function_limits_calculator_arithmetic_and_depth`, `raster::tests::function_limits_public_construction_work_and_depth`, `raster::tests::function_limits_stitching_cycles_are_refused_and_shared_children_work`, `raster::tests::function_limits_invalid_calculator_keeps_the_page_renderable`; (21) `raster::tests::cids_a_substitute_lacks_are_not_drawn_as_other_glyphs`; (26) `raster::tests::accented_letters_the_standard_stand_ins_lack_are_built_from_their_parts`; (22) `raster::tests::variation_sequences_draw_their_kanji_in_a_substitute`; (23) `raster::tests::axial_and_radial_shadings_draw_as_native_gradients`; (24) `raster::tests::tiles_match_full_render`, `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel` (both decode through the cache), `raster::tests::decoded_images_with_identical_dictionaries_keep_their_own_pixels`, `raster::tests::decoded_images_under_a_transfer_function_are_not_shared`; (25) `text::tests::actual_text_replaces_the_glyphs_it_marks_in_every_writing_direction`, `text::tests::actual_text_follows_the_pdf_rules_for_marked_content`; (27) `text::tests::truetype_without_an_encoding_reads_its_codes_as_win_ansi`; (28) `raster::tests::spot_colour_images_run_their_tint_transform_once_per_sample_value`, `raster::tests::images_converted_through_a_table_keep_every_samples_colour`, `raster::tests::a_malformed_tint_for_an_unused_sample_keeps_the_used_colours`; (29) `raster::tests::reading_colors_keep_highlight_annotations_readable`; (30) `text::tests::the_annex_d2_alias_codes_read_as_an_ordinary_hyphen_and_space`; (31) `raster::tests::shadings_with_large_sampled_functions_paint_quickly`, `raster::tests::sampled_functions_interpolate_their_grid_as_before`; (32) `raster::tests::decoded_images_serve_every_size_they_are_drawn_at`; (33) `raster::tests::patch_meshes_keep_their_corner_colours`; (37) `raster::tests::no_rotate_notes_stay_upright_at_their_corner_on_rotated_pages` | | 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 `.notdef` for a CID of an Adobe character collection (Japan1, GB1, CNS1, Korea1) that it can't find by Unicode: hayro took the CID for a glyph index of the substitute, which numbers its glyphs its own way, so the character was drawn as an unrelated one (𠮷 as "Ｑ" in BIZ UDPGothic); (22) a Type 0 font drawn with a substitute finds a character given with a variation selector (`/ToUnicode`, or Adobe-Japan1-UCS2, which maps the JIS X 0208 forms of the kanji JIS X 0213:2004 redrew that way: CID 1247 is 噂 U+E0100) through the substitute's variation sequences, else as the character alone: such a CID found no glyph and was drawn as an unrelated one; (23) backport of hayro 0.8's `gradient` module: `EncodedShadingPattern::as_svg_gradient` turns an axial or radial shading into stops of an sRGB gradient (within 0.01 per channel, a stop at every stitching-function bound via `Function::stitching_bounds`) when a native gradient draws it exactly (extended past the box drawn in; a radial one with no background colour); (24) `DecodedImageCache`: decoded image pixels shared across renders through the interpreter cache within a 64 MiB budget (an image-heavy page decoded its thousands of images again per tile and per scale), keyed by every decode input (`DecodedImageKey`: the stream's place and dictionary bytes, its object, mask/stencil/raster output, interpolation, decode target size, preview or final); images under a transfer function and inline images are not cached, an image over the budget is never kept, and a poisoned lock decodes uncached; (25) `Device::actual_text` hands the device a marked content sequence's `/ActualText` (inline or named in `/Properties`), so text extraction keeps the Thai, Indic and Arabic clusters Chrome marks this way (their shaped tone marks and vowels have no Unicode of their own); (26) non-embedded simple fonts drawn with a standard-font stand-in build the Latin Extended-A letters the stand-in lacks (`/Gbreve`, `/Idotaccent`, `/Scedilla`, `/aogonek`, `/ccaron`…) from the base letter and accent it has (`StandardKind::composite_glyph`, `ACCENTED_LETTERS`): PDFium's Foxit stand-ins hold only the Standard/WinAnsi/Mac Roman glyphs, so iText letters with Times New Roman unembedded showed blanks for Ğ, İ, Ş, ğ and ş; (27) an embedded TrueType font without a base encoding reads its codes as WinAnsi for text extraction (`EmbeddedKind::code_to_unicode_name`), matching pdftotext's output, instead of Standard: OCR text layers of scans write WinAnsi codes with no `/Encoding`, and Standard has no ü, ö or ß and reads é as Ø (glyph selection unchanged); (28) a transparency group that is neither isolated nor knockout, entered at full opacity with no soft mask and Normal blending, is drawn straight onto the page instead of into a layer of its own (§11.4.7: its elements composite with the backdrop; the layer made a `/BM /Multiply` highlight appearance an opaque block, #1968); (29) `Device::begin_annotations`/`end_annotations` bracket the page's annotations, so a device can tell comment paints from document paints (reading colours recolour the document, never a comment, #1548) | (1) `raster::tests::widget_appearance_states`; (2, 3) `xtask text-oracle` on corpus/pdfjs issue918; (4) `raster::tests::layer_override_hides_content`; (5) `raster::tests::soft_mask_without_group_colour_space_is_applied`; (6) `raster::tests::self_referencing_tiling_pattern_terminates`, `raster::tests::self_referencing_type3_glyph_terminates`; (7) `raster::tests::huge_cid_width_ranges_terminate`; (8) `raster::tests::hiding_comments_keeps_fields`; (9) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (10) `raster::tests::type3_glyphs_that_fan_out_into_themselves_finish`; (11) `raster::tests::type1_font_programs_with_long_runs_of_numbers_load_quickly`; (12) `raster::tests::degenerate_annotation_appearances_are_skipped`; (13) `raster::tests::highlights_without_an_appearance_are_drawn`; (14) `raster::tests::malformed_simple_font_widths_keep_later_widths_and_the_font`, `raster::tests::huge_simple_font_first_char_terminates`; (15) `raster::tests::annotation_appearance_box_maps_onto_rect`; (16) `raster::tests::clipping_path_applies_after_any_painting_operator`; (17) `raster::tests::dropping_the_pool_cancels_its_renders`, `raster::tests::the_watchdog_stops_the_render_it_gives_up_on`; (18) `raster::tests::page_contents_and_nested_forms_share_one_budget`, `raster::tests::content_painted_again_and_again_keeps_drawing`, `raster::tests::sampled_functions_with_billions_of_samples_are_refused`; (19) `raster::tests::mesh_shadings_far_off_the_page_render`, `raster::tests::many_overlapping_mesh_triangles_finish`, `raster::tests::mesh_shadings_with_many_patches_or_no_rows_finish`; (20) `raster::tests::function_limits_calculator_arithmetic_and_depth`, `raster::tests::function_limits_public_construction_work_and_depth`, `raster::tests::function_limits_stitching_cycles_are_refused_and_shared_children_work`, `raster::tests::function_limits_invalid_calculator_keeps_the_page_renderable`; (21) `raster::tests::cids_a_substitute_lacks_are_not_drawn_as_other_glyphs`; (26) `raster::tests::accented_letters_the_standard_stand_ins_lack_are_built_from_their_parts` |; (22) `raster::tests::variation_sequences_draw_their_kanji_in_a_substitute`; (23) `raster::tests::axial_and_radial_shadings_draw_as_native_gradients`; (24) `raster::tests::tiles_match_full_render`, `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel` (both decode through the cache), `raster::tests::decoded_images_with_identical_dictionaries_keep_their_own_pixels`, `raster::tests::decoded_images_under_a_transfer_function_are_not_shared`; (25) `text::tests::actual_text_replaces_the_glyphs_it_marks_in_every_writing_direction`, `text::tests::actual_text_follows_the_pdf_rules_for_marked_content`; (27) `text::tests::truetype_without_an_encoding_reads_its_codes_as_win_ansi`; (28) `raster::tests::a_multiply_highlight_inside_a_transparency_group_keeps_the_text`; (29) `raster::tests::reading_colors_keep_highlight_annotations_readable` | | 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 `.notdef` for a CID of an Adobe character collection (Japan1, GB1, CNS1, Korea1) that it can't find by Unicode: hayro took the CID for a glyph index of the substitute, which numbers its glyphs its own way, so the character was drawn as an unrelated one (𠮷 as "Ｑ" in BIZ UDPGothic); (22) a Type 0 font drawn with a substitute finds a character given with a variation selector (`/ToUnicode`, or Adobe-Japan1-UCS2, which maps the JIS X 0208 forms of the kanji JIS X 0213:2004 redrew that way: CID 1247 is 噂 U+E0100) through the substitute's variation sequences, else as the character alone: such a CID found no glyph and was drawn as an unrelated one; (23) backport of hayro 0.8's `gradient` module: `EncodedShadingPattern::as_svg_gradient` turns an axial or radial shading into stops of an sRGB gradient (within 0.01 per channel, a stop at every stitching-function bound via `Function::stitching_bounds`) when a native gradient draws it exactly (extended past the box drawn in; a radial one with no background colour); (24) `DecodedImageCache`: decoded image pixels shared across renders through the interpreter cache within a 64 MiB budget (an image-heavy page decoded its thousands of images again per tile and per scale), keyed by every decode input (`DecodedImageKey`: the stream's place and dictionary bytes, its object, mask/stencil/raster output, interpolation, decode target size, preview or final); images under a transfer function and inline images are not cached, an image over the budget is never kept, and a poisoned lock decodes uncached; (25) `Device::actual_text` hands the device a marked content sequence's `/ActualText` (inline or named in `/Properties`), so text extraction keeps the Thai, Indic and Arabic clusters Chrome marks this way (their shaped tone marks and vowels have no Unicode of their own); (26) non-embedded simple fonts drawn with a standard-font stand-in build the Latin Extended-A letters the stand-in lacks (`/Gbreve`, `/Idotaccent`, `/Scedilla`, `/aogonek`, `/ccaron`…) from the base letter and accent it has (`StandardKind::composite_glyph`, `ACCENTED_LETTERS`): PDFium's Foxit stand-ins hold only the Standard/WinAnsi/Mac Roman glyphs, so iText letters with Times New Roman unembedded showed blanks for Ğ, İ, Ş, ğ and ş; (27) text in a Type 3 font is drawn filled whatever the text rendering mode, except mode 3 (invisible): only that mode has any effect on Type 3 fonts (ISO 32000-1 §9.3.6). Stroking ran the glyph procedure a second time, so `2 Tr` made bitmap fonts look bold, `1 Tr` painted an uncoloured glyph in the stroking colour, and `7 Tr` hid it; (29) `Device::begin_annotations`/`end_annotations` bracket the page's annotations, so a device can tell comment paints from document paints (reading colours recolour the document, never a comment, #1548) | (1) `raster::tests::widget_appearance_states`; (2, 3) `xtask text-oracle` on corpus/pdfjs issue918; (4) `raster::tests::layer_override_hides_content`; (5) `raster::tests::soft_mask_without_group_colour_space_is_applied`; (6) `raster::tests::self_referencing_tiling_pattern_terminates`, `raster::tests::self_referencing_type3_glyph_terminates`; (7) `raster::tests::huge_cid_width_ranges_terminate`; (8) `raster::tests::hiding_comments_keeps_fields`; (9) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (10) `raster::tests::type3_glyphs_that_fan_out_into_themselves_finish`; (11) `raster::tests::type1_font_programs_with_long_runs_of_numbers_load_quickly`; (12) `raster::tests::degenerate_annotation_appearances_are_skipped`; (13) `raster::tests::highlights_without_an_appearance_are_drawn`; (14) `raster::tests::malformed_simple_font_widths_keep_later_widths_and_the_font`, `raster::tests::huge_simple_font_first_char_terminates`; (15) `raster::tests::annotation_appearance_box_maps_onto_rect`; (16) `raster::tests::clipping_path_applies_after_any_painting_operator`; (17) `raster::tests::dropping_the_pool_cancels_its_renders`, `raster::tests::the_watchdog_stops_the_render_it_gives_up_on`; (18) `raster::tests::page_contents_and_nested_forms_share_one_budget`, `raster::tests::content_painted_again_and_again_keeps_drawing`, `raster::tests::sampled_functions_with_billions_of_samples_are_refused`; (19) `raster::tests::mesh_shadings_far_off_the_page_render`, `raster::tests::many_overlapping_mesh_triangles_finish`, `raster::tests::mesh_shadings_with_many_patches_or_no_rows_finish`; (20) `raster::tests::function_limits_calculator_arithmetic_and_depth`, `raster::tests::function_limits_public_construction_work_and_depth`, `raster::tests::function_limits_stitching_cycles_are_refused_and_shared_children_work`, `raster::tests::function_limits_invalid_calculator_keeps_the_page_renderable`; (21) `raster::tests::cids_a_substitute_lacks_are_not_drawn_as_other_glyphs`; (26) `raster::tests::accented_letters_the_standard_stand_ins_lack_are_built_from_their_parts` |; (22) `raster::tests::variation_sequences_draw_their_kanji_in_a_substitute`; (23) `raster::tests::axial_and_radial_shadings_draw_as_native_gradients`; (24) `raster::tests::tiles_match_full_render`, `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel` (both decode through the cache), `raster::tests::decoded_images_with_identical_dictionaries_keep_their_own_pixels`, `raster::tests::decoded_images_under_a_transfer_function_are_not_shared`; (25) `text::tests::actual_text_replaces_the_glyphs_it_marks_in_every_writing_direction`, `text::tests::actual_text_follows_the_pdf_rules_for_marked_content`; (27) `raster::tests::text_rendering_modes_other_than_invisible_draw_type3_glyphs_as_filled`; (29) `raster::tests::reading_colors_keep_highlight_annotations_readable` | | 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 `.notdef` for a CID of an Adobe character collection (Japan1, GB1, CNS1, Korea1) that it can't find by Unicode: hayro took the CID for a glyph index of the substitute, which numbers its glyphs its own way, so the character was drawn as an unrelated one (𠮷 as "Ｑ" in BIZ UDPGothic); (22) a Type 0 font drawn with a substitute finds a character given with a variation selector (`/ToUnicode`, or Adobe-Japan1-UCS2, which maps the JIS X 0208 forms of the kanji JIS X 0213:2004 redrew that way: CID 1247 is 噂 U+E0100) through the substitute's variation sequences, else as the character alone: such a CID found no glyph and was drawn as an unrelated one; (23) backport of hayro 0.8's `gradient` module: `EncodedShadingPattern::as_svg_gradient` turns an axial or radial shading into stops of an sRGB gradient (within 0.01 per channel, a stop at every stitching-function bound via `Function::stitching_bounds`) when a native gradient draws it exactly (extended past the box drawn in; a radial one with no background colour); (24) `DecodedImageCache`: decoded image pixels shared across renders through the interpreter cache within a 64 MiB budget (an image-heavy page decoded its thousands of images again per tile and per scale), keyed by every decode input (`DecodedImageKey`: the stream's place and dictionary bytes, its object, mask/stencil/raster output, interpolation, decode target size, preview or final); images under a transfer function and inline images are not cached, an image over the budget is never kept, and a poisoned lock decodes uncached; (25) Lab colour spaces honour their own `/WhitePoint`: L\*a\*b\* goes to XYZ relative to that white, is Bradford-adapted to D65 and converted to sRGB (ISO 32000-2 §8.6.5.4) instead of through `LAB.icc` as if every white were D50 (a D65 blue came out 45 levels too red; the profile is removed); components normalised to [0, 1] (`manual_scale`, an `/Indexed` palette over Lab) map onto L\* 0..100 and the a\*/b\* `/Range`; (29) `Device::begin_annotations`/`end_annotations` bracket the page's annotations, so a device can tell comment paints from document paints (reading colours recolour the document, never a comment, #1548) | (1) `raster::tests::widget_appearance_states`; (2, 3) `xtask text-oracle` on corpus/pdfjs issue918; (4) `raster::tests::layer_override_hides_content`; (5) `raster::tests::soft_mask_without_group_colour_space_is_applied`; (6) `raster::tests::self_referencing_tiling_pattern_terminates`, `raster::tests::self_referencing_type3_glyph_terminates`; (7) `raster::tests::huge_cid_width_ranges_terminate`; (8) `raster::tests::hiding_comments_keeps_fields`; (9) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (10) `raster::tests::type3_glyphs_that_fan_out_into_themselves_finish`; (11) `raster::tests::type1_font_programs_with_long_runs_of_numbers_load_quickly`; (12) `raster::tests::degenerate_annotation_appearances_are_skipped`; (13) `raster::tests::highlights_without_an_appearance_are_drawn`; (14) `raster::tests::malformed_simple_font_widths_keep_later_widths_and_the_font`, `raster::tests::huge_simple_font_first_char_terminates`; (15) `raster::tests::annotation_appearance_box_maps_onto_rect`; (16) `raster::tests::clipping_path_applies_after_any_painting_operator`; (17) `raster::tests::dropping_the_pool_cancels_its_renders`, `raster::tests::the_watchdog_stops_the_render_it_gives_up_on`; (18) `raster::tests::page_contents_and_nested_forms_share_one_budget`, `raster::tests::content_painted_again_and_again_keeps_drawing`, `raster::tests::sampled_functions_with_billions_of_samples_are_refused`; (19) `raster::tests::mesh_shadings_far_off_the_page_render`, `raster::tests::many_overlapping_mesh_triangles_finish`, `raster::tests::mesh_shadings_with_many_patches_or_no_rows_finish`; (20) `raster::tests::function_limits_calculator_arithmetic_and_depth`, `raster::tests::function_limits_public_construction_work_and_depth`, `raster::tests::function_limits_stitching_cycles_are_refused_and_shared_children_work`, `raster::tests::function_limits_invalid_calculator_keeps_the_page_renderable`; (21) `raster::tests::cids_a_substitute_lacks_are_not_drawn_as_other_glyphs` |; (22) `raster::tests::variation_sequences_draw_their_kanji_in_a_substitute`; (23) `raster::tests::axial_and_radial_shadings_draw_as_native_gradients`; (24) `raster::tests::tiles_match_full_render`, `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel` (both decode through the cache), `raster::tests::decoded_images_with_identical_dictionaries_keep_their_own_pixels`, `raster::tests::decoded_images_under_a_transfer_function_are_not_shared`; (25) `raster::tests::lab_honours_the_space_white_point`, `raster::tests::indexed_palettes_over_lab_keep_their_colours`; (29) `raster::tests::reading_colors_keep_highlight_annotations_readable` | diff --git a/vendor/hayro-interpret/src/interpret/mod.rs b/vendor/hayro-interpret/src/interpret/mod.rs index fe08fc284..a74f417f3 100644 --- a/vendor/hayro-interpret/src/interpret/mod.rs +++ b/vendor/hayro-interpret/src/interpret/mod.rs @@ -21,7 +21,7 @@ use hayro_syntax::object::dict::keys::{ ACTUAL_TEXT, ANNOTS, AP, AS, C, CA, F, MCID, N, OC, QUADPOINTS, RECT, SUBTYPE, }; use hayro_syntax::object::{Array, Dict, Object, Rect, Stream, dict_or_stream}; -use hayro_syntax::page::{ContentReader, Page, Resources}; +use hayro_syntax::page::{ContentReader, Page, Resources, Rotation}; use hayro_syntax::MAX_DECODED_STREAM; use kurbo::{Affine, BezPath, Point, Shape}; use smallvec::smallvec; @@ -249,6 +249,13 @@ pub fn interpret_page<'a>( annot_rect.x0 - sx * transformed_rect.x0, annot_rect.y0 - sy * transformed_rect.y0, ]); + // PdfCraft patch (37): NoRotate (bit 5) keeps the appearance upright as the page is + // displayed (ISO 32000-2 §12.5.3). + let affine = if flags & 16 != 0 && no_rotate_applies(&annot) { + no_rotate_turn(page.rotation(), &annot_rect) * affine + } else { + affine + }; // PdfCraft patch: a /BBox or /Matrix that collapses the appearance box to a line // or a point leaves nothing to map onto /Rect: the scales above divide by zero, @@ -282,6 +289,31 @@ pub fn interpret_page<'a>( } } +/// PdfCraft patch (37): whether the NoRotate flag turns this annotation's appearance. Form fields +/// and links are left in `/Rect`: PdfCraft takes their click targets from it. +fn no_rotate_applies(annot: &Dict<'_>) -> bool { + !matches!( + annot.get::>(SUBTYPE).as_deref(), + Some(b"Widget" | b"Link") + ) +} + +/// PdfCraft patch (37): the user-space turn that keeps a NoRotate annotation upright on a page +/// with `/Rotate`. The page is shown turned clockwise, so the appearance (already fitted to +/// `/Rect`) is turned as far counterclockwise about the rectangle's upper-left corner, which stays +/// where it is on the page (ISO 32000-2 §12.5.3). Its width then runs across the displayed page +/// and its height down from that corner. +fn no_rotate_turn(rotation: Rotation, rect: &kurbo::Rect) -> Affine { + let (cos, sin) = match rotation { + Rotation::None => return Affine::IDENTITY, + Rotation::Horizontal => (0.0, 1.0), + Rotation::Flipped => (-1.0, 0.0), + Rotation::FlippedHorizontal => (0.0, -1.0), + }; + let (x, y) = (rect.x0.min(rect.x1), rect.y0.max(rect.y1)); + Affine::new([cos, sin, -sin, cos, x - (cos * x - sin * y), y - (sin * x + cos * y)]) +} + /// PdfCraft patch: the normal appearance of a Highlight annotation that has none (ISO 32000-2 /// §12.5.6.10), drawn as PdfCraft draws its own highlights (pdfcraft-annot): each /QuadPoints /// quadrilateral filled with /C at /CA opacity, blended with Multiply so the marked-up content From cefbefbef933c111231bb9878e57686428ed0882 Mon Sep 17 00:00:00 2001 From: Guillem Hermida Date: Fri, 23 Oct 2026 11:25:36 +0200 Subject: [PATCH 2/3] deps: ctutils 0.4.5 for RUSTSEC-2026-0385 --- Cargo.lock | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Cargo.lock b/Cargo.lock index 47d73be44..9129e85d6 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -1291,9 +1291,9 @@ dependencies = [ [[package]] name = "ctutils" -version = "0.4.2" +version = "0.4.5" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "7d5515a3834141de9eafb9717ad39eea8247b5674e6066c404e8c4b365d2a29e" +checksum = "38d28fb6aa7dd680b3ffd1a3e64e115c177bdba63c0ae7f9cd48129ff05c4fa0" dependencies = [ "cmov", "subtle", From 07f0faf4d582dc828a8e9bd0c983bcb17a635287 Mon Sep 17 00:00:00 2001 From: Guillem Hermida Date: Fri, 23 Oct 2026 12:13:07 +0200 Subject: [PATCH 3/3] tests: the NoRotate test's annotation has main's points field too --- crates/render/src/inspect.rs | 1 + 1 file changed, 1 insertion(+) diff --git a/crates/render/src/inspect.rs b/crates/render/src/inspect.rs index 24564ca44..e4a3fb041 100644 --- a/crates/render/src/inspect.rs +++ b/crates/render/src/inspect.rs @@ -2356,6 +2356,7 @@ mod tests { intent: None, fill_sign: false, no_rotate: true, + points: Vec::new(), }; let rect = note.rect.map(f64::from); for rotation in [0u16, 90, 180, 270, 45] {