diff --git a/.github/dependabot.yml b/.github/dependabot.yml
new file mode 100644
index 0000000000..1910ea4bd3
--- /dev/null
+++ b/.github/dependabot.yml
@@ -0,0 +1,32 @@
+version: 2
+updates:
+ # Maintain Rust / Cargo dependencies
+ - package-ecosystem: "cargo"
+ directories:
+ - "/"
+ - "/crates/filters/fuzz"
+ schedule:
+ interval: "daily"
+ time: "06:00"
+ timezone: "America/New_York"
+ # Consolidate updates into a single PR rather than one PR per crate
+ groups:
+ cargo-dependencies:
+ patterns:
+ - "*"
+ commit-message:
+ prefix: "chore(deps)"
+
+ # Maintain GitHub Actions workflows
+ - package-ecosystem: "github-actions"
+ directory: "/"
+ schedule:
+ interval: "daily"
+ time: "06:00"
+ timezone: "America/New_York"
+ groups:
+ actions:
+ patterns:
+ - "*"
+ commit-message:
+ prefix: "ci(deps)"
diff --git a/.github/workflows/nix.yml b/.github/workflows/nix.yml
index 18392587a8..ce817da129 100644
--- a/.github/workflows/nix.yml
+++ b/.github/workflows/nix.yml
@@ -23,7 +23,7 @@ jobs:
timeout-minutes: 15
steps:
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
- - uses: cachix/install-nix-action@13d8dd58da0234aa297dedd986986ccb8e7f3e24 # v31.11.1
+ - uses: cachix/install-nix-action@448cde739abec1eaf0646249e3d25bb8d9bc7702 # v31.11.2
- name: nix flake check --no-build
run: |
if [ -f flake.nix ]; then
diff --git a/Cargo.lock b/Cargo.lock
index c6f4f49d69..53e0f097d5 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -4026,6 +4026,7 @@ dependencies = [
"pdfcraft-cos",
"pdfcraft-engine",
"pdfcraft-platform",
+ "pdfcraft-preflight",
"pdfcraft-render",
"rustix 1.1.5",
"serde_json",
@@ -4135,6 +4136,7 @@ dependencies = [
"pdfcraft-xfdf",
"png 0.18.1",
"thiserror 2.0.21",
+ "tiff",
"unicode-segmentation",
"url",
]
diff --git a/README.md b/README.md
index bd33ad23be..2a65b66457 100644
--- a/README.md
+++ b/README.md
@@ -11,7 +11,7 @@
PdfCraft
- The PDF workbench; an open-source, clean-room reimplementation of Adobe Acrobat, rebuilt in pure Rust. The PDF workbench; an open-source, modern PDF reader and editor, built in pure Rust.
@@ -417,16 +417,16 @@ Logs, environment variables and other development notes are in [docs/development
## What's next
-PdfCraft is young and moving fast. The aim is a workbench where you can view, organize, annotate, fill, sign and edit PDFs, at parity with Acrobat Pro.
+PdfCraft is young and moving fast. The aim is a workbench where you can view, organize, annotate, fill, sign and edit PDFs, for professional and everyday workflows.
-**Where it stands (October 2026), honestly:** PdfCraft is **alpha**. Most of Acrobat Pro's offline features exist in some form (57% of all tracked features, 94% of the must-haves). For everyday work (viewing, commenting, filling and signing forms, organizing pages) it is about 55% of the way to replacing Acrobat; against all of Acrobat Pro, about 49%; for someone who only needs the essentials, about 66%. Every everyday workflow can be completed, but converting and sharing is weak: no Office-to-PDF, and Word export that loses layout.
+**Where it stands (October 2026), honestly:** PdfCraft is **alpha**. Most essential offline PDF features exist in some form (57% of the features tracked in [docs/target-app-parity.md](docs/target-app-parity.md), 94% of core essentials). For everyday work (viewing, commenting, filling and signing forms, organizing pages) readiness is about 55%; across all professional workflows, about 49%; for someone who only needs the essentials, about 66%. Every everyday workflow can be completed, but converting and sharing is weak: no office-document-to-PDF, and .docx export that loses layout.
- **Good today:** viewing and search; organizing, combining and splitting; most kinds of comment; filling and authoring forms (with sandboxed JavaScript); passwords, redaction and sanitizing; basic digital signatures; printing; the Accessibility Checker; agent control through the CLI and MCP.
- **Still borrowed:** pages are drawn by the `hayro` crate while our own renderer is built.
-- **Thin or missing:** reliable editing of existing text (especially CJK and right-to-left), OCR beyond Latin script, Office import/export, signature tokens and online long-term validation, Preflight and PDF/X/UA, printing from the web build, physical printer fidelity, and full XFA forms.
-- **Hardening:** fuzzing still turns up crashes and hangs on hostile files; each one is fixed with a regression test. Quality has not yet been compared with Acrobat side by side.
+- **Thin or missing:** reliable editing of existing text (especially CJK and right-to-left), OCR beyond Latin script, office document import/export, signature tokens and online long-term validation, Preflight and PDF/X/UA, printing from the web build, physical printer fidelity, and full XFA forms.
+- **Hardening:** fuzzing still turns up crashes and hangs on hostile files; each one is fixed with a regression test. Quality has not yet been benchmarked side by side.
-**Next, in order:** Office-to-PDF and a Word export that keeps layout, the reported bugs and the Chinese interface font, a fidelity harness against Acrobat and the renderer, editing existing content, then the Pro workflows (OCR, Office, Preflight, signatures, accessibility tagging) and 1.0 polish.
+**Next, in order:** office-document-to-PDF and a .docx export that keeps layout, the reported bugs and the Chinese interface font, a fidelity harness against a reference renderer, editing existing content, then professional workflows (OCR, office formats, Preflight, signatures, accessibility tagging) and 1.0 polish.
The summary is in **[ROADMAP.md](ROADMAP.md)**; the full assessment by dimension and area is in [docs/target-app-parity.md](docs/target-app-parity.md), the ranked gaps in [docs/gaps.md](docs/gaps.md), and the milestones in [docs/roadmap.md](docs/roadmap.md).
@@ -581,8 +581,6 @@ ArtCraft Team and are not covered by this license. They may be used only unmodif
part of this repository and PdfCraft, under [`docs/brand/LICENSE-brand.txt`](docs/brand/LICENSE-brand.txt).
Forks and modified versions must remove them.
-Adobe, Photoshop, Illustrator, Premiere Pro, Lightroom, Acrobat, After Effects and InDesign are trademarks or registered trademarks of Adobe Inc. in the United States and/or other countries. PdfCraft is an independent, open-source project and is not affiliated with, sponsored by or endorsed by Adobe Inc.; these names are used only to describe the workflows it is compatible with.
-
Made by the ArtCraft team and community.
diff --git a/ROADMAP.md b/ROADMAP.md
index 6d67ad4418..8be3485b12 100644
--- a/ROADMAP.md
+++ b/ROADMAP.md
@@ -3,13 +3,18 @@
**Stage: alpha** · next: beta, ~25% (ready for real work 50% → 75%) and ~400–700 h away
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M7: the Edit text caret is as tall as the row it is on; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
+> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M6: CSV tab-containing headers; estimates unchanged)
> **Last reviewed:** 2026-10-18 · **Last updated:** 2026-10-18 · **Change:** minor (M3, #3149: discard render demand after page geometry changes and check the GPU texture limit before upload; estimates unchanged)
+> **Last reviewed:** 2026-10-18 · **Last updated:** 2026-10-18 · **Change:** minor (M4, #1919: repeated page indices accumulate rotations; M3, #3149: discard render demand after page geometry changes and check the GPU texture limit before upload; estimates unchanged)
> **Last reviewed:** 2026-10-18 · **Last updated:** 2026-10-18 · **Change:** minor (M8, #1350: named replacement descriptions retired during Sanitize and across shared resources; estimates unchanged)
+> **Last reviewed:** 2026-10-17 · **Last updated:** 2026-10-17 · **Change:** minor (M4 #2997: Ctrl-drag (Option on macOS) in the Organize grid copies the pages into the drop gap; `page_duplicate` takes `at`; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M6: AcroForm XML text; estimates unchanged)
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M2/gap 12: the Annex D.2 alias codes read as the ink they draw in both tables; the text oracle scores reading order; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M14/#2383 GPU-limit candidate; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
+> **Last reviewed:** 2026-10-12 · **Last updated:** 2026-10-13 · **Change:** minor (M4: indirect page rotations; M13: CLI check visits each directory once through aliases and cycles; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-13 · **Last updated:** 2026-10-13 · **Change:** minor (M6: preserve XML/XFDF attribute whitespace; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-14 · **Last updated:** 2026-10-14 · **Change:** minor (M1: raw-deflate fallback preserves decoding-limit errors; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
+> **Last reviewed:** 2026-10-16 | **Last updated:** 2026-10-16 | **Change:** minor (M10 #1201 partial: lossless image Decode; estimates unchanged) | **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-13 · **Last updated:** 2026-10-14 · **Change:** minor (M14: the Italian catalog has every entry the other catalogs had and the code looks up; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-14 · **Last updated:** 2026-10-14 · **Change:** minor (M3 #2080 candidate: explicit search jumps join per-tab history; rail and control-channel Back/Forward; estimates unchanged)
> **Last reviewed:** 2026-10-13 · **Last updated:** 2026-10-13 · **Change:** minor (M4 #1910: Organize Pages opens on the current page; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
@@ -34,6 +39,7 @@
> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-11 · **Change:** minor (#594 paragraph drags validated in the Linux test environment; localized shortcut labels, #496; readiness table with hours per audience; full number as weighted sum 49%; earlier: second look: ready 40% → 47%, mainstream 55% added; core-workflow gate applied → alpha) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-11 · **Change:** minor (architectural paper controls from #599; localized shortcut labels, #496; readiness table with hours per audience; full number as weighted sum 49%; earlier: second look: ready 40% → 47%, mainstream 55% added; core-workflow gate applied → alpha) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-11 · **Change:** minor (closed issues synced: #774, #756 and the #799–#821 batch; earlier: localized shortcut labels, #496; readiness table with hours per audience; full number as weighted sum 49%; earlier: second look: ready 40% → 47%, mainstream 55% added; core-workflow gate applied → alpha) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
+> **Last reviewed:** 2026-10-18 - **Last updated:** 2026-10-18 - **Change:** minor (M6 #2892: Push-button normal labels are editable per widget through Options and form_set_props, independently of field names; rendering, undo/redo and incremental/full save/reopen are covered. Estimates unchanged)
PdfCraft's one-page summary: how close it is to Adobe Acrobat Pro's offline feature set, and what
comes next. Adobe's cloud services (Document Cloud, Acrobat Sign requests, shared review, the AI
@@ -164,17 +170,34 @@ 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
+- **2026-10-18 (M6, #2892):** Push-button normal labels are editable per widget through Options and form_set_props, independently of field names; rendering, undo/redo and incremental/full save/reopen are covered. Estimates unchanged.
+
+- **2026-10-18 (M7, #3212):** Headers, Bates numbers, watermarks and added text stay visible after unfinished marked content, with original content preserved.
+
- **2026-10-18 (M2, #3116):** Distinguish clipping-only text from invisible OCR so image-filtered extraction retains clipped headings and ActualText. Vendor changes and regression coverage are documented; estimates unchanged.
+
- **2026-10-15 (M7, Edit text caret):** In a text region that mixes sizes (a 24 pt title over 12 pt lines), the inline editor's caret on a small row was as tall as the region's first size and reached into the row below: egui draws the caret at least one row of the text field's font tall, and that font had the paragraph's first size. The field's font is now half the region's smallest size, so each row's own height decides. Kittest regression; checked in the app on a real document. Estimates unchanged.
- **2026-10-16 (M3, #2781):** The Pages panel follows the current page, as Acrobat's page thumbnails do: going to a page whose thumbnail is out of sight (scrolling, Page Down, Go to page, a bookmark) scrolls the panel just enough to show it. A thumbnail already in view, or clicked, leaves the panel where it is, and a page change while the user scrolls the panel or drags thumbnails does not move it. Estimates unchanged.
- **2026-10-16 (M12, #1645):** CLI info, text and render use the engine's dynamic XFA layout rather than placeholder pages. Synthetic two-page and encrypted-form regressions cover counts, text, pixels, copy permissions and unchanged input bytes. Estimates unchanged.
+
+- **2026-10-18 (M4, PR #1919 review):** repeated page indices apply each rotation to the latest page-local value, preserving indirect/inherited angles. Synthetic adjacent/interleaved duplicates retain siblings and page contents through incremental and full saves. Estimates unchanged.
+- **2026-10-15 (M4, #1919 refresh):** Integrated upstream `99941daa` while retaining the indirect page rotation correction and all upstream progress. Only independent documentation conflicts required resolution; focused production and test files are unchanged from the previously green CI head. Estimates unchanged.
- **2026-10-18 (M3, #3149):** Page replacement cancels render demand collected for the previous geometry; all viewer raster upload paths reject empty or oversized textures before cache eviction. Synthetic regressions reproduce the reported 1880×2350 upload at a 2048 limit, retain cached images and unsaved edits, and resume bounded backdrops and tiles. Estimates unchanged.
- **2026-10-18 (M8, #1350):** Integrated current main while preserving content-stream comment boundaries. Hidden-text cleanup retires unused named replacement descriptions on rewritten pages and forms, including inherited/shared resource holders, without removing descriptions still used by surviving content. Synthetic saved-byte and preservation regressions cover the cleanup. Estimates unchanged.
+- **2026-10-17 (M4, #2997):** Dragging page thumbnails in the Organize grid with Ctrl held (Option on macOS) drops copies at the gap instead of moving the pages: a "+" badge sits by the pointer, the copies land at the gap selected, and one undo step ("Duplicate pages") takes them out again. `page_duplicate` takes `at` (1-based, where the first copy goes; left out, after the last of them) so agents can drive the same thing, and the UI control channel's `ui.drag` modifiers now reach gestures that read the input state. Estimates unchanged.
- **2026-10-15 (M6, AcroForm XML text):** Plain AcroForm XML imports retain all direct text around comments and processing instructions, including CDATA and character references. Synthetic regressions verify exact values and unknown catalog data after incremental and full save/reopen. Nested-element and XFA interpretation are unchanged. Estimates unchanged.
- **2026-10-15 (M6, #1917 refresh):** Integrated upstream `99941daa` while retaining the XML/XFDF attribute whitespace correction and all upstream progress. Only independent documentation conflicts required resolution; focused production and test files are unchanged from the previously green CI head. Estimates unchanged.
- **2026-10-15 (M1, #2267 refresh):** Integrated upstream `99941daa` while retaining the raw-deflate fallback correction and all upstream progress. Only independent documentation conflicts required resolution; focused production and test files are unchanged from the previously green CI head. Estimates unchanged.
+- **2026-10-16 (M10, #1201 partial):** Lossless image export applies each component's Decode pair. Indexed values round and clip before palette lookup. Color-key masks keep raw samples. The existing 16-bit precision limit remains. JPEG and mask Decode, ICC transforms, and calibrated colors remain outside this fix. Estimates unchanged.
+- **2026-10-16 (M3, #2739):** Wrapped revision buttons stay above Binding in Document Properties. A synthetic regression checks control bounds and opens a saved revision at two viewport widths. 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.
+- **2026-10-15 (M12, XFA, #2574):** A `pageArea` with no `` repeats without limit instead of taking a subform's default of one use, so a later page area with `occur max="1"` is no longer drawn (with a copy of its fields) over every body page. A page area's `occur min` is honoured too: one the body never reached gets the pages it still needs after the body (the once-only closing page of the public ISCIII form, with its button, is page 7 of 7). Acrobat Pro shows the form as 7 pages with that page last, as here; when printing, Acrobat keeps that page but leaves it blank (`relevant="-print"` on the page area), which PdfCraft does not do yet. Synthetic regression tests. Estimates unchanged.
+- **2026-10-15 (M8, #2447):** Removing links or comments (Remove Hidden Information, Sanitize) also drops the tagged-PDF structure tree's object references to them, so the full rewrite no longer keeps a removed link's URL or a comment's text. Regression seen to fail without the fix. Estimates unchanged.
+- **2026-10-15 (M6, CSV tab-containing headers):** CSV export and spreadsheet merging quote tabs, and import distinguishes quoted tabs from TSV delimiters without changing field identities. Synthetic export/import and incremental/full save regressions cover header tabs, quotes and line breaks, plus ordinary TSV controls. Formula-cell policy is unchanged. Estimates unchanged.
+- **2026-10-15 (M6, AcroForm XML text):** Plain AcroForm XML imports retain all direct text around comments and processing instructions, including CDATA and character references. Synthetic regressions verify exact values and unknown catalog data after incremental and full save/reopen. Nested-element and XFA interpretation are unchanged. Estimates unchanged.
+- **2026-10-13 (M6, data exchange fidelity):** XML and XFDF exports preserve tabs and line breaks in field names and comment metadata using character references. Saved-PDF regressions cover distinct names that XML would normalize to the same attribute value, restored field values and replacement of the original comment identities. Estimates unchanged.
+- **2026-10-15 (M12, XFA, #2574):** A `pageArea` with no `` repeats without limit instead of taking a subform's default of one use, so a later page area with `occur max="1"` is no longer drawn (with a copy of its fields) over every body page. Synthetic regression test; the page area's `occur min` (a once-only page of its own) is still not laid out, recorded in `form.xfa-dynamic`. Estimates unchanged.
- **2026-10-13 (community, M3, #1569 review):** Pixel-only Wayland wheel, wheel-tilt and continuous input use standalone moved events, so Ctrl zoom does not remain latched after the modifier is released (#991); vendored winit patch (13), with a policy test. Estimates unchanged.
- **2026-10-15 (M14, #2383 candidate):** Device-limit requests now clamp unsupported maxima and minimum alignments before enabling the adapter's full 2D texture extent. This addresses the reported V3D request of 16 shader variables on an adapter supporting 15. Synthetic regressions fail before the clamp and pass afterward, along with the existing HiDPI case and Windows GPU-required application tests. Raspberry Pi startup remains unverified; milestone and overall estimates are unchanged.
- **2026-10-12 (M6):** XFDF field import and data-file extraction join all text nodes in a value, preserving text after XML comments and processing instructions. Synthetic regressions cover Unicode, CDATA, empty values and PDF save/reopen. Estimates unchanged.
@@ -186,9 +209,8 @@ Ranked; detail and the full milestone table in [docs/roadmap.md](docs/roadmap.md
- **2026-10-13 (G14, Vietnamese localization):** Added Vietnamese (`vi`, Tiếng Việt) as an interface and date language with a 2,591-entry catalog (`plural_none`) covering 100% of the current localizable key union (plus `theme`/`Appearance`). Locale tags `vi` / `vi-VN` / `vi_VN.UTF-8` resolve; calendar months use numeric `Tháng 1`…`12` / `tháng 1`…`12`. Native-speaker community review recorded. The twelve-language localization mean moves from ≈59% to ≈68%; weighted readiness ≈49% → ≈50%. Status stays `partial`: engine/OS details, document contents, filenames, CLI/MCP and external help stay outside the catalog by design.
- **2026-10-13 (M3, typed zoom):** The right rail's zoom percentage is an editable field: type 150, 150% or 62.5 % and press Enter, Escape discards, and nonsense keeps the current zoom. The control channel's zoom option shares its parser, which now refuses zero, negative and non-finite values.
-- **2026-10-13 (M6, data exchange fidelity):** XML and XFDF exports preserve tabs and line breaks in field names and comment metadata using character references. Saved-PDF regressions cover distinct names that XML would normalize to the same attribute value, restored field values and replacement of the original comment identities. Estimates unchanged.
+- **2026-10-13 (M4, indirect page rotation):** Rotation queries and edits resolve indirect integer values on pages and their page-tree ancestors. Turning a page already at 270° by 90° now saves it at 0°, including when its rotation is inherited. Synthetic full/incremental save-and-reopen tests cover nested inheritance, sibling preservation, negative/large angles and invalid-value fallback. Estimates unchanged.
- **2026-10-14 (M1, Flate fallback):** Raw deflate whose first bytes resemble a zlib header no longer loses a decoding-limit error during fallback. Strict and tolerant filter regressions and a COS stream regression cover refusal at the caller limit and successful decoding within it. No parity status or estimate changes.
-
- **2026-10-12 (M6, XML field names):** AcroForm XML data export constrains generated element names to XML 1.0 NCNames, so Latin-1 names such as `Area²` and `Aµ` produce readable data files. Original PDF field names travel through the existing `xfdf:original` attribute; previously valid tags, including Greek and CJK, keep their spelling. Saved-PDF regressions cover invalid prefixes, combining letters and normalized-name collisions. Estimates unchanged.
- **2026-10-12 (M6, CSV carriage returns):** Form-data CSV export quotes carriage returns in field names and values, so a trailing carriage return cannot turn into a record separator and disappear on reimport. Saved-PDF round-trip regressions cover CR, LF, CRLF, commas and quotes; estimates unchanged.
- **2026-10-13 (community, M14, #1553 review):** Relative Fontconfig includes resolve beside each including config, including nested configuration selected through `FONTCONFIG_PATH`. Directory includes accept only digit-prefixed `*.conf` in lexical order. Isolated regressions reproduce both review findings; configuration size/count/depth limits, cycles, symlinked config and valid reset ordering are covered. Estimates unchanged.
@@ -519,11 +541,15 @@ Newest first: the date, what moved, and the new overall estimate.
| Date | Change | Summary |
|---|---|---|
| 2026-10-15 | minor | M7: the Edit text caret is as tall as the row it is on in a region with several sizes; estimates unchanged. |
+| 2026-10-18 | minor | M4, PR #1919 review: repeated page indices apply each rotation to the latest page-local value, preserving indirect/inherited angles. Synthetic adjacent/interleaved duplicates retain siblings and page contents through incremental and full saves. Estimates unchanged. |
| 2026-10-18 | minor | M3, #3149: page replacement cancels render demand collected for the previous geometry; all viewer raster upload paths reject empty or oversized textures before cache eviction. Synthetic regressions reproduce the reported 1880×2350 upload at a 2048 limit, retain cached images and unsaved edits, and resume bounded backdrops and tiles. Estimates unchanged. |
| 2026-10-15 | minor | M6: Plain AcroForm XML imports retain all direct text around comments and processing instructions, including CDATA and character references. Estimates unchanged. |
| 2026-10-13 | minor | M6: XML/XFDF attribute whitespace survives form and comment export/import; estimates unchanged |
| 2026-10-14 | minor | M1: raw-deflate fallback preserves decoding-limit errors in strict and tolerant decoding; synthetic filter and COS stream regressions. Estimates unchanged. |
| 2026-10-13 | minor | M9: separate page validation overlays preserve signature artwork, and snapshot tabs reuse instead of multiplying; baked captions remain unchanged (#79 open); estimates unchanged |
+| 2026-10-16 | minor | M10 #1201 partial: lossless image export applies component Decode pairs and Indexed mapping. Estimates unchanged. |
+| 2026-10-15 | minor | M6: CSV export and spreadsheet merging quote tabs, and import distinguishes quoted tabs from TSV delimiters without changing field identities. Estimates unchanged. |
+| 2026-10-13 | minor | M4: rotation queries and edits resolve indirect page-tree values; inherited, negative and large rotations verified through both saves. Estimates unchanged. |
| 2026-10-12 | minor | M6: valid AcroForm XML element names with original PDF field names preserved; saved-PDF collision and Unicode regressions. Estimates unchanged. |
| 2026-10-12 | minor | M6: preserve carriage returns in exported CSV field names and values; saved-PDF round-trip coverage. Estimates unchanged. |
| 2026-10-13 | minor | M4: Insert-from-file preserves source bookmarks; integration tests retained with current main. Estimates unchanged. |
diff --git a/apps/pdfcraft-cli/Cargo.toml b/apps/pdfcraft-cli/Cargo.toml
index d7b52fbd66..bab8c28a88 100644
--- a/apps/pdfcraft-cli/Cargo.toml
+++ b/apps/pdfcraft-cli/Cargo.toml
@@ -18,6 +18,7 @@ serde_json = { workspace = true }
pdfcraft-engine = { workspace = true }
pdfcraft-audit = { workspace = true }
pdfcraft-cos = { workspace = true }
+pdfcraft-preflight = { workspace = true }
pdfcraft-automation = { workspace = true }
pdfcraft-platform = { workspace = true }
base64 = "0.22"
diff --git a/apps/pdfcraft-cli/src/main.rs b/apps/pdfcraft-cli/src/main.rs
index 119aaf4acb..5888e51114 100644
--- a/apps/pdfcraft-cli/src/main.rs
+++ b/apps/pdfcraft-cli/src/main.rs
@@ -57,6 +57,7 @@ macro_rules! mcp_usage {
const USAGE: &str = concat!(
"\
pdfcraft-cli info [--password PW] document summary as JSON
+ (includes any declared PDF/A or PDF/UA level)
pdfcraft-cli render --page N [--dpi 96] --out x.png (.png, .jpg, .tif or .pam)
pdfcraft-cli text [--page N] extracted text (pages separated by form feeds)
pdfcraft-cli edit --out out.pdf [--rotate 1,3:90] [--delete 2,4] [--move 5:1]
@@ -315,6 +316,10 @@ fn info(args: &[String]) -> Result<(), CliError> {
})
})
.flatten();
+ // What the document declares as a standard, read the way the Standards panel reads it, so
+ // `info` and the app agree. A file that declares nothing leaves these null and empty.
+ let declared =
+ pdfcraft_cos::Document::open_with_password(bytes.clone(), password).map(|doc| pdfcraft_preflight::declared(&doc)).unwrap_or_default();
let json = serde_json::json!({
"file": path,
"pdf_version": info.pdf_version,
@@ -324,11 +329,15 @@ fn info(args: &[String]) -> Result<(), CliError> {
"first_page_pt": info.pages.first().map(|p| [p.width, p.height]),
"title": info.title, "author": info.author, "producer": info.producer, "creator": info.creator,
"encrypted": info.encrypted, "tagged": info.tagged, "javascript": info.has_javascript,
+ "pdfa": declared.pdfa.as_ref().map(|(part, conformance)| format!("PDF/A-{part}{}", conformance.to_lowercase())),
+ "pdfua": declared.pdfua,
+ "output_intents": declared.output_intents,
"xfa": info.xfa.map(|x| match x { pdfcraft_render::Xfa::Static => "static", pdfcraft_render::Xfa::Dynamic => "dynamic" }),
"xfa_layout": xfa_layout,
"bookmarks": info.outline.len(), "annotations": info.annotations.len(), "fields": info.fields.len(),
"links": info.links.len(), "layers": info.layers.len(), "attachments": info.attachments.len(),
"transitions": info.transitions.len(),
+ "custom_properties": info.custom.len(),
"page_labels": info.pages.iter().take(8).map(|p| p.label.clone()).collect::>(),
"warnings": info.warnings,
});
@@ -613,7 +622,8 @@ fn render(args: &[String]) -> Result<(), CliError> {
return Err(e.into());
}
let bytes = match format {
- Some(f) => pdfcraft_engine::export::encode_image(p.width, p.height, &p.rgba, f)?,
+ // With `reject_oversize` the page is drawn at `--dpi` or not at all, so the file can say so.
+ Some(f) => pdfcraft_engine::export::encode_image(p.width, p.height, &p.rgba, f, f64::from(dpi))?,
None => {
// PAM (netpbm RGB_ALPHA): the raw premultiplied pixels, for tools that read it.
let mut pam = format!("P7\nWIDTH {}\nHEIGHT {}\nDEPTH 4\nMAXVAL 255\nTUPLTYPE RGB_ALPHA\nENDHDR\n", p.width, p.height).into_bytes();
diff --git a/apps/pdfcraft-cli/tests/info_standards.rs b/apps/pdfcraft-cli/tests/info_standards.rs
new file mode 100644
index 0000000000..68be86228c
--- /dev/null
+++ b/apps/pdfcraft-cli/tests/info_standards.rs
@@ -0,0 +1,87 @@
+//! `info` reports the standard a document declares, as the Standards panel does (#3319). A file
+//! carrying `pdfaid:part` and `pdfaid:conformance` in its XMP metadata was summarised with no
+//! mention of PDF/A at all, though the app reads the same declaration and shows it.
+
+use std::process::Command;
+
+const BIN: &str = env!("CARGO_BIN_EXE_pdfcraft-cli");
+
+/// A one-page PDF, with `metadata` as the catalog's XMP stream when given.
+fn fixture(metadata: Option<&str>) -> Vec {
+ let content = b"0 0 0 rg 10 10 40 20 re f";
+ let catalog = match metadata {
+ Some(_) => "<< /Type /Catalog /Pages 2 0 R /Metadata 5 0 R >>".to_string(),
+ None => "<< /Type /Catalog /Pages 2 0 R >>".to_string(),
+ };
+ let mut objs: Vec> = vec![
+ catalog.into_bytes(),
+ b"<< /Type /Pages /Kids [3 0 R] /Count 1 >>".to_vec(),
+ b"<< /Type /Page /Parent 2 0 R /MediaBox [0 0 120 60] /Contents 4 0 R >>".to_vec(),
+ format!("<< /Length {} >> stream\n", content.len()).into_bytes(),
+ ];
+ objs[3].extend_from_slice(content);
+ objs[3].extend_from_slice(b"\nendstream");
+ if let Some(xmp) = metadata {
+ let mut stream = format!("<< /Type /Metadata /Subtype /XML /Length {} >> stream\n", xmp.len()).into_bytes();
+ stream.extend_from_slice(xmp.as_bytes());
+ stream.extend_from_slice(b"\nendstream");
+ objs.push(stream);
+ }
+ let mut out = b"%PDF-1.7\n".to_vec();
+ let mut offsets = Vec::new();
+ for (i, o) in objs.iter().enumerate() {
+ offsets.push(out.len());
+ out.extend_from_slice(format!("{} 0 obj\n", i + 1).as_bytes());
+ out.extend_from_slice(o);
+ out.extend_from_slice(b"\nendobj\n");
+ }
+ let xref = out.len();
+ out.extend_from_slice(format!("xref\n0 {}\n0000000000 65535 f \n", objs.len() + 1).as_bytes());
+ for off in &offsets {
+ out.extend_from_slice(format!("{off:010} 00000 n \n").as_bytes());
+ }
+ out.extend_from_slice(format!("trailer\n<< /Size {} /Root 1 0 R >>\nstartxref\n{xref}\n%%EOF\n", objs.len() + 1).as_bytes());
+ out
+}
+
+fn xmp(part: &str, conformance: &str) -> String {
+ format!(
+ "\n\
+\n \n\
+ \n\
+ {part} \n {conformance} \n\
+ \n \n \n"
+ )
+}
+
+fn info_of(bytes: &[u8], name: &str) -> serde_json::Value {
+ let dir = std::env::temp_dir().join(format!("pdfcraft-info-standards-{}-{name}", std::process::id()));
+ std::fs::create_dir_all(&dir).unwrap();
+ let path = dir.join("f.pdf");
+ std::fs::write(&path, bytes).unwrap();
+ let out = Command::new(BIN).args(["info", path.to_str().unwrap()]).output().unwrap();
+ assert!(out.status.success(), "{}", String::from_utf8_lossy(&out.stderr));
+ let json = serde_json::from_slice(&out.stdout).expect("info prints JSON");
+ let _ = std::fs::remove_dir_all(&dir);
+ json
+}
+
+#[test]
+fn info_names_the_declared_pdfa_level() {
+ let declared = info_of(&fixture(Some(&xmp("2", "B"))), "pdfa2b");
+ assert_eq!(declared["pdfa"], serde_json::json!("PDF/A-2b"), "{declared}");
+ assert_eq!(declared["pdfua"], serde_json::Value::Null);
+ assert_eq!(declared["output_intents"], serde_json::json!([]));
+ // Lower-case conformance in the file still reads as the level's own spelling.
+ assert_eq!(info_of(&fixture(Some(&xmp("3", "a"))), "pdfa3a")["pdfa"], serde_json::json!("PDF/A-3a"));
+}
+
+#[test]
+fn info_declares_nothing_for_a_plain_pdf() {
+ let plain = info_of(&fixture(None), "plain");
+ assert_eq!(plain["pdfa"], serde_json::Value::Null, "{plain}");
+ assert_eq!(plain["pdfua"], serde_json::Value::Null);
+ // The rest of the summary is unaffected.
+ assert_eq!(plain["pages"], serde_json::json!(1));
+ assert_eq!(plain["pdf_version"], serde_json::json!("1.7"));
+}
diff --git a/apps/pdfcraft-cli/tests/render.rs b/apps/pdfcraft-cli/tests/render.rs
index b0ab245601..695431b65b 100644
--- a/apps/pdfcraft-cli/tests/render.rs
+++ b/apps/pdfcraft-cli/tests/render.rs
@@ -56,6 +56,17 @@ fn jpeg_and_tiff_names_get_those_formats() {
assert!(bytes.starts_with(b"II*\0") || bytes.starts_with(b"MM\0*"), "not a TIFF: {:?}", &bytes[..4]);
}
+#[test]
+fn the_file_records_the_dpi_it_was_rendered_at() {
+ // The engine's tests check each format; this checks that `--dpi` reaches the file.
+ let pdf = fixture("dpi");
+ let jpg = tmp("dpi", "p1.jpg");
+ assert!(render(&pdf, &jpg).status.success());
+ let bytes = std::fs::read(&jpg).unwrap();
+ assert_eq!(&bytes[6..11], b"JFIF\0");
+ assert_eq!(bytes[13..18], [1, 0, 36, 0, 36], "JFIF density in dots per inch");
+}
+
#[test]
fn a_pam_name_still_gets_the_netpbm_stream() {
let pdf = fixture("pam");
diff --git a/apps/pdfcraft-web/src/main.rs b/apps/pdfcraft-web/src/main.rs
index a46a4674b9..73d14f4a6c 100644
--- a/apps/pdfcraft-web/src/main.rs
+++ b/apps/pdfcraft-web/src/main.rs
@@ -70,6 +70,10 @@ fn main() {
app.host_save = Some(bridge.save_fn());
app.host_dirty = Some(bridge.dirty_fn());
bridge.post_ready();
+ } else if query_param("host").as_deref() == Some("parent") {
+ eframe::web_sys::console::warn_1(
+ &"PdfCraft: ?host=parent needs a framing page on the same origin; saves download instead".into(),
+ );
}
Ok(Box::new(app))
}),
@@ -130,7 +134,8 @@ async fn fetch_bytes(url: &str) -> Result, String> {
/// - `pdfcraft:dirty` (to the parent): `{ dirty }` whenever unsaved work appears or goes away.
/// - `pdfcraft:failed` (to the parent): `{ error }` when the app couldn't start.
///
-/// Only the parent window on the app's own origin is heard or answered.
+/// Only the parent window on the app's own origin is heard or answered. With a parent on
+/// another origin there is no bridge: the app runs as if unframed and saves download.
#[cfg(target_arch = "wasm32")]
mod host {
use eframe::wasm_bindgen::{JsCast, JsValue, closure::Closure};
@@ -143,7 +148,9 @@ mod host {
}
impl Bridge {
- /// `None` when the page isn't framed (there is no parent to talk to).
+ /// `None` when the page isn't framed (there is no parent to talk to), or when the parent
+ /// is on another origin: messages only go to this origin, so the browser would drop
+ /// every save without an error while the document showed as saved (#2167).
pub fn new() -> Option {
let window = web_sys::window()?;
let parent = window.parent().ok().flatten()?;
@@ -151,6 +158,10 @@ mod host {
return None;
}
let origin = window.location().origin().ok()?;
+ // Reading another origin's location throws, which is `None` too.
+ if parent.location().origin().ok()? != origin {
+ return None;
+ }
Some(Self { window, parent, origin })
}
diff --git a/apps/pdfcraft/src/main.rs b/apps/pdfcraft/src/main.rs
index b274dc53a8..721b7da1b3 100644
--- a/apps/pdfcraft/src/main.rs
+++ b/apps/pdfcraft/src/main.rs
@@ -380,6 +380,7 @@ fn app_creator<'a>(
if let Some(json) = cc.storage.and_then(|s| s.get_string("pdfcraft").or_else(|| s.get_string(LEGACY_STORAGE_KEY))) {
app.restore(&json);
}
+ app.window_state = pdfcraft_ui_egui::window_state::WindowState::started_with(cc.storage);
app.integrated_titlebar = integrated;
app.window_controls = window_controls;
app.system_text_scale = text_scale::system();
diff --git a/contributors/people.toml b/contributors/people.toml
index da3cfc6a77..cd8ef99f72 100644
--- a/contributors/people.toml
+++ b/contributors/people.toml
@@ -145,3 +145,6 @@ sync_github_name = false
[people.Silvertree2010]
display_name = "Netter"
sync_github_name = false
+
+[people.MortenHusted]
+sync_github_name = true
diff --git a/crates/annot/src/lib.rs b/crates/annot/src/lib.rs
index f4973aa021..b7f9d62697 100644
--- a/crates/annot/src/lib.rs
+++ b/crates/annot/src/lib.rs
@@ -2282,6 +2282,11 @@ fn line_endings_of(d: &Dict) -> Option> {
/// Replace Text (Acrobat's proposal): strike out `quads` and add a caret at the end of the
/// struck text holding `replacement`, grouped with the strikeout (`/IRT` + `/RT /Group`) so they
/// move, list and delete as one. Returns the strikeout's index.
+///
+/// The two annotations need their own `/NM`: XFDF names a reply by the `/NM` of what it replies
+/// to, so one id for both writes `inreplyto` equal to the caret's own `name`, and a reader that
+/// resolves names (Acrobat, and our own importer) loses the pair. `caret_meta` is a second id
+/// from the caller, so the edit stays deterministic.
#[allow(clippy::too_many_arguments)]
pub fn add_text_replacement(
doc: &mut Document,
@@ -2292,6 +2297,7 @@ pub fn add_text_replacement(
strike: &Style,
caret: &Style,
meta: &Meta,
+ caret_meta: &Meta,
) -> Result {
let last = *quads.last().ok_or_else(|| AnnotError::Invalid("select the text to replace first".into()))?;
let s = add_annotation(
@@ -2317,7 +2323,7 @@ pub fn add_text_replacement(
let c = add_annotation(
doc,
&NewAnnotation { page, shape: Shape::Caret { rect }, style: caret.clone(), contents: replacement.into(), author: author.into() },
- meta,
+ caret_meta,
)?;
let (_, cr) = annot_ref(doc, page, c)?;
doc.update_dict(cr, |d| {
diff --git a/crates/annot/src/links.rs b/crates/annot/src/links.rs
index 2e45b4e4d4..4d3eef632b 100644
--- a/crates/annot/src/links.rs
+++ b/crates/annot/src/links.rs
@@ -178,6 +178,19 @@ fn write_style(d: &mut Dict, s: &LinkStyle) {
d.set(b"H".to_vec(), Object::name(s.highlight.code()));
}
+/// Bytes of `u` as a 7-bit ASCII URI string (§12.6.4.8).
+fn uri_bytes(u: &str) -> Vec {
+ let mut out = Vec::with_capacity(u.len());
+ for &b in u.as_bytes() {
+ if (0x21..=0x7e).contains(&b) {
+ out.push(b);
+ } else {
+ out.extend_from_slice(format!("%{b:02X}").as_bytes());
+ }
+ }
+ out
+}
+
fn write_action(doc: &Document, d: &mut Dict, a: &LinkAction) -> Result<(), AnnotError> {
match a {
LinkAction::Page(p) => {
@@ -194,7 +207,7 @@ fn write_action(doc: &Document, d: &mut Dict, a: &LinkAction) -> Result<(), Anno
let mut act = Dict::new();
act.set(b"S".to_vec(), Object::name("URI"));
// URIs are 7-bit ASCII strings (§12.6.4.8).
- act.set(b"URI".to_vec(), PdfString::literal(u.as_bytes().to_vec()));
+ act.set(b"URI".to_vec(), PdfString::literal(uri_bytes(u)));
d.set(b"A".to_vec(), Object::Dict(act));
}
LinkAction::Other(_) => {}
diff --git a/crates/annot/src/tests.rs b/crates/annot/src/tests.rs
index faed409abb..d3d448e520 100644
--- a/crates/annot/src/tests.rs
+++ b/crates/annot/src/tests.rs
@@ -588,6 +588,25 @@ fn links_are_added_edited_listed_and_removed() {
assert!(links::add(&mut doc, 0, [0.0, 0.0, 1.0, 1.0], &LinkAction::Page(0), &LinkStyle::default()).is_err(), "too small");
}
+#[test]
+fn link_uri_is_percent_encoded_to_7_bit_ascii() {
+ use crate::links::{self, LinkAction, LinkStyle};
+ let mut doc = fixture();
+ for url in ["https://example.com/ação", "https://example.com/a%20b"] {
+ links::add(&mut doc, 0, [10.0, 10.0, 110.0, 30.0], &LinkAction::Uri(url.into()), &LinkStyle::default()).unwrap();
+ }
+ let doc = reopen(&doc);
+ let urls: Vec = links::list(&doc)
+ .into_iter()
+ .filter_map(|link| match link.action {
+ LinkAction::Uri(url) => Some(url),
+ _ => None,
+ })
+ .collect();
+ assert_eq!(urls, ["https://example.com/a%C3%A7%C3%A3o", "https://example.com/a%20b"]);
+ assert!(urls.iter().all(|url| url.is_ascii()));
+}
+
#[test]
fn link_edit_rejects_non_finite_and_overflowed_rectangles_without_mutating_the_pdf() {
use crate::links::{self, LinkAction, LinkStyle};
@@ -784,7 +803,7 @@ fn replace_text_groups_a_strikeout_and_a_caret() {
let quads = [[100.0, 720.0, 200.0, 720.0, 100.0, 706.0, 200.0, 706.0]];
let red = Style { color: [0.89, 0.13, 0.13], opacity: 1.0, width: 1.0, fill: None };
let blue = Style { color: [0.0, 0.47, 0.84], opacity: 1.0, width: 1.0, fill: None };
- let i = add_text_replacement(&mut doc, 0, &quads, "new words", "Ada", &red, &blue, &meta("r")).unwrap();
+ let i = add_text_replacement(&mut doc, 0, &quads, "new words", "Ada", &red, &blue, &meta("r"), &meta("r-caret")).unwrap();
let doc = reopen(&doc);
let a = list(&doc, 0);
let strike = &a[i];
@@ -792,13 +811,21 @@ fn replace_text_groups_a_strikeout_and_a_caret() {
let caret = a.iter().find(|d| d.name(b"Subtype") == Some(b"Caret")).unwrap();
assert_eq!((text(caret, b"Contents").as_str(), caret.name(b"RT")), ("new words", Some(&b"Group"[..])));
assert!(caret.reference(b"IRT").is_some());
+ // #2666: the pair needs its own `/NM` each. XFDF names a reply by the `/NM` of what it
+ // replies to, so sharing one id writes `inreplyto` equal to the caret's own `name` and the
+ // pair does not survive an export and import.
+ assert_eq!(text(caret, b"NM").as_str(), "r-caret");
+ assert_eq!(text(strike, b"NM").as_str(), "r");
+ assert_ne!(text(caret, b"NM"), text(strike, b"NM"), "the caret must not carry the strikeout's name");
+ let irt = caret.reference(b"IRT").unwrap();
+ assert_eq!(text(doc.get(irt).as_dict().unwrap(), b"NM").as_str(), "r", "the caret replies to the strikeout");
let r = rect(caret);
assert!(r[0] < 200.0 && r[2] > 200.0 && r[1] < 706.0, "at the end of the line, below its baseline: {r:?}");
// Deleting the strikeout takes the caret with it (the group).
let mut doc = doc;
delete_annotation(&mut doc, 0, i).unwrap();
assert!(!list(&doc, 0).iter().any(|d| d.name(b"Subtype") == Some(b"Caret")));
- assert!(add_text_replacement(&mut doc, 0, &[], "x", "Ada", &red, &blue, &meta("r")).is_err());
+ assert!(add_text_replacement(&mut doc, 0, &[], "x", "Ada", &red, &blue, &meta("r"), &meta("r-caret")).is_err());
}
#[test]
diff --git a/crates/automation/README.md b/crates/automation/README.md
index 4e27702a32..669ea61463 100644
--- a/crates/automation/README.md
+++ b/crates/automation/README.md
@@ -64,6 +64,20 @@ semantics; they never silently act on partial text.
## Shared field properties
+For an individual push button, `form_set_props` also edits its visible normal label:
+
+```json
+{"doc": 1, "field": "clear_button", "caption": "Clear Form", "widget": 1}
+```
+
+`widget` is a 1-based index (default 1), used only with `caption`; an empty caption clears the
+label. Neither renames the field. `form_fields` returns the first widget's `caption` and all
+`captions` as `[{"widget": 1, "caption": "Clear Form"}]`. Other widgets and icon-only layouts
+keep their settings. CLI and MCP use the same edit, including undo/redo and save/reopen.
+The desktop exposes it as **Button Properties → Options → Label**. Rollover/down labels and
+icon layout are not edited here; the existing caption font and unsupported-character warning
+policy apply.
+
`form_set_props` accepts either `field` (one name) or `fields` (1–1000 unique names):
```json
@@ -75,7 +89,7 @@ The batch is atomic and creates one undo step. A missing or locked field, invali
unknown argument or malformed appearance refuses the whole edit. Set `locked: false` to
unlock and change fields together. Only supplied properties change: each field/widget keeps
its other colours, fonts, values and unknown PDF dictionary keys. `"none"` removes a border
-or fill colour. Names, rectangles and rotations require the single-field form. Results keep
+or fill colour. Names, rectangles, rotations and captions require the single-field form. Results keep
`field` for single edits and return `fields` for batches. CLI and MCP use the same tool.
## MCP server
diff --git a/crates/automation/src/forms.rs b/crates/automation/src/forms.rs
index 7c60425a8c..f2eab5ddc0 100644
--- a/crates/automation/src/forms.rs
+++ b/crates/automation/src/forms.rs
@@ -233,6 +233,19 @@ impl Automation {
o.insert("unsupported_scripts".into(), json!(f.actions.unsupported));
}
match f.kind {
+ FormFieldKind::PushButton => {
+ o.insert("caption".into(), json!(w.and_then(|w| w.caption.as_deref()).unwrap_or_default()));
+ o.insert(
+ "captions".into(),
+ json!(
+ f.widgets
+ .iter()
+ .enumerate()
+ .map(|(i, w)| { json!({"widget": i.saturating_add(1), "caption": w.caption.as_deref().unwrap_or_default()}) })
+ .collect::>()
+ ),
+ );
+ }
FormFieldKind::Radio => {
o.insert("options".into(), json!((0..f.widgets.len()).filter_map(|i| f.export_of(i)).collect::>()));
}
@@ -327,6 +340,9 @@ impl Automation {
}
pub(crate) fn form_set_props(&mut self, a: &Args) -> Result {
+ if a.get("widget").is_some() && a.get("caption").is_none() {
+ return Err(bad("widget requires caption"));
+ }
let single = a.opt_str("field")?;
let bulk = a.get("fields").is_some();
if single.is_some() == bulk {
@@ -344,7 +360,7 @@ impl Automation {
return Err(bad(format!("field {name:?} is listed more than once")));
}
}
- for key in ["name", "rect", "rotation"] {
+ for key in ["name", "rect", "rotation", "caption", "widget"] {
if a.get(key).is_some() {
return Err(bad(format!("{key} applies to one field; use field instead of fields")));
}
@@ -398,6 +414,15 @@ impl Automation {
rect,
name: a.opt_str("name")?.map(str::to_owned),
tooltip: a.opt_str("tooltip")?.map(str::to_owned),
+ caption: a
+ .opt_str("caption")?
+ .map(|caption| -> Result<_> {
+ let wi = a.opt_int("widget")?.unwrap_or(1);
+ let wi =
+ usize::try_from(wi).ok().and_then(|wi| wi.checked_sub(1)).ok_or_else(|| bad("widget must be a positive 1-based index"))?;
+ Ok((wi, caption.to_owned()))
+ })
+ .transpose()?,
read_only: a.opt_bool("read_only")?,
required: a.opt_bool("required")?,
multiline: a.opt_bool("multiline")?,
diff --git a/crates/automation/src/lib.rs b/crates/automation/src/lib.rs
index 8a4dcec16c..0e2e0dd698 100644
--- a/crates/automation/src/lib.rs
+++ b/crates/automation/src/lib.rs
@@ -380,7 +380,8 @@ impl Automation {
}
"page_duplicate" => {
let pages = self.pages(&a, "pages")?;
- self.apply(&a, Edit::DuplicatePages { pages })?
+ let at = self.opt_position(&a, "at")?;
+ self.apply(&a, Edit::DuplicatePages { pages, at })?
}
"page_set_box" => self.page_set_box(&a)?,
"page_transition_set" => self.page_transition_set(&a)?,
@@ -1753,16 +1754,21 @@ impl Automation {
let mut used = std::collections::HashSet::new();
for (i, (first, last, bytes)) in parts.iter().enumerate() {
let safe = |t: &str| t.chars().map(|c| if c.is_alphanumeric() || " -_.,()".contains(c) { c } else { '_' }).collect::();
- let file = match titles.iter().find(|t| t.0 + 1 == *first) {
- Some((_, t)) => format!("{stem}-{}.pdf", safe(t)),
- None => format!("{stem}-part{}.pdf", i + 1),
+ let mut base = match titles.iter().find(|t| t.0 + 1 == *first) {
+ Some((_, t)) => format!("{stem}-{}", safe(t)),
+ None => format!("{stem}-part{}", i + 1),
};
+ // A file name is at most 255 bytes: keep the base to 200, cut between characters, which leaves room for ".pdf" and a -N suffix.
+ if let Some((end, _)) = base.char_indices().find(|(at, c)| at + c.len_utf8() > 200) {
+ base.truncate(end);
+ }
+ let file = format!("{base}.pdf");
// Equal titles (or titles that sanitize alike) get -2, -3, ... so no part overwrites another; compared case-insensitively.
- let mut unique = file.clone();
+ let mut unique = file;
let mut n = 1;
while !used.insert(unique.to_lowercase()) {
n += 1;
- unique = format!("{}-{n}.pdf", file.trim_end_matches(".pdf"));
+ unique = format!("{base}-{n}.pdf");
}
let path = child(&dir, &unique);
write_atomic(&path, bytes)?;
@@ -1788,12 +1794,17 @@ impl Automation {
/// A 1-based insertion position (1 = before the first page, n+1 = after the last) → 0-based.
fn position(&self, a: &Args, key: &str) -> Result {
+ self.opt_position(a, key)?.ok_or_else(|| Args::missing(key))
+ }
+
+ /// `position`, but `None` when the key is absent.
+ fn opt_position(&self, a: &Args, key: &str) -> Result> {
+ let Some(at) = a.opt_int(key)? else { return Ok(None) };
let n = self.doc(a)?.info.pages.len();
- let at = a.int(key)?;
if at < 1 || at as usize > n + 1 {
return Err(ToolError::InvalidArgs(format!("{key} must be between 1 and {} (the document has {n} pages)", n + 1)));
}
- Ok(at as usize - 1)
+ Ok(Some(at as usize - 1))
}
fn renderer(&mut self, id: DocId) -> Result<&mut PageRenderer> {
@@ -2401,6 +2412,9 @@ fn info(d: &Document) -> Value {
"pdf_version": i.pdf_version,
"file_size": i.file_size,
"title": i.title, "author": i.author, "subject": i.subject, "keywords": i.keywords,
+ // Custom document properties (Document Properties ▸ Custom): /Info entries beyond the
+ // ones the spec names, set and cleared with doc_set_info.
+ "custom": i.custom.iter().map(|(name, value)| json!({ "name": name, "value": value })).collect::>(),
"creator": i.creator, "producer": i.producer,
"tagged": i.tagged,
"has_javascript": i.has_javascript,
diff --git a/crates/automation/src/tools.rs b/crates/automation/src/tools.rs
index 6b7b8a306a..91cd489c0c 100644
--- a/crates/automation/src/tools.rs
+++ b/crates/automation/src/tools.rs
@@ -156,7 +156,7 @@ pub fn tools() -> Vec {
}),
&["doc"],
)),
- t("doc_set_info", "Set document metadata", "Set a document information entry such as Title, Author, Subject or Keywords; an empty value removes it. When the file has XMP metadata, the matching XMP property (dc:title, dc:creator, dc:description, pdf:Keywords, xmp:CreatorTool, pdf:Producer) is changed too, so viewers that read XMP first show the new value. Undoable.")
+ t("doc_set_info", "Set document metadata", "Set a document information entry such as Title, Author, Subject or Keywords, or a custom property under any other name (no spaces, delimiters, / or #; at most 127 bytes; doc_info lists them under custom); an empty value removes it. When the file has XMP metadata, the matching XMP property (dc:title, dc:creator, dc:description, pdf:Keywords, xmp:CreatorTool, pdf:Producer) is changed too, so viewers that read XMP first show the new value. Undoable.")
.with(schema(json!({ "doc": doc(), "key": { "type": "string" }, "value": { "type": "string" } }), &["doc", "key", "value"])),
t("page_render", "Render a page", "Render one page to a PNG image (default 96 dpi, at most 600).")
.ro()
@@ -321,7 +321,7 @@ pub fn tools() -> Vec {
}),
&["doc", "from", "to"],
)),
- t("form_fields", "List form fields", "Every interactive form field: name, type (text, checkbox, radio, combo, list, button, signature), value, options, page, rect (top-left-origin points) and rotation (0, 90, 180 or 270 degrees counterclockwise), read-only and required flags.")
+ t("form_fields", "List form fields", "Every interactive form field: name, type (text, checkbox, radio, combo, list, button, signature), value, options, page, rect (top-left-origin points) and rotation (0, 90, 180 or 270 degrees counterclockwise), read-only and required flags. Push buttons include caption (first widget's normal label) and captions [{widget, caption}] for all widgets (1-based).")
.ro()
.with(schema(json!({ "doc": doc() }), &["doc"])),
t(
@@ -374,7 +374,7 @@ pub fn tools() -> Vec {
t(
"form_set_props",
"Field properties",
- "Change field properties: give field for one field, or fields for a list of unique field names (1–1000). A bulk change is atomic and one Undo step; an invalid or locked field leaves every field unchanged. Only the given properties change, and partial appearance changes preserve each field's other colours and font. name (rename), rect (position) and rotation require a single field. General and Options: tooltip, read_only, required, multiline, max_length (0 = no limit), options (combo/list items), font_size (0 = auto), format, validate, calculate (Acrobat's native AF rules). Options: align (left, center, right), default (Reset form's value; check boxes/radio buttons use their on state, \"\" for off), flags { scroll, rich_text, password, file_select, spell_check, comb (needs max_length), sort, editable, multi_select, commit_immediately }, check_style (check boxes/radio buttons: check, circle, cross, diamond, square, star). rotation is 0, 90, 180 or 270 degrees counterclockwise and swaps width/height when the axis changes. Undoable.",
+ "Change field properties: give field for one field, or fields for a list of unique field names (1–1000). A bulk change is atomic and one Undo step; an invalid or locked field leaves every field unchanged. Only the given properties change, and partial appearance changes preserve each field's other colours and font. name (rename), rect (position), rotation and caption require a single field. caption changes one push-button normal label (widget: 1-based index, default 1); an empty string clears it. Other widgets and icon-only layouts are preserved. General and Options: tooltip, read_only, required, multiline, max_length (0 = no limit), options (combo/list items), font_size (0 = auto), format, validate, calculate (Acrobat's native AF rules). Options: align (left, center, right), default (Reset form's value; check boxes/radio buttons use their on state, \"\" for off), flags { scroll, rich_text, password, file_select, spell_check, comb (needs max_length), sort, editable, multi_select, commit_immediately }, check_style (check boxes/radio buttons: check, circle, cross, diamond, square, star). rotation is 0, 90, 180 or 270 degrees counterclockwise and swaps width/height when the axis changes. Undoable.",
)
.with(schema(
json!({
@@ -383,6 +383,8 @@ pub fn tools() -> Vec {
"fields": { "type": "array", "items": { "type": "string" }, "minItems": 1, "maxItems": 1000, "uniqueItems": true, "description": "Apply the same changes to these fields in one atomic edit. Use either field or fields." },
"name": { "type": "string" },
"tooltip": { "type": "string" },
+ "caption": { "type": "string", "description": "Normal push-button label; an empty string clears it. Requires a single field." },
+ "widget": { "type": "integer", "minimum": 1, "description": "Widget whose caption changes (1-based, default 1). Requires caption. See form_fields.captions." },
"read_only": { "type": "boolean" },
"required": { "type": "boolean" },
"multiline": { "type": "boolean" },
@@ -825,9 +827,9 @@ pub fn tools() -> Vec {
t("page_replace", "Replace pages", "Replace the content of pages with pages of another PDF (same count). Links, comments, form fields and bookmarks on the original pages stay, as in Acrobat. Undoable.")
.cmd("page.replace")
.with(schema(json!({ "doc": doc(), "pages": pages("to replace"), "path": path_arg(), "from_pages": pages("of the other file, in order (default: its first pages)") }), &["doc", "pages", "path"])),
- t("page_duplicate", "Duplicate pages", "Insert copies of pages after the last of them (fonts and images are shared, not copied). Undoable.")
+ t("page_duplicate", "Duplicate pages", "Insert copies of pages so the first copy becomes page `at`, or after the last of them when `at` is left out (fonts and images are shared, not copied). Undoable.")
.cmd("page.duplicate")
- .with(schema(json!({ "doc": doc(), "pages": pages("to duplicate") }), &["doc", "pages"])),
+ .with(schema(json!({ "doc": doc(), "pages": pages("to duplicate"), "at": { "type": "integer", "minimum": 1 } }), &["doc", "pages"])),
t(
"page_transition_set",
"Set page transitions",
diff --git a/crates/automation/tests/automation.rs b/crates/automation/tests/automation.rs
index 30658fdfd7..f3bcfd21a4 100644
--- a/crates/automation/tests/automation.rs
+++ b/crates/automation/tests/automation.rs
@@ -214,6 +214,98 @@ fn page_transitions_are_set_and_removed_through_the_tools() {
);
}
+/// #1937 through the tools: a damaged file is repaired on open (`doc_info` lists the repair) and
+/// `doc_save` over it writes the rebuilt file, so opening it again needs no repair.
+#[test]
+fn saving_a_repaired_file_puts_the_repair_on_disk() {
+ let dir = workdir("repaired-save");
+ let good = fixture(2);
+ let mut broken = b"%PDF-1.7\n%shifted\n".to_vec();
+ broken.extend_from_slice(&good[b"%PDF-1.7\n".len()..]);
+ std::fs::write(dir.join("broken.pdf"), &broken).unwrap();
+ let mut a = auto(&dir);
+ let doc = ok(&mut a, "doc_open", json!({ "path": "broken.pdf" }))["doc"].as_u64().unwrap();
+ let info = ok(&mut a, "doc_info", json!({ "doc": doc }));
+ assert!(info["repairs"].as_array().unwrap().iter().any(|r| r.as_str().unwrap_or("").contains("rebuilt")), "{}", info["repairs"]);
+ let saved = ok(&mut a, "doc_save", json!({ "doc": doc }));
+ assert_eq!(saved["incremental"], false, "{saved}");
+ let bytes = std::fs::read(dir.join("broken.pdf")).unwrap();
+ assert_ne!(bytes, broken, "the damaged bytes were written back");
+ let reopened = pdfcraft_cos::Document::open(std::sync::Arc::new(bytes)).unwrap();
+ assert!(reopened.repair_log().is_empty(), "{:?}", reopened.repair_log());
+ let info = ok(&mut a, "doc_info", json!({ "doc": doc }));
+ assert_eq!(info["repairs"].as_array().unwrap().len(), 0, "{}", info["repairs"]);
+ assert_eq!(info["pages"].as_array().unwrap().len(), 2);
+}
+
+/// Custom document properties through the tools: set and cleared with doc_set_info under any
+/// name the spec does not use, listed by doc_info apart from the standard entries, bad names
+/// refused without a change.
+#[test]
+fn custom_document_properties_are_set_listed_and_cleared_through_the_tools() {
+ let dir = workdir("custom-info");
+ std::fs::write(dir.join("a.pdf"), fixture(1)).unwrap();
+ let mut a = auto(&dir);
+ let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap();
+ ok(&mut a, "doc_set_info", json!({ "doc": doc, "key": "Title", "value": "Report" }));
+ ok(&mut a, "doc_set_info", json!({ "doc": doc, "key": "Project", "value": "Apollo" }));
+ ok(&mut a, "doc_set_info", json!({ "doc": doc, "key": "Revision", "value": "7" }));
+ let info = ok(&mut a, "doc_info", json!({ "doc": doc }));
+ assert_eq!(info["title"], "Report");
+ let mut custom: Vec<(String, String)> =
+ info["custom"].as_array().unwrap().iter().map(|c| (c["name"].as_str().unwrap().into(), c["value"].as_str().unwrap().into())).collect();
+ custom.sort();
+ assert_eq!(custom, [("Project".to_string(), "Apollo".to_string()), ("Revision".to_string(), "7".to_string())], "{}", info["custom"]);
+ assert!(a.call("doc_set_info", &json!({ "doc": doc, "key": "Bad Name", "value": "x" })).is_err(), "a name with a space is refused");
+ assert!(a.call("doc_set_info", &json!({ "doc": doc, "key": "", "value": "x" })).is_err());
+ ok(&mut a, "doc_set_info", json!({ "doc": doc, "key": "Project", "value": "" }));
+ let info = ok(&mut a, "doc_info", json!({ "doc": doc }));
+ assert_eq!(info["custom"].as_array().unwrap().len(), 1, "{}", info["custom"]);
+ assert_eq!(info["custom"][0]["name"], "Revision");
+ ok(&mut a, "doc_save", json!({ "doc": doc, "path": "out/custom.pdf" }));
+ let mut b = auto(&dir);
+ let re = ok(&mut b, "doc_open", json!({ "path": "out/custom.pdf" }))["doc"].as_u64().unwrap();
+ let info = ok(&mut b, "doc_info", json!({ "doc": re }));
+ assert_eq!(info["custom"][0]["value"], "7", "{}", info["custom"]);
+}
+
+/// The custom list `doc_info` shows when a file opens is the one it shows after an unrelated
+/// edit: a name value, a blank or NUL-padded string and a number are judged the same way by the
+/// open-time reader and by the engine's refresh.
+#[test]
+fn custom_document_properties_do_not_change_after_an_unrelated_edit() {
+ let dir = workdir("custom-info-stable");
+ let objs = [
+ "<< /Type /Catalog /Pages 2 0 R >>",
+ "<< /Type /Pages /Kids [3 0 R] /Count 1 >>",
+ "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 200 200] >>",
+ "<< /Producer (Test) /Project (Apollo) /Dept /RnD /Blank ( ) /Padded (Lab\\000\\000) /Revision 7 /Empty () /Mixed ( \\000Lab\\000 ) /Pad ( \\000 ) >>",
+ ];
+ let mut pdf = b"%PDF-1.7\n".to_vec();
+ let mut offsets = Vec::new();
+ for (i, o) in objs.iter().enumerate() {
+ offsets.push(pdf.len());
+ pdf.extend_from_slice(format!("{} 0 obj\n{o}\nendobj\n", i + 1).as_bytes());
+ }
+ let xref = pdf.len();
+ pdf.extend_from_slice(format!("xref\n0 {}\n0000000000 65535 f \n", objs.len() + 1).as_bytes());
+ for o in offsets {
+ pdf.extend_from_slice(format!("{o:010} 00000 n \n").as_bytes());
+ }
+ pdf.extend_from_slice(format!("trailer\n<< /Size {} /Root 1 0 R /Info 4 0 R >>\nstartxref\n{xref}\n%%EOF\n", objs.len() + 1).as_bytes());
+ std::fs::write(dir.join("custom.pdf"), pdf).unwrap();
+ let mut a = auto(&dir);
+ let doc = ok(&mut a, "doc_open", json!({ "path": "custom.pdf" }))["doc"].as_u64().unwrap();
+ let opened = ok(&mut a, "doc_info", json!({ "doc": doc }))["custom"].clone();
+ let names: Vec<&str> = opened.as_array().unwrap().iter().map(|c| c["name"].as_str().unwrap()).collect();
+ assert_eq!(names, ["Project", "Padded", "Mixed"], "{opened}");
+ assert_eq!(opened[1]["value"], "Lab", "{opened}");
+ assert_eq!(opened[2]["value"], "Lab", "{opened}");
+ ok(&mut a, "doc_set_info", json!({ "doc": doc, "key": "Title", "value": "Report" }));
+ let edited = ok(&mut a, "doc_info", json!({ "doc": doc }))["custom"].clone();
+ assert_eq!(edited, opened);
+}
+
#[test]
fn autotag_preview_proposes_text_without_changing_the_pdf() {
let dir = workdir("autotag-preview");
@@ -2097,6 +2189,19 @@ fn reversing_pages_through_tools() {
assert!(matches!(a.call("page_reverse", &json!({ "doc": doc, "pages": [2] })), Err(ToolError::Failed(_))), "one page has nothing to reverse");
}
+#[test]
+fn page_duplicate_at_inserts_the_copies_there() {
+ let dir = workdir("duplicate-at");
+ let mut a = auto(&dir);
+ let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap();
+ let r = ok(&mut a, "page_duplicate", json!({ "doc": doc, "pages": [2], "at": 1 }));
+ assert_eq!(r["pages"], 4);
+ assert_eq!(page_text(&mut a, doc), ["Page 2", "Page 1", "Page 2", "Page 3"]);
+ // `at` is 1-based and within 1..=pages+1, as for the other page inserts.
+ assert!(matches!(a.call("page_duplicate", &json!({ "doc": doc, "pages": [1], "at": 0 })), Err(ToolError::InvalidArgs(_))), "1-based");
+ assert!(matches!(a.call("page_duplicate", &json!({ "doc": doc, "pages": [1], "at": 6 })), Err(ToolError::InvalidArgs(_))), "past the end");
+}
+
#[test]
fn duplicating_and_cropping_through_tools() {
let dir = workdir("boxes");
@@ -3839,6 +3944,35 @@ fn bookmark_split_with_equal_titles_keeps_every_part() {
}
}
+#[test]
+fn bookmark_split_with_long_titles_shortens_the_file_names() {
+ let dir = workdir("organize_long_titles");
+ let mut a = auto(&dir);
+ let doc = ok(&mut a, "doc_open", json!({ "path": "a.pdf" }))["doc"].as_u64().unwrap();
+ // Two equal titles far over the 255-byte name limit, then a multibyte one (3 bytes a character).
+ let long = format!("Chapter {}", "x".repeat(300));
+ ok(&mut a, "bookmark_add", json!({ "doc": doc, "title": long, "page": 1 }));
+ ok(&mut a, "bookmark_add", json!({ "doc": doc, "title": long, "page": 2 }));
+ ok(&mut a, "bookmark_add", json!({ "doc": doc, "title": "章".repeat(100), "page": 3 }));
+ let s = ok(&mut a, "doc_split", json!({ "doc": doc, "bookmarks": true, "out_dir": "parts" }));
+ let files: Vec = s["files"]
+ .as_array()
+ .unwrap()
+ .iter()
+ .map(|f| std::path::Path::new(f["path"].as_str().unwrap()).file_name().unwrap().to_string_lossy().into_owned())
+ .collect();
+ assert_eq!(files.len(), 3, "{files:?}");
+ let distinct: std::collections::HashSet<_> = files.iter().map(|f| f.to_lowercase()).collect();
+ assert_eq!(distinct.len(), 3, "{files:?}");
+ for f in &files {
+ assert!(f.len() <= 255, "{f} is {} bytes", f.len());
+ assert!(f.ends_with(".pdf"), "{f}");
+ assert!(dir.join("parts").join(f).exists(), "{f} missing");
+ }
+ assert!(files[0].starts_with("a-Chapter x"), "{files:?}");
+ assert!(files[2].starts_with("a-章"), "{files:?}");
+}
+
#[test]
fn organizing_with_filters_bookmark_splits_and_extract_options() {
let dir = workdir("organize2");
diff --git a/crates/automation/tests/button_caption.rs b/crates/automation/tests/button_caption.rs
new file mode 100644
index 0000000000..cb1a22a026
--- /dev/null
+++ b/crates/automation/tests/button_caption.rs
@@ -0,0 +1,91 @@
+//! Editing a button's visible label through the CLI/MCP tool table (#2892).
+
+use pdfcraft_automation::{Automation, Content};
+use pdfcraft_engine::{DocId, Edit, NewField, Session};
+use serde_json::{Value, json};
+
+fn tool(a: &mut Automation, name: &str, args: Value) -> Value {
+ match a.call(name, &args).unwrap().remove(0) {
+ Content::Json(value) => value,
+ other => panic!("expected JSON: {other:?}"),
+ }
+}
+
+#[test]
+fn button_caption_can_be_changed_after_creation() {
+ let dir = std::env::temp_dir().join(format!("pdfcraft-button-caption-{}", std::process::id()));
+ std::fs::create_dir_all(&dir).unwrap();
+ let mut session = Session::new();
+ let bytes = session.create_blank(300.0, 400.0, 1).unwrap();
+ let id = session.open("buttons.pdf", None, bytes, None).unwrap();
+ session
+ .apply(
+ id,
+ Edit::AddField {
+ page: 0,
+ rect: [40.0, 300.0, 200.0, 330.0],
+ kind: NewField::Button { caption: String::new() },
+ name: Some("clear_button".into()),
+ },
+ )
+ .unwrap();
+ std::fs::write(dir.join("buttons.pdf"), session.save_bytes(id).unwrap().as_slice()).unwrap();
+ let mut a = Automation::new().with_root(&dir).unwrap().with_clock(|| 1_700_000_000);
+ let id = tool(&mut a, "doc_open", json!({"path": "buttons.pdf"}))["doc"].as_u64().unwrap();
+ let before = a.session().save_bytes(DocId(id)).unwrap();
+ let (_, _, empty_pixels) = a.render_pixels(id, 1, 1.0).unwrap();
+ tool(&mut a, "form_set_props", json!({"doc": id, "field": "clear_button", "caption": "Clear Form"}));
+ let fields = tool(&mut a, "form_fields", json!({"doc": id}));
+ assert_eq!(fields["fields"][0]["caption"], "Clear Form");
+ assert_eq!(fields["fields"][0]["captions"], json!([{"widget": 1, "caption": "Clear Form"}]));
+ assert_eq!(a.session().get(DocId(id)).unwrap().can_undo(), Some("Change field properties"));
+ let (_, _, labelled_pixels) = a.render_pixels(id, 1, 1.0).unwrap();
+ assert_ne!(empty_pixels, labelled_pixels, "the label must reach the actual page appearance");
+ tool(&mut a, "edit_undo", json!({"doc": id}));
+ assert_eq!(a.session().save_bytes(DocId(id)).unwrap(), before);
+ tool(&mut a, "edit_redo", json!({"doc": id}));
+ tool(&mut a, "doc_save", json!({"doc": id, "path": "labelled.pdf"}));
+ let saved = std::sync::Arc::new(std::fs::read(dir.join("labelled.pdf")).unwrap());
+ let doc = pdfcraft_cos::Document::open(saved).unwrap();
+ let root = doc.get(doc.root().unwrap());
+ let form = doc.resolve(root.as_dict().unwrap().get(b"AcroForm").unwrap());
+ let fields = doc.resolve(form.as_dict().unwrap().get(b"Fields").unwrap());
+ let button = doc.resolve(&fields.as_array().unwrap()[0]);
+ let mk = doc.resolve(button.as_dict().unwrap().get(b"MK").unwrap());
+ assert_eq!(doc.resolve(mk.as_dict().unwrap().get(b"CA").unwrap()).as_string().unwrap().to_text(), "Clear Form");
+ assert_eq!(doc.resolve(button.as_dict().unwrap().get(b"T").unwrap()).as_string().unwrap().to_text(), "clear_button");
+ let reopened = tool(&mut a, "doc_open", json!({"path": "labelled.pdf"}))["doc"].as_u64().unwrap();
+ let fields = tool(&mut a, "form_fields", json!({"doc": reopened}));
+ assert_eq!(fields["fields"][0]["caption"], "Clear Form");
+ let (_, _, pixels) = a.render_pixels(reopened, 1, 1.0).unwrap();
+ assert_eq!(pixels, labelled_pixels);
+ let out = tool(&mut a, "form_set_props", json!({"doc": id, "field": "clear_button", "caption": "提交"}));
+ assert!(out["warnings"].as_array().unwrap().iter().filter_map(Value::as_str).any(|w| w.contains("U+63D0")));
+ let fields = tool(&mut a, "form_fields", json!({"doc": id}));
+ assert_eq!(fields["fields"][0]["caption"], "提交", "unsupported characters are retained and reported");
+ tool(&mut a, "form_set_props", json!({"doc": id, "field": "clear_button", "caption": "Clear Form"}));
+ for invalid in [
+ json!({"doc": id, "field": "clear_button", "caption": "bad", "widget": 0}),
+ json!({"doc": id, "field": "clear_button", "caption": "bad", "widget": -1}),
+ json!({"doc": id, "field": "clear_button", "caption": "bad", "widget": 2}),
+ json!({"doc": id, "field": "clear_button", "caption": "bad", "widget": i64::MAX}),
+ json!({"doc": id, "field": "clear_button", "caption": 42}),
+ json!({"doc": id, "field": "clear_button", "widget": 1, "tooltip": "bad"}),
+ json!({"doc": id, "fields": ["clear_button"], "caption": "bad"}),
+ ] {
+ let before = a.session().save_bytes(DocId(id)).unwrap();
+ assert!(a.call("form_set_props", &invalid).is_err(), "{invalid}");
+ assert_eq!(a.session().save_bytes(DocId(id)).unwrap(), before, "{invalid}");
+ }
+ tool(&mut a, "form_set_props", json!({"doc": id, "field": "clear_button", "locked": true}));
+ let before = a.session().save_bytes(DocId(id)).unwrap();
+ assert!(a.call("form_set_props", &json!({"doc": id, "field": "clear_button", "caption": "bad"})).is_err());
+ assert_eq!(a.session().save_bytes(DocId(id)).unwrap(), before);
+ tool(&mut a, "form_set_props", json!({"doc": id, "field": "clear_button", "locked": false}));
+ tool(&mut a, "form_set_props", json!({"doc": id, "field": "clear_button", "caption": ""}));
+ let fields = tool(&mut a, "form_fields", json!({"doc": id}));
+ assert_eq!(fields["fields"][0]["caption"], "");
+ let (_, _, pixels) = a.render_pixels(id, 1, 1.0).unwrap();
+ assert_eq!(pixels, empty_pixels);
+ std::fs::remove_dir_all(dir).unwrap();
+}
diff --git a/crates/cos/src/document.rs b/crates/cos/src/document.rs
index e943555905..639edf582c 100644
--- a/crates/cos/src/document.rs
+++ b/crates/cos/src/document.rs
@@ -82,6 +82,8 @@ struct XrefSection {
row_count: usize,
max_num: u32,
has_entries: bool,
+ /// A table subsection was read one number lower than it says ([`subsection_start`]).
+ renumbered: bool,
}
#[derive(Clone, Debug)]
@@ -307,6 +309,10 @@ pub struct Revision {
pub xref_offset: u64,
/// `true` if the section is a cross-reference stream.
pub is_stream: bool,
+ /// `true` if `xref_offset` counts from the `%PDF-` header rather than from the start of the
+ /// file. The two differ only in a file with bytes before its header, and an update appended
+ /// to such a file must count the same way as the section it points back to.
+ pub header_relative: bool,
}
#[derive(Clone, Debug)]
@@ -857,6 +863,19 @@ impl Document {
&self.revisions
}
+ /// Position of the `%PDF-` header in the file: 0 unless bytes come before it.
+ pub fn header_offset(&self) -> usize {
+ self.header_offset
+ }
+
+ /// There is no cross-reference chain to append to: the index was rebuilt from object
+ /// headers because the file's own was unusable (see [`Self::repair_log`]), or the document
+ /// was built from scratch ([`Self::new_empty`]). A save writes such a document in full, even
+ /// when nothing was edited: for a repaired file that is what puts the repair on disk.
+ pub fn reconstructed(&self) -> bool {
+ self.revisions.is_empty()
+ }
+
/// Where each revision of the file ends (oldest first): the byte after the `%%EOF` (and its
/// end-of-line) that closes its cross-reference section. `data[..end]` is that revision as it
/// was saved. A linearized file's first-page section belongs to the revision it precedes, so
@@ -891,9 +910,10 @@ impl Document {
self.repair_log = Arc::new(log);
}
- /// `true` when there are unsaved edits.
+ /// `true` when there are unsaved edits. Removing an `/Encrypt` dictionary written inline in
+ /// the trailer (as MuPDF writes it) changes no object, and is an edit all the same.
pub fn is_modified(&self) -> bool {
- !self.overlay.is_empty()
+ !self.overlay.is_empty() || self.encryption_changed
}
/// Object numbers changed since the document was opened (or last saved).
@@ -1397,13 +1417,13 @@ impl Document {
log.push(format!("cross-reference chain loops at offset {off}; stopped"));
break;
}
- let (primary, sect_trailer, is_stream) = match self.read_compact_section(off as usize) {
- Ok(Some(v)) => v,
+ let ((primary, sect_trailer, is_stream), header_relative) = match self.read_compact_section(off as usize) {
+ Ok(Some(v)) => (v, false),
Ok(None) => return Ok(None),
Err(e) if self.header_offset > 0 => {
log.push(format!("xref at {off} unreadable ({e}); retrying relative to the header"));
match self.read_compact_section(off as usize + self.header_offset)? {
- Some(v) => v,
+ Some(v) => (v, true),
None => return Ok(None),
}
}
@@ -1411,7 +1431,10 @@ impl Document {
// materialized reader logs it and the file is rebuilt from object headers (#1656).
Err(e) => return Err(e),
};
- revisions.push(Revision { xref_offset: off, is_stream });
+ revisions.push(Revision { xref_offset: off, is_stream, header_relative });
+ if primary.renumbered {
+ log.push(format!("xref subsection at {off} is numbered from 1 but begins with object 0's entry; read from 0"));
+ }
let Some(rows) = retained_rows.checked_add(primary.row_count).filter(|rows| *rows <= MAX_COMPACT_XREF_ROWS) else { return Ok(None) };
let Some(bytes) = retained_bytes.checked_add(primary.retained_bytes()).filter(|bytes| *bytes <= MAX_COMPACT_XREF_BYTES) else {
return Ok(None);
@@ -1486,6 +1509,7 @@ impl Document {
let mut has_entries = false;
let mut previous_end = None;
let mut row_count = 0usize;
+ let mut renumbered = false;
loop {
lx.skip_ws();
if lx.eat_keyword(b"trailer") {
@@ -1494,7 +1518,7 @@ impl Document {
if !layout_ok {
return Ok(None);
}
- return Ok(Some((XrefSection { data: XrefSectionData::Table, ranges, row_count, max_num, has_entries }, d, false)));
+ return Ok(Some((XrefSection { data: XrefSectionData::Table, ranges, row_count, max_num, has_entries, renumbered }, d, false)));
}
let (Ok(Object::Int(start)), Ok(Object::Int(count))) = (lx.object(), lx.object()) else {
return Err(CosError::Syntax { offset: lx.pos, detail: "bad xref subsection header".into() });
@@ -1502,6 +1526,9 @@ impl Document {
if !valid_xref_range(start, count) || count > 10_000_000 {
return Err(CosError::Syntax { offset: lx.pos, detail: "invalid xref object range".into() });
}
+ let shifted = subsection_start(data, lx.pos, start, count);
+ renumbered |= shifted != start;
+ let start = shifted;
let Some(next_rows) = usize::try_from(count).ok().and_then(|count| row_count.checked_add(count)) else { return Ok(None) };
if next_rows > MAX_COMPACT_XREF_ROWS {
return Ok(None);
@@ -1654,7 +1681,14 @@ impl Document {
}
let max_num = ranges.iter().filter_map(|range| range.start.checked_add(range.count.saturating_sub(1))).max().unwrap_or(0);
let row_count = row_start;
- let section = XrefSection { data: XrefSectionData::Stream { bytes: Arc::new(raw), widths }, ranges, row_count, max_num, has_entries: false };
+ let section = XrefSection {
+ data: XrefSectionData::Stream { bytes: Arc::new(raw), widths },
+ ranges,
+ row_count,
+ max_num,
+ has_entries: false,
+ renumbered: false,
+ };
let has_entries = section.has_any_entry();
let section = XrefSection { has_entries, ..section };
Ok(Some((section, stream.dict.clone(), true)))
@@ -1688,13 +1722,13 @@ impl Document {
let mut sect = BTreeMap::new();
// Object numbers whose entry in this section was set aside ([`Self::xref_oddities`]).
let mut odd = Vec::new();
- let (sect_trailer, is_stream) = match self.read_section(off as usize, &mut sect, &mut odd) {
- Ok(v) => v,
+ let ((sect_trailer, is_stream), header_relative) = match self.read_section(off as usize, &mut sect, &mut odd, log) {
+ Ok(v) => (v, false),
Err(e) if self.header_offset > 0 => {
log.push(format!("xref at {off} unreadable ({e}); retrying relative to the header"));
sect.clear();
odd.clear();
- self.read_section(off as usize + self.header_offset, &mut sect, &mut odd)?
+ (self.read_section(off as usize + self.header_offset, &mut sect, &mut odd, log)?, true)
}
Err(e) => {
// An unreadable older section loses every object only it locates (the page
@@ -1705,12 +1739,12 @@ impl Document {
return Err(e);
}
};
- revisions.push(Revision { xref_offset: off, is_stream });
+ revisions.push(Revision { xref_offset: off, is_stream, header_relative });
// Hybrid files (§7.5.8.4): the /XRefStm entries supplement the table and replace the
// free entries it gives objects only a cross-reference stream can locate.
if let Some(x) = sect_trailer.int(b"XRefStm").filter(|x| *x >= 0) {
let mut stm = BTreeMap::new();
- match self.read_section(x as usize, &mut stm, &mut odd) {
+ match self.read_section(x as usize, &mut stm, &mut odd, log) {
Ok(_) => {
for (num, e) in stm {
match sect.get(&num) {
@@ -1745,7 +1779,13 @@ impl Document {
/// Read one section at `off`, adding entries not already known (newer sections win). `odd`
/// collects the object numbers whose entry here is set aside ([`Self::xref_oddities`]).
- fn read_section(&self, off: usize, entries: &mut BTreeMap, odd: &mut Vec) -> Result<(Dict, bool), CosError> {
+ fn read_section(
+ &self,
+ off: usize,
+ entries: &mut BTreeMap,
+ odd: &mut Vec,
+ log: &mut Vec,
+ ) -> Result<(Dict, bool), CosError> {
let data = &self.data;
let mut lx = Lexer::new(data, off);
if lx.eat_keyword(b"xref") {
@@ -1762,6 +1802,11 @@ impl Document {
if !valid_xref_range(start, count) || count > 10_000_000 {
return Err(CosError::Syntax { offset: lx.pos, detail: "invalid xref object range".into() });
}
+ let shifted = subsection_start(data, lx.pos, start, count);
+ if shifted != start {
+ log.push(format!("xref subsection at {off} is numbered from 1 but begins with object 0's entry; read from 0"));
+ }
+ let start = shifted;
for i in 0..count {
lx.skip_ws();
let o = lx.token();
@@ -1972,6 +2017,34 @@ fn generated_id(seed: &[u8; 32]) -> Vec {
out
}
+/// The object number a classic xref subsection really starts at. A common producer bug heads the
+/// table `1 n` and then writes object 0's free entry (`0000000000 65535 f`) as its first row, so
+/// every row would be read one number too high and object 1 as free. Such a subsection is read
+/// from 0 when its second row, read that way, locates an object that is indeed numbered 1. `rows`
+/// is the position just after the subsection header.
+fn subsection_start(data: &[u8], rows: usize, start: i64, count: i64) -> i64 {
+ if start != 1 || count < 2 {
+ return start;
+ }
+ let mut lx = Lexer::new(data, rows);
+ let mut row = || {
+ lx.skip_ws();
+ let offset = std::str::from_utf8(lx.token()).ok().and_then(|s| s.parse::().ok());
+ lx.skip_ws();
+ let generation = std::str::from_utf8(lx.token()).ok().and_then(|s| s.parse::().ok());
+ lx.skip_ws();
+ (offset, generation, lx.token() == b"f")
+ };
+ let (Some(0), Some(65535), true) = row() else { return start };
+ let (Some(second), _, false) = row() else { return start };
+ let Some(at) = usize::try_from(second).ok().filter(|at| *at < data.len()) else { return start };
+ let mut obj = Lexer::new(data, at);
+ match (obj.object(), obj.object()) {
+ (Ok(Object::Int(1)), Ok(Object::Int(_))) if obj.eat_keyword(b"obj") => 0,
+ _ => start,
+ }
+}
+
/// Object numbers must fit the representation used by references and the xref map.
/// Validate the whole subsection before adding entries, so neither arithmetic overflow nor
/// truncation can turn a damaged range into entries for unrelated objects.
@@ -2159,6 +2232,71 @@ mod tests {
assert!(doc.xref_entry_is_exact(1) && doc.xref_entry_is_exact(2));
}
+ /// A classic table with `/Size` rows and object 0's free entry first, its subsection numbered
+ /// `1 n` instead of `0 n` (#2676): every row is one number off.
+ fn misnumbered_table(start: u32) -> Vec {
+ let bodies = ["<< /Marker (first object) >>", "<< /Type /Pages /Kids [] /Count 0 >>", "<< /Type /Catalog /Pages 2 0 R >>"];
+ let mut out = b"%PDF-1.7
+"
+ .to_vec();
+ let mut offsets = Vec::new();
+ for (i, body) in bodies.iter().enumerate() {
+ offsets.push(out.len());
+ out.extend_from_slice(
+ format!(
+ "{} 0 obj
+{body}
+endobj
+",
+ i + 1
+ )
+ .as_bytes(),
+ );
+ }
+ let xref = out.len();
+ out.extend_from_slice(format!("xref\n{start} 4\n0000000000 65535 f \n").as_bytes());
+ for o in offsets {
+ out.extend_from_slice(format!("{o:010} 00000 n \n").as_bytes());
+ }
+ out.extend_from_slice(format!("trailer\n<< /Size 4 /Root 3 0 R >>\nstartxref\n{xref}\n%%EOF\n").as_bytes());
+ out
+ }
+
+ #[test]
+ fn a_table_numbered_from_1_that_begins_with_object_0_is_read_from_0() {
+ let doc = Document::open(Arc::new(misnumbered_table(1))).unwrap();
+ let first = doc.get(ObjRef::new(1, 0)).as_dict().cloned().expect("object 1 is not free");
+ assert!(first.contains(b"Marker"));
+ assert_eq!(doc.root(), Some(ObjRef::new(3, 0)));
+ assert!(doc.repair_log().iter().any(|l| l.contains("numbered from 1")), "{:?}", doc.repair_log());
+ // A correct table reads the same and logs nothing.
+ let fine = Document::open(Arc::new(misnumbered_table(0))).unwrap();
+ assert!(fine.get(ObjRef::new(1, 0)).as_dict().is_some_and(|d| d.contains(b"Marker")));
+ assert!(!fine.repair_log().iter().any(|l| l.contains("numbered from 1")));
+ }
+
+ #[test]
+ fn a_table_that_really_starts_at_1_keeps_its_numbers() {
+ // Object 1 deleted for good (generation 65535) in a subsection that does start at 1: its
+ // second row locates object 2, so nothing is shifted.
+ let mut bytes = b"%PDF-1.7\n".to_vec();
+ let pages = bytes.len();
+ bytes.extend_from_slice(b"2 0 obj\n<< /Type /Pages /Kids [] /Count 0 >>\nendobj\n");
+ let catalog = bytes.len();
+ bytes.extend_from_slice(b"3 0 obj\n<< /Type /Catalog /Pages 2 0 R >>\nendobj\n");
+ let xref = bytes.len();
+ bytes.extend_from_slice(
+ format!(
+ "xref\n0 1\n0000000000 65535 f \n1 3\n0000000000 65535 f \n{pages:010} 00000 n \n{catalog:010} 00000 n \ntrailer\n<< /Size 4 /Root 3 0 R >>\nstartxref\n{xref}\n%%EOF\n"
+ )
+ .as_bytes(),
+ );
+ let doc = Document::open(Arc::new(bytes)).unwrap();
+ assert!(doc.get(ObjRef::new(2, 0)).as_dict().is_some_and(|d| d.name(b"Type") == Some(b"Pages")));
+ assert!(doc.get(ObjRef::new(1, 0)).as_dict().is_none(), "object 1 stays free");
+ assert!(!doc.repair_log().iter().any(|l| l.contains("numbered from 1")));
+ }
+
#[test]
fn opening_an_exhausted_classic_xref_repairs_and_allocates() {
// The extra in-use row need not be loaded to exhaust the next object number.
diff --git a/crates/cos/src/parser.rs b/crates/cos/src/parser.rs
index 8ac8a195bf..c9689d7fee 100644
--- a/crates/cos/src/parser.rs
+++ b/crates/cos/src/parser.rs
@@ -337,11 +337,14 @@ impl<'a> Lexer<'a> {
}
return Ok(Object::Real(0.0));
}
- match s.trim_start_matches('+').parse::() {
+ // Tolerate a doubled sign ("--280") like the real branch does.
+ let t = s.trim_start_matches('+');
+ let cleaned = t.strip_prefix('-').filter(|r| r.starts_with('-')).unwrap_or(t);
+ match cleaned.parse::() {
Ok(v) => Ok(Object::Int(v)),
Err(_) => {
// Out-of-range integers are treated as reals (§7.3.3 allows implementation limits).
- s.parse::().map(Object::Real).map_err(|_| CosError::Syntax { offset: start, detail: format!("bad number {s:?}") })
+ cleaned.parse::().map(Object::Real).map_err(|_| CosError::Syntax { offset: start, detail: format!("bad number {s:?}") })
}
}
}
@@ -475,6 +478,20 @@ mod tests {
Lexer::new(s.as_bytes(), 0).object().unwrap_or_else(|e| panic!("{s}: {e}"))
}
+ #[test]
+ fn doubled_sign_integer_reads_like_the_real_form() {
+ // Some writers emit a TJ kerning as "--280"; the real form "--280.0" is already read.
+ assert_eq!(obj("--280.0"), Object::Real(-280.0));
+ assert_eq!(obj("--280"), Object::Int(-280));
+ match obj("[(Hello) --280 (World)]") {
+ Object::Array(items) => {
+ assert_eq!(items.len(), 3);
+ assert_eq!(items[1], Object::Int(-280));
+ }
+ other => panic!("not an array: {other:?}"),
+ }
+ }
+
#[test]
fn scalars() {
assert_eq!(obj("true"), Object::Bool(true));
diff --git a/crates/cos/src/writer.rs b/crates/cos/src/writer.rs
index f8d8fd3d80..331f97d443 100644
--- a/crates/cos/src/writer.rs
+++ b/crates/cos/src/writer.rs
@@ -183,7 +183,7 @@ pub fn write_incremental(doc: &Document, opts: &SaveOptions) -> Result,
}
// Reconstructed files have no chain to append to; added or removed encryption must
// rewrite every object; redaction must not leave the old revision behind. All need a full save.
- if doc.revisions().is_empty() || doc.encryption_changed() || doc.full_save_required() {
+ if doc.reconstructed() || doc.encryption_changed() || doc.full_save_required() {
return write_full(doc, opts);
}
let mut doc = doc.clone();
@@ -193,7 +193,11 @@ pub fn write_incremental(doc: &Document, opts: &SaveOptions) -> Result,
// allocated once at its exact size: growing a buffer that holds a large original past a
// guessed capacity would copy the whole file again and keep up to as much again unused.
let newline = !matches!(original.last(), Some(b'\n' | b'\r'));
- let base = (original.len() + usize::from(newline)) as u64;
+ // A file with bytes before its `%PDF-` header may count its offsets from the header. The new
+ // section then has to as well: readers take `/Prev` and `startxref` in one convention, and a
+ // mix sends them into reconstruction (#2364).
+ let shift = if doc.revisions().last().is_some_and(|r| r.header_relative) { doc.header_offset() } else { 0 };
+ let base = (original.len() + usize::from(newline)).saturating_sub(shift) as u64;
let mut out = Vec::new();
let mut offsets: BTreeMap = BTreeMap::new(); // num → (offset, gen, in use)
for (num, generation, obj) in doc.overlay_entries() {
diff --git a/crates/cos/tests/encryption.rs b/crates/cos/tests/encryption.rs
index af92cf67c7..e05161558d 100644
--- a/crates/cos/tests/encryption.rs
+++ b/crates/cos/tests/encryption.rs
@@ -336,3 +336,41 @@ fn an_earlier_revision_opens_with_the_documents_own_credentials() {
assert_eq!(content(¤t.open_with_same_credentials(Arc::new(fixture())).unwrap()), SECRET, "{alg:?}");
}
}
+
+/// `protect`, with the `/Encrypt` dictionary written inline in the trailer, as MuPDF writes it,
+/// instead of as an indirect object.
+fn protect_inline(alg: Algorithm, user: &str, owner: &str) -> Vec {
+ let mut doc = Document::open(Arc::new(fixture())).unwrap();
+ doc.set_encryption(&NewEncryption {
+ algorithm: alg,
+ user_password: user,
+ owner_password: owner,
+ permissions: -1,
+ encrypt_metadata: true,
+ seed: [9; 32],
+ })
+ .unwrap();
+ let dict = doc.get(doc.trailer().reference(b"Encrypt").unwrap()).as_dict().cloned().unwrap();
+ doc.trailer_mut().set(b"Encrypt".to_vec(), Object::Dict(dict));
+ write_full(&doc, &SaveOptions::default()).unwrap()
+}
+
+/// MuPDF writes `/Encrypt` inline in the trailer, so removing it changes no object. It is still
+/// an unsaved change, and saving must write a file that opens without a password (#2661).
+#[test]
+fn removing_an_inline_encrypt_dictionary_leaves_an_unencrypted_file() {
+ for alg in ALL {
+ let bytes = Arc::new(protect_inline(alg, "user", "owner"));
+ assert!(matches!(Document::open(bytes.clone()), Err(CosError::NeedsPassword)), "{alg:?}");
+ let mut doc = Document::open_with_password(bytes, Some("owner")).unwrap();
+ assert!(matches!(doc.trailer().get(b"Encrypt"), Some(Object::Dict(_))), "{alg:?}: inline");
+ assert!(!doc.is_modified(), "{alg:?}");
+ doc.remove_encryption();
+ assert!(doc.is_modified(), "{alg:?}: removing security is an unsaved change");
+ let plain = write_incremental(&doc, &SaveOptions::default()).unwrap();
+ let again = Document::open(Arc::new(plain.clone())).unwrap_or_else(|e| panic!("{alg:?}: {e:?}"));
+ assert!(again.security().is_none() && !again.trailer().contains(b"Encrypt"), "{alg:?}");
+ assert_eq!(content(&again), SECRET, "{alg:?}");
+ assert!(hayro_syntax::Pdf::new(plain).is_ok(), "{alg:?}");
+ }
+}
diff --git a/crates/cos/tests/writer.rs b/crates/cos/tests/writer.rs
index ec18b13f36..ac04be2e94 100644
--- a/crates/cos/tests/writer.rs
+++ b/crates/cos/tests/writer.rs
@@ -392,3 +392,67 @@ fn full_save_preserves_unknown_trailer_data() {
assert_eq!(trailer_note(&back), trailer_note(&doc));
assert_eq!(back.trailer().get(b"Version"), doc.trailer().get(b"Version"));
}
+
+/// The `startxref` value at the end of `bytes`.
+fn last_startxref(bytes: &[u8]) -> usize {
+ let at = bytes.windows(9).rposition(|w| w == b"startxref").unwrap();
+ String::from_utf8_lossy(&bytes[at + 9..]).split_whitespace().next().unwrap().parse().unwrap()
+}
+
+#[test]
+fn incremental_saves_of_a_file_with_bytes_before_its_header_count_from_the_header() {
+ // Offsets written before the prefix was added count from `%PDF-`, as in the KillerPDF file
+ // of #2364; the appended section must count the same way as the `/Prev` it points back to.
+ let prefix = vec![b'#'; 128];
+ let as_stream = write_full(&Document::open(Arc::new(fixture())).unwrap(), &SaveOptions::default()).unwrap();
+ for (name, body, marker) in [("table", classic_table(), &b"xref"[..]), ("stream", as_stream, &b" 0 obj"[..])] {
+ let input = [prefix.as_slice(), &body].concat();
+ let mut doc = Document::open(Arc::new(input.clone())).unwrap();
+ assert!(doc.revisions().last().unwrap().header_relative, "{name}");
+ doc.update_dict(ObjRef::new(1, 0), |d| d.set(b"Lang".to_vec(), Object::String(pdfcraft_cos::PdfString::literal(b"en".to_vec())))).unwrap();
+ let saved = write_incremental(&doc, &SaveOptions::default()).unwrap();
+ assert!(saved.starts_with(&input), "{name}");
+
+ let xref = last_startxref(&saved) + prefix.len();
+ let section = saved.get(xref..).unwrap_or_default();
+ assert!(section.windows(marker.len()).position(|w| w == marker).is_some_and(|p| p < 16), "{name}: startxref misses its section");
+ let back = Document::open(Arc::new(saved.clone())).unwrap();
+ assert_eq!(back.revisions().len(), doc.revisions().len() + 1, "{name}");
+ assert!(back.revisions().iter().all(|r| r.header_relative), "{name}: one convention across the chain");
+ let edited = back.get(ObjRef::new(1, 0));
+ assert_eq!(edited.as_dict().unwrap().get(b"Lang").and_then(|o| o.as_string()).map(|s| s.bytes.as_slice()), Some(b"en".as_slice()));
+ assert!(back.xref_entry_is_exact(1), "{name}");
+ assert_qpdf_clean(&saved);
+ }
+}
+
+#[test]
+fn incremental_saves_of_a_file_whose_offsets_count_from_its_first_byte_keep_counting_so() {
+ // The other convention: the same file, but its offsets include the bytes before the header.
+ let prefix = vec![b'#'; 128];
+ let mut input = [prefix.as_slice(), &classic_table()].concat();
+ let xref = input.windows(6).rposition(|w| w == b"\nxref\n").unwrap() + 1;
+ let table: String = String::from_utf8_lossy(&input[xref..])
+ .lines()
+ .map(|line| match line.strip_suffix(" 00000 n ").and_then(|o| o.parse::().ok()) {
+ Some(o) => format!("{:010} 00000 n \n", o + prefix.len()),
+ None if line.parse::().is_ok() => format!("{xref}\n"),
+ None => format!("{line}\n"),
+ })
+ .collect();
+ input.truncate(xref);
+ input.extend_from_slice(table.as_bytes());
+
+ let mut doc = Document::open(Arc::new(input.clone())).unwrap();
+ assert!(doc.repair_log().is_empty(), "{:?}", doc.repair_log());
+ assert!(!doc.revisions().last().unwrap().header_relative);
+ doc.update_dict(ObjRef::new(1, 0), |d| d.set(b"Lang".to_vec(), Object::String(pdfcraft_cos::PdfString::literal(b"en".to_vec())))).unwrap();
+ let saved = write_incremental(&doc, &SaveOptions::default()).unwrap();
+ assert!(saved.get(last_startxref(&saved)..).unwrap_or_default().starts_with(b"xref"));
+ let back = Document::open(Arc::new(saved.clone())).unwrap();
+ assert!(back.repair_log().is_empty(), "{:?}", back.repair_log());
+ assert_eq!(back.revisions().len(), 2);
+ assert!(back.xref_entry_is_exact(1));
+ // No qpdf check here: qpdf counts every offset from the header in a file with bytes before
+ // it, so it already reconstructs this input before anything is appended.
+}
diff --git a/crates/create/README.md b/crates/create/README.md
index 88c6da2ea5..499081059a 100644
--- a/crates/create/README.md
+++ b/crates/create/README.md
@@ -23,3 +23,9 @@ Layer L4. Create a PDF (Acrobat's Create a PDF tool), execution plan M10.2:
line says so (`ESCAPED_NOTE`).
Every function returns a `pdfcraft_cos::Document`; the caller writes it.
+
+Image export and Save Image As apply each lossless image's Decode pair before converting its color components.
+Indexed Decode values select palette entries after rounding and clipping. Color-key masks compare the original samples.
+Missing, null or malformed Decode arrays keep the default mapping.
+The existing 16-bit path still uses each sample's high byte. JPEG and mask Decode handling remain limited.
+ICC transforms and calibrated color conversion remain outside this export path.
diff --git a/crates/create/src/extract.rs b/crates/create/src/extract.rs
index 1034489458..768da4d1f1 100644
--- a/crates/create/src/extract.rs
+++ b/crates/create/src/extract.rs
@@ -126,13 +126,13 @@ fn image(doc: &Document, s: &Stream) -> Result<(&'static str, Vec), String>
// A stencil mask (`ImageMask true`) has no /SMask or /Mask of its own (8.9.6.2).
let mut alpha = if mask { None } else { soft_mask(doc, s, w, h).or_else(|| stencil_mask(doc, s, w, h)) };
let key = if mask { None } else { colour_key(doc, s) };
- // Decode arrays: only inversion ([1 0] per component) is honoured, the common case.
- let inverted = s
- .dict
- .get(b"Decode")
- .map(|d| doc.resolve(d))
- .and_then(|d| d.as_array().map(|a| a.first().and_then(Object::as_f64) > a.get(1).and_then(Object::as_f64)))
- .unwrap_or(false);
+ // Keep the existing JPEG and stencil-mask Decode handling separate from lossless colour images.
+ let inverted = (jpeg || mask)
+ && s.dict
+ .get(b"Decode")
+ .map(|d| doc.resolve(d))
+ .and_then(|d| d.as_array().map(|a| a.first().and_then(Object::as_f64) > a.get(1).and_then(Object::as_f64)))
+ .unwrap_or(false);
// In a PDF a CMYK JPEG's samples are ink amounts unless /Decode inverts them, while image
// viewers read a CMYK JPEG file as inverted (Adobe's convention): as is, it would show as
// a negative.
@@ -161,6 +161,7 @@ fn image(doc: &Document, s: &Stream) -> Result<(&'static str, Vec), String>
(unpack(&data, w, h, space.components(), bpc).ok_or("the image data is shorter than its size says")?, bpc)
};
let n = space.components();
+ let decode = if jpeg || mask { None } else { decode(doc, s, n) };
if alpha.is_none() {
alpha = key.and_then(|k| keyed_alpha(&samples, n, bpc, &k));
}
@@ -168,10 +169,21 @@ fn image(doc: &Document, s: &Stream) -> Result<(&'static str, Vec), String>
return as_is();
}
let invert = inverted ^ mask; // a stencil mask paints its 0 samples
+ // `unpack` keeps only the high byte at 16 bits, so Decode uses an effective maximum of 255 too.
let max = (1u32 << bpc.min(8)) - 1;
let mut rgb = Vec::with_capacity(w * h * 3);
for px in samples.chunks_exact(n) {
+ let decoded = |i: usize| {
+ decode.as_ref().and_then(|pairs| pairs.get(i)).map(|&[lo, hi]| {
+ let t = f64::from(px.get(i).copied().unwrap_or(0)) / f64::from(max);
+ // Avoid overflowing hi - lo for finite endpoints with opposite signs.
+ (1.0 - t) * lo + t * hi
+ })
+ };
let v = |i: usize| {
+ if let Some(value) = decoded(i) {
+ return (value.clamp(0.0, 1.0) * 255.0).round() as u8;
+ }
let x = (px.get(i).copied().unwrap_or(0) as u32 * 255 / max) as u8;
if invert { 255 - x } else { x }
};
@@ -180,8 +192,12 @@ fn image(doc: &Document, s: &Stream) -> Result<(&'static str, Vec), String>
Space::Rgb => rgb.extend_from_slice(&[v(0), v(1), v(2)]),
Space::Cmyk => rgb.extend_from_slice(&cmyk(v(0), v(1), v(2), v(3))),
Space::Indexed(base, hival, table) => {
- // Indexed samples outside the declared range use the nearest endpoint (ISO 32000-1, 8.6.6.3).
- let i = px.first().copied().unwrap_or(0).min(*hival) as usize;
+ // Explicit Decode maps into index units, rounded then clipped (8.6.6.3).
+ // Without Decode, retain raw indices rather than normalizing to hival.
+ let i = match decoded(0) {
+ Some(value) => value.round().clamp(0.0, f64::from(*hival)) as usize,
+ None => px.first().copied().unwrap_or(0).min(*hival) as usize,
+ };
let k = base.components();
let e = table.get(i * k..i * k + k).unwrap_or(&[0, 0, 0, 0][..k]);
rgb.extend_from_slice(&match **base {
@@ -198,6 +214,24 @@ fn image(doc: &Document, s: &Stream) -> Result<(&'static str, Vec), String>
png(w as u32, h as u32, &rgb, alpha.as_deref()).map(|p| ("png", p))
}
+/// The explicit Decode intervals for a lossless colour image (8.9.5.2, Table 90).
+///
+/// A malformed array (not an array, the wrong length, a non-number or non-finite entry) gives
+/// the default mapping, so the image still exports.
+fn decode(doc: &Document, s: &Stream, n: usize) -> Option> {
+ let value = doc.resolve(s.dict.get(b"Decode")?);
+ let values = value.as_array()?;
+ if Some(values.len()) != n.checked_mul(2) {
+ return None;
+ }
+ let endpoint = |value: &Object| doc.resolve(value).as_f64().filter(|v| v.is_finite());
+ let mut pairs = Vec::with_capacity(n);
+ for [lo, hi] in values.as_chunks::<2>().0 {
+ pairs.push([endpoint(lo)?, endpoint(hi)?]);
+ }
+ Some(pairs)
+}
+
/// Width and height as sizes, when both are positive and the image is at most `MAX_PIXELS`.
fn size(w: i64, h: i64) -> Option<(usize, usize)> {
let (w, h) = (usize::try_from(w).ok()?, usize::try_from(h).ok()?);
diff --git a/crates/create/src/tests.rs b/crates/create/src/tests.rs
index a996b703f3..6356105699 100644
--- a/crates/create/src/tests.rs
+++ b/crates/create/src/tests.rs
@@ -124,6 +124,18 @@ fn text_is_wrapped_and_paginated() {
assert!(content(p.last().unwrap()).contains("(After a form feed) Tj"), "a form feed starts a page");
}
+#[test]
+fn text_keeps_leading_spaces_and_tabs() {
+ let doc = reopen(&from_text("t", " indented\n\ttabbed\nplain", LETTER, 11.0).unwrap());
+ let p = pages(&doc);
+ let c = doc.resolve(p[0].get(b"Contents").unwrap());
+ let Object::Stream(s) = &*c else { panic!() };
+ let content = String::from_utf8_lossy(&s.decoded().unwrap()).into_owned();
+ assert!(content.contains("( indented) Tj"), "{content}");
+ assert!(content.contains("( tabbed) Tj"), "{content}");
+ assert!(content.contains("(plain) Tj"), "{content}");
+}
+
/// Text Helvetica can't draw is refused with its line and character, never set as `?`; text
/// PdfCraft generates writes such characters as [U+XXXX] and says so.
#[test]
@@ -875,6 +887,193 @@ fn indexed_hival_keeps_existing_short_palette_fallback() {
assert_eq!(exported_png_pixels(&bytes), vec![255, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]);
}
+/// Save and reopen a synthetic image before checking both export entry points.
+fn decode_fixture(dict: Dict, samples: Vec) -> (Document, pdfcraft_cos::ObjRef) {
+ let mut doc = from_images(&[("flat.png".into(), png_bytes(false))]).unwrap();
+ with_image(&mut doc, |_| Stream::from_raw(dict, samples));
+ let doc = reopen(&doc);
+ let page = pdfcraft_model::pages(&doc)[0].clone();
+ let res = doc.resolve(page.dict.get(b"Resources").unwrap());
+ let xo = doc.resolve(res.as_dict().unwrap().get(b"XObject").unwrap());
+ let image = xo.as_dict().unwrap().iter().next().unwrap().1.as_ref().unwrap();
+ (doc, image)
+}
+
+fn assert_decode_pixels(dict: Dict, samples: Vec, expected: &[u8]) {
+ let (doc, image) = decode_fixture(dict, samples);
+ let out = extract_images(&doc, &[0], 0);
+ assert!(out.skipped.is_empty(), "{:?}", out.skipped);
+ assert_eq!(out.images.len(), 1);
+ assert_eq!(exported_png_pixels(&out.images[0].data), expected);
+ let (ext, bytes) = image_file(&doc, image).unwrap();
+ assert_eq!(ext, "png");
+ assert_eq!(exported_png_pixels(&bytes), expected);
+}
+
+fn decode_array(values: &[f64]) -> Object {
+ Object::Array(values.iter().copied().map(Object::Real).collect())
+}
+
+#[test]
+fn lossless_decode_maps_each_component_before_colour_conversion() {
+ for (space, samples, decode, expected) in [
+ ("DeviceRGB", vec![0, 64, 255], vec![1., 0., 0., 1., 0., 1.], [255, 64, 255]),
+ ("DeviceRGB", vec![0, 64, 255], vec![0., 1., 1., 0., 0., 1.], [0, 191, 255]),
+ ("DeviceGray", vec![0], vec![0.2, 0.8], [51, 51, 51]),
+ ("DeviceGray", vec![128], vec![-1., 2.], [129, 129, 129]),
+ ("DeviceCMYK", vec![0, 0, 0, 0], vec![0., 1., 1., 0., 0., 1., 0., 1.], [255, 0, 255]),
+ ] {
+ let mut dict = image_dict(1, 1, space, 8);
+ dict.set(b"Decode".to_vec(), decode_array(&decode));
+ assert_decode_pixels(dict, samples, &expected);
+ }
+}
+
+#[test]
+fn lossless_decode_scales_packed_samples_by_their_bit_depth() {
+ let mut dict = image_dict(4, 1, "DeviceGray", 2);
+ dict.set(b"Decode".to_vec(), decode_array(&[0.2, 0.8]));
+ assert_decode_pixels(dict, vec![0x1b], &[51, 51, 51, 102, 102, 102, 153, 153, 153, 204, 204, 204]);
+}
+
+#[test]
+fn lossless_decode_identity_and_defaults_keep_depth_and_row_padding() {
+ for (bpc, samples, expected) in [
+ (1, vec![0x7f, 0xff], vec![0, 0, 0, 255, 255, 255]),
+ (2, vec![0x7f, 0xbf], vec![85, 85, 85, 170, 170, 170]),
+ (4, vec![0x5f, 0xaf], vec![85, 85, 85, 170, 170, 170]),
+ (8, vec![85, 170], vec![85, 85, 85, 170, 170, 170]),
+ (16, vec![85, 255, 170, 0], vec![85, 85, 85, 170, 170, 170]),
+ ] {
+ for decode in [None, Some(Object::Null), Some(decode_array(&[0., 1.]))] {
+ let mut dict = image_dict(1, 2, "DeviceGray", bpc);
+ if let Some(decode) = decode {
+ dict.set(b"Decode".to_vec(), decode);
+ }
+ assert_decode_pixels(dict, samples.clone(), &expected);
+ }
+ }
+}
+
+#[test]
+fn lossless_indexed_decode_rounds_in_index_units_then_clamps() {
+ let palette = vec![10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120];
+ for (bpc, samples, decode, expected) in [
+ (2, vec![0x1b], Some(vec![3., 0.]), vec![100, 110, 120, 70, 80, 90, 40, 50, 60, 10, 20, 30]),
+ (2, vec![0x1b], Some(vec![-1., 5.]), vec![10, 20, 30, 40, 50, 60, 100, 110, 120, 100, 110, 120]),
+ (2, vec![0x1b], Some(vec![0.5, 0.5]), vec![40, 50, 60, 40, 50, 60, 40, 50, 60, 40, 50, 60]),
+ (8, vec![0, 1, 2, 255], None, vec![10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120]),
+ // Existing 16-bit policy retains the high byte, 42 / 255 * 3 < 0.5.
+ (16, vec![0x2a, 0xab, 0, 0, 255, 255, 128, 0], Some(vec![0., 3.]), vec![10, 20, 30, 10, 20, 30, 100, 110, 120, 70, 80, 90]),
+ ] {
+ let (mut doc, image) = indexed_export_fixture(Object::Int(3), palette.clone(), false, false, bpc, samples);
+ if let Some(decode) = decode {
+ let Object::Stream(mut stream) = (*doc.get(image)).clone() else { panic!() };
+ stream.dict.set(b"Decode".to_vec(), decode_array(&decode));
+ doc.set(image, Object::Stream(stream));
+ }
+ let doc = reopen(&doc);
+ let out = extract_images(&doc, &[0], 0);
+ assert!(out.skipped.is_empty(), "{:?}", out.skipped);
+ assert_eq!(exported_png_pixels(&out.images[0].data), expected);
+ let (_, bytes) = image_file(&doc, out.images[0].object).unwrap();
+ assert_eq!(exported_png_pixels(&bytes), expected);
+ }
+}
+
+#[test]
+fn lossless_decode_keeps_colour_key_comparisons_in_raw_samples() {
+ let mut dict = image_dict(2, 1, "DeviceGray", 8);
+ dict.set(b"Decode".to_vec(), decode_array(&[0.2, 0.8]));
+ dict.set(b"Mask".to_vec(), Object::Array(vec![Object::Int(0), Object::Int(0)]));
+ assert_decode_pixels(dict, vec![0, 255], &[51, 51, 51, 0, 204, 204, 204, 255]);
+
+ let (mut doc, image) = indexed_export_fixture(Object::Int(1), vec![255, 0, 0, 0, 0, 255], false, false, 2, vec![0x1b]);
+ let Object::Stream(mut stream) = (*doc.get(image)).clone() else { panic!() };
+ stream.dict.set(b"Decode".to_vec(), decode_array(&[3., 0.]));
+ stream.dict.set(b"Mask".to_vec(), Object::Array(vec![Object::Int(2), Object::Int(2)]));
+ doc.set(image, Object::Stream(stream));
+ let doc = reopen(&doc);
+ let out = extract_images(&doc, &[0], 0);
+ let expected = [0, 0, 255, 255, 0, 0, 255, 255, 0, 0, 255, 0, 255, 0, 0, 255];
+ assert_eq!(exported_png_pixels(&out.images[0].data), expected);
+ let (_, bytes) = image_file(&doc, out.images[0].object).unwrap();
+ assert_eq!(exported_png_pixels(&bytes), expected);
+}
+
+#[test]
+fn lossless_decode_resolves_indirect_arrays_and_endpoints() {
+ let mut doc = from_images(&[("flat.png".into(), png_bytes(false))]).unwrap();
+ with_image(&mut doc, |doc| {
+ let lo = doc.add(Object::Real(0.2));
+ let hi = doc.add(Object::Real(0.8));
+ let decode = doc.add(Object::Array(vec![Object::Ref(lo), Object::Ref(hi)]));
+ let mut dict = image_dict(1, 1, "DeviceGray", 8);
+ dict.set(b"Decode".to_vec(), Object::Ref(decode));
+ Stream::from_raw(dict, vec![0])
+ });
+ let doc = reopen(&doc);
+ let out = extract_images(&doc, &[0], 0);
+ assert_eq!(exported_png_pixels(&out.images[0].data), [51, 51, 51]);
+ let (_, bytes) = image_file(&doc, out.images[0].object).unwrap();
+ assert_eq!(exported_png_pixels(&bytes), [51, 51, 51]);
+
+ let mut doc = from_images(&[("flat.png".into(), png_bytes(false))]).unwrap();
+ with_image(&mut doc, |doc| {
+ let decode = doc.add(Object::Null);
+ let mut dict = image_dict(1, 1, "DeviceGray", 8);
+ dict.set(b"Decode".to_vec(), Object::Ref(decode));
+ Stream::from_raw(dict, vec![64])
+ });
+ let doc = reopen(&doc);
+ let out = extract_images(&doc, &[0], 0);
+ assert_eq!(exported_png_pixels(&out.images[0].data), [64, 64, 64]);
+ let (_, bytes) = image_file(&doc, out.images[0].object).unwrap();
+ assert_eq!(exported_png_pixels(&bytes), [64, 64, 64]);
+}
+
+#[test]
+fn lossless_decode_interpolation_avoids_endpoint_difference_overflow() {
+ let (mut doc, image) = decode_fixture(image_dict(3, 1, "DeviceGray", 8), vec![0, 64, 255]);
+ let Object::Stream(mut stream) = (*doc.get(image)).clone() else { panic!() };
+ stream.dict.set(b"Decode".to_vec(), decode_array(&[-f64::MAX, f64::MAX]));
+ doc.set(image, Object::Stream(stream));
+ let expected = [0, 0, 0, 0, 0, 0, 255, 255, 255];
+ let out = extract_images(&doc, &[0], 0);
+ assert_eq!(exported_png_pixels(&out.images[0].data), expected);
+ let (_, bytes) = image_file(&doc, image).unwrap();
+ assert_eq!(exported_png_pixels(&bytes), expected);
+}
+
+#[test]
+fn lossless_decode_falls_back_to_the_default_mapping_when_malformed() {
+ for (space, samples, decode, expected) in [
+ ("DeviceRGB", vec![0, 64, 255], decode_array(&[1., 0.]), [0, 64, 255]),
+ ("DeviceGray", vec![64], Object::name("bad"), [64, 64, 64]),
+ ("DeviceGray", vec![64], decode_array(&[]), [64, 64, 64]),
+ ("DeviceGray", vec![64], decode_array(&[0.]), [64, 64, 64]),
+ ("DeviceGray", vec![64], decode_array(&[1., 0., 0., 1.]), [64, 64, 64]),
+ ("DeviceGray", vec![64], Object::Array(vec![Object::Int(1), Object::name("bad")]), [64, 64, 64]),
+ ("DeviceGray", vec![64], Object::Array(vec![Object::Int(1), Object::Null]), [64, 64, 64]),
+ ] {
+ let mut dict = image_dict(1, 1, space, 8);
+ dict.set(b"Decode".to_vec(), decode);
+ assert_decode_pixels(dict, samples, &expected);
+ }
+ // Non-finite reals cannot be serialized as PDF numbers. Exercise the in-memory input.
+ for value in [f64::NAN, f64::INFINITY, f64::NEG_INFINITY] {
+ let mut dict = image_dict(1, 1, "DeviceGray", 8);
+ dict.set(b"Decode".to_vec(), decode_array(&[1., value]));
+ let (mut doc, image) = decode_fixture(image_dict(1, 1, "DeviceGray", 8), vec![64]);
+ doc.set(image, Object::Stream(Stream::from_raw(dict, vec![64])));
+ let out = extract_images(&doc, &[0], 0);
+ assert!(out.skipped.is_empty(), "{:?}", out.skipped);
+ assert_eq!(exported_png_pixels(&out.images[0].data), [64, 64, 64]);
+ let (_, bytes) = image_file(&doc, image).unwrap();
+ assert_eq!(exported_png_pixels(&bytes), [64, 64, 64]);
+ }
+}
+
/// An 8×8 baseline JPEG of one flat colour with four components, stored as `samples`, and an
/// Adobe APP14 segment with `transform` (0: CMYK, 2: YCCK).
fn flat_four_component_jpeg(samples: [u8; 4], transform: u8) -> Vec {
diff --git a/crates/crypt/src/lib.rs b/crates/crypt/src/lib.rs
index fd600cb9af..82b598688a 100644
--- a/crates/crypt/src/lib.rs
+++ b/crates/crypt/src/lib.rs
@@ -203,8 +203,22 @@ impl SecurityHandler {
self.recovered_user.clone()
}
+ /// For R5/R6, `/P` is stored in clear and again in `/Perms`, encrypted with the file key
+ /// (Algorithm 13). Where the two differ, a permission counts only when both grant it, so an
+ /// edited `/P` grants nothing extra. Without a readable `/Perms`, `/P` applies as before.
pub fn permissions(&self) -> Permissions {
- Permissions { bits: self.dict.p, owner: self.auth == Auth::Owner }
+ let bits = self.perms_p().map_or(self.dict.p, |p| self.dict.p & p);
+ Permissions { bits, owner: self.auth == Auth::Owner }
+ }
+
+ /// `/P` as `/Perms` holds it: AES-256 ECB of one block (= CBC with a zero IV), marked `adb`.
+ fn perms_p(&self) -> Option {
+ if !matches!(self.dict.r, 5 | 6) {
+ return None;
+ }
+ let plain = aes_cbc_decrypt(&self.key, &[0; 16], self.dict.perms.first_chunk::<16>()?, false);
+ let p = i32::from_le_bytes(*plain.first_chunk::<4>()?);
+ (plain.get(9..12) == Some(&b"adb"[..])).then_some(p)
}
pub fn dict(&self) -> &EncryptDict {
diff --git a/crates/crypt/src/tests.rs b/crates/crypt/src/tests.rs
index 7272af9ea1..5110e79998 100644
--- a/crates/crypt/src/tests.rs
+++ b/crates/crypt/src/tests.rs
@@ -212,6 +212,28 @@ fn permission_bits_follow_table_22() {
assert!(p.modify(), "owners ignore /P");
}
+#[test]
+fn perms_limits_what_an_edited_p_grants() {
+ let made = new(Algorithm::Aes256, "user pw", "owner pw", 0b0100); // print only
+ let open = |d: &EncryptDict, pw| SecurityHandler::open(d.clone(), ID, Some(pw)).unwrap().permissions();
+ let mut d = made.dict().clone();
+ assert_eq!(open(&d, "user pw").bits, d.p, "/P and /Perms agree");
+ d.p = -4; // /P edited to allow everything, /Perms as written
+ let p = open(&d, "user pw");
+ assert!(p.print() && !p.modify() && !p.copy() && !p.annotate() && !p.fill_forms() && !p.assemble());
+ assert!(open(&d, "owner pw").modify(), "owners ignore /P");
+ d.p = -3904; // a stricter /P still applies
+ assert!(!open(&d, "user pw").print());
+ // No /Perms, or one that does not decrypt to an `adb` block: /P as before.
+ for perms in [Vec::new(), vec![0; 16], made.dict().perms[..15].to_vec()] {
+ let d = EncryptDict { p: -4, perms, ..made.dict().clone() };
+ assert!(open(&d, "user pw").modify());
+ }
+ // R2 to R4 have no /Perms: /P is part of their file key, so an edited /P does not open.
+ let d = EncryptDict { p: -4, ..new(Algorithm::Aes128, "user pw", "owner pw", 0b0100).dict().clone() };
+ assert_eq!(SecurityHandler::open(d, ID, Some("user pw")).unwrap_err(), CryptError::WrongPassword);
+}
+
proptest::proptest! {
#[test]
fn malformed_dictionaries_never_panic(o in proptest::collection::vec(proptest::num::u8::ANY, 0..60),
diff --git a/crates/edit/src/images.rs b/crates/edit/src/images.rs
index 62bef8bc5e..79f2c2cd2a 100644
--- a/crates/edit/src/images.rs
+++ b/crates/edit/src/images.rs
@@ -4,8 +4,8 @@
//! or delete. Only the stream that draws it is rewritten, as a new object.
use kurbo::{BezPath, Point, Shape};
-use pdfcraft_content::{Matrix, Op, parse, serialize_ops};
-use pdfcraft_cos::{Dict, Document, ObjRef, Object, Stream};
+use pdfcraft_content::{Matrix, Op, Pieces, parse};
+use pdfcraft_cos::{Dict, Document, ObjRef, Object};
use std::sync::Arc;
use crate::EditError;
@@ -30,6 +30,8 @@ pub struct PageImage {
/// Pixel size; zero for vector/grouped Form artwork.
pub width: u32,
pub height: u32,
+ /// The content stream its `Do` (its first token) is in, and the `Do`'s index among the
+ /// operators of the page's streams joined ([`Pieces`]).
stream: usize,
op: usize,
/// CTM before `Do`, excluding a Form's own Matrix.
@@ -90,6 +92,11 @@ fn streams(doc: &Document, page: &Dict) -> Vec<(Object, Vec)> {
.collect()
}
+/// The page's streams joined into the one stream they are: an operator may span two of them.
+fn joined(streams: &[(Object, Vec)]) -> Pieces {
+ Pieces::join(&streams.iter().map(|(_, data)| data.as_slice()).collect::>())
+}
+
fn page_of(doc: &Document, page: usize) -> Result {
pdfcraft_model::pages(doc).into_iter().nth(page).ok_or(EditError::NoSuchPage(page))
}
@@ -214,72 +221,71 @@ pub fn page_images(doc: &Document, page: usize) -> Result, EditEr
let p = page_of(doc, page)?;
let xo = xobjects(doc, &p.dict);
let mut out = Vec::new();
- // The graphics state carries over from one stream to the next.
+ // The streams are one stream in pieces, so the graphics state and operators carry over.
let mut state = ImageState::new(Matrix::IDENTITY, p.crop(doc), Vec::new());
- for (si, (_, data)) in streams(doc, &p.dict).into_iter().enumerate() {
- for (i, op) in parse(&data).ops.iter().enumerate() {
- state.apply(op)?;
- let ctm = state.placement.matrix;
- if op.is("Do") {
- let Some(name) = op.name(0) else { continue };
- let Some(resource) = xo.get(name) else { continue };
- let obj = doc.resolve(resource);
- let Object::Stream(s) = &*obj else { continue };
- let is_form = s.dict.name(b"Subtype") == Some(b"Form");
- let (matrix, bounds) = if is_form {
- let Some(bounds) = s.dict.get(b"BBox").map(|v| doc.resolve(v)).and_then(|v| {
- v.as_array().and_then(|a| {
- if a.len() != 4 {
- return None;
- }
- let mut b = [0.0; 4];
- for (n, v) in b.iter_mut().zip(a) {
- *n = v.as_f64()?;
- }
- Some(b)
- })
- }) else {
- continue;
- };
- let matrix = match s.dict.get(b"Matrix") {
- Some(v) => {
- let v = doc.resolve(v);
- let Some(m) = v.as_array().and_then(|a| Matrix::from_operands(a)) else { continue };
- m
+ let pieces = joined(&streams(doc, &p.dict));
+ for (i, op) in pieces.parse().ops.iter().enumerate() {
+ state.apply(op)?;
+ let ctm = state.placement.matrix;
+ if op.is("Do") {
+ let Some(name) = op.name(0) else { continue };
+ let Some(resource) = xo.get(name) else { continue };
+ let obj = doc.resolve(resource);
+ let Object::Stream(s) = &*obj else { continue };
+ let is_form = s.dict.name(b"Subtype") == Some(b"Form");
+ let (matrix, bounds) = if is_form {
+ let Some(bounds) = s.dict.get(b"BBox").map(|v| doc.resolve(v)).and_then(|v| {
+ v.as_array().and_then(|a| {
+ if a.len() != 4 {
+ return None;
}
- None => Matrix::IDENTITY,
- };
- (matrix.then(&ctm), bounds)
- } else if s.dict.name(b"Subtype") == Some(b"Image") {
- (ctm, [0.0, 0.0, 1.0, 1.0])
- } else {
+ let mut b = [0.0; 4];
+ for (n, v) in b.iter_mut().zip(a) {
+ *n = v.as_f64()?;
+ }
+ Some(b)
+ })
+ }) else {
continue;
};
- let rect = matrix.bbox(bounds);
- if !bounds.iter().chain(matrix.0.iter()).chain(rect.iter()).all(|v| v.is_finite())
- || bounds[2] <= bounds[0]
- || bounds[3] <= bounds[1]
- || rect[2] <= rect[0]
- || rect[3] <= rect[1]
- {
- continue;
- }
- out.push(PageImage {
- rect,
- visible_rect: intersection(rect, state.placement.clip),
- clip_rect: Some(state.placement.clip),
- matrix: matrix.0,
- name: String::from_utf8_lossy(name).into_owned(),
- object: resource.as_ref(),
- is_form,
- width: if is_form { 0 } else { s.dict.int(b"Width").unwrap_or(0).clamp(0, u32::MAX as i64) as u32 },
- height: if is_form { 0 } else { s.dict.int(b"Height").unwrap_or(0).clamp(0, u32::MAX as i64) as u32 },
- stream: si,
- op: i,
- draw_matrix: ctm.0,
- clips: state.placement.clips.clone(),
- });
+ let matrix = match s.dict.get(b"Matrix") {
+ Some(v) => {
+ let v = doc.resolve(v);
+ let Some(m) = v.as_array().and_then(|a| Matrix::from_operands(a)) else { continue };
+ m
+ }
+ None => Matrix::IDENTITY,
+ };
+ (matrix.then(&ctm), bounds)
+ } else if s.dict.name(b"Subtype") == Some(b"Image") {
+ (ctm, [0.0, 0.0, 1.0, 1.0])
+ } else {
+ continue;
+ };
+ let rect = matrix.bbox(bounds);
+ if !bounds.iter().chain(matrix.0.iter()).chain(rect.iter()).all(|v| v.is_finite())
+ || bounds[2] <= bounds[0]
+ || bounds[3] <= bounds[1]
+ || rect[2] <= rect[0]
+ || rect[3] <= rect[1]
+ {
+ continue;
}
+ out.push(PageImage {
+ rect,
+ visible_rect: intersection(rect, state.placement.clip),
+ clip_rect: Some(state.placement.clip),
+ matrix: matrix.0,
+ name: String::from_utf8_lossy(name).into_owned(),
+ object: resource.as_ref(),
+ is_form,
+ width: if is_form { 0 } else { s.dict.int(b"Width").unwrap_or(0).clamp(0, u32::MAX as i64) as u32 },
+ height: if is_form { 0 } else { s.dict.int(b"Height").unwrap_or(0).clamp(0, u32::MAX as i64) as u32 },
+ stream: pieces.piece_of(op.span.start),
+ op: i,
+ draw_matrix: ctm.0,
+ clips: state.placement.clips.clone(),
+ });
}
}
Ok(out)
@@ -298,15 +304,15 @@ pub fn reading_images(doc: &Document, page: usize) -> Result, Edi
}
fn walk(
doc: &Document,
- data: &[u8],
+ ops: &[Op],
resources: &Dict,
st: &mut ImageState,
- stream: usize,
+ stream: &dyn Fn(&Op) -> usize,
forms: &mut Forms,
out: &mut Vec,
) -> Result<(), EditError> {
let xo = resources.get(b"XObject").map(|x| doc.resolve(x)).and_then(|x| x.as_dict().cloned()).unwrap_or_default();
- for (i, op) in parse(data).ops.iter().enumerate() {
+ for (i, op) in ops.iter().enumerate() {
st.apply(op)?;
let ctm = st.placement.matrix;
if op.is("Do") {
@@ -325,7 +331,8 @@ pub fn reading_images(doc: &Document, page: usize) -> Result, Edi
);
let clip = bounds.map_or(st.placement.clip, |b| intersection(st.placement.clip, b).unwrap_or([0.0; 4]));
let mut inner = ImageState::new(form.ctm, clip, st.placement.clips.clone());
- walk(doc, &form.data, &form.resources, &mut inner, stream, forms, out)?;
+ let piece = stream(op);
+ walk(doc, &parse(&form.data).ops, &form.resources, &mut inner, &|_| piece, forms, out)?;
if form.obj.is_some() {
forms.path.pop();
}
@@ -347,7 +354,7 @@ pub fn reading_images(doc: &Document, page: usize) -> Result, Edi
is_form: false,
width: s.dict.int(b"Width").unwrap_or(0).max(0) as u32,
height: s.dict.int(b"Height").unwrap_or(0).max(0) as u32,
- stream,
+ stream: stream(op),
op: i,
draw_matrix: ctm.0,
clips: st.placement.clips.clone(),
@@ -361,9 +368,8 @@ pub fn reading_images(doc: &Document, page: usize) -> Result, Edi
let mut out = Vec::new();
let mut st = ImageState::new(Matrix::IDENTITY, p.crop(doc), Vec::new());
let mut forms = Forms { path: Vec::new(), visits: 0 };
- for (si, (_, data)) in streams(doc, &p.dict).into_iter().enumerate() {
- walk(doc, &data, &res, &mut st, si, &mut forms, &mut out)?;
- }
+ let pieces = joined(&streams(doc, &p.dict));
+ walk(doc, &pieces.parse().ops, &res, &mut st, &|op| pieces.piece_of(op.span.start), &mut forms, &mut out)?;
Ok(out)
}
@@ -450,9 +456,9 @@ pub fn change_image(doc: &mut Document, page: usize, index: usize, change: &Imag
let img = images.get(index).cloned().ok_or_else(|| EditError::Invalid(format!("page {} has no image {}", page + 1, index + 1)))?;
let p = page_of(doc, page)?;
let all = streams(doc, &p.dict);
- let (obj, data) = all.get(img.stream).cloned().ok_or_else(|| EditError::Invalid("the image stream is missing".into()))?;
- let mut ops = parse(&data).ops;
- if ops.get(img.op).is_none() {
+ let pieces = joined(&all);
+ let ops = pieces.parse().ops;
+ if !ops.get(img.op).is_some_and(|op| op.is("Do")) {
return Err(EditError::Invalid("the image operation is missing".into()));
}
let mut resources = None;
@@ -479,16 +485,10 @@ pub fn change_image(doc: &mut Document, page: usize, index: usize, change: &Imag
vec![Op::new("Do", vec![Object::name(&name)])]
}
};
- ops.splice(img.op..=img.op, replacement);
- let mut dict = match &*doc.resolve(&obj) {
- Object::Stream(s) => s.dict.clone(),
- _ => Dict::new(),
- };
- dict.remove(b"Length");
- let new = doc.add(Object::Stream(Stream::flate(dict, &serialize_ops(&ops))));
- let contents: Vec = all.iter().enumerate().map(|(i, (o, _))| if i == img.stream { Object::Ref(new) } else { o.clone() }).collect();
+ let changed = crate::text::splice(&pieces, &ops, |index| if index == img.op { (replacement.clone(), false) } else { (Vec::new(), true) });
+ let contents = crate::text::rewritten_contents(doc, &all, changed);
doc.update_dict(p.obj, |d| {
- d.set(b"Contents".to_vec(), if let [one] = contents.as_slice() { one.clone() } else { Object::Array(contents) });
+ d.set(b"Contents".to_vec(), contents);
if let Some(res) = resources {
d.set(b"Resources".to_vec(), Object::Dict(res));
}
@@ -599,22 +599,12 @@ fn transform_image(
if !transform.iter().all(|v| v.is_finite() && v.abs() <= 1e12) {
return Err(EditError::Invalid("the image transform must be finite and no larger than 1e12".into()));
}
- let mut offset = 0_usize;
- for (_, data) in streams.iter().take(image.stream) {
- offset = offset
- .checked_add(data.len())
- .and_then(|n| n.checked_add(1))
- .ok_or_else(|| EditError::Invalid("the image content is too large".into()))?;
- }
- let data = streams.get(image.stream).ok_or_else(|| EditError::Invalid("the image stream is missing".into()))?;
- let op = parse(&data.1).ops.get(image.op).cloned().ok_or_else(|| EditError::Invalid("the image operation is missing".into()))?;
- let span = offset.checked_add(op.span.start).ok_or_else(|| EditError::Invalid("the image content is too large".into()))?;
- let pieces = pdfcraft_content::Pieces::join(&streams.iter().map(|(_, data)| data.as_slice()).collect::>());
+ let pieces = joined(streams);
let ops = pieces.parse().ops;
- let target = ops
- .iter()
- .position(|op| op.span.start == span && op.is("Do"))
- .ok_or_else(|| EditError::Invalid("the image operation could not be located".into()))?;
+ let target = image.op;
+ if !ops.get(target).is_some_and(|op| op.is("Do")) {
+ return Err(EditError::Invalid("the image operation could not be located".into()));
+ }
// A private path crop moves with its image regardless of intervening state operators.
// Shared crops retain the existing fixed-clip behavior, preserving neighbouring paint.
let (start, end, placement) = image_transform_scope(&ops, target)?.unwrap_or((target, target, Matrix(image.draw_matrix)));
@@ -677,22 +667,12 @@ fn crop_image(
if x1 - x0 < 1e-6 || y1 - y0 < 1e-6 {
return Err(EditError::Invalid("the crop box has no area".into()));
}
- let mut offset = 0_usize;
- for (_, data) in streams.iter().take(image.stream) {
- offset = offset
- .checked_add(data.len())
- .and_then(|n| n.checked_add(1))
- .ok_or_else(|| EditError::Invalid("the image content is too large".into()))?;
- }
- let data = streams.get(image.stream).ok_or_else(|| EditError::Invalid("the image stream is missing".into()))?;
- let op = parse(&data.1).ops.get(image.op).cloned().ok_or_else(|| EditError::Invalid("the image operation is missing".into()))?;
- let span = offset.checked_add(op.span.start).ok_or_else(|| EditError::Invalid("the image content is too large".into()))?;
- let pieces = pdfcraft_content::Pieces::join(&streams.iter().map(|(_, data)| data.as_slice()).collect::>());
+ let pieces = joined(streams);
let ops = pieces.parse().ops;
- let target = ops
- .iter()
- .position(|op| op.span.start == span && op.is("Do"))
- .ok_or_else(|| EditError::Invalid("the image operation could not be located".into()))?;
+ let target = image.op;
+ if !ops.get(target).is_some_and(|op| op.is("Do")) {
+ return Err(EditError::Invalid("the image operation could not be located".into()));
+ }
let (start, end, placement) = image_transform_scope(&ops, target)?.unwrap_or((target, target, Matrix(image.draw_matrix)));
let inverse = placement.invert().ok_or_else(|| EditError::Invalid("the image has no area".into()))?;
let corners = [(x0, y0), (x1, y0), (x1, y1), (x0, y1)].map(|(x, y)| inverse.apply(x, y));
@@ -758,76 +738,72 @@ pub(crate) fn translate_images(doc: &mut Document, page: usize, indexes: &[usize
if !movement.0.iter().all(|v| v.is_finite() && v.abs() <= 1e12) {
return Err(EditError::Invalid("the image move is too large".into()));
}
- plans.insert((image.stream, image.op), (movement, placement, offset));
+ plans.insert(image.op, (movement, placement, offset));
}
- let mut data = Vec::with_capacity(all.len());
+ let pieces = joined(&all);
+ let ops = pieces.parse().ops;
let mut inside_text = false;
- for (stream, (_, bytes)) in all.iter().enumerate() {
- let pieces = pdfcraft_content::Pieces::join(&[bytes.as_slice()]);
- let ops = pieces.parse().ops;
- for (index, op) in ops.iter().enumerate() {
- if op.is("BT") {
- inside_text = true;
- } else if op.is("ET") {
- inside_text = false;
- } else if inside_text && plans.contains_key(&(stream, index)) {
- return Err(EditError::Invalid("an image inside a text object cannot be moved in a group".into()));
- }
+ for (index, op) in ops.iter().enumerate() {
+ if op.is("BT") {
+ inside_text = true;
+ } else if op.is("ET") {
+ inside_text = false;
+ } else if inside_text && plans.contains_key(&index) {
+ return Err(EditError::Invalid("an image inside a text object cannot be moved in a group".into()));
}
- // Images alone in their own `q re W n cm Do Q` take the clip with them.
- let mut clips = std::collections::BTreeMap::new();
- for index in 0..ops.len() {
- let Some(&(_, placement, offset)) = plans.get(&(stream, index)) else { continue };
- let Some((at, cm)) = exclusive_clip(&ops, index) else { continue };
- let translation = Matrix([1.0, 0.0, 0.0, 1.0, offset[0], offset[1]]);
- let Some(c) = clip_matrix(placement, cm, translation).filter(|c| c.0.iter().all(|v| v.is_finite() && v.abs() <= 1e12)) else {
- return Err(EditError::Invalid("the image move is too large".into()));
- };
- plans.remove(&(stream, index));
- // A `cm` an earlier move put before the clip takes this move too: `c · prev` is the
- // same CTM as `prev` followed by another `cm c`, without one more operator per move.
- let folded = at
- .checked_sub(1)
- .and_then(|i| Some((i, c.then(&Matrix(ops.get(i).filter(|op| op.is("cm"))?.nums::<6>()?)))))
- .filter(|(_, m)| m.0.iter().all(|v| v.is_finite() && v.abs() <= 1e12));
- match folded {
- Some((i, m)) => clips.insert((stream, i), (m, false)),
- None => clips.insert((stream, at), (c, true)),
- };
- }
- // An image already alone in `q cm Do Q` moves by changing that `cm`, not by nesting another.
- let mut folds = std::collections::BTreeMap::new();
- for index in 0..ops.len() {
- let Some(&(movement, _, _)) = plans.get(&(stream, index)) else { continue };
- let Some((at, cm)) = enclosing_transform(&ops, index, index) else { continue };
- let folded = movement.then(&cm);
- if !folded.0.iter().all(|v| v.is_finite() && v.abs() <= 1e12) {
- return Err(EditError::Invalid("the image move is too large".into()));
- }
- plans.remove(&(stream, index));
- folds.insert(at, folded);
- }
- let changed = crate::text::splice(&pieces, &ops, |index| {
- if let Some((c, keep)) = clips.remove(&(stream, index)) {
- return (vec![Op::new("cm", c.0.map(Object::Real).to_vec())], keep);
- }
- if let Some(m) = folds.remove(&index) {
- return (vec![Op::new("cm", m.0.map(Object::Real).to_vec())], false);
- }
- let Some((matrix, _, _)) = plans.remove(&(stream, index)) else { return (Vec::new(), true) };
- let original = ops.get(index).cloned().into_iter();
- // Local translations can be tiny under a large CTM. Preserve their precision
- // rather than rounding them to zero or amplifying an error on the page.
- let mut out = vec![Op::new("q", vec![]), Op::new("cm", matrix.0.map(Object::Real).to_vec())];
- out.extend(original);
- out.push(Op::new("Q", vec![]));
- (out, false)
- });
- data.push(changed.into_iter().next().unwrap_or_default());
}
- if !plans.is_empty() {
+ // Images alone in their own `q re W n cm Do Q` take the clip with them.
+ let mut clips = std::collections::BTreeMap::new();
+ for index in 0..ops.len() {
+ let Some(&(_, placement, offset)) = plans.get(&index) else { continue };
+ let Some((at, cm)) = exclusive_clip(&ops, index) else { continue };
+ let translation = Matrix([1.0, 0.0, 0.0, 1.0, offset[0], offset[1]]);
+ let Some(c) = clip_matrix(placement, cm, translation).filter(|c| c.0.iter().all(|v| v.is_finite() && v.abs() <= 1e12)) else {
+ return Err(EditError::Invalid("the image move is too large".into()));
+ };
+ plans.remove(&index);
+ // A `cm` an earlier move put before the clip takes this move too: `c · prev` is the
+ // same CTM as `prev` followed by another `cm c`, without one more operator per move.
+ let folded = at
+ .checked_sub(1)
+ .and_then(|i| Some((i, c.then(&Matrix(ops.get(i).filter(|op| op.is("cm"))?.nums::<6>()?)))))
+ .filter(|(_, m)| m.0.iter().all(|v| v.is_finite() && v.abs() <= 1e12));
+ match folded {
+ Some((i, m)) => clips.insert(i, (m, false)),
+ None => clips.insert(at, (c, true)),
+ };
+ }
+ // An image already alone in `q cm Do Q` moves by changing that `cm`, not by nesting another.
+ let mut folds = std::collections::BTreeMap::new();
+ for index in 0..ops.len() {
+ let Some(&(movement, _, _)) = plans.get(&index) else { continue };
+ let Some((at, cm)) = enclosing_transform(&ops, index, index) else { continue };
+ let folded = movement.then(&cm);
+ if !folded.0.iter().all(|v| v.is_finite() && v.abs() <= 1e12) {
+ return Err(EditError::Invalid("the image move is too large".into()));
+ }
+ plans.remove(&index);
+ folds.insert(at, folded);
+ }
+ if plans.keys().any(|index| !ops.get(*index).is_some_and(|op| op.is("Do"))) {
return Err(EditError::Invalid("a selected image operation is missing".into()));
}
+ let data = crate::text::splice(&pieces, &ops, |index| {
+ if let Some((c, keep)) = clips.remove(&index) {
+ return (vec![Op::new("cm", c.0.map(Object::Real).to_vec())], keep);
+ }
+ if let Some(m) = folds.remove(&index) {
+ return (vec![Op::new("cm", m.0.map(Object::Real).to_vec())], false);
+ }
+ let Some((matrix, _, _)) = plans.remove(&index) else { return (Vec::new(), true) };
+ let original = ops.get(index).cloned().into_iter();
+ // Local translations can be tiny under a large CTM. Preserve their precision
+ // rather than rounding them to zero or amplifying an error on the page.
+ let mut out = vec![Op::new("q", vec![]), Op::new("cm", matrix.0.map(Object::Real).to_vec())];
+ out.extend(original);
+ out.push(Op::new("Q", vec![]));
+ (out, false)
+ });
let contents = crate::text::rewritten_contents(doc, &all, data);
doc.update_dict(p.obj, |d| d.set(b"Contents".to_vec(), contents))?;
Ok(())
diff --git a/crates/edit/src/lib.rs b/crates/edit/src/lib.rs
index cef8f3c9d2..fcf2f19975 100644
--- a/crates/edit/src/lib.rs
+++ b/crates/edit/src/lib.rs
@@ -8,7 +8,8 @@
//! `/PCMark` entry (also in the stream dictionary) names the kind of mark, which is how Update and
//! Remove find them again without decompressing page content. When
//! anything is drawn on top, the page's original content is wrapped in `q … Q` (two tiny streams)
-//! so its graphics state can't leak into the mark.
+//! so its graphics state cannot reach the mark. The closing stream also ends known unfinished
+//! marked-content sequences before the new mark.
//!
//! Text uses standard Helvetica (WinAnsi) added to the page resources as `/PCHelv`.
@@ -332,12 +333,16 @@ const WRAP_OPEN: &[u8] = b"q %PdfCraft\n";
const WRAP_CLOSE: &[u8] = b"\nQ %PdfCraft\n";
const OLD_WRAP_CLOSE: &[u8] = b"Q %PdfCraft\n";
-/// Whether `bytes` are a wrapper's close: [`WRAP_CLOSE`] after a `Q` for each `q` the wrapped
-/// content left open, or [`OLD_WRAP_CLOSE`].
+/// Whether `bytes` are a wrapper's close: canonical `EMC` closures, then the existing
+/// graphics closures. Legacy closes without the leading newline remain valid on their own.
fn is_wrap_close(bytes: &[u8]) -> bool {
let mut rest = bytes;
+ while let Some(shorter) = rest.strip_prefix(b"\nEMC".as_slice()) {
+ rest = shorter;
+ }
+ let legacy = rest.len() == bytes.len();
loop {
- if rest == WRAP_CLOSE || rest == OLD_WRAP_CLOSE {
+ if rest == WRAP_CLOSE || (legacy && rest == OLD_WRAP_CLOSE) {
return true;
}
match rest.strip_prefix(b"\nQ".as_slice()) {
@@ -387,11 +392,13 @@ enum Isolation {
/// `suffix`: the stream from which the content leaves graphics state set (or a comment open)
/// that a mark on top must not inherit, so it needs a wrapper from there to the end.
/// `wrappers`: PdfCraft wrapper pairs (open, close) as stream indexes.
- Clean { suffix: Option, wrappers: Vec<(usize, usize)> },
+ /// `marked`: unfinished marked-content sequences to close before a new mark.
+ Clean { suffix: Option, wrappers: Vec<(usize, usize)>, marked: usize },
/// The content could not be read that precisely (it doesn't decode, is too large, or has
/// stray bytes, unbalanced `q`/`Q` or an unfinished text object or path). `unclosed` is
- /// the count of `q` left open at the end, when known.
- Unknown { unclosed: usize },
+ /// the count of `q` left open at the end, when known. `marked` counts known unfinished
+ /// marked-content sequences independently of graphics malformation, or zero when unknown.
+ Unknown { unclosed: usize, marked: usize },
}
const ISOLATION_MAX_STREAMS: usize = 100_000;
@@ -400,7 +407,7 @@ const ISOLATION_MAX_OPS: usize = 1_000_000;
const ISOLATION_MAX_DEPTH: usize = 10_000;
fn isolation(doc: &Document, list: &[Object]) -> Isolation {
- let unknown = Isolation::Unknown { unclosed: 0 };
+ let unknown = Isolation::Unknown { unclosed: 0, marked: 0 };
if list.len() > ISOLATION_MAX_STREAMS {
return unknown;
}
@@ -422,14 +429,17 @@ fn isolation(doc: &Document, list: &[Object]) -> Isolation {
let edges = pieces.comment_edges(&parsed);
let (mut text_open, mut path_open, mut dirty, mut malformed) = (false, false, false, false);
let mut stack: Vec = Vec::new();
+ let (mut marked, mut marked_known) = (0usize, true);
let mut split = 0;
let mut previous_last = None;
let mut wrappers = Vec::new();
for op in &parsed.ops {
let (first, last) = pieces.pieces_of(op);
- // A wrapper can only go in before a whole stream, at the top level, between operators,
- // and where no comment runs into the stream (it would swallow the wrapper's `q`).
+ // A wrapper can only go in before a whole stream, outside graphics and marked-content
+ // scopes, between operators, where no comment would swallow the wrapper's `q`.
if stack.is_empty()
+ && marked_known
+ && marked == 0
&& !text_open
&& !path_open
&& !dirty
@@ -439,6 +449,26 @@ fn isolation(doc: &Document, list: &[Object]) -> Isolation {
split = first;
}
previous_last = Some(last);
+ // Match the reader's tolerant operands: BMC uses the last name, BDC the last
+ // name/object pair, and EMC ignores surplus operands. Invalid begins are ignored.
+ let begin_marked = match op.op.as_slice() {
+ b"BMC" => op.operands.last().is_some_and(|o| matches!(o, Object::Name(_))),
+ b"BDC" => op.operands.len().checked_sub(2).and_then(|i| op.operands.get(i)).is_some_and(|o| matches!(o, Object::Name(_))),
+ _ => false,
+ };
+ if marked_known {
+ if begin_marked {
+ if marked < ISOLATION_MAX_DEPTH {
+ marked += 1;
+ } else {
+ // Once the budget is exceeded, later EMCs cannot restore a known depth.
+ marked_known = false;
+ marked = 0;
+ }
+ } else if op.is("EMC") {
+ marked = marked.saturating_sub(1);
+ }
+ }
match op.op.as_slice() {
b"BT" if !text_open => text_open = true,
b"ET" if text_open => text_open = false,
@@ -468,11 +498,11 @@ fn isolation(doc: &Document, list: &[Object]) -> Isolation {
_ => {}
}
}
- if malformed || text_open || path_open || !stack.is_empty() {
- return Isolation::Unknown { unclosed: if malformed { 0 } else { stack.len() } };
+ if malformed || text_open || path_open || !stack.is_empty() || !marked_known {
+ return Isolation::Unknown { unclosed: if malformed { 0 } else { stack.len() }, marked };
}
let comment_open = edges.last().is_some_and(|edge| edge.1);
- Isolation::Clean { suffix: (dirty || comment_open).then_some(split), wrappers }
+ Isolation::Clean { suffix: (dirty || comment_open || marked != 0).then_some(split), wrappers, marked }
}
/// Put a mark's content on a page: behind (prepended) or on top (appended, after wrapping the
@@ -484,21 +514,22 @@ fn place(doc: &mut Document, page: &pdfcraft_model::Page, kind: MarkKind, conten
/// Put content on a page, tagged `tag` (see [`place`]). On top, whatever graphics state the
/// content before leaves set (a `cm`, a clip, text state…) must not reach the mark, including
/// content appended after an earlier PdfCraft wrapper by another tool (#3084): the stretch of
-/// content that leaves state set is wrapped in q/Q, once.
+/// content that leaves state set is wrapped in q/Q, once. Any known unfinished marked-content
+/// sequences are closed at the end too, so hidden optional content cannot hide the new mark.
fn place_tagged(doc: &mut Document, page: &pdfcraft_model::Page, tag: &str, content: Vec, behind: bool) -> Result<(), EditError> {
let mut list = contents(doc, page)?;
let wrap = if behind {
None
} else {
match isolation(doc, &list) {
- Isolation::Clean { suffix, .. } => suffix.map(|split| (split, 0)),
+ Isolation::Clean { suffix, marked, .. } => suffix.map(|split| (split, 0, marked)),
// Content that can't be read precisely is wrapped whole, as before: after the
// behind-marks at the start. A `q` it leaves open is closed at its end, where
// closing it changes nothing already drawn.
- Isolation::Unknown { unclosed } => {
+ Isolation::Unknown { unclosed, marked } => {
let behind_kinds = [MarkKind::Background.tag(), MarkKind::Watermark.tag()];
let split = list.iter().position(|o| !stream_tag(doc, o).is_some_and(|t| behind_kinds.contains(&t.as_str()))).unwrap_or(list.len());
- (split < list.len()).then_some((split, unclosed))
+ (split < list.len()).then_some((split, unclosed, marked))
}
}
};
@@ -506,8 +537,9 @@ fn place_tagged(doc: &mut Document, page: &pdfcraft_model::Page, tag: &str, cont
if behind {
list.insert(0, mark);
} else {
- if let Some((split, unclosed)) = wrap {
- let mut close_bytes = b"\nQ".repeat(unclosed);
+ if let Some((split, unclosed, marked)) = wrap {
+ let mut close_bytes = b"\nEMC".repeat(marked);
+ close_bytes.extend_from_slice(&b"\nQ".repeat(unclosed));
close_bytes.extend_from_slice(WRAP_CLOSE);
let open = Object::Ref(doc.add(Object::Stream(Stream::from_raw(tagged("Wrap"), WRAP_OPEN.to_vec()))));
let close = Object::Ref(doc.add(Object::Stream(Stream::from_raw(tagged("Wrap"), close_bytes))));
@@ -836,5 +868,7 @@ pub use added::{
update_content,
};
+#[cfg(test)]
+mod marked_isolation_tests;
#[cfg(test)]
mod tests;
diff --git a/crates/edit/src/marked_isolation_tests.rs b/crates/edit/src/marked_isolation_tests.rs
new file mode 100644
index 0000000000..6da8322296
--- /dev/null
+++ b/crates/edit/src/marked_isolation_tests.rs
@@ -0,0 +1,122 @@
+use super::*;
+
+fn fixture(parts: &[&[u8]]) -> Document {
+ let mut doc = Document::new_empty();
+ let pages = doc.get(doc.root().unwrap()).as_dict().unwrap().reference(b"Pages").unwrap();
+ let streams = parts.iter().map(|data| Object::Ref(doc.add(Object::Stream(Stream::from_raw(Dict::new(), data.to_vec()))))).collect();
+ let mut page = Dict::new();
+ page.set(b"Type".to_vec(), Object::name("Page"));
+ page.set(b"Parent".to_vec(), Object::Ref(pages));
+ page.set(b"MediaBox".to_vec(), Object::Array(vec![0.into(), 0.into(), 600.into(), 800.into()]));
+ page.set(b"Contents".to_vec(), Object::Array(streams));
+ let page = doc.add(page);
+ doc.update_dict(pages, |d| {
+ d.set(b"Kids".to_vec(), Object::Array(vec![Object::Ref(page)]));
+ d.set(b"Count".to_vec(), Object::Int(1));
+ })
+ .unwrap();
+ doc
+}
+
+fn placed(parts: &[&[u8]]) -> (Document, Vec) {
+ let mut doc = fixture(parts);
+ stamp(&mut doc, 0, "Probe", b"q Q\n".to_vec()).unwrap();
+ let page = page_list(&doc).remove(0);
+ let list = contents(&mut doc, &page).unwrap();
+ (doc, list)
+}
+
+fn close(parts: &[&[u8]]) -> Vec {
+ let (doc, list) = placed(parts);
+ let object = &list[list.len() - 2];
+ assert_eq!(stream_tag(&doc, object).as_deref(), Some("Wrap"));
+ let bytes = stream_bytes(&doc, object).unwrap();
+ assert_eq!(wrapper_role(&doc, object), Some(false));
+ bytes
+}
+
+#[test]
+fn closes_only_marked_operators_not_strings_names_comments_or_image_samples() {
+ for tail in
+ [b"/Span << /Text (BMC BDC EMC) >> DP".as_slice(), b"/BMC /EMC DP", b"% BMC BDC EMC", b"BI /W 11 /H 1 /BPC 8 /CS /G ID BMC BDC EMC EI\n"]
+ {
+ let mut input = b"/OC /L1 BDC\n".to_vec();
+ input.extend_from_slice(tail);
+ assert_eq!(close(&[&input]), b"\nEMC\nQ %PdfCraft\n", "{tail:?}");
+ }
+}
+
+#[test]
+fn safe_suffix_stays_outside_marked_content_across_balanced_graphics() {
+ let (doc, list) = placed(&[b"q /OC /L1 BDC Q", b"50 Tz"]);
+ assert_eq!(stream_bytes(&doc, &list[0]).unwrap(), WRAP_OPEN);
+ assert_eq!(stream_bytes(&doc, &list[1]).unwrap(), b"q /OC /L1 BDC Q");
+ assert_eq!(stream_bytes(&doc, &list[list.len() - 2]).unwrap(), b"\nEMC\nQ %PdfCraft\n");
+ assert_eq!(close(&[b"q /OC /L1 BDC Q"]), b"\nEMC\nQ %PdfCraft\n");
+}
+
+#[test]
+fn known_marked_depth_survives_unknown_graphics_and_tolerant_operands() {
+ for (input, expected) in [
+ (b"/OC /L1 BDC q".as_slice(), b"\nEMC\nQ\nQ %PdfCraft\n".as_slice()),
+ (b"Q /OC /L1 BDC", b"\nEMC\nQ %PdfCraft\n"),
+ (b"/OC /L1 BDC BT", b"\nEMC\nQ %PdfCraft\n"),
+ (b"/OC /L1 BDC 0 0 m", b"\nEMC\nQ %PdfCraft\n"),
+ (b"/OC /L1 BDC 123 BMC /Artifact BDC", b"\nEMC\nQ %PdfCraft\n"),
+ (b"EMC 123 /OC /L1 BDC 123 /Artifact BMC", b"\nEMC\nEMC\nQ %PdfCraft\n"),
+ (b"/OC /L1 BDC 123 EMC", b"\nQ %PdfCraft\n"),
+ (b"/Span 123 BDC", b"\nEMC\nQ %PdfCraft\n"),
+ ] {
+ assert_eq!(close(&[input]), expected, "{input:?}");
+ }
+}
+
+#[test]
+fn closes_cross_stream_begins_without_closing_balanced_sequences_twice() {
+ assert_eq!(close(&[b"/OC", b"/L1 BDC"]), b"\nEMC\nQ %PdfCraft\n");
+ assert_eq!(close(&[b"/OC /L1 BDC", b"EMC"]), WRAP_CLOSE);
+ assert_eq!(close(&[b"/OC /L1 BDC % EMC", b"EMC\n"]), b"\nEMC\nQ %PdfCraft\n");
+}
+
+#[test]
+fn recognizes_only_canonical_new_closes_and_retains_legacy_closes() {
+ for bytes in [WRAP_CLOSE, OLD_WRAP_CLOSE, b"\nQ\nQ %PdfCraft\n", b"\nEMC\nQ %PdfCraft\n", b"\nEMC\nEMC\nQ\nQ %PdfCraft\n"] {
+ assert!(is_wrap_close(bytes), "{bytes:?}");
+ }
+ for bytes in [
+ b"\nEMCQ %PdfCraft\n".as_slice(),
+ b"\nEMC\nQQ %PdfCraft\n",
+ b"\nQ\nEMC\nQ %PdfCraft\n",
+ b"EMC\nQ %PdfCraft\n",
+ b"\nEMC 1 w\nQ %PdfCraft\n",
+ b"\nEMC\nQ %PdfCraft\nq Q",
+ ] {
+ assert!(!is_wrap_close(bytes), "{bytes:?}");
+ }
+}
+
+#[test]
+fn depth_budget_bounds_emitted_closures_and_overflow_stays_unknown() {
+ let maximum = b"/Artifact BMC\n".repeat(ISOLATION_MAX_DEPTH);
+ let mut expected = b"\nEMC".repeat(ISOLATION_MAX_DEPTH);
+ expected.extend_from_slice(WRAP_CLOSE);
+ assert_eq!(close(&[&maximum]), expected);
+ let mut combined = maximum.clone();
+ combined.extend_from_slice(&b"q\n".repeat(ISOLATION_MAX_DEPTH));
+ let mut expected = b"\nEMC".repeat(ISOLATION_MAX_DEPTH);
+ expected.extend_from_slice(&b"\nQ".repeat(ISOLATION_MAX_DEPTH));
+ expected.extend_from_slice(WRAP_CLOSE);
+ assert_eq!(close(&[&combined]), expected);
+ assert!(expected.len() < 64 * 1024, "maximum generated close remains recognizable for removal");
+ let mut doc = fixture(&[&combined]);
+ let page = page_list(&doc).remove(0);
+ let original = contents(&mut doc, &page).unwrap();
+ place_tagged(&mut doc, &page, MarkKind::HeaderFooter.tag(), b"q Q\n".to_vec(), false).unwrap();
+ assert_eq!(remove_marks(&mut doc, &[0], MarkKind::HeaderFooter).unwrap(), 1);
+ assert_eq!(contents(&mut doc, &page).unwrap(), original, "maximum combined close can be removed without nesting wrappers");
+ for suffix in [b"/Artifact BMC\n".as_slice(), b"/Artifact BMC\nEMC\n", b"/Artifact BMC\nEMC\nEMC\n"] {
+ let mut over = maximum.clone();
+ over.extend_from_slice(suffix);
+ assert_eq!(close(&[&over]), WRAP_CLOSE, "overflow cannot resume speculative depth tracking after EMC");
+ }
+}
diff --git a/crates/edit/src/tests.rs b/crates/edit/src/tests.rs
index 299356b293..f2def2fd6a 100644
--- a/crates/edit/src/tests.rs
+++ b/crates/edit/src/tests.rs
@@ -1473,6 +1473,65 @@ fn skia_subset_page() -> Document {
doc
}
+/// #2616: a subset whose ToUnicode has no space (as LaTeX and Sphinx output often has, spacing
+/// words with TJ kerning) makes finding the space width read every font on the page. That ran
+/// for every character of a multi-row draft, so each keystroke in a long paragraph took half a
+/// second. It runs once per font now, however long the paragraph.
+#[test]
+fn a_long_draft_in_a_font_without_a_space_works_out_the_space_width_once() {
+ // Two rows of one paragraph (14 pt apart) in the Chrome-style subset of `skia_subset_page`,
+ // whose ToUnicode here has no entry for the space glyph.
+ let mut doc =
+ text_page("BT /F3 12 Tf 72 700 Td <0177031602E001FB03050255> Tj ET BT /F3 12 Tf 72 686 Td <008D0266035402FE01FB030503E6> Tj ET");
+ let program = doc.add(Object::Stream(Stream::flate(Dict::new(), include_bytes!("../../../assets/fonts/Inter-Regular.ttf"))));
+ let unicode = doc.add(Object::Stream(Stream::flate(
+ Dict::new(),
+ b"1 begincodespacerange <0000> endcodespacerange 12 beginbfchar <0177> <0050> <0316> <006F> <02E0> <006C> <01FB> <0061> <0305> <006E> <0255> <0064> <008D> <0047> <0266> <0065> <0354> <0072> <02FE> <006D> <03E6> <0079> <1000> <007A> endbfchar",
+ )));
+ let widths: Vec =
+ [(0x8D, 746), (0x177, 639), (0x1FB, 551), (0x255, 612), (0x266, 583), (0x2E0, 242), (0x2FE, 876), (0x305, 591), (0x316, 600)]
+ .into_iter()
+ .chain([(0x354, 376), (0x3E6, 515)])
+ .flat_map(|(cid, width)| [Object::Int(cid), Object::Array(vec![Object::Int(width)])])
+ .collect();
+ let descriptor = Dict::from_iter([
+ (b"Type".to_vec(), Object::name("FontDescriptor")),
+ (b"FontName".to_vec(), Object::name("BAAAAA+SampleSans")),
+ (b"Flags".to_vec(), Object::Int(32)),
+ (b"FontFile2".to_vec(), Object::Ref(program)),
+ ]);
+ let cid_font = Dict::from_iter([
+ (b"Type".to_vec(), Object::name("Font")),
+ (b"Subtype".to_vec(), Object::name("CIDFontType2")),
+ (b"BaseFont".to_vec(), Object::name("BAAAAA+SampleSans")),
+ (b"CIDToGIDMap".to_vec(), Object::name("Identity")),
+ (b"DW".to_vec(), Object::Int(562)),
+ (b"W".to_vec(), Object::Array(widths)),
+ (b"FontDescriptor".to_vec(), Object::Dict(descriptor)),
+ ]);
+ let font = Dict::from_iter([
+ (b"Type".to_vec(), Object::name("Font")),
+ (b"Subtype".to_vec(), Object::name("Type0")),
+ (b"BaseFont".to_vec(), Object::name("BAAAAA+SampleSans")),
+ (b"Encoding".to_vec(), Object::name("Identity-H")),
+ (b"DescendantFonts".to_vec(), Object::Array(vec![Object::Dict(cid_font)])),
+ (b"ToUnicode".to_vec(), Object::Ref(unicode)),
+ ]);
+ let p = pdfcraft_model::pages(&doc)[0].obj;
+ let fonts = Dict::from_iter([(b"F3".to_vec(), Object::Dict(font))]);
+ doc.update_dict(p, |d| d.set(b"Resources".to_vec(), Object::Dict(Dict::from_iter([(b"Font".to_vec(), Object::Dict(fonts))])))).unwrap();
+ let blocks = text::text_blocks(&doc, 0).unwrap();
+ assert_eq!(blocks.len(), 1, "one paragraph: {:?}", blocks.iter().map(|b| &b.text).collect::>());
+ let draft = "Poland and Germany
+"
+ .repeat(20);
+ text::SOURCE_SPACE_CALLS.with(|c| c.set(0));
+ let layout = text::text_edit_layout(&doc, 0, 0, draft.trim_end(), &text::BlockStyle::default()).unwrap();
+ let calls = text::SOURCE_SPACE_CALLS.with(std::cell::Cell::get);
+ assert!(layout.draft_rect.is_some_and(|r| r[2] > r[0]), "laid out: {:?}", layout.draft_rect);
+ assert!(calls <= 8, "the space width was worked out {calls} times for {} characters", draft.len());
+}
+
#[test]
fn a_subset_keeps_its_font_for_glyphs_left_to_the_default_width() {
let doc = skia_subset_page();
@@ -2246,6 +2305,72 @@ fn replacing_a_line_keeps_a_distinct_nearby_line() {
}
}
+#[test]
+fn editing_text_keeps_an_overlapping_watermark_and_covered_paragraphs() {
+ let listed = |doc: &Document| text::text_lines(doc, 0).unwrap().into_iter().map(|l| l.text).collect::>();
+ // #1756: a rotated watermark's bounding box covers the middle of the page, and a body line
+ // whose baseline happens to sit near the watermark's starting y is still unrelated text.
+ for rotation in [35.0, 45.0, 0.0] {
+ let mut base = text_page("BT /F1 14 Tf 60 700 Td (Heading) Tj ET");
+ add_watermark(
+ &mut base,
+ &[0],
+ &Watermark { text: "TEST WATERMARK".into(), font_size: 48.0, rotation, opacity: 0.25, ..Watermark::default() },
+ false,
+ )
+ .unwrap();
+ let wm_y = text::text_lines(&base, 0).unwrap().into_iter().find(|l| l.text == "TEST WATERMARK").unwrap().origin_baseline();
+ let mut doc = text_page(&format!("BT /F1 14 Tf 1 0 0 1 60 {wm_y:.2} Tm (Body paragraph one with several words here) Tj ET"));
+ add_watermark(
+ &mut doc,
+ &[0],
+ &Watermark { text: "TEST WATERMARK".into(), font_size: 48.0, rotation, opacity: 0.25, ..Watermark::default() },
+ false,
+ )
+ .unwrap();
+ let line_idx = text::text_lines(&doc, 0).unwrap().iter().position(|l| l.text.starts_with("Body")).unwrap();
+ let mut line_edited = doc.clone();
+ text::replace_line(&mut line_edited, 0, line_idx, "Edited body line").unwrap();
+ assert_eq!(listed(&reopen(&line_edited)), ["Edited body line", "TEST WATERMARK"], "rotation {rotation}");
+
+ let block_idx = text::text_blocks(&doc, 0).unwrap().iter().position(|b| b.text.starts_with("Body")).unwrap();
+ let mut preview = doc.clone();
+ text::clear_block(&mut preview, 0, block_idx).unwrap();
+ assert_eq!(listed(&preview), ["TEST WATERMARK"], "clear_block keeps the watermark at rotation {rotation}");
+
+ text::rewrite_block(&mut doc, 0, block_idx, Some("Edited body paragraph"), &text::BlockStyle::default()).unwrap();
+ assert_eq!(listed(&reopen(&doc)), ["Edited body paragraph", "TEST WATERMARK"], "rotation {rotation}");
+ }
+
+ // #1368: dragging one paragraph onto another paragraph's baseline and moving it on again
+ // leaves the covered paragraph intact.
+ let mut doc = text_page("BT /F1 12 Tf 72 700 Td (First paragraph) Tj ET BT /F1 12 Tf 72 600 Td (Second paragraph) Tj ET");
+ text::rewrite_block(&mut doc, 0, 0, None, &text::BlockStyle { offset: Some([0.0, -100.0]), ..Default::default() }).unwrap();
+ let second = text::text_blocks(&doc, 0).unwrap().iter().position(|b| b.text == "Second paragraph");
+ assert!(second.is_some(), "the covered paragraph stays on the page: {:?}", listed(&doc));
+ let first = text::text_blocks(&doc, 0).unwrap().iter().position(|b| b.text == "First paragraph").unwrap();
+ text::rewrite_block(&mut doc, 0, first, None, &text::BlockStyle { offset: Some([0.0, -100.0]), ..Default::default() }).unwrap();
+ assert_eq!(listed(&reopen(&doc)), ["First paragraph", "Second paragraph"]);
+
+ // Even when a large horizontal watermark shows the same word as a body line at the same y,
+ // or a rotated 12 pt line crosses a horizontal 12 pt line of the same word at the same origin,
+ // editing one keeps the other.
+ let mut same_word = text_page("BT /F1 12 Tf 1 0 0 1 200 400 Tm (DRAFT) Tj ET BT /F1 48 Tf 1 0 0 1 150 400 Tm (DRAFT) Tj ET");
+ text::replace_line(&mut same_word, 0, 0, "FINAL").unwrap();
+ let after = text::text_lines(&reopen(&same_word), 0).unwrap();
+ assert_eq!(after.iter().map(|l| (&l.text[..], l.size as i32)).collect::>(), [("FINAL", 12), ("DRAFT", 48)]);
+
+ let mut crossed = text_page("BT /F1 12 Tf 1 0 0 1 100 500 Tm (Crossed) Tj ET BT /F1 12 Tf 0.707 0.707 -0.707 0.707 100 500 Tm (Crossed) Tj ET");
+ text::replace_line(&mut crossed, 0, 0, "Horizontal").unwrap();
+ assert_eq!(listed(&reopen(&crossed)), ["Horizontal", "Crossed"]);
+
+ // An invisible OCR text layer (`3 Tr`) on the same baseline and size as a visible line is
+ // still removed when the visible line is edited, even if OCR misread a letter.
+ let mut ocr = text_page("BT /F1 12 Tf 0 Tr 1 0 0 1 72 700 Tm (Hello world) Tj ET BT /F1 12 Tf 3 Tr 1 0 0 1 72 700 Tm (Rello world) Tj ET");
+ text::replace_line(&mut ocr, 0, 0, "Goodbye world").unwrap();
+ assert_eq!(listed(&reopen(&ocr)), ["Goodbye world"]);
+}
+
#[test]
fn paragraphs_move_and_rewrap_to_a_new_width() {
let para = "BT /F1 10 Tf 12 TL 100 700 Td (One two three four five six seven) Tj T* (eight nine ten eleven twelve) Tj ET";
@@ -3311,6 +3436,55 @@ fn latin_ligature_actual_text_stays_on_its_line() {
}
}
+/// #3110: `/ActualText` may come from a named property list (`/Span /MC0 BDC`) looked up in the
+/// page's or the form's `/Properties`, as well as from an inline one.
+#[test]
+fn actual_text_from_named_property_list() {
+ let properties = "/Properties << /MC0 << /ActualText (Hello World) >> /MC1 << /Lang (en) >> >>";
+ let page_with = |content: &str| {
+ build(&[
+ "<< /Type /Catalog /Pages 2 0 R >>",
+ "<< /Type /Pages /Kids [3 0 R] /Count 1 >>",
+ &format!("<< /Type /Page /Parent 2 0 R /MediaBox [0 0 612 792] /Resources << /Font << /F1 5 0 R >> {properties} >> /Contents 4 0 R >>"),
+ &stream("", content),
+ "<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica /Encoding /WinAnsiEncoding >>",
+ ])
+ };
+ for (content, expected) in [
+ ("/Span /MC0 BDC BT /F1 24 Tf 72 700 Td (Glyphs) Tj ET EMC", "Hello World"),
+ ("/Span <> BDC BT /F1 24 Tf 72 700 Td (Glyphs) Tj ET EMC", "Hello World"),
+ ("/Span /MC1 BDC BT /F1 24 Tf 72 700 Td (Glyphs) Tj ET EMC", "Glyphs"),
+ ("/Span /Missing BDC BT /F1 24 Tf 72 700 Td (Glyphs) Tj ET EMC", "Glyphs"),
+ ] {
+ let doc = page_with(content);
+ for doc in [&doc, &reopen(&doc)] {
+ let lines = text::text_lines(doc, 0).unwrap();
+ assert_eq!(lines.iter().map(|l| l.text.as_str()).collect::>(), [expected], "{content}");
+ assert_eq!(text::text_blocks(doc, 0).unwrap()[0].editor_text, expected, "{content}");
+ let reading: Vec = text::reading_blocks(doc, 0).unwrap().into_iter().map(|b| b.text).collect();
+ assert_eq!(reading, [expected], "{content}");
+ }
+ }
+
+ // A form's named list is looked up in the form's own /Properties, not the page's.
+ let fm = stream(
+ "/Type /XObject /Subtype /Form /BBox [0 0 612 792] /Resources << /Font << /F1 5 0 R >> /Properties << /MC0 << /ActualText (Form text) >> >> >>",
+ "/Span /MC0 BDC BT /F1 24 Tf 72 600 Td (Glyphs) Tj ET EMC",
+ );
+ let doc = build(&[
+ "<< /Type /Catalog /Pages 2 0 R >>",
+ "<< /Type /Pages /Kids [3 0 R] /Count 1 >>",
+ &format!(
+ "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 612 792] /Resources << /Font << /F1 5 0 R >> /XObject << /Fm1 6 0 R >> {properties} >> /Contents 4 0 R >>"
+ ),
+ &stream("", "/Fm1 Do"),
+ "<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica /Encoding /WinAnsiEncoding >>",
+ &fm,
+ ]);
+ let reading: Vec = text::reading_blocks(&doc, 0).unwrap().into_iter().map(|b| b.text).collect();
+ assert_eq!(reading, ["Form text"]);
+}
+
#[test]
fn actual_text_and_visual_rtl_glyphs_keep_their_own_reading_order() {
let hex: String = "שלום".encode_utf16().map(|u| format!("{u:04X}")).collect();
@@ -3754,6 +3928,74 @@ fn private_image_crop_with_intervening_state_resizes_across_streams() {
}
}
+/// [`image_page`] with `pieces` as its `/Contents` array and a Form /Fm0 drawing 72,500–172,600.
+fn split_image_page(pieces: &[&[u8]]) -> Document {
+ let mut doc = image_page();
+ let page = pdfcraft_model::pages(&doc)[0].obj;
+ let mut form = Dict::new();
+ form.set(b"Type".to_vec(), Object::name("XObject"));
+ form.set(b"Subtype".to_vec(), Object::name("Form"));
+ form.set(b"BBox".to_vec(), Object::Array([72, 500, 172, 600].map(Object::Int).to_vec()));
+ let form = doc.add(Object::Stream(Stream::flate(form, b"0.5 g 72 500 100 100 re f")));
+ let image = doc.get(pdfcraft_model::pages(&doc)[0].obj);
+ let mut res = image.as_dict().and_then(|d| d.get(b"Resources")).and_then(|r| r.as_dict().cloned()).unwrap();
+ let mut xo = res.get(b"XObject").and_then(|x| x.as_dict().cloned()).unwrap();
+ xo.set(b"Fm0".to_vec(), Object::Ref(form));
+ res.set(b"XObject".to_vec(), Object::Dict(xo));
+ let refs = pieces.iter().map(|b| Object::Ref(doc.add(Object::Stream(Stream::flate(Dict::new(), b))))).collect();
+ doc.update_dict(page, |d| {
+ d.set(b"Contents".to_vec(), Object::Array(refs));
+ d.set(b"Resources".to_vec(), Object::Dict(res));
+ })
+ .unwrap();
+ doc
+}
+
+#[test]
+fn an_image_whose_operator_starts_the_next_stream_can_be_selected_moved_and_deleted() {
+ // #3113: an operator's operands end one piece and its keyword starts the next.
+ let square = [72.0, 500.0, 172.0, 600.0];
+ let cases: [(&[&[u8]], bool); 4] = [
+ (&[b"q 100 0 0 100 72 500 cm /Im0", b"Do Q"], false),
+ (&[b"q 100 0 0 100 72 500", b"cm /Im0 Do Q"], false),
+ (&[b"q /Fm0", b"Do Q"], true),
+ (&[b"q 100 0 0 100 72 500 cm /Im0 Do Q"], false),
+ ];
+ for (pieces, is_form) in cases {
+ let doc = split_image_page(pieces);
+ let label: Vec<_> = pieces.iter().map(|p| String::from_utf8_lossy(p)).collect();
+ let images = images::page_images(&doc, 0).unwrap();
+ assert_eq!(images.len(), 1, "{label:?}");
+ assert!(close(images[0].rect, square), "{label:?}: {:?}", images[0].rect);
+ assert_eq!(images[0].is_form, is_form);
+ assert!(images[0].contains([122.0, 550.0]), "{label:?}");
+ if !is_form {
+ let read = images::reading_images(&doc, 0).unwrap();
+ assert!(read.len() == 1 && close(read[0].rect, square), "{label:?}: {read:?}");
+ }
+
+ let mut deleted = doc.clone();
+ images::change_image(&mut deleted, 0, 0, &images::ImageChange::Delete).unwrap();
+ let deleted = reopen(&deleted);
+ assert!(images::page_images(&deleted, 0).unwrap().is_empty(), "{label:?}");
+ let (joined, ops) = joined_ops(&deleted);
+ assert!(!ops.iter().any(|o| o.is("Do")), "{joined}");
+
+ let target = [300.0, 500.0, 400.0, 600.0];
+ let mut moved = doc.clone();
+ images::change_image(&mut moved, 0, 0, &images::ImageChange::Transform(images::rect_to_rect(square, target))).unwrap();
+ let moved = reopen(&moved);
+ let after = images::page_images(&moved, 0).unwrap();
+ assert!(after.len() == 1 && close(after[0].rect, target), "{label:?}: {after:?} {:?}", joined_ops(&moved).0);
+
+ let mut shifted = doc.clone();
+ images::translate_images(&mut shifted, 0, &[0], [228.0, 0.0]).unwrap();
+ let shifted = reopen(&shifted);
+ let after = images::page_images(&shifted, 0).unwrap();
+ assert!(after.len() == 1 && close(after[0].rect, target), "{label:?}: {after:?} {:?}", joined_ops(&shifted).0);
+ }
+}
+
#[test]
fn curved_image_clips_have_analytic_bounds_and_holes_do_not_capture_points() {
let mut doc = image_page();
@@ -3932,6 +4174,31 @@ fn turned_paragraph_keeps_its_line_breaks_when_a_word_is_added() {
assert!(text::rewrite_block(&mut doc, 0, 0, Some("Only line"), &style).is_err());
}
+/// A paragraph drawn at a slant rewraps at the width its rows have along their baseline, as the
+/// same paragraph upright does. Its upright bounding box turned into the baseline's frame is wider
+/// by the rows' height times |sin 2θ|, which let a word that doesn't fit stay on its row.
+#[test]
+fn a_paragraph_at_a_slant_rewraps_at_the_width_it_has_upright() {
+ let rows = ["Rotated paragraph, first row of text", "that continues on a second row and", "ends on the third row here."];
+ let draft = "Rotated paragraph, first row of text that continues on a second row and now carries a few extra words \
+ before it ends on the third row here.";
+ let rewrapped = |degrees: f64| {
+ let (sin, cos) = degrees.to_radians().sin_cos();
+ let shown = rows.iter().map(|r| format!("({r}) Tj")).collect::>().join(" T* ");
+ let mut doc = text_page(&format!("BT /F1 12 Tf 14.4 TL {cos:.6} {sin:.6} {:.6} {cos:.6} 300 400 Tm {shown} ET", -sin));
+ let block = text::text_blocks(&doc, 0).unwrap().remove(0);
+ assert_eq!(block.editor_text, rows.join("\n"), "{degrees}°");
+ text::rewrite_block(&mut doc, 0, 0, Some(draft), &text::BlockStyle::default()).unwrap();
+ text::text_lines(&reopen(&doc), 0).unwrap().into_iter().map(|l| l.text).collect::>()
+ };
+ let upright = rewrapped(0.0);
+ // The widest source row is 189.44 pt; "now carries a few extra words before" is 196.07 pt.
+ assert_eq!(upright[2], "now carries a few extra words", "{upright:?}");
+ for degrees in [30.0, 45.0, 60.0, 90.0, 135.0, 210.0, 330.0] {
+ assert_eq!(rewrapped(degrees), upright, "{degrees}°");
+ }
+}
+
/// A paragraph in a Chrome-style subset font: `/W` leaves out the glyphs of the `/DW` width.
/// Adding a word keeps every row as it was and shows the word in the page's own font (#1443).
#[test]
@@ -3986,6 +4253,76 @@ fn default_width_subset_paragraph_keeps_its_rows_when_a_word_is_added() {
}
}
+/// #2578: a single turned row grows along its baseline only up to the page's 36 pt margin and wraps
+/// onto further rows from there, instead of running off the page.
+#[test]
+fn turned_single_row_wraps_at_the_page_margin() {
+ let draft = "A single turned row that keeps growing with many more words added until it is far longer than the page is tall, so it has to stop or wrap somewhere";
+ // Reading upwards (90° CCW) from y = 600 on the 600 × 800 test page: top margin is at y = 764.
+ let before = text_page("BT /F1 12 Tf 0 1 -1 0 300 600 Tm (A single turned row) Tj ET");
+ let layout = text::text_edit_layout(&before, 0, 0, draft, &text::BlockStyle::default()).unwrap();
+ let mut doc = before;
+ text::rewrite_block(&mut doc, 0, 0, Some(draft), &text::BlockStyle::default()).unwrap();
+ let reopened = reopen(&doc);
+ let lines = text::text_lines(&reopened, 0).unwrap();
+ assert!(lines.len() > 1, "expected the turned row to wrap at the top margin: {lines:?}");
+ for line in &lines {
+ assert!((line.rect[1] - 600.0).abs() < 0.5, "row starts at y = 600: {:?}", line.rect);
+ assert!(line.rect[3] <= 800.0 - 36.0 + 0.5, "row stays within the top margin: {:?}", line.rect);
+ }
+ let blocks = text::text_blocks(&reopened, 0).unwrap();
+ assert_eq!(blocks.len(), 1, "wrapped turned rows stay one paragraph: {blocks:?}");
+ assert_eq!(blocks[0].text, draft);
+ assert_eq!(layout.draft_rect, Some(blocks[0].rect));
+
+ // Reading downwards (90° CW) from y = 200: bottom margin is at y = 36.
+ let mut doc = text_page("BT /F1 12 Tf 0 -1 1 0 300 200 Tm (A single turned row) Tj ET");
+ text::rewrite_block(&mut doc, 0, 0, Some(draft), &text::BlockStyle::default()).unwrap();
+ let lines = text::text_lines(&reopen(&doc), 0).unwrap();
+ assert!(lines.len() > 1, "expected 90° CW row to wrap at the bottom margin: {lines:?}");
+ for line in &lines {
+ assert!((line.rect[3] - 200.0).abs() < 0.5, "row starts at y = 200: {:?}", line.rect);
+ assert!(line.rect[1] >= 36.0 - 0.5, "row stays within the bottom margin: {:?}", line.rect);
+ }
+
+ // Reading upside down (180°) from x = 200: left margin is at x = 36.
+ let mut doc = text_page("BT /F1 12 Tf -1 0 0 -1 200 600 Tm (A single turned row) Tj ET");
+ text::rewrite_block(&mut doc, 0, 0, Some(draft), &text::BlockStyle::default()).unwrap();
+ let lines = text::text_lines(&reopen(&doc), 0).unwrap();
+ assert!(lines.len() > 1, "expected 180° row to wrap at the left margin: {lines:?}");
+ for line in &lines {
+ assert!((line.rect[2] - 200.0).abs() < 0.5, "row starts at x = 200: {:?}", line.rect);
+ assert!(line.rect[0] >= 36.0 - 0.5, "row stays within the left margin: {:?}", line.rect);
+ }
+
+ // Slanted (45°) from (400, 600): baseline stops where it first reaches a 36 pt margin.
+ let (sin, cos) = 45.0_f64.to_radians().sin_cos();
+ let mut doc = text_page(&format!("BT /F1 12 Tf {cos:.6} {sin:.6} {:.6} {cos:.6} 400 600 Tm (A single turned row) Tj ET", -sin));
+ text::rewrite_block(&mut doc, 0, 0, Some(draft), &text::BlockStyle::default()).unwrap();
+ let lines = text::text_lines(&reopen(&doc), 0).unwrap();
+ assert!(lines.len() > 1, "expected 45° slanted row to wrap at the margin: {lines:?}");
+ for line in &lines {
+ assert!(line.rect[2] <= 600.0 && line.rect[3] <= 800.0, "slanted row stays on the page: {:?}", line.rect);
+ }
+
+ // Moving a turned single row with `offset` reduces the room left before the margin.
+ let mut doc = text_page("BT /F1 12 Tf 0 1 -1 0 300 500 Tm (A single turned row) Tj ET");
+ let moved = text::BlockStyle { offset: Some([0.0, 150.0]), ..Default::default() };
+ text::rewrite_block(&mut doc, 0, 0, Some(draft), &moved).unwrap();
+ let lines = text::text_lines(&reopen(&doc), 0).unwrap();
+ assert!(lines.len() > 1, "{lines:?}");
+ for line in &lines {
+ assert!((line.rect[1] - 650.0).abs() < 0.5, "moved row starts at y = 650: {:?}", line.rect);
+ assert!(line.rect[3] <= 800.0 - 36.0 + 0.5, "moved row stops at the top margin: {:?}", line.rect);
+ }
+
+ // A turned row that already starts inside the 36 pt margin keeps its own width instead of collapsing.
+ let mut doc = text_page("BT /F1 12 Tf 0 1 -1 0 300 770 Tm (Edge label) Tj ET");
+ text::rewrite_block(&mut doc, 0, 0, Some("Edge label updated"), &text::BlockStyle::default()).unwrap();
+ let lines = text::text_lines(&reopen(&doc), 0).unwrap();
+ assert!(lines.len() >= 2, "{lines:?}");
+}
+
/// An added item is written out as a specific standard-14 face (`/Times-Italic`, `/Helvetica-Bold`
/// …), so it has to be *measured* with that same face. Measuring the upright face while writing
/// the italic one mis-sizes every box: Times-Italic `A` is 611 against Times-Roman's 722.
@@ -4276,6 +4613,106 @@ fn japanese_paragraphs_wrap_between_characters_without_closing_marks_at_a_row_st
}
}
+#[test]
+fn rewritten_paragraph_is_drawn_after_backgrounds_between_its_rows() {
+ // Each row has its own background box drawn immediately before its BT..ET, as cards and
+ // highlighted headings exported from LibreOffice or browsers do (#2377).
+ let card = "0 0.45 0.25 rg 60 696 240 16 re f BT /F1 10 Tf 1 1 1 rg 64 700 Td (SEMINARIO ONLINE) Tj ET \
+ 0 0.45 0.25 rg 60 668 240 24 re f BT /F1 16 Tf 1 1 1 rg 64 674 Td (Dalla teoria alla pratica:) Tj ET \
+ 0 0.45 0.25 rg 60 632 240 32 re f BT /F1 24 Tf 1 1 1 rg 64 640 Td (gestire con metodo) Tj ET \
+ 0 0.45 0.25 rg 60 596 240 32 re f BT /F1 24 Tf 1 1 1 rg 64 604 Td (un progetto complesso) Tj ET";
+ let shown_at = |ops: &[pdfcraft_content::Op], needle: &str| {
+ ops.iter()
+ .position(|op| {
+ matches!(op.op.as_slice(), b"Tj" | b"TJ")
+ && op.operands.iter().any(|o| matches!(o, Object::String(s) if String::from_utf8_lossy(&s.bytes).contains(needle)))
+ })
+ .unwrap()
+ };
+
+ // 1. Editing the title rows emits all rewritten rows after all four background fills.
+ let mut doc = text_page(card);
+ assert_eq!(text::text_blocks(&doc, 0).unwrap().len(), 1);
+ let edited = "SEMINARIO ONLINE\nDalla teoria alla pratica:\ngestire con criterio\nun progetto nuovo";
+ text::rewrite_block(&mut doc, 0, 0, Some(edited), &text::BlockStyle::default()).unwrap();
+ let doc = reopen(&doc);
+ assert_eq!(text::text_blocks(&doc, 0).unwrap()[0].editor_text, edited);
+ let ops = pdfcraft_content::parse(&page_content_bytes(&doc, 0)).ops;
+ let fills: Vec = ops.iter().enumerate().filter_map(|(i, op)| op.is("f").then_some(i)).collect();
+ assert_eq!(fills.len(), 4);
+ assert!(
+ shown_at(&ops, "criterio") > fills[3] && shown_at(&ops, "nuovo") > fills[3],
+ "rewritten title rows must be drawn after their row background fills (fills at {fills:?}): {ops:?}"
+ );
+
+ // 2. Partial edit with a kept suffix row: rewritten rows 0..=2 are drawn after fills[0..=2]
+ // and still before kept row 3, preserving reading order.
+ let mut doc = text_page(card);
+ let partial = "WORKSHOP ONLINE\nDalla teoria alla pratica:\ngestire con criterio\nun progetto complesso";
+ text::rewrite_block(&mut doc, 0, 0, Some(partial), &text::BlockStyle::default()).unwrap();
+ let doc = reopen(&doc);
+ assert_eq!(text::text_blocks(&doc, 0).unwrap()[0].editor_text, partial);
+ let ops = pdfcraft_content::parse(&page_content_bytes(&doc, 0)).ops;
+ let fills: Vec = ops.iter().enumerate().filter_map(|(i, op)| op.is("f").then_some(i)).collect();
+ assert!(
+ shown_at(&ops, "WORKSHOP") > fills[2]
+ && shown_at(&ops, "pratica:") > fills[2]
+ && shown_at(&ops, "criterio") > fills[2]
+ && shown_at(&ops, "criterio") < shown_at(&ops, "complesso")
+ && shown_at(&ops, "complesso") > fills[3],
+ "rewritten prefix rows must follow their fills and precede the kept final row"
+ );
+
+ // 3. Editing row 2 so that it wraps pushes row 3 down; all rewritten rows must follow fills[3].
+ let mut doc = text_page(card);
+ let wrapped_edit = "SEMINARIO ONLINE\nDalla teoria alla pratica:\ngestire con criterio e metodo approfondito\nun progetto complesso";
+ text::rewrite_block(&mut doc, 0, 0, Some(wrapped_edit), &text::BlockStyle::default()).unwrap();
+ let doc = reopen(&doc);
+ let ops = pdfcraft_content::parse(&page_content_bytes(&doc, 0)).ops;
+ let last_fill = ops.iter().rposition(|op| op.is("f")).unwrap();
+ assert!(
+ shown_at(&ops, "approfondito") > last_fill && shown_at(&ops, "complesso") > last_fill,
+ "wrapped rows and pushed-down suffix row must follow all row background fills"
+ );
+
+ // 4. Single-style paragraph path across a CTM change between rows, with movement, underline,
+ // and a trailing unrelated line sharing the second row's BT..ET.
+ let cross_ctm = "0.9 0.9 0.2 rg 60 696 240 18 re f BT /F1 12 Tf 0 0 0 rg 64 700 Td (First highlighted row) Tj ET \
+ q 1 0 0 1 10 12 cm 0.9 0.9 0.2 rg 50 668 240 18 re f \
+ BT /F1 12 Tf 0 0 0 rg 54 672 Td (Second highlighted row) Tj 246 -40 Td (Side footer) Tj ET Q";
+ let mut doc = text_page(cross_ctm);
+ let para = text::text_blocks(&doc, 0).unwrap().into_iter().position(|b| b.text.starts_with("First highlighted")).unwrap();
+ assert_eq!(text::text_blocks(&doc, 0).unwrap()[para].rows.len(), 2);
+ text::rewrite_block(
+ &mut doc,
+ 0,
+ para,
+ Some("Updated first highlighted row and updated second highlighted row"),
+ &text::BlockStyle { offset: Some([6.0, -4.0]), underline: Some(true), ..Default::default() },
+ )
+ .unwrap();
+ let doc = reopen(&doc);
+ let ops = pdfcraft_content::parse(&page_content_bytes(&doc, 0)).ops;
+ let last_fill = ops.iter().rposition(|op| op.is("f")).unwrap();
+ assert!(
+ shown_at(&ops, "Updated first") > last_fill && shown_at(&ops, "Side footer") > shown_at(&ops, "Updated first"),
+ "cross-CTM single-style rewrite must follow both background fills and precede the shared BT's trailing line"
+ );
+ let lines = text::text_lines(&doc, 0).unwrap();
+ let updated_first = lines.iter().find(|l| l.text.starts_with("Updated first")).unwrap();
+ let side_footer = lines.iter().find(|l| l.text == "Side footer").unwrap();
+ assert!(
+ (updated_first.baseline()[0] - 70.0).abs() < 0.01 && (updated_first.baseline()[1] - 696.0).abs() < 0.01,
+ "offset baseline in user space across CTM: {:?}",
+ updated_first.baseline()
+ );
+ assert!(
+ (side_footer.baseline()[0] - 310.0).abs() < 0.01 && (side_footer.baseline()[1] - 644.0).abs() < 0.01,
+ "trailing line in shared BT keeps its tlm across CTM: {:?}",
+ side_footer.baseline()
+ );
+}
+
/// Small graphics-state oracle for synthetic streams: every tracked state change must
/// have been restored when a probe XObject would draw. No external assets are involved.
fn probe_states(doc: &Document) -> Vec, Vec>> {
diff --git a/crates/edit/src/text.rs b/crates/edit/src/text.rs
index 916aaeadd9..32d63d5c80 100644
--- a/crates/edit/src/text.rs
+++ b/crates/edit/src/text.rs
@@ -50,6 +50,10 @@ pub struct TextLine {
pub visual_order: bool,
/// The Japanese or Hebrew fallback face the font subsets, when the line is in one (see [`same_font`]).
face: Option,
+ /// Its box along and across its baseline, `[along0, across0, along1, across1]` (see
+ /// [`TextLine::frame_rect`]). Upright text has `rect`; turned text has its own box, which is
+ /// narrower than `rect` turned into the same frame.
+ frame: [f64; 4],
/// The content stream piece the line starts in, and its operators (indices into the page's
/// streams parsed joined).
stream: usize,
@@ -470,12 +474,7 @@ impl TextLine {
/// Its box in the same frame, `[along0, across0, along1, across1]` (for upright text, `rect`).
fn frame_rect(&self) -> [f64; 4] {
- let dir = self.direction();
- let r = self.rect;
- [(r[0], r[1]), (r[2], r[1]), (r[0], r[3]), (r[2], r[3])]
- .into_iter()
- .map(|p| to_frame(dir, p))
- .fold([f64::MAX, f64::MAX, f64::MIN, f64::MIN], |b, p| [b[0].min(p.0), b[1].min(p.1), b[2].max(p.0), b[3].max(p.1)])
+ self.frame
}
}
@@ -493,6 +492,11 @@ fn to_frame(dir: (f64, f64), p: (f64, f64)) -> (f64, f64) {
(p.0 * dir.0 + p.1 * dir.1, p.1 * dir.0 - p.0 * dir.1)
}
+/// The box covering both `[x0, y0, x1, y1]` boxes.
+fn union(a: [f64; 4], b: [f64; 4]) -> [f64; 4] {
+ [a[0].min(b[0]), a[1].min(b[1]), a[2].max(b[2]), a[3].max(b[3])]
+}
+
/// What replacing a line did.
#[derive(Clone, Debug, PartialEq)]
pub struct LineEdit {
@@ -639,6 +643,8 @@ struct Shown {
bt_state: Ts,
text: String,
rect: [f64; 4],
+ /// Its box along and across its baseline (see [`TextLine::frame`]).
+ frame: [f64; 4],
/// The boxes of its words (see [`TextLine::ink`]).
ink: Vec<[f64; 4]>,
baseline: f64,
@@ -734,7 +740,8 @@ fn interpret(
cache: &mut HashMap, (Rc, Option)>,
carry: &mut Carry,
mut invoked: Option<&mut Vec>,
- logical_text: bool,
+ // `Some(properties)`: read `/ActualText`, looking named property lists up in `properties`.
+ logical_text: Option<&Dict>,
) -> Vec {
let mut out: Vec = Vec::new();
let mut ts = carry.ts.clone();
@@ -751,11 +758,16 @@ fn interpret(
// artifacts and inner spans, which inherit their enclosing region.
let region = matches!(op.name(0), Some(b"P" | b"H" | b"H1" | b"H2" | b"H3" | b"H4" | b"H5" | b"H6" | b"LI" | b"TD" | b"TH"));
marked.push(if region { Some(i) } else { marked.last().copied().flatten() });
- let logical = op.operands.get(1).and_then(Object::as_dict).and_then(|d| d.get(b"ActualText")).and_then(|s| match s {
+ // The property list is inline or named in the resources' /Properties (§14.6.2).
+ let list = logical_text.and_then(|properties| match op.operands.get(1) {
+ Some(Object::Name(n)) => properties.get(n).and_then(|o| doc.dict(o)),
+ other => other.and_then(Object::as_dict).cloned(),
+ });
+ let logical = list.as_ref().and_then(|d| d.get(b"ActualText")).and_then(|s| match s {
Object::String(s) if s.bytes.len() <= 4_000_000 => Some(s.to_text()),
_ => None,
});
- actual.push(if logical_text { logical.map(|s| (out.len(), i, s)) } else { None });
+ actual.push(logical.map(|s| (out.len(), i, s)));
}
b"EMC" => {
marked.pop();
@@ -773,6 +785,7 @@ fn interpret(
first.rect[2].max(shown.rect[2]),
first.rect[3].max(shown.rect[3]),
];
+ first.frame = union(first.frame, shown.frame);
first.ink.extend(shown.ink);
}
first.extra_ops.extend([open, i]);
@@ -936,6 +949,13 @@ fn interpret(
let rect = corners
.iter()
.fold([f64::MAX, f64::MAX, f64::MIN, f64::MIN], |b, p| [b[0].min(p.0), b[1].min(p.1), b[2].max(p.0), b[3].max(p.1)]);
+ // The same corners along and across the baseline: the run's own box, not its
+ // upright bounding box turned (which is wider by the box's height on a slant).
+ let dir = direction(&trm0);
+ let frame = corners
+ .iter()
+ .map(|&p| to_frame(dir, p))
+ .fold([f64::MAX, f64::MAX, f64::MIN, f64::MIN], |b, p| [b[0].min(p.0), b[1].min(p.1), b[2].max(p.0), b[3].max(p.1)]);
let ink = words
.iter()
.map(|&(a, b)| {
@@ -963,6 +983,7 @@ fn interpret(
bt_state: at_bt.1.clone(),
text,
rect,
+ frame,
ink,
baseline: start.1,
start_x: start.0,
@@ -1003,7 +1024,8 @@ pub fn text_lines(doc: &Document, page: usize) -> Result, EditErro
// joined, so an operator whose operands end one piece still has them.
let pieces = Pieces::join(&content_streams(doc, &p.dict).into_iter().map(|(_, d)| d).collect::>());
let ops = pieces.parse().ops;
- let shown = interpret(doc, &ops, &fonts_res, &mut cache, &mut carry, None, true);
+ let properties = res.get(b"Properties").and_then(|o| doc.dict(o)).unwrap_or_default();
+ let shown = interpret(doc, &ops, &fonts_res, &mut cache, &mut carry, None, Some(&properties));
let mut last: Option<(usize, Pen)> = None; // the stream piece and where its last run ended
for mut s in shown {
// A single RTL glyph's ActualText (e.g. lam-alef) is an inline ligature, not a
@@ -1040,6 +1062,7 @@ pub fn text_lines(doc: &Document, page: usize) -> Result, EditErro
}
l.text.push_str(&s.text);
l.rect = [l.rect[0].min(s.rect[0]), l.rect[1].min(s.rect[1]), l.rect[2].max(s.rect[2]), l.rect[3].max(s.rect[3])];
+ l.frame = union(l.frame, s.frame);
l.ink.extend(s.ink.iter().copied());
l.ops.push(s.op);
l.ops.extend(s.extra_ops.iter().copied());
@@ -1051,6 +1074,7 @@ pub fn text_lines(doc: &Document, page: usize) -> Result, EditErro
lines.push(TextLine {
text: s.text.clone(),
rect: s.rect,
+ frame: s.frame,
ink: s.ink.clone(),
font: String::from_utf8_lossy(&s.font).into_owned(),
base_font: s.base_font.clone(),
@@ -1128,6 +1152,7 @@ fn add_shown(lines: &mut Vec, last: &mut Option, s: Shown, stream
}
l.text.push_str(&s.text);
l.rect = [l.rect[0].min(s.rect[0]), l.rect[1].min(s.rect[1]), l.rect[2].max(s.rect[2]), l.rect[3].max(s.rect[3])];
+ l.frame = union(l.frame, s.frame);
l.ink.extend(s.ink.iter().copied());
l.ops.push(s.op);
l.ops.extend(s.extra_ops.iter().copied());
@@ -1138,6 +1163,7 @@ fn add_shown(lines: &mut Vec, last: &mut Option, s: Shown, stream
lines.push(TextLine {
text: s.text.clone(),
rect: s.rect,
+ frame: s.frame,
ink: s.ink.clone(),
font: String::from_utf8_lossy(&s.font).into_owned(),
base_font: s.base_font.clone(),
@@ -1242,7 +1268,8 @@ fn reading_lines(doc: &Document, page: usize) -> Result, EditError
let fonts_res = resources.get(b"Font").map(|f| w.doc.resolve(f)).and_then(|f| f.as_dict().cloned()).unwrap_or_default();
let mut cache = HashMap::new();
let mut invoked = Vec::new();
- let shown = interpret(w.doc, ops, &fonts_res, &mut cache, carry, Some(&mut invoked), true);
+ let properties = resources.get(b"Properties").and_then(|o| w.doc.dict(o)).unwrap_or_default();
+ let shown = interpret(w.doc, ops, &fonts_res, &mut cache, carry, Some(&mut invoked), Some(&properties));
let mut last = None;
let mut last_stream = None;
let mut forms = invoked.into_iter().peekable();
@@ -1642,7 +1669,15 @@ fn source_encodes(metrics: &Metrics, text: &str) -> bool {
text.split(' ').all(|word| metrics.encode(word).is_some())
}
+#[cfg(test)]
+thread_local! {
+ /// How often [`source_space`] ran on this thread, for the test that pins it out of per-character loops.
+ pub(crate) static SOURCE_SPACE_CALLS: std::cell::Cell = const { std::cell::Cell::new(0) };
+}
+
fn source_space(doc: &Document, fonts: &Dict, metrics: &Metrics) -> f64 {
+ #[cfg(test)]
+ SOURCE_SPACE_CALLS.with(|c| c.set(c.get().saturating_add(1)));
let width = |m: &Metrics| m.encode(" ").map(|b| m.codes(&b).iter().map(|(c, _)| m.width(*c)).sum::());
if let Some(w) = width(metrics) {
return w;
@@ -1784,7 +1819,8 @@ pub(crate) fn replace_line_with_limit(doc: &mut Document, page: usize, line: usi
}
// Rebuild: the line's first operator becomes the replacement, its others go. Copies of the
// line drawn on the same baseline go too, so the replacement is all that shows.
- let drops = coincident_ops(&lines, &[&target]);
+ let mut drops = coincident_ops(&lines, &[&target]);
+ drops.extend(target.ops.iter().copied());
let new_data = splice(&pieces, &ops, |i| if i == first { (std::mem::take(&mut replacement), false) } else { (Vec::new(), !drops.contains(&i)) });
let contents = rewritten_contents(doc, &streams, new_data);
let page_ref = p.obj;
@@ -1816,7 +1852,10 @@ fn coincident_ops(lines: &[TextLine], targets: &[&TextLine]) -> std::collections
let area = |r: [f64; 4]| ((r[2] - r[0]) * (r[3] - r[1])).max(0.0);
let mut drop = std::collections::HashSet::new();
for line in lines {
- let copy = targets.iter().any(|target| overlaps_half(line.rect, target.rect, area) && on_same_baseline(line, target));
+ let copy = targets.iter().any(|target| {
+ let same_text = line.text.trim() == target.text.trim() || (line.origin.state.render_mode == 3) != (target.origin.state.render_mode == 3);
+ same_text && overlaps_half(line.rect, target.rect, area) && on_same_baseline(line, target)
+ });
if copy {
drop.extend(line.ops.iter().copied());
}
@@ -1834,13 +1873,16 @@ fn overlaps_half(a: [f64; 4], b: [f64; 4], area: impl Fn([f64; 4]) -> f64) -> bo
/// Perpendicular distance between baselines, in user space. Horizontal text compares height;
/// turned text compares the distance across the writing direction.
fn on_same_baseline(line: &TextLine, target: &TextLine) -> bool {
+ let (dir, target_dir) = (line.direction(), target.direction());
+ let same_dir = (dir.0 - target_dir.0).abs() < 1e-3 && (dir.1 - target_dir.1).abs() < 1e-3;
+ let same_size = line.size.min(target.size) >= line.size.max(target.size) * 0.8;
let angle = target.rotation().to_radians();
let (sin, cos) = if angle.is_finite() { angle.sin_cos() } else { (0.0, 1.0) };
let dx = line.origin.x - target.origin.x;
let dy = line.origin.baseline - target.origin.baseline;
let offset = dx.mul_add(-sin, dy * cos).abs();
let size = line.size.min(target.size).max(1.0);
- offset.is_finite() && offset < size * 0.3
+ same_dir && same_size && offset.is_finite() && offset < size * 0.3
}
/// The paragraphs on a page (0-based).
@@ -1865,7 +1907,7 @@ pub fn clear_block(doc: &mut Document, page: usize, block: usize) -> Result<(),
let streams = content_streams(doc, &p.dict);
let pieces = Pieces::join(&streams.iter().map(|(_, d)| d.as_slice()).collect::>());
let ops = pieces.parse().ops;
- let shown = interpret(doc, &ops, &fonts, &mut HashMap::new(), &mut Carry::new(), None, false);
+ let shown = interpret(doc, &ops, &fonts, &mut HashMap::new(), &mut Carry::new(), None, None);
let mut patches = HashMap::new();
for s in shown.into_iter().filter(|s| drops.contains(&s.op)) {
if s.state.render_mode >= 4 {
@@ -2072,6 +2114,8 @@ fn rewrite_block_from(
// The characters to show, for choosing fonts: the breaks are moves, not glyphs.
let flat = text.replace('\n', " ");
let o = &first.origin;
+ let last = members.last().copied().unwrap_or(first);
+ let emit = &last.origin;
let (font_name, old_size) = o.state.font.clone().ok_or_else(|| EditError::Invalid("the paragraph has no font".into()))?;
let k = o.k.max(1e-6);
// The size in text space: points ÷ the text-to-user scale.
@@ -2132,7 +2176,7 @@ fn rewrite_block_from(
// paragraph leaves the document as it was. The move is in page space, so it is taken back through
// the CTM's linear part into the space the paragraph is drawn in.
let movement = if dx != 0.0 || dy != 0.0 {
- let c = o.ctm;
+ let c = emit.ctm;
let back = Matrix([c[0], c[1], c[2], c[3], 0.0, 0.0])
.invert()
.ok_or_else(|| EditError::Invalid("the paragraph is drawn in a space it can't be moved in".into()))?;
@@ -2140,6 +2184,12 @@ fn rewrite_block_from(
} else {
None
};
+ let tm = if o.ctm == emit.ctm {
+ o.tm
+ } else {
+ let emit_inv = Matrix(emit.ctm).invert().ok_or_else(|| EditError::Invalid("the paragraph has a singular transform".into()))?;
+ Matrix(o.tm).then(&Matrix(o.ctm)).then(&emit_inv).0
+ };
let type3 = if !reuse && needs_type3(&flat) { Some(type3_font(doc, &mut fonts_res, &flat, family, bold, max_glyphs)?) } else { None };
let advance = |s: &str| -> f64 {
let t = match (&metrics, reuse) {
@@ -2161,9 +2211,27 @@ fn rewrite_block_from(
};
t * o_state.scale * k
};
- // A turned single line grows to fit, as far as the largest page PDF allows.
+ // A turned single line grows to fit, up to where its baseline reaches the page's margin.
let width = if !upright && style.width.is_none() && members.len() == 1 {
- text.split('\n').map(&advance).fold(width, f64::max).min(14_400.0)
+ let crop = p.crop(doc);
+ let (x0, y0) = (first.origin.x + dx, first.origin.baseline + dy);
+ let (ux, uy) = first.direction();
+ let tx = if ux > 1e-6 {
+ (crop[2] - 36.0 - x0) / ux
+ } else if ux < -1e-6 {
+ (crop[0] + 36.0 - x0) / ux
+ } else {
+ f64::INFINITY
+ };
+ let ty = if uy > 1e-6 {
+ (crop[3] - 36.0 - y0) / uy
+ } else if uy < -1e-6 {
+ (crop[1] + 36.0 - y0) / uy
+ } else {
+ f64::INFINITY
+ };
+ let cap = width.max(tx.min(ty)).min(14_400.0);
+ text.split('\n').map(&advance).fold(width, f64::max).min(cap)
} else {
width
};
@@ -2217,7 +2285,7 @@ fn rewrite_block_from(
let n = pdfcraft_content::num;
// In its own graphics state, so a new colour (or anything else) stops at the paragraph.
let mut block_ops = vec![Op::new("q", vec![])];
- block_ops.extend(reveal_ops(&b, &lines, style, o.ctm)?);
+ block_ops.extend(reveal_ops(&b, &lines, style, emit.ctm)?);
// Moved: a translation inside that state.
if let Some((ax, ay)) = movement {
block_ops.push(Op::new("cm", vec![n(1.0), n(0.0), n(0.0), n(1.0), n(ax), n(ay)]));
@@ -2232,7 +2300,7 @@ fn rewrite_block_from(
state.fill = vec![Op::new("rg", vec![n(r), n(g), n(bl)])];
}
block_ops.extend(state.ops());
- block_ops.push(Op::new("Tm", o.tm.iter().map(|v| n(*v)).collect()));
+ block_ops.push(Op::new("Tm", tm.iter().map(|v| n(*v)).collect()));
// Alignment: each line's offset from the left edge, in text space.
let offset = |line: &str| -> f64 {
let free = (width - advance(line)).max(0.0) / k;
@@ -2315,7 +2383,7 @@ fn rewrite_block_from(
})
.collect();
block_ops.push(Op::new("q", vec![]));
- block_ops.push(Op::new("cm", o.tm.iter().map(|v| n(*v)).collect()));
+ block_ops.push(Op::new("cm", tm.iter().map(|v| n(*v)).collect()));
block_ops.extend(colour);
block_ops.push(Op::new("w", vec![n(size * 0.06)]));
for (x0, x1, y) in underlines {
@@ -2330,24 +2398,24 @@ fn rewrite_block_from(
// second copy, an invisible text layer): the new text is all that may show.
let mut drop = coincident_ops(&lines, &members);
drop.extend(members.iter().flat_map(|l| l.ops.iter().copied()));
- // Text shown earlier in the same text object stays first: the paragraph then goes where it
- // was, between two halves of the text object, so it keeps its place in reading order.
- let start = first.ops.first().copied().unwrap_or(o.bt_op);
+ // Text shown earlier in the same text object stays first: the paragraph then goes where its
+ // last row was, after any backgrounds between the rows, or between two halves of the text object.
+ let start = last.ops.first().copied().unwrap_or(emit.bt_op);
let split = ops
- .get(o.bt_op..start)
+ .get(emit.bt_op..start)
.unwrap_or_default()
.iter()
.enumerate()
- .any(|(i, op)| matches!(op.op.as_slice(), b"Tj" | b"TJ" | b"'" | b"\"") && !drop.contains(&(o.bt_op + i)));
+ .any(|(i, op)| matches!(op.op.as_slice(), b"Tj" | b"TJ" | b"'" | b"\"") && !drop.contains(&(emit.bt_op + i)));
// What the text after the paragraph expects: the state at its BT (or where it was split).
- let after = if split { &o.state } else { &o.bt_state };
+ let after = if split { &emit.state } else { &emit.bt_state };
block_ops.extend(after.ops().into_iter().filter(|op| !op.is("Tf") || after.font.is_some()));
if split {
block_ops.insert(0, Op::new("ET", vec![]));
block_ops.push(Op::new("BT", vec![]));
- block_ops.push(Op::new("Tm", o.tlm.iter().map(|v| n(*v)).collect()));
+ block_ops.push(Op::new("Tm", emit.tlm.iter().map(|v| n(*v)).collect()));
}
- let at = if split { start } else { o.bt_op };
+ let at = if split { start } else { emit.bt_op };
let new_data = splice(&pieces, &ops, |i| {
let insert = if i == at { std::mem::take(&mut block_ops) } else { Vec::new() };
(insert, !drop.contains(&i))
@@ -2496,7 +2564,7 @@ pub(crate) fn translate_blocks(doc: &mut Document, page: usize, indexes: &[usize
let fonts = res.get(b"Font").map(|r| doc.resolve(r)).and_then(|r| r.as_dict().cloned()).unwrap_or_default();
let pieces = Pieces::join(&streams.iter().map(|(_, d)| d.as_slice()).collect::>());
let ops = pieces.parse().ops;
- let shown = interpret(doc, &ops, &fonts, &mut HashMap::new(), &mut Carry::new(), None, false);
+ let shown = interpret(doc, &ops, &fonts, &mut HashMap::new(), &mut Carry::new(), None, None);
let mut patches = HashMap::new();
// Keep full precision: a tiny rounded text-space change can be magnified by a CTM,
// and restoring rise/spacing must not alter following unselected text.
diff --git a/crates/edit/src/text_regions.rs b/crates/edit/src/text_regions.rs
index 2d7baf519f..e08068d9f5 100644
--- a/crates/edit/src/text_regions.rs
+++ b/crates/edit/src/text_regions.rs
@@ -17,10 +17,6 @@ fn same_plane(a: &TextLine, b: &TextLine) -> bool {
})
}
-fn union(a: [f64; 4], b: [f64; 4]) -> [f64; 4] {
- [a[0].min(b[0]), a[1].min(b[1]), a[2].max(b[2]), a[3].max(b[3])]
-}
-
fn row_rect(row: &[usize], lines: &[TextLine]) -> Option<[f64; 4]> {
row.iter().filter_map(|i| lines.get(*i)).map(|l| l.rect).reduce(union)
}
@@ -493,6 +489,20 @@ pub(super) fn rewrite_styled_block(
};
// A source run may carry TJ kerning/justification that is absent from its text state.
// Keep its measured advance when rebuilding the run instead of making it wrap on entry.
+ // Each font's space width, worked out once: when a font can't show a space, finding one
+ // reads every font on the page, and doing that for every character of a paragraph made each
+ // keystroke in a long paragraph take half a second (#2616).
+ // Only for the fonts the paragraph uses, when first asked: working out every page font's width
+ // up front would scan the page once per font for a short edit on a page of many fonts.
+ let spaces: std::cell::RefCell> = std::cell::RefCell::new(HashMap::new());
+ let space_of = |name: &str, m: &Metrics| {
+ if let Some(w) = spaces.borrow().get(name) {
+ return *w;
+ }
+ let w = source_space(doc, &fonts, m);
+ spaces.borrow_mut().insert(name.to_owned(), w);
+ w
+ };
let mut calibration = HashMap::new();
for (index, source) in lines.iter().enumerate() {
if let Some(m) = metrics.get(&source.font) {
@@ -500,7 +510,7 @@ pub(super) fn rewrite_styled_block(
m,
&source.text,
source.origin.state.font.as_ref().map_or(1.0, |(_, s)| *s),
- source_space(doc, &fonts, m),
+ space_of(&source.font, m),
source.origin.state.char_spacing,
source.origin.state.word_spacing,
) * source.origin.state.scale
@@ -517,7 +527,7 @@ pub(super) fn rewrite_styled_block(
m,
&item.0.to_string(),
size_of(source, item.1),
- source_space(doc, &fonts, m),
+ space_of(&item.2, m),
overrides
.get(&item.1)
.and_then(|r| r.char_spacing)
@@ -620,19 +630,20 @@ pub(super) fn rewrite_styled_block(
let kept_lines: std::collections::HashSet =
kept.iter().enumerate().filter(|(_, k)| **k).filter_map(|(r, _)| block.rows.get(r)).flatten().copied().collect();
let rewritten: Vec<&TextLine> = block.lines.iter().filter(|i| !kept_lines.contains(i)).filter_map(|i| lines.get(*i)).collect();
- // The rewrite goes where the first rewritten row was drawn, so it must share the paragraph's
- // space. A dropped `'` or `"` also moved to the next line, which a kept row relies on.
- let lead = rewritten.first().copied().filter(|lead| {
- lead.origin.ctm.iter().zip(first.origin.ctm).all(|(a, b)| (a - b).abs() < 1e-9)
- && !rewritten.iter().flat_map(|l| &l.ops).any(|i| ops.get(*i).is_some_and(|op| op.is("'") || op.is("\"")))
- });
+ // The rewrite goes where the last rewritten row was drawn, after any backgrounds painted
+ // between the rows. A dropped `'` or `"` also moved to the next line, which a kept row relies on.
+ let lead =
+ rewritten.last().copied().filter(|_| !rewritten.iter().flat_map(|l| &l.ops).any(|i| ops.get(*i).is_some_and(|op| op.is("'") || op.is("\""))));
let (members, lead) = match lead {
Some(lead) if !kept_lines.is_empty() => (rewritten, lead),
_ => {
kept.iter_mut().for_each(|k| *k = false);
- (block.lines.iter().filter_map(|i| lines.get(*i)).collect(), first)
+ let all: Vec<&TextLine> = block.lines.iter().filter_map(|i| lines.get(*i)).collect();
+ let last = all.last().copied().unwrap_or(first);
+ (all, last)
}
};
+ let o = &lead.origin;
for (row, _) in rows.iter().zip(&kept).filter(|(_, k)| !**k) {
let text: String = row.iter().map(|v| v.0).collect();
if pdfcraft_fonts::source_shaping::required(&text) && row.first().is_some_and(|first| row.iter().any(|v| v.1 != first.1 || v.2 != first.2)) {
@@ -668,8 +679,8 @@ pub(super) fn rewrite_styled_block(
}
let n = pdfcraft_content::num;
let mut block_ops = vec![Op::new("q", vec![])];
- block_ops.extend(reveal_ops(block, lines, style, first.origin.ctm)?);
- let back = Matrix([first.origin.ctm[0], first.origin.ctm[1], first.origin.ctm[2], first.origin.ctm[3], 0.0, 0.0])
+ block_ops.extend(reveal_ops(block, lines, style, o.ctm)?);
+ let back = Matrix([o.ctm[0], o.ctm[1], o.ctm[2], o.ctm[3], 0.0, 0.0])
.invert()
.ok_or_else(|| EditError::Invalid("the paragraph is drawn in a space it cannot be edited in".into()))?;
let (tx, ty) = back.apply(dx, dy);
@@ -741,7 +752,7 @@ pub(super) fn rewrite_styled_block(
if let Some([r, g, b]) = overrides.get(&item.1).and_then(|r| r.color).or(style.color) {
state.fill = vec![Op::new("rg", vec![n(r), n(g), n(b)])];
}
- let inverse = Matrix(first.origin.ctm).invert().ok_or_else(|| EditError::Invalid("the paragraph has a singular transform".into()))?;
+ let inverse = Matrix(o.ctm).invert().ok_or_else(|| EditError::Invalid("the paragraph has a singular transform".into()))?;
let mut tm = Matrix(source.origin.tm).then(&Matrix(source.origin.ctm)).then(&inverse).0;
let (ax, ay) = back.apply(x - source.origin.x, baseline - source.origin.baseline);
tm[4] += ax;
@@ -781,7 +792,7 @@ pub(super) fn rewrite_styled_block(
m,
&string,
size_of(source, item.1),
- source_space(doc, &fonts, m),
+ spaces.borrow().get(&item.2).copied().unwrap_or_else(|| source_space(doc, &fonts, m)),
state.char_spacing,
state.word_spacing,
)?);
@@ -790,10 +801,10 @@ pub(super) fn rewrite_styled_block(
if underlines.get(&item.1).copied().or(style.underline) == Some(true) {
let [r, g, b] = overrides.get(&item.1).and_then(|r| r.color).or(style.color).unwrap_or(source.color);
let underline_size = size_of(source, item.1) * source.origin.k;
- let (x0, y0) = back.apply(x - first.origin.ctm[4], baseline - first.origin.ctm[5] - underline_size * 0.12);
- let (x1, y1) = back.apply(x + segment_width - first.origin.ctm[4], baseline - first.origin.ctm[5] - underline_size * 0.12);
+ let (x0, y0) = back.apply(x - o.ctm[4], baseline - o.ctm[5] - underline_size * 0.12);
+ let (x1, y1) = back.apply(x + segment_width - o.ctm[4], baseline - o.ctm[5] - underline_size * 0.12);
block_ops.push(Op::new("RG", vec![n(r), n(g), n(b)]));
- block_ops.push(Op::new("w", vec![n(underline_size * 0.06 / first.origin.ctm[2].hypot(first.origin.ctm[3]).max(1e-6))]));
+ block_ops.push(Op::new("w", vec![n(underline_size * 0.06 / o.ctm[2].hypot(o.ctm[3]).max(1e-6))]));
block_ops.push(Op::new("m", vec![n(x0), n(y0)]));
block_ops.push(Op::new("l", vec![n(x1), n(y1)]));
block_ops.push(Op::new("S", vec![]));
@@ -810,7 +821,6 @@ pub(super) fn rewrite_styled_block(
block_ops.push(Op::new("Q", vec![]));
let mut drop = coincident_ops(lines, &members);
drop.extend(members.iter().flat_map(|l| l.ops.iter().copied()));
- let o = &lead.origin;
let start = lead.ops.first().copied().unwrap_or(o.bt_op);
let split = ops
.get(o.bt_op..start)
diff --git a/crates/engine/Cargo.toml b/crates/engine/Cargo.toml
index 74567eb488..ff5e7611d8 100644
--- a/crates/engine/Cargo.toml
+++ b/crates/engine/Cargo.toml
@@ -37,6 +37,9 @@ pdfcraft-model = { workspace = true }
pdfcraft-geom = { workspace = true }
thiserror = { workspace = true }
png = "0.18"
+# Already in the graph (through image and pdfcraft-create): image's TiffEncoder can't set the
+# resolution or the compression of an exported page.
+tiff = "0.11"
# Already in the graph (url and idna through webbrowser, icu_properties through boa): they show
# the host of a document's link the way the browser reads it (links::display_host).
url = "2.5"
diff --git a/crates/engine/src/export.rs b/crates/engine/src/export.rs
index 6b7aad4fad..0bf4e2f79d 100644
--- a/crates/engine/src/export.rs
+++ b/crates/engine/src/export.rs
@@ -93,14 +93,14 @@ impl Exporter {
encode_png(r.width, r.height, &r.rgba)
}
- /// Page `page` as an image file of `format`.
+ /// Page `page` as an image file of `format`, which says the resolution it was drawn at.
pub fn image(&mut self, page: usize, dpi: f64, format: ImageFormat) -> Result, String> {
self.check(page)?;
let r = self.renderer.render(RenderRequest { page, scale: (dpi.clamp(18.0, 1200.0) / 72.0) as f32, ..Default::default() });
if let Some(e) = r.error {
return Err(format!("page {}: {e}", page + 1));
}
- encode_image(r.width, r.height, &r.rgba, format)
+ encode_image(r.width, r.height, &r.rgba, format, self.dpi_used(page, dpi))
}
/// The reading-order text of a page.
@@ -127,12 +127,14 @@ impl Exporter {
}
}
-/// Premultiplied RGBA → an image file of `format`.
-pub fn encode_image(width: u32, height: u32, premultiplied: &[u8], format: ImageFormat) -> Result, String> {
+/// Premultiplied RGBA → an image file of `format`. A JPEG or TIFF says it is `dpi` dots per inch,
+/// so other programs show and print it at the page's size. A PNG says nothing: its `pHYs` counts
+/// pixels per metre, so 72 dpi would read back as 72.009 and a page made from it would shrink.
+pub fn encode_image(width: u32, height: u32, premultiplied: &[u8], format: ImageFormat, dpi: f64) -> Result, String> {
match format {
ImageFormat::Png => encode_png(width, height, premultiplied),
- ImageFormat::Jpeg { quality } => encode_jpeg(width, height, premultiplied, quality),
- ImageFormat::Tiff => encode_tiff(width, height, premultiplied),
+ ImageFormat::Jpeg { quality } => encode_jpeg(width, height, premultiplied, quality, dpi),
+ ImageFormat::Tiff => encode_tiff(width, height, premultiplied, dpi),
}
}
@@ -199,23 +201,37 @@ fn over_white(premultiplied: &[u8]) -> Vec {
.collect()
}
-/// Premultiplied RGBA → baseline JPEG (pages are opaque: transparency becomes white paper).
-pub fn encode_jpeg(width: u32, height: u32, premultiplied: &[u8], quality: u8) -> Result, String> {
+/// Premultiplied RGBA → baseline JPEG at `dpi` (pages are opaque: transparency becomes white paper).
+pub fn encode_jpeg(width: u32, height: u32, premultiplied: &[u8], quality: u8, dpi: f64) -> Result, String> {
use image::ImageEncoder;
+ use image::codecs::jpeg::{JpegEncoder, PixelDensity};
let rgb = over_white(premultiplied);
let mut out = Vec::new();
- image::codecs::jpeg::JpegEncoder::new_with_quality(&mut out, quality.clamp(1, 100))
- .write_image(&rgb, width, height, image::ExtendedColorType::Rgb8)
- .map_err(|e| e.to_string())?;
+ let mut enc = JpegEncoder::new_with_quality(&mut out, quality.clamp(1, 100));
+ // The JFIF density is whole dots per inch.
+ enc.set_pixel_density(PixelDensity::dpi(dpi.round() as u16));
+ enc.write_image(&rgb, width, height, image::ExtendedColorType::Rgb8).map_err(|e| e.to_string())?;
Ok(out)
}
-/// Premultiplied RGBA → TIFF (RGB, LZW-compressed by the encoder's default).
-pub fn encode_tiff(width: u32, height: u32, premultiplied: &[u8]) -> Result, String> {
- use image::ImageEncoder;
+/// Premultiplied RGBA → TIFF (RGB, LZW-compressed) at `dpi`. Written with `tiff` directly:
+/// `image`'s `TiffEncoder` always says 1 × 1 with no unit and leaves the strips uncompressed.
+/// No predictor: LZW alone is the compression TIFF readers support most widely after none.
+pub fn encode_tiff(width: u32, height: u32, premultiplied: &[u8], dpi: f64) -> Result, String> {
+ use tiff::encoder::{Compression, Rational, TiffEncoder, colortype::RGB8};
+ use tiff::tags::{ResolutionUnit, Tag};
+ let err = |e: tiff::TiffError| e.to_string();
let rgb = over_white(premultiplied);
let mut out = std::io::Cursor::new(Vec::new());
- image::codecs::tiff::TiffEncoder::new(&mut out).write_image(&rgb, width, height, image::ExtendedColorType::Rgb8).map_err(|e| e.to_string())?;
+ let mut tiff = TiffEncoder::new(&mut out).map_err(err)?.with_compression(Compression::Lzw);
+ let mut image = tiff.new_image::(width, height).map_err(err)?;
+ // In hundredths: a page drawn at a lowered resolution (`Exporter::dpi_used`) is rarely a whole number.
+ let resolution = Rational { n: (dpi * 100.0).round() as u32, d: 100 };
+ let tags = image.encoder();
+ tags.write_tag(Tag::ResolutionUnit, ResolutionUnit::Inch).map_err(err)?;
+ tags.write_tag(Tag::XResolution, resolution.clone()).map_err(err)?;
+ tags.write_tag(Tag::YResolution, resolution).map_err(err)?;
+ image.write_data(&rgb).map_err(err)?;
Ok(out.into_inner())
}
diff --git a/crates/engine/src/lib.rs b/crates/engine/src/lib.rs
index 147fc634ec..8f39248abd 100644
--- a/crates/engine/src/lib.rs
+++ b/crates/engine/src/lib.rs
@@ -831,6 +831,11 @@ impl Document {
self.editor.as_ref().and_then(|e| pdfcraft_organize::info(&e.cos, key))
}
+ /// The custom document properties as they are now (name, text), Document Properties ▸ Custom.
+ pub fn custom_info(&self) -> Vec<(String, String)> {
+ self.editor.as_ref().map(|e| pdfcraft_organize::custom_info(&e.cos)).unwrap_or_default()
+ }
+
/// The name shown on the tab and window: the document title when the document asks for it
/// (Initial View ▸ Show: Document Title) and has a real one, else the file name.
pub fn display_name(&self) -> String {
@@ -1276,9 +1281,11 @@ pub enum Edit {
width: f64,
height: f64,
},
- /// Copies of `pages` inserted after the last of them.
+ /// Copies of `pages` inserted at `at` (an index in the page order as it is now), or after
+ /// the last of them when `at` is `None`.
DuplicatePages {
pages: Vec,
+ at: Option,
},
/// Replace Pages: the content of `pages` is replaced by `src_pages` of another PDF; the
/// pages' links, comments and bookmarks stay.
@@ -1943,7 +1950,7 @@ impl Edit {
Edit::MovePages { pages, .. } => plural("Move page", "Move pages", pages.len()),
Edit::ReversePages { .. } => Label::of("Reverse pages"),
Edit::InsertBlankPage { .. } => Label::of("Insert blank page"),
- Edit::DuplicatePages { pages } => plural("Duplicate page", "Duplicate pages", pages.len()),
+ Edit::DuplicatePages { pages, .. } => plural("Duplicate page", "Duplicate pages", pages.len()),
Edit::ReplacePages { pages, .. } => plural("Replace page", "Replace pages", pages.len()),
Edit::SetPageBox { pages, which: PageBox::Crop, .. } => plural("Crop page", "Crop pages", pages.len()),
Edit::SetPageBox { .. } => Label::of("Set page boxes"),
@@ -2330,7 +2337,7 @@ fn run_edit(doc: &mut pdfcraft_cos::Document, edit: &Edit, cx: &mut EditCtx) ->
pdfcraft_organize::insert_blank_page(doc, *at, *width, *height)?;
}
Edit::SetInfo { key, value } => pdfcraft_organize::set_info(doc, key, value)?,
- Edit::DuplicatePages { pages } => pdfcraft_organize::duplicate_pages(doc, pages)?,
+ Edit::DuplicatePages { pages, at } => pdfcraft_organize::duplicate_pages(doc, pages, *at)?,
Edit::ReplacePages { pages, name, bytes, src_pages } => {
let src = open_source(name, bytes)?;
pdfcraft_organize::replace_pages(doc, pages, &src, src_pages)?;
@@ -2402,7 +2409,10 @@ fn run_edit(doc: &mut pdfcraft_cos::Document, edit: &Edit, cx: &mut EditCtx) ->
pdfcraft_annot::erase_ink(doc, *page, *index, path, *radius, &cx.meta())?;
}
Edit::ReplaceText { page, quads, text, author, strike, caret } => {
- pdfcraft_annot::add_text_replacement(doc, *page, quads, text, author, strike, caret, &cx.meta())?;
+ // One `/NM` each: the caret replies to the strikeout, and XFDF names that reply by
+ // the strikeout's `/NM` (#2666).
+ let (strike_meta, caret_meta) = (cx.meta(), cx.meta());
+ pdfcraft_annot::add_text_replacement(doc, *page, quads, text, author, strike, caret, &strike_meta, &caret_meta)?;
}
Edit::MarkAnnotation { page, index, marked, author } => {
pdfcraft_annot::set_marked(doc, *page, *index, *marked, author, &cx.meta())?;
@@ -3902,6 +3912,7 @@ impl Session {
doc.info.keywords = info("Keywords");
doc.info.creator = info("Creator");
doc.info.producer = info("Producer");
+ doc.info.custom = pdfcraft_organize::custom_info(&editor.cos);
doc.info.file_size = bytes.len();
if !doc.signatures.is_empty() {
let signatures = signatures_of(&editor.cos, &bytes, &doc.trust, &doc.sig_cache);
@@ -4078,11 +4089,16 @@ impl Session {
/// The bytes to write for Save: an incremental update of the file as opened/last saved,
/// with `/ModDate` stamped. Call `mark_saved` after writing them successfully.
+ ///
+ /// A file whose cross-reference data had to be rebuilt is written in full even when nothing
+ /// was edited: its bytes as opened are the damaged file, and this save is how the repair the
+ /// user was told about reaches the disk (#1937). Not a signed one: a full rewrite would
+ /// break its signatures, so with no edits it keeps its bytes, damage and all.
pub fn save_bytes(&self, id: DocId) -> Result>, EditError> {
let doc = self.get(id).ok_or(EditError::NoDocument)?;
let Some(editor) = doc.editor.as_ref() else { return Ok(doc.bytes.clone()) };
let on_saved = Arc::ptr_eq(editor.cos.bytes(), &editor.saved);
- if !editor.cos.is_modified() && on_saved {
+ if !editor.cos.is_modified() && on_saved && (!editor.cos.reconstructed() || doc.is_signed()) {
return Ok(editor.saved.clone());
}
let opts = SaveOptions { mod_date: self.now().map(pdfcraft_cos::pdf_date), ..SaveOptions::default() };
diff --git a/crates/engine/src/tests.rs b/crates/engine/src/tests.rs
index 1c473b7807..f125c41dac 100644
--- a/crates/engine/src/tests.rs
+++ b/crates/engine/src/tests.rs
@@ -519,6 +519,63 @@ fn save_is_incremental_stamps_mod_date_and_rebases() {
assert_eq!(reopened.revisions().len(), 3);
}
+/// #1937: a file whose cross-reference data had to be rebuilt saved as its original bytes when
+/// nothing was edited, so the damage Document Properties reported as repaired was still in the
+/// file. Save writes the rebuilt file in full; reopened, it needs no repair.
+#[test]
+fn a_repaired_file_is_rewritten_by_save_even_without_edits() {
+ let good = fixture(2);
+ // Every offset in the table is off by the inserted line: the chain reads but leads nowhere.
+ let mut shifted = b"%PDF-1.7\n%shifted\n".to_vec();
+ shifted.extend_from_slice(&good[b"%PDF-1.7\n".len()..]);
+ // No cross-reference table at all.
+ let mut lost = good.clone();
+ lost.truncate(good.windows(9).rposition(|w| w == b"startxref").unwrap());
+ lost.extend_from_slice(b"startxref\n999999\n%%EOF\n");
+ for (name, damaged) in [("shifted", shifted), ("lost", lost)] {
+ let mut s = Session::new().with_clock(|| 1_700_000_000);
+ let id = s.open("broken.pdf", Some("/tmp/broken.pdf".into()), Arc::new(damaged.clone()), None).unwrap();
+ let doc = s.get(id).unwrap();
+ assert!(doc.repair_log().iter().any(|l| l.contains("rebuilt cross-reference index")), "{name}: {:?}", doc.repair_log());
+ assert_eq!(doc.info.pages.len(), 2, "{name}");
+ assert!(!doc.dirty, "{name}: a repair is not an edit");
+ let saved = s.save_bytes(id).unwrap();
+ assert_ne!(&saved[..], &damaged[..], "{name}: the damaged bytes are not what Save writes");
+ let reopened = pdfcraft_cos::Document::open(saved.clone()).unwrap();
+ assert!(reopened.repair_log().is_empty(), "{name}: {:?}", reopened.repair_log());
+ assert_eq!(reopened.revisions().len(), 1, "{name}");
+ s.mark_saved(id, saved.clone(), None).unwrap();
+ let doc = s.get(id).unwrap();
+ assert!(doc.repair_log().is_empty(), "{name}: {:?}", doc.repair_log());
+ assert_eq!(doc.info.pages.len(), 2, "{name}");
+ // Sound now: saving again without edits writes the same bytes.
+ assert_eq!(s.save_bytes(id).unwrap(), saved, "{name}");
+ }
+}
+
+/// A damaged file that is signed keeps its bytes on a Save without edits: rewriting it in full
+/// to put the repair on disk would invalidate its signatures without a word.
+#[test]
+fn a_signed_repaired_file_keeps_its_bytes_when_saved_without_edits() {
+ let p12 = std::fs::read(concat!(env!("CARGO_MANIFEST_DIR"), "/../sign/tests/data/ec-p256.p12")).unwrap();
+ let digital_id = sign::pkcs12::open(&p12, "test").unwrap();
+ // After the test certificate's start date (2026-10-02).
+ let mut s = Session::new().with_clock(|| 1_800_000_000);
+ let id = s.open("fixture.pdf", None, Arc::new(fixture(1)), None).unwrap();
+ let opts = SignOptions { page: 0, rect: Some([20.0, 20.0, 180.0, 60.0]), ..SignOptions::default() };
+ let signed = s.sign(id, &digital_id, opts).unwrap();
+ // A trailing `startxref` that points nowhere: the signed revision's bytes are untouched, but
+ // the file's cross-reference data has to be rebuilt.
+ let mut damaged = signed.to_vec();
+ damaged.extend_from_slice(b"\nstartxref\n999999\n%%EOF\n");
+ let mut t = Session::new().with_clock(|| 1_800_000_000);
+ let tid = t.open("damaged.pdf", Some("/tmp/damaged.pdf".into()), Arc::new(damaged.clone()), None).unwrap();
+ let doc = t.get(tid).unwrap();
+ assert!(doc.repair_log().iter().any(|l| l.contains("rebuilt cross-reference index")), "{:?}", doc.repair_log());
+ assert!(doc.is_signed());
+ assert_eq!(&t.save_bytes(tid).unwrap()[..], &damaged[..], "a signed file is not rewritten without edits");
+}
+
#[test]
fn undo_still_works_after_save() {
let (mut s, id) = session_with(2);
@@ -813,7 +870,7 @@ fn form_with_field(name: &str) -> Vec {
fn duplicating_a_page_with_a_field_keeps_the_field_name() {
let mut s = Session::new().with_clock(|| 1_700_000_000);
let id = s.open("a.pdf", None, Arc::new(form_with_field("Text1")), None).unwrap();
- s.apply(id, Edit::DuplicatePages { pages: vec![0] }).unwrap();
+ s.apply(id, Edit::DuplicatePages { pages: vec![0], at: None }).unwrap();
let doc = s.get(id).unwrap();
assert_eq!(doc.info.pages.len(), 2);
assert!(doc.form.iter().all(|f| f.name == "Text1"), "{:?}", doc.form.iter().map(|f| &f.name).collect::>());
@@ -1186,6 +1243,22 @@ fn the_permissions_password_unlocks_an_open_document() {
assert!(pdfcraft_cos::Document::open(plain).unwrap().security().is_none(), "security removed");
}
+#[test]
+fn remove_security_from_a_file_with_its_encrypt_dictionary_in_the_trailer() {
+ // MuPDF writes /Encrypt inline in the trailer rather than as an object (#2661).
+ let mut doc = pdfcraft_cos::Document::open_with_password(protected("user", "owner", -1), Some("owner")).unwrap();
+ let r = doc.trailer().reference(b"Encrypt").unwrap();
+ let dict = doc.get(r).as_dict().cloned().unwrap();
+ doc.trailer_mut().set(b"Encrypt".to_vec(), pdfcraft_cos::Object::Dict(dict));
+ let bytes = Arc::new(pdfcraft_cos::write_full(&doc, &Default::default()).unwrap());
+ let mut s = Session::new();
+ let id = s.open("m.pdf", None, bytes, Some("owner")).unwrap();
+ s.apply(id, Edit::RemoveProtection).unwrap();
+ assert!(!s.get(id).unwrap().info.encrypted);
+ let plain = s.save_bytes(id).unwrap();
+ assert!(pdfcraft_cos::Document::open(plain).unwrap().security().is_none(), "security removed");
+}
+
#[test]
fn unlocking_needs_the_permissions_password_not_the_open_password() {
let mut s = Session::new();
@@ -2110,7 +2183,7 @@ fn crop_and_duplicate_pages_show_in_the_viewer() {
assert_eq!(d.info.pages[0].crop, [10.0, 20.0, 170.0, 260.0]);
assert_eq!((d.info.pages[0].width, d.info.pages[0].height), (160.0, 240.0), "the viewer shows the cropped size");
assert_eq!(d.page_boxes()[1][1], [0.0, 0.0, 200.0, 300.0]);
- s.apply(id, Edit::DuplicatePages { pages: vec![1] }).unwrap();
+ s.apply(id, Edit::DuplicatePages { pages: vec![1], at: None }).unwrap();
assert_eq!(page_texts(&s, id), ["Page 1", "Page 2", "Page 2"]);
assert_eq!(s.get(id).unwrap().can_undo(), Some("Duplicate page"));
}
@@ -2501,6 +2574,48 @@ fn pages_export_as_png_jpeg_and_tiff() {
let tif = ex.image(0, 144.0, ImageFormat::Tiff).unwrap();
let decoded = image::load_from_memory_with_format(&tif, image::ImageFormat::Tiff).unwrap();
assert_eq!((decoded.width(), decoded.height()), (400, 600));
+ let mut tags = tiff::decoder::Decoder::new(std::io::Cursor::new(&tif)).unwrap();
+ let compression = tags.get_tag_u32(tiff::tags::Tag::Compression).unwrap();
+ let predictor = tags.get_tag_u32(tiff::tags::Tag::Predictor).unwrap();
+ assert_eq!((compression, predictor), (5, 1), "LZW, without a predictor");
+}
+
+/// A TIFF's resolution unit (2 = inch) and its horizontal and vertical resolution.
+fn tiff_resolution(bytes: &[u8]) -> (u32, f64, f64) {
+ use tiff::{decoder::ifd::Value, tags::Tag};
+ let mut d = tiff::decoder::Decoder::new(std::io::Cursor::new(bytes)).unwrap();
+ let mut res = |tag| match d.get_tag(tag).unwrap() {
+ Value::Rational(n, d) => f64::from(n) / f64::from(d),
+ v => panic!("{tag:?} is {v:?}"),
+ };
+ let (x, y) = (res(Tag::XResolution), res(Tag::YResolution));
+ (d.get_tag_u32(Tag::ResolutionUnit).unwrap(), x, y)
+}
+
+/// An exported TIFF or JPEG says the resolution it was drawn at, so other programs show and print it
+/// at the page's size. A TIFF said 1 × 1 with no unit, and a JPEG said nothing.
+#[test]
+fn exported_pages_record_the_resolution_they_were_drawn_at() {
+ use crate::export::{Exporter, ImageFormat};
+ let (s, id) = session_with(1);
+ let mut ex = Exporter::new(s.get(id).unwrap());
+ assert_eq!(tiff_resolution(&ex.image(0, 144.0, ImageFormat::Tiff).unwrap()), (2, 144.0, 144.0));
+ let jpg = ex.image(0, 144.0, ImageFormat::Jpeg { quality: 85 }).unwrap();
+ assert_eq!(&jpg[6..11], b"JFIF\0");
+ assert_eq!(jpg[13..18], [1, 0, 144, 0, 144], "JFIF density in dots per inch");
+ // pHYs counts pixels per metre: 72 dpi would read back as 72.009, and a page made from the PNG
+ // (doc_create from images) would come out 0.01% small.
+ let png = ex.image(0, 72.0, ImageFormat::Png).unwrap();
+ assert!(png::Decoder::new(std::io::Cursor::new(png)).read_info().unwrap().info().pixel_dims.is_none(), "a PNG states no resolution");
+
+ // A page too large for the renderer at the asked resolution says the one it was drawn at.
+ let mut s = Session::new().with_clock(|| 1_700_000_000);
+ let bytes = s.create_blank(14_400.0, 792.0, 1).unwrap();
+ let id = s.open("wide.pdf", None, bytes, None).unwrap();
+ let mut ex = Exporter::new(s.get(id).unwrap());
+ let used = ex.dpi_used(0, 600.0);
+ let (unit, x, y) = tiff_resolution(&ex.image(0, 600.0, ImageFormat::Tiff).unwrap());
+ assert!(unit == 2 && (x - used).abs() < 0.01 && (y - used).abs() < 0.01, "{x} × {y}, drawn at {used}");
}
#[test]
diff --git a/crates/engine/tests/marked_content_isolation.rs b/crates/engine/tests/marked_content_isolation.rs
new file mode 100644
index 0000000000..32a42d517a
--- /dev/null
+++ b/crates/engine/tests/marked_content_isolation.rs
@@ -0,0 +1,274 @@
+//! Contributor-original pages. Added marks must not inherit an unfinished hidden layer.
+use std::sync::Arc;
+
+use pdfcraft_cos::{Dict, Document, Object, SaveOptions, Stream, write_incremental};
+use pdfcraft_engine::{AddedText, Edit, HeaderFooter, Session, Watermark};
+use pdfcraft_render::{PageRenderer, RenderRequest, RequestKind};
+
+const VISIBLE: &str = "BT /F1 18 Tf 72 700 Td (Visible text) Tj ET\n";
+const HIDDEN: &str = "BT /F1 18 Tf 72 600 Td (Hidden note) Tj ET\n";
+
+fn fixture(parts: &[String]) -> Arc> {
+ let mut doc = Document::new_empty();
+ let pages = doc.get(doc.root().unwrap()).as_dict().unwrap().reference(b"Pages").unwrap();
+ let mut font = Dict::new();
+ font.set(b"Type".to_vec(), Object::name("Font"));
+ font.set(b"Subtype".to_vec(), Object::name("Type1"));
+ font.set(b"BaseFont".to_vec(), Object::name("Helvetica"));
+ let mut fonts = Dict::new();
+ fonts.set(b"F1".to_vec(), Object::Ref(doc.add(font)));
+ let mut group = Dict::new();
+ group.set(b"Type".to_vec(), Object::name("OCG"));
+ group.set(b"Name".to_vec(), Object::String(pdfcraft_cos::PdfString::text("Hidden layer")));
+ let group = doc.add(group);
+ let mut properties = Dict::new();
+ properties.set(b"L1".to_vec(), Object::Ref(group));
+ let mut resources = Dict::new();
+ resources.set(b"Font".to_vec(), Object::Dict(fonts));
+ resources.set(b"Properties".to_vec(), Object::Dict(properties));
+ let mut config = Dict::new();
+ config.set(b"OFF".to_vec(), Object::Array(vec![Object::Ref(group)]));
+ let mut oc = Dict::new();
+ oc.set(b"OCGs".to_vec(), Object::Array(vec![Object::Ref(group)]));
+ oc.set(b"D".to_vec(), Object::Dict(config));
+ doc.update_dict(doc.root().unwrap(), |d| d.set(b"OCProperties".to_vec(), Object::Dict(oc))).unwrap();
+ let streams = parts.iter().map(|s| Object::Ref(doc.add(Object::Stream(Stream::from_raw(Dict::new(), s.as_bytes().to_vec()))))).collect();
+ let mut page = Dict::new();
+ page.set(b"Type".to_vec(), Object::name("Page"));
+ page.set(b"Parent".to_vec(), Object::Ref(pages));
+ page.set(b"MediaBox".to_vec(), Object::Array(vec![0.into(), 0.into(), 612.into(), 792.into()]));
+ page.set(b"Resources".to_vec(), Object::Dict(resources));
+ page.set(b"Contents".to_vec(), Object::Array(streams));
+ let page = doc.add(page);
+ doc.update_dict(pages, |d| {
+ d.set(b"Kids".to_vec(), Object::Array(vec![Object::Ref(page)]));
+ d.set(b"Count".to_vec(), Object::Int(1));
+ })
+ .unwrap();
+ Arc::new(write_incremental(&doc, &SaveOptions::default()).unwrap())
+}
+
+fn render(bytes: Arc>) -> (Vec, String) {
+ let mut renderer = PageRenderer::new(bytes, Default::default());
+ let pixels = renderer.render(RenderRequest { page: 0, scale: 1.0, ..Default::default() });
+ assert!(pixels.error.is_none(), "{:?}", pixels.error);
+ let text = renderer.render(RenderRequest { page: 0, kind: RequestKind::Text, ..Default::default() });
+ assert!(text.error.is_none(), "{:?}", text.error);
+ let text = text.text.unwrap();
+ (pixels.rgba.to_vec(), text.text_of(0..text.glyphs.len()))
+}
+
+fn edit(kind: &str) -> Edit {
+ match kind {
+ "HEADER" | "000001" => {
+ let mut settings = HeaderFooter::default();
+ if kind == "HEADER" {
+ settings.text[1] = "HEADER".into();
+ } else {
+ settings.text[5] = "<>".into();
+ }
+ Edit::AddHeaderFooter { pages: vec![0], settings, replace: true }
+ }
+ "WATERMARK" => Edit::AddWatermark {
+ pages: vec![0],
+ settings: Watermark { text: kind.into(), opacity: 1.0, rotation: 0.0, font_size: 18.0, ..Default::default() },
+ replace: true,
+ file: None,
+ },
+ "ADDED" => Edit::AddText { page: 0, text: AddedText { text: kind.into(), rect: [72.0, 300.0, 250.0, 330.0], ..Default::default() } },
+ _ => panic!("unknown mark"),
+ }
+}
+
+fn add_and_save(input: Arc>, kind: &str) -> Arc> {
+ let mut session = Session::new();
+ let id = session.open("fixture.pdf", None, input.clone(), None).unwrap();
+ session.apply(id, edit(kind)).unwrap();
+ let saved = session.save_bytes(id).unwrap();
+ assert!(saved.starts_with(&input), "incremental save preserves the original file bytes");
+ assert_original_streams(&input, &saved);
+ let mut reopened = Session::new();
+ let id = reopened.open("saved.pdf", None, saved.clone(), None).unwrap();
+ assert_eq!(reopened.save_bytes(id).unwrap().as_slice(), saved.as_slice());
+ saved
+}
+
+#[test]
+fn all_four_mark_paths_remain_visible_after_saved_unfinished_marked_content() {
+ let mut failures = Vec::new();
+ for (case, opening, hidden) in [("hidden", "/OC /L1 BDC\n", true), ("artifact", "/Artifact BMC\n", false), ("balanced", "/OC /L1 BDC\n", true)] {
+ let closed = fixture(&[format!("{VISIBLE}{opening}{HIDDEN}EMC\n")]);
+ let input = if case == "balanced" { closed.clone() } else { fixture(&[format!("{VISIBLE}{opening}{HIDDEN}")]) };
+ let original = render(input.clone());
+ assert!(original.1.contains("Visible text"));
+ assert_eq!(original.1.contains("Hidden note"), !hidden);
+ for kind in ["HEADER", "000001", "WATERMARK", "ADDED"] {
+ let saved = add_and_save(input.clone(), kind);
+
+ let actual = render(saved);
+ let expected_bytes = add_and_save(closed.clone(), kind);
+
+ let expected = render(expected_bytes);
+ let mut wanted = vec!["Visible text", kind];
+ if !hidden {
+ wanted.push("Hidden note");
+ }
+ wanted.sort_unstable();
+ let pass = actual.0 == expected.0 && actual.1 == expected.1 && lines(&actual.1) == wanted;
+ eprintln!("{case}/{kind}: visible={} closed-control-pixels={}", actual.1.contains(kind), actual.0 == expected.0);
+ if !pass {
+ failures.push(format!("{case}/{kind}: {:?}", actual.1));
+ }
+ }
+ }
+ assert!(failures.is_empty(), "{}", failures.join("\n"));
+}
+
+fn stream_list(doc: &Document) -> Vec {
+ let page = pdfcraft_model::pages(doc).remove(0);
+ let contents = doc.resolve(page.dict.get(b"Contents").unwrap());
+ match &*contents {
+ Object::Array(items) => items.clone(),
+ _ => vec![page.dict.get(b"Contents").unwrap().clone()],
+ }
+}
+
+fn assert_original_streams(before: &Arc>, after: &Arc>) {
+ let original = Document::open(before.clone()).unwrap();
+ let saved = Document::open(after.clone()).unwrap();
+ let saved_list = stream_list(&saved);
+ for reference in stream_list(&original) {
+ assert!(saved_list.contains(&reference), "the original stream stays on the page");
+ assert_eq!(original.resolve(&reference), saved.resolve(&reference), "original encoded stream object remains unchanged");
+ }
+}
+
+fn lines(text: &str) -> Vec<&str> {
+ let mut lines: Vec<_> = text.lines().filter(|line| !line.is_empty()).collect();
+ lines.sort_unstable();
+ lines
+}
+
+#[test]
+fn nested_stream_edges_and_malformed_graphics_keep_hidden_content_hidden() {
+ let cases: Vec<(&str, Vec, usize)> = vec![
+ ("nested", vec![format!("{VISIBLE}/OC /L1 BDC /Artifact BMC {HIDDEN}")], 2),
+ ("split operands", vec![format!("{VISIBLE}/OC"), format!("/L1 BDC {HIDDEN}")], 1),
+ ("split close", vec![format!("{VISIBLE}/OC /L1 BDC {HIDDEN}"), "EMC\n".into()], 0),
+ ("trailing comment", vec![format!("{VISIBLE}/OC /L1 BDC {HIDDEN}% EMC")], 1),
+ ("cross-stream comment", vec![format!("{VISIBLE}/OC /L1 BDC {HIDDEN}% EMC"), "EMC BMC\n".into()], 1),
+ ("unclosed graphics", vec![format!("{VISIBLE}/OC /L1 BDC {HIDDEN}q")], 1),
+ ("balanced graphics open marked", vec![format!("q {VISIBLE}/OC /L1 BDC {HIDDEN}Q")], 1),
+ ("malformed graphics", vec![format!("Q {VISIBLE}/OC /L1 BDC {HIDDEN}")], 1),
+ ("invalid begins", vec![format!("{VISIBLE}/OC /L1 BDC {HIDDEN}123 BMC /Artifact BDC")], 1),
+ ("extra begins", vec![format!("{VISIBLE}123 /OC /L1 BDC {HIDDEN}123 /Artifact BMC")], 2),
+ ("extra close", vec![format!("{VISIBLE}/OC /L1 BDC {HIDDEN}123 EMC")], 0),
+ ("underflow", vec![format!("EMC {VISIBLE}/OC /L1 BDC {HIDDEN}")], 1),
+ ("operator strings", vec![format!("{VISIBLE}/OC /L1 BDC {HIDDEN}/Note <> DP")], 1),
+ ("inline samples", vec![format!("{VISIBLE}/OC /L1 BDC {HIDDEN}BI /W 11 /H 1 /BPC 8 /CS /G ID BMC BDC EMC EI\n")], 1),
+ ];
+ for (case, parts, close_count) in cases {
+ let input = fixture(&parts);
+ let mut closed = parts.clone();
+ closed.push(format!("\n{}", "EMC\n".repeat(close_count)));
+ let actual = render(add_and_save(input, "HEADER"));
+ let expected = render(add_and_save(fixture(&closed), "HEADER"));
+ assert_eq!(lines(&actual.1), lines("Visible text\nHEADER"), "{case}");
+ assert!(actual.0 == expected.0, "{case}: pixels equal the explicitly closed control");
+ }
+}
+
+#[test]
+fn replacing_and_removing_marks_restores_the_original_stream_list() {
+ let input = fixture(&[format!("{VISIBLE}/OC /L1 BDC {HIDDEN}")]);
+ let original = Document::open(input.clone()).unwrap();
+ for kind in ["HEADER", "000001", "WATERMARK"] {
+ let mut session = Session::new();
+ let mut id = session.open("fixture.pdf", None, input.clone(), None).unwrap();
+ for _ in 0..3 {
+ session.apply(id, edit(kind)).unwrap();
+ let saved = session.save_bytes(id).unwrap();
+ let page = Document::open(saved.clone()).unwrap();
+ assert_eq!(stream_list(&page).len(), 4, "{kind}: updating does not add wrappers");
+ assert_original_streams(&input, &saved);
+ assert_eq!(lines(&render(saved.clone()).1), lines(&format!("Visible text\n{kind}")));
+ session = Session::new();
+ id = session.open("saved.pdf", None, saved, None).unwrap();
+ }
+ let mark_kind = if kind == "WATERMARK" { pdfcraft_edit::MarkKind::Watermark } else { pdfcraft_edit::MarkKind::HeaderFooter };
+ session.apply(id, Edit::RemoveMarks { kind: mark_kind }).unwrap();
+ let saved = session.save_bytes(id).unwrap();
+ let page = Document::open(saved.clone()).unwrap();
+ assert_eq!(stream_list(&page), stream_list(&original), "{kind}: removal restores original references");
+ assert_original_streams(&input, &saved);
+ assert!(render(saved) == render(input.clone()), "{kind}: removal restores original pixels and text");
+ session.undo(id).unwrap();
+ assert_eq!(lines(&render(session.save_bytes(id).unwrap()).1), lines(&format!("Visible text\n{kind}")));
+ session.redo(id).unwrap();
+ assert!(render(session.save_bytes(id).unwrap()) == render(input.clone()), "redo removal restores the original rendering");
+ }
+}
+
+#[test]
+fn added_update_delete_and_readd_preserve_existing_wrapper_behavior() {
+ let input = fixture(&[format!("{VISIBLE}/OC /L1 BDC {HIDDEN}")]);
+ let mut session = Session::new();
+ let mut id = session.open("fixture.pdf", None, input.clone(), None).unwrap();
+ for _ in 0..3 {
+ session.apply(id, edit("ADDED")).unwrap();
+ assert_eq!(session.get(id).unwrap().added().len(), 1);
+ let text = AddedText { text: "UPDATED".into(), rect: [72.0, 300.0, 250.0, 330.0], ..Default::default() };
+ session.apply(id, Edit::UpdateContent { page: 0, index: 0, content: pdfcraft_edit::Content::Text(text) }).unwrap();
+ let saved = session.save_bytes(id).unwrap();
+ assert_eq!(lines(&render(saved.clone()).1), lines("Visible text\nUPDATED"));
+ assert_eq!(stream_list(&Document::open(saved.clone()).unwrap()).len(), 4);
+ assert_original_streams(&input, &saved);
+ session = Session::new();
+ id = session.open("saved.pdf", None, saved, None).unwrap();
+ session.apply(id, Edit::DeleteContent { page: 0, index: 0 }).unwrap();
+ let saved = session.save_bytes(id).unwrap();
+ assert_eq!(stream_list(&Document::open(saved.clone()).unwrap()).len(), 3, "Added deletion retains the established wrappers");
+ assert!(render(saved.clone()) == render(input.clone()), "deleting Added restores the original rendering");
+ assert_original_streams(&input, &saved);
+ session = Session::new();
+ id = session.open("saved.pdf", None, saved, None).unwrap();
+ }
+}
+
+#[test]
+fn an_existing_behind_watermark_does_not_inherit_or_reveal_hidden_content() {
+ let input = fixture(&[format!("{VISIBLE}/OC /L1 BDC {HIDDEN}")]);
+ let mut session = Session::new();
+ let id = session.open("fixture.pdf", None, input.clone(), None).unwrap();
+ session
+ .apply(
+ id,
+ Edit::AddWatermark {
+ pages: vec![0],
+ settings: Watermark { text: "BEHIND".into(), behind: true, font_size: 18.0, opacity: 1.0, rotation: 0.0, ..Default::default() },
+ replace: false,
+ file: None,
+ },
+ )
+ .unwrap();
+ session.apply(id, edit("HEADER")).unwrap();
+ let saved = session.save_bytes(id).unwrap();
+ assert_eq!(lines(&render(saved.clone()).1), lines("Visible text\nBEHIND\nHEADER"));
+ assert_original_streams(&input, &saved);
+}
+
+#[test]
+fn new_mark_closes_hidden_content_appended_after_an_old_wrapper() {
+ let bytes = fixture(&["q %PdfCraft\n".into(), VISIBLE.into(), "Q %PdfCraft\n".into(), format!("/OC /L1 BDC {HIDDEN}")]);
+ let mut doc = Document::open(bytes).unwrap();
+ let list = stream_list(&doc);
+ for index in [0, 2] {
+ let reference = list[index].as_ref().unwrap();
+ let Object::Stream(mut stream) = doc.get(reference).as_ref().clone() else { panic!("stream") };
+ stream.dict.set(b"PCMark".to_vec(), Object::name("Wrap"));
+ doc.set(reference, Object::Stream(stream));
+ }
+ let input = Arc::new(write_incremental(&doc, &SaveOptions::default()).unwrap());
+ let saved = add_and_save(input, "HEADER");
+ assert_eq!(lines(&render(saved).1), lines("Visible text\nHEADER"));
+}
diff --git a/crates/export/src/lib.rs b/crates/export/src/lib.rs
index 16dd2b06d2..30299b2d76 100644
--- a/crates/export/src/lib.rs
+++ b/crates/export/src/lib.rs
@@ -114,6 +114,34 @@ const COL_TOL: f64 = 4.0;
/// its columns, so a hostile page laid out as a staircase of blocks would otherwise make
/// (blocks / 2)² cells.
const MAX_COLS: usize = 63;
+/// Two blocks in a row this close (as a fraction of their font size) are one line of running
+/// text that changed font or colour, such as a link inside a sentence, not two table cells.
+const INLINE_GAP: f64 = 0.5;
+/// How many blocks ending near a block's start are looked at to find the one it continues; a
+/// page laid out so that many end at the same place stays quick.
+const CONTINUATION_PROBE: usize = 64;
+/// Per block, whether it continues another on the same line: it starts where that one ends (up to
+/// [`INLINE_GAP`] later, or a point earlier for rounding), so it is the same line of text running
+/// on (a link inside a sentence, a superscript), not a cell of its own.
+fn continuations(blocks: &[Block]) -> Vec {
+ let mut by_end: Vec<(f64, usize)> = blocks.iter().enumerate().map(|(i, b)| (b.rect[2], i)).collect();
+ by_end.sort_by(|a, b| a.0.total_cmp(&b.0));
+ blocks
+ .iter()
+ .enumerate()
+ .map(|(bi, b)| {
+ let k = by_end.partition_point(|(end, _)| *end < b.rect[0] - INLINE_GAP * b.size);
+ by_end
+ .get(k..)
+ .unwrap_or_default()
+ .iter()
+ .take_while(|(end, _)| *end <= b.rect[0] + 1.0)
+ .take(CONTINUATION_PROBE)
+ .filter_map(|(_, ai)| blocks.get(*ai).filter(|_| *ai != bi))
+ .any(|a| a.rect[3].min(b.rect[3]) - a.rect[1].max(b.rect[1]) > 0.5 * (a.rect[3] - a.rect[1]).min(b.rect[3] - b.rect[1]))
+ })
+ .collect()
+}
/// A cell is short: taller blocks are body text (or a multi-column layout), not table cells.
fn is_cell_like(b: &Block) -> bool {
(b.rect[3] - b.rect[1]) <= b.size * 5.0 && !b.text.trim().is_empty()
@@ -264,13 +292,20 @@ pub(crate) fn tables(blocks: &[Block]) -> (Vec, Vec) {
es.dedup_by(|a, b| (*a - *b).abs() <= COL_TOL);
es
};
+ let cont = continuations(blocks);
let qual: Vec = rows
.iter()
.map(|row| {
- let mut cs: Vec = row.iter().map(|(c, _)| *c).collect();
+ // A block that continues another adds no column of its own, and the cells of a table row
+ // sit side by side: blocks that overlap are a paragraph's lines, not cells (#2806).
+ let cells: Vec<(usize, usize)> = row.iter().copied().filter(|(_, bi)| !cont.get(*bi).copied().unwrap_or(false)).collect();
+ let mut spans: Vec<[f64; 2]> = cells.iter().filter_map(|(_, bi)| blocks.get(*bi)).map(|b| [b.rect[0], b.rect[2]]).collect();
+ spans.sort_by(|a, b| a[0].total_cmp(&b[0]));
+ let apart = spans.windows(2).all(|w| w[1][0] >= w[0][1] - COL_TOL);
+ let mut cs: Vec = cells.iter().map(|(c, _)| *c).collect();
cs.sort_unstable();
cs.dedup();
- cs.len() >= 2
+ apart && cs.len() >= 2
})
.collect();
let mut i = 0;
@@ -1186,6 +1221,75 @@ mod tests {
assert_eq!(grid_cols(&grid(64)), 0);
}
+ fn block(t: &str, x0: f64, y0: f64, x1: f64, y1: f64) -> Block {
+ Block { text: t.into(), rect: [x0, y0, x1, y1], size: 12.0, bold: false, italic: false, color: [0.0; 3], rotation: 0.0 }
+ }
+
+ /// #2806: a sentence whose link is set in another colour or font runs on as separate blocks
+ /// on one line. They are not table columns, even when several such lines share left edges.
+ #[test]
+ fn a_link_inside_a_sentence_is_not_a_table_column() {
+ let blue = |t: &str, x0: f64, y0: f64, x1: f64, y1: f64| Block { color: [0.02, 0.388, 0.757], ..block(t, x0, y0, x1, y1) };
+ let p = Page {
+ width: 612.0,
+ height: 792.0,
+ blocks: vec![
+ block("6.3. 2. All content shown at", 121.7, 710.3, 272.0, 724.9),
+ blue("http", 273.5, 710.3, 294.1, 724.9),
+ blue("s", 294.1, 709.8, 298.8, 725.8),
+ blue("://women", 298.8, 710.3, 348.0, 724.9),
+ blue("g", 348.1, 709.8, 353.7, 725.8),
+ blue("randprix.fide.com", 353.7, 710.3, 443.2, 724.9),
+ block("shall be reviewed and approved by the Press Officer.", 164.3, 622.5, 533.1, 724.9),
+ block("6.3. 3. The Organiser shall warrant that", 121.7, 601.8, 345.1, 616.5),
+ blue("https://womengrandprix.fide.com", 350.2, 601.3, 519.9, 617.4),
+ block("is managed in a professional way and furthermore that it is capable of", 164.3, 557.9, 533.0, 616.5),
+ block("6. 3. 4. The Press Officer shall report to the FIDE Head of PR.", 121.7, 516.6, 530.3, 551.9),
+ block("conducted during a WGP tournament.", 164.3, 502.0, 352.5, 516.6),
+ ],
+ images: Vec::new(),
+ };
+ let h = html(std::slice::from_ref(&p), "Regulations");
+ assert!(!h.contains(""), "running text became a table: {h}");
+ for b in &p.blocks {
+ assert!(h.contains(&b.text), "{} was lost: {h}", b.text);
+ }
+ }
+
+ /// #2806: lines that each run on into a link starting at the same place are sentences, not a
+ /// two-column table, even three in a row.
+ #[test]
+ fn sentences_that_run_on_into_links_are_not_a_table() {
+ let blue = |t: &str, x0: f64, y0: f64, x1: f64, y1: f64| Block { color: [0.02, 0.388, 0.757], ..block(t, x0, y0, x1, y1) };
+ let mut blocks = Vec::new();
+ for (n, lead) in ["Rules are published at", "Pairings are posted at", "Results are listed at"].iter().enumerate() {
+ let y = 700.0 - 16.0 * n as f64;
+ blocks.push(block(lead, 72.0, y, 211.0, y + 12.0));
+ blocks.push(blue("https://example.org/event", 212.5, y, 360.0, y + 12.0));
+ }
+ let p = Page { width: 612.0, height: 792.0, blocks, images: Vec::new() };
+ let h = html(std::slice::from_ref(&p), "Event");
+ assert!(!h.contains(""), "sentences became a table: {h}");
+ assert_eq!(h.matches("https://example.org/event").count(), 3, "{h}");
+ }
+
+ /// #2806: a superscript right after a number ("1" "st") continues its cell; the prize table
+ /// around it stays a table.
+ #[test]
+ fn a_superscript_beside_a_number_keeps_its_table() {
+ let mut blocks = vec![block("PLACE", 100.0, 700.0, 140.0, 712.0), block("PRIZE (euros)", 300.0, 700.0, 380.0, 712.0)];
+ for (n, (row, sup)) in [("1", "st"), ("2", "nd"), ("3", "rd"), ("4", "th")].iter().enumerate() {
+ let y = 680.0 - 20.0 * n as f64;
+ blocks.push(block(row, 100.0, y, 106.0, y + 12.0));
+ blocks.push(Block { size: 8.0, ..block(sup, 106.0, y + 5.0, 113.0, y + 13.0) });
+ blocks.push(block(&format!("{},000", 30 - 5 * n), 300.0, y, 340.0, y + 12.0));
+ }
+ let p = Page { width: 612.0, height: 792.0, blocks, images: Vec::new() };
+ let h = html(std::slice::from_ref(&p), "Prizes");
+ assert!(h.contains(""), "the prize table was lost: {h}");
+ assert!(h.contains("30,000") && h.contains("15,000"), "{h}");
+ }
+
#[test]
fn two_column_body_text_is_not_mistaken_for_a_table() {
let para = |t: &str, x: f64, y: f64| Block {
diff --git a/crates/fonts/src/embedding.rs b/crates/fonts/src/embedding.rs
index 84801d5fd2..a5bb011e00 100644
--- a/crates/fonts/src/embedding.rs
+++ b/crates/fonts/src/embedding.rs
@@ -195,7 +195,7 @@ pub fn embed_mapping(
// page; source subset programs stay untouched.
let existing = written(doc, fonts).find(|w| w.has_program(doc, bytes)).map(|w| w.program).or_else(|| {
doc.scan_objects().take(8192).find_map(|(r, o)| match o.as_ref() {
- Object::Stream(s) if s.dict.get(b"PCFullFont").is_some() && s.raw.as_ref() == bytes => Some(Object::Ref(r)),
+ Object::Stream(s) if is_program(doc, s, bytes) => Some(Object::Ref(r)),
_ => None,
})
});
@@ -206,7 +206,7 @@ pub fn embed_mapping(
d.set(b"Subtype".to_vec(), Object::name("OpenType"));
}
d.set(b"PCFullFont".to_vec(), Object::Bool(true));
- Object::Ref(doc.add(Object::Stream(Stream::from_raw(d, bytes.to_vec()))))
+ Object::Ref(doc.add(Object::Stream(Stream::flate(d, bytes))))
});
descriptor.set(if cff { b"FontFile3".to_vec() } else { b"FontFile2".to_vec() }, program);
let desc = Object::Ref(doc.add(Object::Dict(descriptor)));
@@ -264,15 +264,20 @@ struct Written {
}
impl Written {
- /// Whether its program is `bytes`. The length it was written with is compared before decoding.
+ /// Whether its program is `bytes`.
fn has_program(&self, doc: &Document, bytes: &[u8]) -> bool {
- let Object::Stream(s) = doc.resolve(&self.program).as_ref().clone() else { return false };
- s.dict.get(b"PCFullFont").is_some()
- && s.dict.get(b"Length1").and_then(|l| doc.resolve(l).as_int()) == i64::try_from(bytes.len()).ok()
- && s.decoded_within(17 << 20).is_ok_and(|data| data == bytes)
+ matches!(doc.resolve(&self.program).as_ref(), Object::Stream(s) if is_program(doc, s, bytes))
}
}
+/// Whether `s` is a full program written for `bytes`, compressed or not. The length it was
+/// written with is compared before decoding.
+fn is_program(doc: &Document, s: &Stream, bytes: &[u8]) -> bool {
+ s.dict.get(b"PCFullFont").is_some()
+ && s.dict.get(b"Length1").and_then(|l| doc.resolve(l).as_int()) == i64::try_from(bytes.len()).ok()
+ && s.decoded_within(17 << 20).is_ok_and(|data| data == bytes)
+}
+
/// The `PCFull` resources in `fonts`, in order.
fn written<'a>(doc: &'a Document, fonts: &'a Dict) -> impl Iterator- + 'a {
fonts.iter().filter(|(key, _)| key.starts_with(b"PCFull")).take(256).filter_map(move |(key, value)| {
@@ -379,4 +384,31 @@ mod tests {
assert!(!fonts.contains(medium.as_bytes()), "{medium} is new");
assert_eq!(programs(&doc), 2);
}
+
+ #[test]
+ fn the_full_program_is_written_compressed_and_another_page_reuses_it() {
+ let bytes = include_bytes!("../../../assets/fonts/Inter-Regular.ttf");
+ let mut doc = Document::new_empty();
+ let mut first = Dict::new();
+ embed(&mut doc, &mut first, bytes, "Inter").unwrap();
+ let program = || {
+ doc.scan_objects()
+ .find_map(|(_, o)| match o.as_ref() {
+ Object::Stream(s) if s.dict.get(b"PCFullFont").is_some() => Some(s.clone()),
+ _ => None,
+ })
+ .unwrap()
+ };
+ let stream = program();
+ assert_eq!(stream.dict.name(b"Filter"), Some(b"FlateDecode".as_slice()));
+ assert!(stream.raw.len() < bytes.len(), "{} of {} bytes", stream.raw.len(), bytes.len());
+ assert_eq!(stream.decoded().unwrap(), bytes);
+ // Another page's resources don't have it, so it is found among the document's objects.
+ let objects = doc.scan_objects().count();
+ let mut second = Dict::new();
+ embed(&mut doc, &mut second, bytes, "Inter").unwrap();
+ let programs = doc.scan_objects().filter(|(_, o)| matches!(o.as_ref(), Object::Stream(s) if s.dict.get(b"PCFullFont").is_some())).count();
+ assert_eq!(programs, 1, "one program for both pages");
+ assert!(doc.scan_objects().count() > objects, "the second page gets its own resource");
+ }
}
diff --git a/crates/fonts/src/lib.rs b/crates/fonts/src/lib.rs
index cd0d8358e5..a1f2ff2779 100644
--- a/crates/fonts/src/lib.rs
+++ b/crates/fonts/src/lib.rs
@@ -80,13 +80,19 @@ pub fn wrap_with(text: &str, size: f64, width: f64, measure: impl Fn(&str, f64)
let mut lines = Vec::new();
for para in text.split(['\n', '\r']) {
let mut line = String::new();
+ // No word is on the line yet; an empty word (a leading space) still starts it.
+ let mut fresh = true;
for word in para.split(' ') {
- let candidate = if line.is_empty() { word.to_string() } else { format!("{line} {word}") };
- if measure(&candidate, size) <= width || line.is_empty() && measure(word, size) <= width {
+ let candidate = if fresh { word.to_string() } else { format!("{line} {word}") };
+ if measure(&candidate, size) <= width || fresh && measure(word, size) <= width {
line = candidate;
+ fresh = false;
continue;
}
- if !line.is_empty() {
+ // Indentation alone is not a line: drop it before an over-long word.
+ if line.trim().is_empty() {
+ line.clear();
+ } else {
lines.push(std::mem::take(&mut line));
}
for ch in word.chars() {
@@ -95,6 +101,7 @@ pub fn wrap_with(text: &str, size: f64, width: f64, measure: impl Fn(&str, f64)
}
line.push(ch);
}
+ fresh = line.is_empty();
}
lines.push(line);
}
@@ -256,6 +263,20 @@ pub fn literal(bytes: &[u8]) -> Vec
{
mod tests {
use super::*;
+ #[test]
+ fn wrap_keeps_leading_spaces() {
+ assert_eq!(wrap(" indented", 11.0, 500.0), [" indented"]);
+ assert_eq!(wrap(" a b\n c", 11.0, 500.0), [" a b", " c"]);
+ assert_eq!(wrap("", 11.0, 500.0), [""]);
+ // A single leading space before a word that must be broken leaves no blank first line.
+ let broken = wrap(" abcdefghijklmnopqrstuvwxyz", 11.0, 40.0);
+ assert!(broken.len() > 1 && broken.iter().all(|l| !l.is_empty()), "{broken:?}");
+ // So does indentation of two or more spaces: no whitespace-only line, same as one space.
+ let broken = wrap(" abcdefghijklmnopqrstuvwxyz", 11.0, 40.0);
+ assert!(broken.len() > 1 && broken.iter().all(|l| !l.trim().is_empty()), "{broken:?}");
+ assert_eq!(broken, wrap(" abcdefghijklmnopqrstuvwxyz", 11.0, 40.0));
+ }
+
#[test]
fn win_ansi_encodability_matches_the_encoder() {
for c in ['a', 'é', 'ÿ', '€', '—', '™', '\t', '?'] {
diff --git a/crates/forms/README.md b/crates/forms/README.md
index 3280cc2431..66fb919fe7 100644
--- a/crates/forms/README.md
+++ b/crates/forms/README.md
@@ -56,6 +56,13 @@ delete_field(&mut doc, "email")?;
Button1, Date1, Signature1; `/Helv` is added to `/DR`; appearances are drawn at once. Radio
buttons joining a group become kid widgets with their own export state. Renaming changes only
the last part of a dotted name.
+- `Widget::caption` reads a push button's normal label (`/MK /CA`, including indirect values).
+ `FieldProps { caption: Some((widget_index, text)), ..Default::default() }` changes or clears
+ that label, independently of the field name. The index is 0-based. A label-only edit redraws
+ just that widget, copying shared/indirect `/MK` and appearance dictionaries before editing.
+ Other widgets, actions, icons, rotation and unknown entries are preserved. Existing icon-only
+ layouts keep hiding the caption; caption rendering retains the existing Helvetica/WinAnsi
+ limits and reports unavailable characters as above.
## Not yet
diff --git a/crates/forms/src/af.rs b/crates/forms/src/af.rs
index b50ed4c1ed..129f7fa751 100644
--- a/crates/forms/src/af.rs
+++ b/crates/forms/src/af.rs
@@ -587,7 +587,9 @@ pub fn parse_date(text: &str, fmt: &str) -> Option {
return None;
}
let lower = t.to_lowercase();
- let pm = lower.contains("pm") || lower.ends_with('p');
+ // A bare trailing "p" ("5:30p") only counts when it follows a digit, not a word like "Sep".
+ let bare_p = lower.strip_suffix('p').is_some_and(|rest| rest.trim_end().ends_with(|c: char| c.is_ascii_digit()));
+ let pm = lower.contains("pm") || bare_p;
let am = lower.contains("am");
// Tokens: numbers and month names.
let mut nums: Vec = Vec::new();
@@ -1285,3 +1287,25 @@ mod never_crash_tests {
let _ = format_date(dt, "dddd ddd");
}
}
+
+#[cfg(test)]
+mod pm_marker_tests {
+ use super::*;
+
+ /// A month name ending in "p" ("Sep") is not a PM marker; only a "p" attached to a time is.
+ #[test]
+ fn month_name_ending_in_p_is_not_pm() {
+ let dt = parse_date("2026 05 Sep", "yyyy dd mmm HH:MM").expect("valid date");
+ assert_eq!((dt.m, dt.d, dt.hh, dt.mm), (9, 5, 0, 0));
+ let f = Format::Date("yyyy dd mmm HH:MM".into());
+ assert_eq!(format_value(&f, "2026 05 Sep"), "2026 05 Sep 00:00");
+ assert!(keystroke(&f, "date", "2026 05 Sep").is_ok());
+ }
+
+ #[test]
+ fn bare_p_after_a_time_is_still_pm() {
+ assert_eq!(parse_date("05/09/2026 5:30p", "mm/dd/yyyy h:MM").map(|d| d.hh), Some(17));
+ assert_eq!(parse_date("05/09/2026 5:30 p", "mm/dd/yyyy h:MM").map(|d| d.hh), Some(17));
+ assert_eq!(parse_date("05/09/2026 5:30 pm", "mm/dd/yyyy h:MM").map(|d| d.hh), Some(17));
+ }
+}
diff --git a/crates/forms/src/author.rs b/crates/forms/src/author.rs
index bb45f38d9f..521d1f5278 100644
--- a/crates/forms/src/author.rs
+++ b/crates/forms/src/author.rs
@@ -63,6 +63,9 @@ impl NewField {
pub struct FieldProps {
pub name: Option,
pub tooltip: Option,
+ /// Normal push-button label (`/MK /CA`) of one widget (0-based index).
+ /// An empty string clears the label; other widgets and icon layouts are preserved.
+ pub caption: Option<(usize, String)>,
pub read_only: Option,
pub required: Option,
pub multiline: Option,
@@ -657,6 +660,7 @@ pub fn add_field(doc: &mut Document, page: usize, rect: [f64; 4], kind: &NewFiel
locked: false,
hidden: false,
rotation: 0,
+ caption: None,
list: None,
};
let ap = appearance::check_box_states(doc, &w, FieldKind::Radio, export);
@@ -677,9 +681,16 @@ pub fn add_field(doc: &mut Document, page: usize, rect: [f64; 4], kind: &NewFiel
/// (`pdfcraft_fonts::UNDRAWN_MARK`) and other viewers are asked to redraw the form (see
/// `ask_viewers_to_redraw`), as when a value is set.
pub fn redraw_field(doc: &mut Document, name: &str) -> Result<(), FormError> {
+ redraw_widgets(doc, name, None)
+}
+
+fn redraw_widgets(doc: &mut Document, name: &str, only: Option) -> Result<(), FormError> {
let f = fields(doc).into_iter().find(|f| f.name == name).ok_or_else(|| FormError::NoSuchField(name.into()))?;
let mut undrawn = false;
- for w in &f.widgets {
+ for (index, w) in f.widgets.iter().enumerate() {
+ if only.is_some_and(|only| only != index) {
+ continue;
+ }
let mut missing = None;
let ap = match f.kind {
FieldKind::Text | FieldKind::Combo | FieldKind::List => {
@@ -861,6 +872,15 @@ pub fn set_props(doc: &mut Document, name: &str, props: &FieldProps) -> Result> = None;
@@ -1178,7 +1198,18 @@ pub fn set_props(doc: &mut Document, name: &str, props: &FieldProps) -> Result,
/// List boxes: where the options are drawn, so a click can pick the one under the pointer.
pub list: Option,
}
@@ -802,6 +804,13 @@ fn walk(
locked: annot_flags & 128 != 0,
hidden: annot_flags & (2 | 32) != 0,
rotation: appearance::mk_rotation(doc, wd),
+ caption: (kind == FieldKind::PushButton)
+ .then(|| {
+ let mk = doc.resolve(wd.get(b"MK")?);
+ let value = doc.resolve(mk.as_dict()?.get(b"CA")?);
+ value.as_string().map(|s| s.to_text())
+ })
+ .flatten(),
list: (kind == FieldKind::List).then(|| appearance::list_rows_of(doc, inh.da.as_deref().unwrap_or("/Helv 0 Tf 0 g"), Some(d), wd)),
})
})
diff --git a/crates/forms/src/tests.rs b/crates/forms/src/tests.rs
index 60066e4739..fef5bcfbd8 100644
--- a/crates/forms/src/tests.rs
+++ b/crates/forms/src/tests.rs
@@ -87,6 +87,108 @@ fn ap(doc: &Document, w: &Widget) -> String {
String::from_utf8_lossy(&s.decoded().unwrap()).into_owned()
}
+#[test]
+fn button_caption_preserves_other_widgets_and_shared_dictionaries() {
+ for indirect_caption in [false, true] {
+ let mut doc = one_page();
+ let a = add_field(&mut doc, 0, [40.0, 600.0, 200.0, 630.0], &NewField::Button { caption: "Original".into() }, Some("first")).unwrap();
+ let b = add_field(&mut doc, 0, [40.0, 500.0, 200.0, 530.0], &NewField::Button { caption: "Original".into() }, Some("second")).unwrap();
+ let all = fields(&doc);
+ let (first, second) = (field(&all, &a).obj, field(&all, &b).obj);
+ let mut parent = Dict::new();
+ parent.set(b"FT".to_vec(), Object::name("Btn"));
+ parent.set(b"Ff".to_vec(), Object::Int(flags::PUSH_BUTTON as i64));
+ parent.set(b"T".to_vec(), PdfString::text("clear_button"));
+ parent.set(b"Kids".to_vec(), Object::Array(vec![Object::Ref(first), Object::Ref(second)]));
+ let mut action = Dict::new();
+ action.set(b"S".to_vec(), Object::name("ResetForm"));
+ parent.set(b"A".to_vec(), Object::Dict(action.clone()));
+ let parent = doc.add(parent);
+ let mut mk = Dict::new();
+ let ca = Object::String(PdfString::text("Original"));
+ let ca = if indirect_caption { Object::Ref(doc.add(ca)) } else { ca };
+ mk.set(b"CA".to_vec(), ca);
+ mk.set(b"TP".to_vec(), Object::Int(0));
+ mk.set(b"R".to_vec(), Object::Int(90));
+ mk.set(b"VendorNote".to_vec(), PdfString::text("kept"));
+ let mk = doc.add(mk);
+ let mut shared_ap = doc.get(first).as_dict().unwrap().get(b"AP").unwrap().as_dict().unwrap().clone();
+ let normal = shared_ap.get(b"N").unwrap().clone();
+ shared_ap.set(b"R".to_vec(), normal.clone());
+ shared_ap.set(b"D".to_vec(), normal);
+ shared_ap.set(b"VendorAP".to_vec(), PdfString::text("kept"));
+ let shared_ap = doc.add(shared_ap);
+ for widget in [first, second] {
+ doc.update_dict(widget, |d| {
+ for key in [b"FT".as_slice(), b"Ff", b"T"] {
+ d.remove(key);
+ }
+ d.set(b"Parent".to_vec(), Object::Ref(parent));
+ d.set(b"MK".to_vec(), Object::Ref(mk));
+ d.set(b"AP".to_vec(), Object::Ref(shared_ap));
+ })
+ .unwrap();
+ }
+ let af = doc.get(doc.root().unwrap()).as_dict().unwrap().reference(b"AcroForm").unwrap();
+ doc.update_dict(af, |d| d.set(b"Fields".to_vec(), Object::Array(vec![Object::Ref(parent)]))).unwrap();
+ let original = pdfcraft_cos::write_full(&doc, &SaveOptions::default()).unwrap();
+ let original = Document::open(Arc::new(original)).unwrap();
+ // Full saves can renumber objects: reacquire references from the reopened model.
+ let f = field(&fields(&original), "clear_button").clone();
+ let first = f.widgets[0].obj;
+ let mk = original.get(first).as_dict().unwrap().reference(b"MK").unwrap();
+ for full in [false, true] {
+ let mut doc = original.clone();
+ let untouched = doc.get(first).clone();
+ let shared_mk = doc.get(mk).clone();
+ assert_eq!(field(&fields(&doc), "clear_button").widgets[1].caption.as_deref(), Some("Original"));
+ set_props(&mut doc, "clear_button", &FieldProps { caption: Some((1, "Clear Form".into())), ..Default::default() }).unwrap();
+ assert_eq!(doc.get(first), untouched, "the other widget and all its appearances are untouched");
+ assert_eq!(doc.get(mk), shared_mk, "shared MK is not patched in place");
+ let bytes = if full {
+ pdfcraft_cos::write_full(&doc, &SaveOptions::default()).unwrap()
+ } else {
+ let bytes = write_incremental(&doc, &SaveOptions::default()).unwrap();
+ assert!(bytes.starts_with(original.bytes().as_slice()));
+ assert!(bytes.len() > original.bytes().len());
+ bytes
+ };
+ let doc = Document::open(Arc::new(bytes)).unwrap();
+ let f = field(&fields(&doc), "clear_button").clone();
+ assert_eq!(f.widgets.iter().map(|w| w.caption.as_deref()).collect::>(), [Some("Original"), Some("Clear Form")]);
+ assert_eq!(f.button, Some(af::ButtonAction::Reset { fields: Vec::new(), exclude: false }));
+ assert_eq!(doc.get(f.obj).as_dict().unwrap().get(b"A"), Some(&Object::Dict(action.clone())));
+ let w = doc.get(f.widgets[1].obj);
+ let mk = w.as_dict().unwrap().get(b"MK").unwrap().as_dict().unwrap();
+ assert_eq!(mk.get(b"VendorNote").unwrap().as_string().unwrap().to_text(), "kept");
+ assert_eq!(mk.get(b"R").unwrap().as_int(), Some(90));
+ assert_eq!(mk.get(b"TP").unwrap().as_int(), Some(0));
+ let apd = w.as_dict().unwrap().get(b"AP").unwrap().as_dict().unwrap();
+ assert!(apd.contains(b"VendorAP"));
+ assert!(!apd.contains(b"R") && !apd.contains(b"D"), "stale hover/pressed text must not survive");
+ assert!(ap(&doc, &f.widgets[1]).contains("(Clear Form) Tj"));
+ }
+ }
+}
+
+#[test]
+fn button_caption_rejects_wrong_type_and_index_before_mutation() {
+ let mut doc = fixture();
+ for (name, wi) in [("name", 0), ("agree", 0), ("go", usize::MAX)] {
+ let before = pdfcraft_cos::write_full(&doc, &SaveOptions::default()).unwrap();
+ let props = FieldProps { caption: Some((wi, "Clear Form".into())), name: Some("renamed".into()), ..Default::default() };
+ assert!(set_props(&mut doc, name, &props).is_err());
+ assert_eq!(pdfcraft_cos::write_full(&doc, &SaveOptions::default()).unwrap(), before);
+ }
+ set_props(&mut doc, "go", &FieldProps { caption: Some((0, "提交".into())), ..Default::default() }).unwrap();
+ assert_eq!(field(&fields(&doc), "go").widgets[0].caption.as_deref(), Some("提交"));
+ assert_eq!(undrawn_fields(&doc), [("go".to_string(), '提')]);
+ set_props(&mut doc, "go", &FieldProps { caption: Some((0, String::new())), ..Default::default() }).unwrap();
+ assert_eq!(field(&fields(&doc), "go").widgets[0].caption.as_deref(), Some(""));
+ assert!(undrawn_fields(&doc).is_empty());
+ assert!(!ap(&doc, &field(&fields(&doc), "go").widgets[0]).contains(" Tj"));
+}
+
/// Read the drawn strings together with the appearance's actual font after saving and reopening.
fn appearance_text(doc: &Document, w: &Widget, font_name: &[u8]) -> (Object, Vec>) {
let widget = doc.get(w.obj);
diff --git a/crates/js/src/formcalc.rs b/crates/js/src/formcalc.rs
index 059276e936..dc32d457bf 100644
--- a/crates/js/src/formcalc.rs
+++ b/crates/js/src/formcalc.rs
@@ -712,6 +712,13 @@ fn num_text(n: f64) -> String {
if s.is_empty() || s == "-0" { "0".into() } else { s.to_string() }
}
+/// `v` with `decimals` places, rounding exact halves away from zero (Rust's formatter rounds them to even).
+fn fixed(v: f64, decimals: usize) -> String {
+ let k = 10f64.powi(decimals.min(20) as i32);
+ let scaled = (v * k).round() / k;
+ format!("{:.decimals$}", if scaled.is_finite() { scaled } else { v })
+}
+
/// The number a string holds: its leading numeric part (`12abc` → 12), `None` for none.
fn parse_num(s: &str) -> Option {
let t = s.trim();
@@ -1926,6 +1933,8 @@ fn days_in(y: i64, m: u32) -> u32 {
}
}
+const WEEKDAYS: [&str; 7] = ["Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday"];
+
fn format_date(y: i64, m: u32, d: u32, picture: &str) -> String {
let pic: Vec = picture.chars().collect();
let mut out = String::new();
@@ -1948,6 +1957,19 @@ fn format_date(y: i64, m: u32, d: u32, picture: &str) -> String {
('D', 1) => out.push_str(&d.to_string()),
(_, _) => out.push_str(&format!("{d:02}")),
}
+ } else if pic[i] == 'E' {
+ let mut run = 1;
+ while i + run < pic.len() && pic[i + run] == 'E' {
+ run += 1;
+ }
+ i += run;
+ // Day 1 (1900-01-01) was a Monday, so the serial modulo 7 indexes from Sunday.
+ let weekday = days_from_civil(y, m, d).rem_euclid(7) as usize;
+ match run {
+ 1 | 2 => out.push_str(&(weekday + 1).to_string()),
+ 3 => out.push_str(WEEKDAYS.get(weekday).map_or("", |n| &n[..3])),
+ _ => out.push_str(WEEKDAYS.get(weekday).copied().unwrap_or("")),
+ }
} else {
out.push(pic[i]);
i += 1;
@@ -2166,7 +2188,7 @@ impl Interp<'_> {
let v = self.arg_num(args, 0)?;
let width = if args.len() > 1 { self.arg_num(args, 1)?.clamp(0.0, 1000.0) as usize } else { 10 };
let prec = if args.len() > 2 { self.arg_num(args, 2)?.clamp(0.0, 20.0) as usize } else { 0 };
- let text = format!("{v:.prec$}");
+ let text = fixed(v, prec);
Val::Str(if text.len() > width { "*".repeat(width) } else { format!("{text:>width$}") })
}
"substr" => {
@@ -2255,7 +2277,7 @@ impl Interp<'_> {
let num = self.to_num(&v);
let decimals = pic.split_once('.').map_or(0, |(_, f)| f.chars().filter(|c| matches!(c, '9' | 'z' | 'Z')).count()).min(20);
let grouped = pic.contains(',');
- let mut text = format!("{:.decimals$}", num.abs());
+ let mut text = fixed(num.abs(), decimals);
if grouped {
let (int, frac) = text.split_once('.').map_or((text.as_str(), ""), |(a, b)| (a, b));
let mut g = String::new();
diff --git a/crates/js/src/tests.rs b/crates/js/src/tests.rs
index 55fa45760d..61d3a71e51 100644
--- a/crates/js/src/tests.rs
+++ b/crates/js/src/tests.rs
@@ -636,6 +636,34 @@ mod formcalc {
assert_eq!(calc("; a comment\n// another\nSum(1,\n 2,\n 3)"), "6");
}
+ #[test]
+ fn date_pictures_render_the_weekday_letters() {
+ assert_eq!(calc(r#"Num2Date(45350, DateFmt(4))"#), "Thursday, February 29, 2024");
+ assert_eq!(calc(r#"Num2Date(45350, "EEE, YYYY-MM-DD")"#), "Thu, 2024-02-29");
+ assert_eq!(calc(r#"Num2Date(45350, "EEEE")"#), "Thursday");
+ assert_eq!(calc(r#"Num2Date(45350, "E")"#), "5");
+ assert_eq!(calc(r#"Num2Date(45351, "EEEE")"#), "Friday");
+ assert_eq!(calc(r#"Num2Date(45352, "EEEE")"#), "Saturday");
+ assert_eq!(calc(r#"Num2Date(45353, "EEEE")"#), "Sunday");
+ assert_eq!(calc(r#"Num2Date(1, "EEEE")"#), "Monday");
+ assert_eq!(calc(r#"Format("EEEE, YYYY-MM-DD", 45350)"#), "Thursday, 2024-02-29");
+ // Other letters are still literal text.
+ assert_eq!(calc(r#"Num2Date(45350, "YYYY-MM-DD x")"#), "2024-02-29 x");
+ }
+
+ #[test]
+ fn str_and_format_round_exact_ties_away_from_zero() {
+ assert_eq!(calc("Str(2.5, 3, 0)"), " 3");
+ assert_eq!(calc("Str(0.5, 3, 0)"), " 1");
+ assert_eq!(calc("Str(-2.5, 4, 0)"), " -3");
+ assert_eq!(calc("Str(1.125, 6, 2)"), " 1.13");
+ assert_eq!(calc(r#"Format("9", 2.5)"#), "3");
+ assert_eq!(calc(r#"Format("9.99", 0.125)"#), "0.13");
+ assert_eq!(calc(r#"Format("9.9", 0.25)"#), "0.3");
+ assert_eq!(calc(r#"Format("9.9", -0.25)"#), "-0.3");
+ assert_eq!(calc(r#"Format("9.99", 2.344)"#), "2.34");
+ }
+
#[test]
fn built_in_functions_cover_the_common_set() {
assert_eq!(calc("Abs(-2.5)"), "2.5");
diff --git a/crates/organize/src/lib.rs b/crates/organize/src/lib.rs
index 133bc03ecf..99c228c60f 100644
--- a/crates/organize/src/lib.rs
+++ b/crates/organize/src/lib.rs
@@ -113,13 +113,17 @@ pub fn pages(doc: &Document) -> Result, OrganizeError> {
pub fn page_rotation(doc: &Document, index: usize) -> Result {
let pages = walk(doc)?;
let (_, attrs) = pages.get(index).ok_or(OrganizeError::NoSuchPage(index))?;
- Ok(attrs.int(b"Rotate").unwrap_or(0).rem_euclid(360))
+ Ok(resolved_rotation(doc, attrs).rem_euclid(360))
}
pub fn page_count(doc: &Document) -> Result {
Ok(walk(doc)?.len())
}
+fn resolved_rotation(doc: &Document, attrs: &Dict) -> i64 {
+ attrs.get(b"Rotate").and_then(|value| doc.resolve(value).as_int()).unwrap_or(0)
+}
+
/// Rebuild the page tree as one flat node with `order` as its kids (attributes pushed down).
fn rebuild(doc: &mut Document, order: &[(ObjRef, Dict)]) -> Result<(), OrganizeError> {
let root = pages_root(doc)?;
@@ -201,7 +205,10 @@ pub fn rotate_pages(doc: &mut Document, indices: &[usize], degrees: i64) -> Resu
let delta = degrees.rem_euclid(360);
for &i in indices {
let (page, inherited) = &all[i];
- let current = doc.get(*page).as_dict().and_then(|d| d.int(b"Rotate")).or_else(|| inherited.int(b"Rotate")).unwrap_or(0);
+ // Repeated indices must see earlier edits instead of the page walk's snapshot.
+ let current_page = doc.get(*page);
+ let current_attrs = current_page.as_dict().filter(|attrs| attrs.get(b"Rotate").is_some()).unwrap_or(inherited);
+ let current = resolved_rotation(doc, current_attrs);
let new = (current.rem_euclid(360) + delta).rem_euclid(360);
doc.update_dict(*page, |d| d.set(b"Rotate".to_vec(), Object::Int(new)))?;
}
@@ -492,9 +499,10 @@ pub fn reverse_pages(doc: &mut Document, indices: &[usize]) -> Result<(), Organi
rebuild(doc, &all)
}
-/// Duplicate pages: copies of `indices` (in order) are inserted after the last of them, sharing
+/// Duplicate pages: copies of `indices` (in order) are inserted at `at` in the page order
+/// (clamped to the document's end), or after the last of them when `at` is `None`, sharing
/// fonts and images with the originals.
-pub fn duplicate_pages(doc: &mut Document, indices: &[usize]) -> Result<(), OrganizeError> {
+pub fn duplicate_pages(doc: &mut Document, indices: &[usize], at: Option) -> Result<(), OrganizeError> {
let n = page_count(doc)?;
check(indices, n)?;
let Some(&last) = indices.iter().max() else { return Ok(()) };
@@ -502,7 +510,7 @@ pub fn duplicate_pages(doc: &mut Document, indices: &[usize]) -> Result<(), Orga
order.sort_unstable();
order.dedup();
let src = doc.clone();
- import_pages(doc, &src, &order, last + 1).map(|_| ())
+ import_pages(doc, &src, &order, at.unwrap_or(last + 1)).map(|_| ())
}
/// Replace Pages: the content of `targets` (in order) is replaced by that of `src_pages` of
@@ -556,6 +564,54 @@ pub fn insert_blank_page(doc: &mut Document, at: usize, width: f64, height: f64)
/// Document information keys editable in Document Properties ▸ Description.
pub const INFO_KEYS: [&str; 4] = ["Title", "Author", "Subject", "Keywords"];
+/// The document-information entries the spec names (ISO 32000-2 table 349). Any other key in
+/// `/Info` is a custom property (Document Properties ▸ Custom).
+pub const STANDARD_INFO_KEYS: [&str; 9] = ["Title", "Author", "Subject", "Keywords", "Creator", "Producer", "CreationDate", "ModDate", "Trapped"];
+
+/// Longest custom property name accepted, in bytes: the name length every reader is asked to
+/// accept (§7.3.5, Annex C).
+pub const MAX_CUSTOM_KEY_LEN: usize = 127;
+
+/// The custom document properties: every `/Info` entry whose key the spec does not name and whose
+/// value is text, in the dictionary's order, as (name, text), the text without NUL or whitespace
+/// padding and entries that are only padding left out. At most 256 are listed.
+pub fn custom_info(doc: &Document) -> Vec<(String, String)> {
+ let Some(info) = doc.trailer().get(b"Info").map(|o| doc.resolve(o)) else { return Vec::new() };
+ let Some(d) = info.as_dict() else { return Vec::new() };
+ d.iter()
+ .filter(|(k, _)| !STANDARD_INFO_KEYS.iter().any(|s| s.as_bytes() == k.as_slice()))
+ .filter_map(|(k, v)| {
+ // Padding (\0s and whitespace, in any mix) is not part of the value, and a value that
+ // is only padding is no property. Filtered before the cap, so blank entries can't
+ // crowd real ones out.
+ let text = doc.resolve(v).as_string()?.to_text();
+ let text = text.trim_matches(|c: char| c == '\0' || c.is_whitespace());
+ (!text.is_empty()).then(|| (String::from_utf8_lossy(k).into_owned(), text.to_owned()))
+ })
+ .take(256)
+ .collect()
+}
+
+/// Whether `name` may be a custom property's key: not one the spec names (those are the
+/// Description fields), not empty, at most [`MAX_CUSTOM_KEY_LEN`] bytes, and without the
+/// characters a name cannot carry unescaped (whitespace, the delimiters `()<>[]{}/%` and `#`),
+/// so every reader shows the same name. Acrobat refuses the standard and empty names too.
+pub fn check_custom_key(name: &str) -> Result<(), OrganizeError> {
+ if name.is_empty() {
+ return Err(OrganizeError::Invalid("a custom property needs a name".into()));
+ }
+ if STANDARD_INFO_KEYS.contains(&name) {
+ return Err(OrganizeError::Invalid(format!("{name} is a standard document property, not a custom one")));
+ }
+ if name.len() > MAX_CUSTOM_KEY_LEN {
+ return Err(OrganizeError::Invalid(format!("a custom property name is at most {MAX_CUSTOM_KEY_LEN} bytes")));
+ }
+ if let Some(c) = name.chars().find(|c| c.is_whitespace() || c.is_control() || "()<>[]{}/%#".contains(*c)) {
+ return Err(OrganizeError::Invalid(format!("a custom property name can't contain {c:?}")));
+ }
+ Ok(())
+}
+
/// Read a document-information entry as text.
pub fn info(doc: &Document, key: &str) -> Option {
let info = doc.trailer().get(b"Info").map(|o| doc.resolve(o))?;
@@ -568,6 +624,9 @@ pub fn info(doc: &Document, key: &str) -> Option {
/// in the XMP packet, and viewers that read XMP first show its value: the packet's property is
/// changed (or removed) to match (see `xmp`).
pub fn set_info(doc: &mut Document, key: &str, value: &str) -> Result<(), OrganizeError> {
+ if !STANDARD_INFO_KEYS.contains(&key) {
+ check_custom_key(key)?;
+ }
let value = value.trim();
let entry = (!value.is_empty()).then(|| Object::String(PdfString::text(value)));
set_info_entry(doc, key.as_bytes(), entry)?;
diff --git a/crates/organize/src/tests.rs b/crates/organize/src/tests.rs
index c857cd4da7..9bd3eb396a 100644
--- a/crates/organize/src/tests.rs
+++ b/crates/organize/src/tests.rs
@@ -16,6 +16,53 @@ fn page_rotation_resolves_inheritance_overrides_and_missing_pages() {
assert_eq!(page_rotation(&doc, 3), Err(OrganizeError::NoSuchPage(3)));
}
+#[test]
+fn indirect_page_rotation_is_read_and_updated_from_its_effective_value() {
+ let large = i64::MAX - i64::MAX % 90;
+ for (current, degrees, expected) in [(270, 90, 0), (-90, 90, 0), (large, 90, 90), (90, -180, 270)] {
+ for level in 0..3 {
+ let mut doc = open(fixture());
+ let rotation = doc.add(Object::Int(current));
+ let page = pages(&doc).unwrap()[0].obj;
+ let target = match level {
+ 0 => page,
+ 1 => doc.get(page).as_dict().unwrap().reference(b"Parent").unwrap(),
+ _ => pages_root(&doc).unwrap(),
+ };
+ doc.update_dict(target, |d| d.set(b"Rotate".to_vec(), Object::Ref(rotation))).unwrap();
+ let bytes = write_full(&doc, &SaveOptions::default()).unwrap();
+ let mut doc = open(bytes);
+ let rotation = walk(&doc).unwrap()[0].1.reference(b"Rotate").unwrap();
+ let initial = page_rotation(&doc, 0).unwrap();
+ rotate_pages(&mut doc, &[0], degrees).unwrap();
+ assert_eq!(*doc.get(rotation), Object::Int(current), "shared rotation object unchanged");
+ for saved in [write_incremental(&doc, &SaveOptions::default()).unwrap(), write_full(&doc, &SaveOptions::default()).unwrap()] {
+ let reopened = open(saved.clone());
+ let pdf = hayro_syntax::Pdf::new(saved).unwrap();
+ let dimensions = if expected % 180 == 0 { (300.0, 400.0) } else { (400.0, 300.0) };
+ assert_eq!(pdf.pages()[0].render_dimensions(), dimensions, "written PDF has the correct orientation");
+ assert_eq!(page_rotation(&reopened, 0).unwrap(), expected);
+ assert_eq!(page_rotation(&reopened, 1).unwrap(), if level == 0 { 90 } else { current.rem_euclid(360) }, "sibling unchanged");
+ assert_eq!(page_rotation(&reopened, 2).unwrap(), if level == 2 { current.rem_euclid(360) } else { 90 }, "other subtree unchanged");
+ }
+ assert_eq!(initial, current.rem_euclid(360), "effective indirect rotation at level {level}");
+ }
+ }
+}
+
+#[test]
+fn malformed_indirect_rotations_keep_the_zero_fallback() {
+ for value in [Object::Null, Object::name("invalid"), Object::Ref(ObjRef::new(99_999, 0))] {
+ let mut doc = open(fixture());
+ let rotation = doc.add(value);
+ let page = pages(&doc).unwrap()[0].obj;
+ doc.update_dict(page, |d| d.set(b"Rotate".to_vec(), Object::Ref(rotation))).unwrap();
+ assert_eq!(page_rotation(&doc, 0).unwrap(), 0);
+ rotate_pages(&mut doc, &[0], 90).unwrap();
+ assert_eq!(page_rotation(&doc, 0).unwrap(), 90);
+ }
+}
+
#[test]
fn rotate_rejects_angles_that_are_not_multiples_of_90() {
// The page tree only turns in quarter steps, so anything else used to be silently truncated
@@ -336,6 +383,59 @@ fn page_transitions_are_set_removed_and_checked() {
assert_eq!(page_dict(&doc, 2).get(b"Dur").and_then(Object::as_f64), Some(5.0));
}
+/// Custom document properties (Document Properties ▸ Custom): any `/Info` key the spec does not
+/// name, set and cleared like the standard ones, listed apart from them, and with the names
+/// Acrobat refuses refused here too.
+#[test]
+fn custom_document_properties_are_listed_set_removed_and_their_names_checked() {
+ let mut doc = open(fixture());
+ assert!(custom_info(&doc).is_empty());
+ set_info(&mut doc, "Title", "Report").unwrap();
+ set_info(&mut doc, "Project", "Apollo").unwrap();
+ set_info(&mut doc, "Revision", "7").unwrap();
+ let mut listed = custom_info(&doc);
+ listed.sort();
+ assert_eq!(listed, [("Project".to_string(), "Apollo".to_string()), ("Revision".to_string(), "7".to_string())], "standard keys are not custom");
+ assert_eq!(info(&doc, "Project").as_deref(), Some("Apollo"));
+ set_info(&mut doc, "Project", "").unwrap();
+ let doc = save_and_reopen(&doc);
+ assert_eq!(custom_info(&doc), [("Revision".to_string(), "7".to_string())], "cleared, and the rest survives a save");
+ assert_eq!(info(&doc, "Title").as_deref(), Some("Report"));
+
+ for bad in ["", "Title", "ModDate", "Bad Name", "a/b", "tab\tbed", "hash#tag", "(x)"] {
+ assert!(check_custom_key(bad).is_err(), "{bad:?} must be refused");
+ }
+ let long = "x".repeat(MAX_CUSTOM_KEY_LEN + 1);
+ assert!(check_custom_key(&long).is_err());
+ for good in ["Project", "Révision", "部門", "key-2.0", &"y".repeat(MAX_CUSTOM_KEY_LEN)] {
+ assert!(check_custom_key(good).is_ok(), "{good:?} is a usable name");
+ }
+ let mut doc = open(fixture());
+ assert!(set_info(&mut doc, "Bad Name", "x").is_err(), "set_info checks custom names");
+ assert!(custom_info(&doc).is_empty());
+}
+
+/// Padding is not part of a custom value, in any mix of NULs and whitespace, a value that is only
+/// padding is not listed, and such entries don't count towards the 256 listed: a real property
+/// after 300 blank ones is still there.
+#[test]
+fn custom_document_properties_drop_padding_before_the_cap() {
+ let mut doc = open(fixture());
+ set_info(&mut doc, "Title", "Report").unwrap();
+ let Some(Object::Ref(info_ref)) = doc.trailer().get(b"Info").cloned() else { panic!("an /Info reference") };
+ let text = |s: &str| Object::String(pdfcraft_cos::PdfString { bytes: s.as_bytes().to_vec(), hex: false });
+ doc.update_dict(info_ref, |d| {
+ d.set(b"Mixed".to_vec(), text(" \0Lab\0 "));
+ d.set(b"Pad".to_vec(), text(" \0 \0"));
+ for i in 0..300 {
+ d.set(format!("Blank{i}").into_bytes(), text(" "));
+ }
+ d.set(b"Keep".to_vec(), text("yes"));
+ })
+ .unwrap();
+ assert_eq!(custom_info(&doc), [("Mixed".to_string(), "Lab".to_string()), ("Keep".to_string(), "yes".to_string())]);
+}
+
/// A one-page document whose catalog has `xmp` as its (Flate-compressed) XMP packet, and the
/// same title and author in `/Info`, as LibreOffice writes them.
fn doc_with_xmp(xmp: &[u8]) -> Document {
@@ -1871,11 +1971,24 @@ fn page_boxes_default_inherit_and_set() {
#[test]
fn duplicate_pages_inserts_copies_after_the_last() {
let mut doc = open(fixture());
- duplicate_pages(&mut doc, &[0, 1]).unwrap();
+ duplicate_pages(&mut doc, &[0, 1], None).unwrap();
let doc = open(write_full(&doc, &SaveOptions::default()).unwrap());
assert_eq!(labels(&doc), ["Page 1", "Page 2", "Page 1", "Page 2", "Page 3"]);
}
+#[test]
+fn duplicate_pages_at_inserts_the_copies_there() {
+ let mut doc = open(fixture());
+ duplicate_pages(&mut doc, &[2], Some(0)).unwrap();
+ let doc = open(write_full(&doc, &SaveOptions::default()).unwrap());
+ assert_eq!(labels(&doc), ["Page 3", "Page 1", "Page 2", "Page 3"]);
+ // An index past the end lands at the end.
+ let mut doc = open(fixture());
+ duplicate_pages(&mut doc, &[0], Some(99)).unwrap();
+ let doc = open(write_full(&doc, &SaveOptions::default()).unwrap());
+ assert_eq!(labels(&doc), ["Page 1", "Page 2", "Page 3", "Page 1"]);
+}
+
#[test]
fn replace_pages_swaps_content_but_keeps_annotations() {
// Page 2 of the target gets an annotation; replacing its content keeps it.
@@ -2482,7 +2595,7 @@ fn import_can_merge_into_same_named_fields_instead() {
#[test]
fn duplicating_a_page_with_a_field_keeps_the_field_name() {
let mut doc = doc_with_field("Text1");
- crate::duplicate_pages(&mut doc, &[0]).unwrap();
+ crate::duplicate_pages(&mut doc, &[0], None).unwrap();
assert_eq!(crate::page_count(&doc).unwrap(), 2);
assert_eq!(form_field_names(&doc), ["Text1", "Text1"], "nothing renamed");
}
@@ -2676,3 +2789,35 @@ fn a_chain_of_objects_too_deep_to_copy_is_refused_and_leaves_the_destination_alo
.join();
assert!(run.is_ok(), "the copy must not overflow the stack");
}
+
+#[test]
+fn rotate_repeated_indices_accumulate_from_the_current_page_value() {
+ for level in 0..3 {
+ for indices in [vec![0, 0], vec![0, 1, 0]] {
+ let mut doc = open(fixture());
+ let rotation = doc.add(Object::Int(90));
+ let page = pages(&doc).unwrap()[0].obj;
+ let target = match level {
+ 0 => page,
+ 1 => doc.get(page).as_dict().unwrap().reference(b"Parent").unwrap(),
+ _ => pages_root(&doc).unwrap(),
+ };
+ doc.update_dict(target, |d| d.set(b"Rotate".to_vec(), Object::Ref(rotation))).unwrap();
+ let mut doc = open(write_full(&doc, &SaveOptions::default()).unwrap());
+ let rotation = walk(&doc).unwrap()[0].1.reference(b"Rotate").unwrap();
+ let original_labels = labels(&doc);
+ rotate_pages(&mut doc, &indices, 90).unwrap();
+ assert_eq!(page_rotation(&doc, 0).unwrap(), 270, "level {level}, indices {indices:?}");
+ assert_eq!(*doc.get(rotation), Object::Int(90), "shared source angle is unchanged");
+ for saved in [write_incremental(&doc, &SaveOptions::default()).unwrap(), write_full(&doc, &SaveOptions::default()).unwrap()] {
+ let reopened = open(saved.clone());
+ assert_eq!(page_rotation(&reopened, 0).unwrap(), 270);
+ assert_eq!(page_rotation(&reopened, 1).unwrap(), if indices.len() == 3 { 180 } else { 90 });
+ assert_eq!(page_rotation(&reopened, 2).unwrap(), 90);
+ assert_eq!(labels(&reopened), original_labels);
+ let independent = hayro_syntax::Pdf::new(saved).unwrap();
+ assert_eq!(independent.pages()[0].render_dimensions(), (400.0, 300.0));
+ }
+ }
+ }
+}
diff --git a/crates/platform/src/staging.rs b/crates/platform/src/staging.rs
index f7200efb9e..fee0a16336 100644
--- a/crates/platform/src/staging.rs
+++ b/crates/platform/src/staging.rs
@@ -27,6 +27,20 @@ pub fn create_staging(
name: &str,
form: StagingName,
suffixes: impl IntoIterator- ,
+) -> std::io::Result<(PathBuf, std::fs::File)> {
+ create_staging_as(dir, name, form, suffixes, None)
+}
+
+/// [`create_staging`], giving the new file the permissions `keep` on Unix. It is created readable
+/// and writable by its owner only and then given `keep`, so its mode is never more open than that
+/// of the file it will replace, not even while it is still empty (a handle opened then would keep
+/// reading what is written later). Only the mode is copied: the owner and group are the saver's.
+fn create_staging_as(
+ dir: &Path,
+ name: &str,
+ form: StagingName,
+ suffixes: impl IntoIterator
- ,
+ keep: Option<&std::fs::Permissions>,
) -> std::io::Result<(PathBuf, std::fs::File)> {
// At most 128 bytes of the target's name, cut between characters, so the staging name fits
// the 255-byte (Linux, macOS) and 255-unit (Windows) limits however long that name is.
@@ -42,8 +56,22 @@ pub fn create_staging(
StagingName::SuffixThenTag => format!(".{stem}.{suffix:016x}.pdfcraft-tmp"),
StagingName::TagThenSuffix => format!(".{stem}.pdfcraft-{suffix:016x}.tmp"),
});
- match std::fs::OpenOptions::new().write(true).create_new(true).open(&tmp) {
- Ok(file) => return Ok((tmp, file)),
+ let mut options = std::fs::OpenOptions::new();
+ options.write(true).create_new(true);
+ #[cfg(unix)]
+ if keep.is_some() {
+ use std::os::unix::fs::OpenOptionsExt as _;
+ options.mode(0o600);
+ }
+ match options.open(&tmp) {
+ Ok(file) => {
+ if let Some(keep) = keep {
+ // A volume that can't change modes (FAT, some network shares) leaves the file
+ // owner-only: stricter than the file it replaces, never looser.
+ let _ = file.set_permissions(keep.clone());
+ }
+ return Ok((tmp, file));
+ }
Err(e) if e.kind() == std::io::ErrorKind::AlreadyExists => {}
// Windows reports a folder at the name as "access denied"; it is taken all the same.
Err(e) if e.kind() == std::io::ErrorKind::PermissionDenied && tmp.symlink_metadata().is_ok() => {}
@@ -69,14 +97,17 @@ pub fn staging_suffixes() -> impl Iterator
- {
/// crash, a full disk or a refused flush never leaves a half-written file where a good one was,
/// and the target is untouched until the new file is complete. The target's folder must exist;
/// a folder at the target is refused by the rename. Flushing is the staging file's own
-/// (`sync_all`); parent-directory durability after a power loss is not promised.
+/// (`sync_all`); parent-directory durability after a power loss is not promised. On Unix a file
+/// that is replaced keeps its permissions, so a private file stays private; a new file gets the
+/// usual mode for a new file. On Windows and macOS a replaced file keeps its creation date, where
+/// the volume stores one.
pub fn write_atomically(path: &Path, bytes: &[u8], form: StagingName) -> std::io::Result<()> {
write_atomically_with(path, bytes, form, staging_suffixes(), std::fs::File::sync_all)
}
/// [`write_atomically`] with the staging-name suffixes and the flush given, so tests can plant
/// names and refuse flushes. A failure at any step removes the staging file and leaves the
-/// target as it was.
+/// target as it was. A read-only target is refused before anything is written.
pub fn write_atomically_with(
path: &Path,
bytes: &[u8],
@@ -85,14 +116,30 @@ pub fn write_atomically_with(
sync_all: impl FnOnce(&std::fs::File) -> std::io::Result<()>,
) -> std::io::Result<()> {
use std::io::Write as _;
+ // A read-only file is not replaced. Windows refuses the rename anyway, but on Unix a rename
+ // ignores the target's permissions: the file would be overwritten without a word, and come
+ // out writable (#2658).
+ if std::fs::metadata(path).is_ok_and(|m| m.is_file() && m.permissions().readonly()) {
+ return Err(std::io::Error::new(std::io::ErrorKind::PermissionDenied, "the file is read-only; save a copy with Save As"));
+ }
let dir = path.parent().filter(|d| !d.as_os_str().is_empty()).unwrap_or(Path::new("."));
let name = path.file_name().and_then(|n| n.to_str()).unwrap_or("save");
- let (tmp, file) = create_staging(dir, name, form, suffixes)?;
+ // A file that is replaced keeps its permissions (#1790). Unix modes only: Windows is as before.
+ #[cfg(unix)]
+ let keep = std::fs::metadata(path).ok().map(|m| m.permissions());
+ #[cfg(not(unix))]
+ let keep = None;
+ // A saved file is the same document: it keeps its creation date (#2658).
+ #[cfg(any(windows, target_os = "macos"))]
+ let created = std::fs::metadata(path).ok().filter(|m| m.is_file()).and_then(|m| m.created().ok());
+ let (tmp, file) = create_staging_as(dir, name, form, suffixes, keep.as_ref())?;
let result = (|| {
// Closed at the end of the block, before the rename (required on Windows).
{
let mut file = file;
file.write_all(bytes)?;
+ #[cfg(any(windows, target_os = "macos"))]
+ keep_created(&file, created);
sync_staged(&file, sync_all)?;
}
std::fs::rename(&tmp, path)
@@ -103,6 +150,20 @@ pub fn write_atomically_with(
result
}
+/// Give the staging file the creation date of the file it replaces, where the system keeps one
+/// that a program can set (Windows, macOS). Best effort: a volume that can't store it leaves the
+/// new file's own date, as before.
+#[cfg(any(windows, target_os = "macos"))]
+fn keep_created(file: &std::fs::File, created: Option
) {
+ if let Some(created) = created {
+ #[cfg(target_os = "macos")]
+ use std::os::macos::fs::FileTimesExt as _;
+ #[cfg(windows)]
+ use std::os::windows::fs::FileTimesExt as _;
+ let _ = file.set_times(std::fs::FileTimes::new().set_created(created));
+ }
+}
+
/// Flush the staging file to the disk before it replaces the target. On macOS `sync_all` asks
/// for a full flush (`F_FULLFSYNC`), which SMB and FUSE volumes, such as a mounted Cryptomator
/// vault, refuse as unsupported: there a plain `fsync` is the strongest flush the volume offers.
@@ -162,6 +223,33 @@ mod tests {
let _ = std::fs::remove_dir_all(&dir);
}
+ /// #2658: a read-only file is refused before anything is staged, on Unix (where a rename would
+ /// replace it) as on Windows (where the rename fails).
+ #[test]
+ fn a_read_only_target_is_refused_and_kept() {
+ let dir = test_dir("readonly");
+ let target = dir.join("out.pdf");
+ std::fs::write(&target, "OLD").unwrap();
+ let mut perms = std::fs::metadata(&target).unwrap().permissions();
+ perms.set_readonly(true);
+ std::fs::set_permissions(&target, perms.clone()).unwrap();
+ let staged = std::cell::Cell::new(false);
+ let e = write_atomically_with(&target, b"NEW", StagingName::TagThenSuffix, [7], |_: &std::fs::File| {
+ staged.set(true);
+ Ok(())
+ })
+ .unwrap_err();
+ assert_eq!(e.kind(), std::io::ErrorKind::PermissionDenied, "{e}");
+ assert!(!staged.get(), "nothing was staged");
+ assert_eq!(std::fs::read(&target).unwrap(), b"OLD");
+ assert!(std::fs::metadata(&target).unwrap().permissions().readonly(), "still read-only");
+ assert_eq!(std::fs::read_dir(&dir).unwrap().count(), 1, "no staging file is left");
+ #[allow(clippy::permissions_set_readonly_false)]
+ perms.set_readonly(false);
+ std::fs::set_permissions(&target, perms).unwrap();
+ let _ = std::fs::remove_dir_all(&dir);
+ }
+
#[test]
fn running_out_of_suffixes_refuses_and_creates_nothing() {
let dir = test_dir("empty");
@@ -215,4 +303,62 @@ mod tests {
assert_eq!(names(&dir), ["out.pdf"]);
let _ = std::fs::remove_dir_all(&dir);
}
+
+ /// #1790: a replaced file keeps its permissions, and its staging file already has them while
+ /// it is written; a new file gets the same mode as any other new file.
+ #[cfg(unix)]
+ #[test]
+ fn a_replaced_file_keeps_its_permissions() {
+ use std::os::unix::fs::PermissionsExt as _;
+ let dir = test_dir("modes");
+ let mode = |p: &Path| std::fs::metadata(p).unwrap().permissions().mode() & 0o777;
+ let target = dir.join("out.pdf");
+ for keep in [0o600, 0o640, 0o664] {
+ std::fs::write(&target, b"OLD").unwrap();
+ std::fs::set_permissions(&target, std::fs::Permissions::from_mode(keep)).unwrap();
+ let mut staged = 0;
+ write_atomically_with(&target, b"NEW", StagingName::SuffixThenTag, [keep as u64], |f| {
+ staged = f.metadata()?.permissions().mode() & 0o777;
+ Ok(())
+ })
+ .unwrap();
+ assert_eq!(staged, keep, "the staging file has the target's mode before it is flushed");
+ assert_eq!(mode(&target), keep, "the saved file keeps mode {keep:o}");
+ assert_eq!(std::fs::read(&target).unwrap(), b"NEW");
+ }
+ let fresh = dir.join("new.pdf");
+ write_atomically(&fresh, b"NEW", StagingName::TagThenSuffix).unwrap();
+ let plain = dir.join("plain");
+ std::fs::write(&plain, b"").unwrap();
+ assert_eq!(mode(&fresh), mode(&plain), "a new file is not made private");
+ let _ = std::fs::remove_dir_all(&dir);
+ }
+
+ /// #2658: a save replaces the file, but the document is the same one: it keeps its creation
+ /// date where the system lets a program set it.
+ #[cfg(any(windows, target_os = "macos"))]
+ #[test]
+ fn a_replaced_file_keeps_its_creation_date() {
+ #[cfg(target_os = "macos")]
+ use std::os::macos::fs::FileTimesExt as _;
+ #[cfg(windows)]
+ use std::os::windows::fs::FileTimesExt as _;
+ let dir = test_dir("created");
+ let target = dir.join("doc.pdf");
+ std::fs::write(&target, b"OLD").unwrap();
+ // 2020-01-02 03:04:05 UTC.
+ let then = std::time::SystemTime::UNIX_EPOCH + std::time::Duration::from_secs(1_577_934_245);
+ let file = std::fs::OpenOptions::new().write(true).open(&target).unwrap();
+ file.set_times(std::fs::FileTimes::new().set_created(then)).unwrap();
+ drop(file);
+ assert_eq!(std::fs::metadata(&target).unwrap().created().unwrap(), then, "the fixture's date is set");
+ write_atomically(&target, b"NEW", StagingName::SuffixThenTag).unwrap();
+ assert_eq!(std::fs::read(&target).unwrap(), b"NEW");
+ assert_eq!(std::fs::metadata(&target).unwrap().created().unwrap(), then, "the saved file keeps its creation date");
+ // A new file gets a date of its own.
+ let fresh = dir.join("new.pdf");
+ write_atomically(&fresh, b"NEW", StagingName::SuffixThenTag).unwrap();
+ assert!(std::fs::metadata(&fresh).unwrap().created().unwrap() > then);
+ let _ = std::fs::remove_dir_all(&dir);
+ }
}
diff --git a/crates/redact/src/sanitize.rs b/crates/redact/src/sanitize.rs
index 4dd098d58f..0c9228ac02 100644
--- a/crates/redact/src/sanitize.rs
+++ b/crates/redact/src/sanitize.rs
@@ -271,6 +271,12 @@ pub fn remove_hidden(doc: &mut Document, which: &[Hidden]) -> Result 0) || (layers && count(Hidden::HiddenLayers) > 0) {
content_pass(doc, text, layers, true)?;
}
+ // Every annotation the pages list now: whatever is no longer listed at the end (flattened
+ // widgets included) leaves the structure tree too.
+ let listed = |doc: &Document| -> std::collections::HashSet {
+ pdfcraft_model::pages(doc).iter().flat_map(|p| annots_of(doc, &p.dict)).filter_map(|a| a.as_ref()).collect()
+ };
+ let listed_before = listed(doc);
if which.contains(&Hidden::FormFields) && count(Hidden::FormFields) > 0 {
// Fields without appearances get one first, so their values stay visible.
for f in pdfcraft_forms::fields(doc) {
@@ -291,16 +297,16 @@ pub fn remove_hidden(doc: &mut Document, which: &[Hidden]) -> Result = pdfcraft_model::pages(doc)
+ .iter()
+ .flat_map(|p| annots_of(doc, &p.dict))
+ .filter_map(|a| a.as_ref())
+ .filter(|&r| doc.get(r).as_dict().is_some_and(drop_annot))
+ .collect();
for p in pdfcraft_model::pages(doc) {
let list = annots_of(doc, &p.dict);
- let mut removed: Vec = Vec::new();
- for a in &list {
- if let Some(r) = a.as_ref()
- && doc.get(r).as_dict().is_some_and(drop_annot)
- {
- removed.push(r);
- }
- }
let mut kept = Vec::new();
for a in list {
let Some(r) = a.as_ref() else {
@@ -340,6 +346,12 @@ pub fn remove_hidden(doc: &mut Document, which: &[Hidden]) -> Result = listed_before.difference(&still).copied().collect();
+ crate::tags::drop_object_refs(doc, &gone)?;
let off = hidden_layers(doc);
let mut cat = doc.get(root).as_dict().cloned().unwrap_or_default();
let mut names = sub(doc, &cat, b"Names");
diff --git a/crates/redact/src/tags.rs b/crates/redact/src/tags.rs
index 02aae12aa3..b70cf23ab9 100644
--- a/crates/redact/src/tags.rs
+++ b/crates/redact/src/tags.rs
@@ -202,3 +202,111 @@ pub(crate) fn clean(
}
Ok(n)
}
+
+/// Whether `o` is an object reference (`/Type /OBJR`, direct or indirect) to one of `gone`.
+fn refers_to(doc: &Document, o: &Object, gone: &HashSet) -> bool {
+ doc.resolve(o)
+ .as_dict()
+ .is_some_and(|d| d.name(b"Type") == Some(b"OBJR") && d.get(b"Obj").and_then(Object::as_ref).is_some_and(|r| gone.contains(&r)))
+}
+
+/// Drop the structure tree's object references (OBJR) to `gone`, annotations removed from their
+/// pages (#2447). A tagged PDF's tree points at each link and comment it tags; left there, the
+/// reference keeps the annotation, its URL or text included, in a full rewrite after the page
+/// no longer lists it. Returns how many references were dropped.
+pub(crate) fn drop_object_refs(doc: &mut Document, gone: &HashSet) -> Result {
+ drop_object_refs_within(doc, gone, MAX_STRUCT_REFS)
+}
+
+/// The most structure-tree references [`drop_object_refs`] visits.
+const MAX_STRUCT_REFS: usize = 2_000_000;
+
+/// [`drop_object_refs`], visiting at most `limit` references. A tree too large to walk in full
+/// could keep a reference the walk never reached, so then the whole structure tree goes instead
+/// (with `/MarkInfo`, which would claim tags there no longer are): fail-closed, at the cost of
+/// the document's tags.
+pub(crate) fn drop_object_refs_within(doc: &mut Document, gone: &HashSet, limit: usize) -> Result {
+ if gone.is_empty() {
+ return Ok(0);
+ }
+ let Some(root) = doc
+ .root()
+ .and_then(|r| doc.get(r).as_dict().cloned())
+ .and_then(|c| c.get(b"StructTreeRoot").map(|s| doc.resolve(s)).and_then(|s| s.as_dict().cloned()))
+ else {
+ return Ok(0);
+ };
+ let mut todo: Vec = root.get(b"K").cloned().into_iter().collect();
+ let mut seen = HashSet::new();
+ let mut hits: Vec = Vec::new();
+ while let Some(k) = todo.pop() {
+ match k {
+ Object::Array(a) => todo.extend(a),
+ Object::Ref(r) => {
+ if !seen.insert(r) {
+ continue;
+ }
+ if seen.len() > limit {
+ let Some(catalog) = doc.root() else { return Ok(0) };
+ doc.update_dict(catalog, |c| {
+ c.remove(b"StructTreeRoot");
+ c.remove(b"MarkInfo");
+ })?;
+ return Ok(0);
+ }
+ let object = doc.get(r);
+ if let Object::Array(a) = &*object {
+ todo.extend(a.iter().cloned());
+ continue;
+ }
+ let Some(d) = object.as_dict() else { continue };
+ if !d.contains(b"S") {
+ continue;
+ }
+ let kids: Vec = match d.get(b"K").map(|k| doc.resolve(k)) {
+ Some(o) => match &*o {
+ Object::Array(a) => a.clone(),
+ _ => d.get(b"K").cloned().into_iter().collect(),
+ },
+ None => Vec::new(),
+ };
+ if kids.iter().any(|x| refers_to(doc, x, gone)) {
+ hits.push(r);
+ }
+ todo.extend(kids);
+ }
+ _ => {}
+ }
+ }
+ let mut n = 0;
+ for r in hits {
+ let Some(d) = doc.get(r).as_dict().cloned() else { continue };
+ let kids = d.get(b"K").map(|k| doc.resolve(k));
+ let array = kids.as_ref().is_some_and(|k| matches!(&**k, Object::Array(_)));
+ let list: Vec = match kids.as_deref() {
+ Some(Object::Array(a)) => a.clone(),
+ Some(_) => d.get(b"K").cloned().into_iter().collect(),
+ None => Vec::new(),
+ };
+ let before = list.len();
+ let keep: Vec = list.into_iter().filter(|x| !refers_to(doc, x, gone)).collect();
+ n += before - keep.len();
+ doc.update_dict(r, |e| {
+ if array {
+ e.set(b"K".to_vec(), Object::Array(keep.clone()));
+ } else if let Some(k) = keep.first().cloned() {
+ e.set(b"K".to_vec(), k);
+ } else {
+ e.remove(b"K");
+ }
+ // An element left with nothing to tag keeps no text describing what it tagged:
+ // a link's alternate text often repeats the URL, an annotation's its comment.
+ if keep.is_empty() {
+ for key in [&b"Alt"[..], b"ActualText", b"T", b"E"] {
+ e.remove(key);
+ }
+ }
+ })?;
+ }
+ Ok(n)
+}
diff --git a/crates/redact/src/tests.rs b/crates/redact/src/tests.rs
index a4e3530aba..a7e11bcaf1 100644
--- a/crates/redact/src/tests.rs
+++ b/crates/redact/src/tests.rs
@@ -708,6 +708,140 @@ fn hidden_information_is_counted_and_removed() {
assert!(shown, "the field's value stays visible as page content");
}
+/// Issue #2447: in a tagged PDF the structure tree points at the link and the comment it tags
+/// (`/OBJR`). Removing them must drop those references too, or the full rewrite keeps the
+/// annotations, the URL and the comment text included, while the page no longer lists them.
+#[test]
+fn removed_links_and_comments_leave_the_structure_tree_and_the_saved_file() {
+ use crate::sanitize::{Hidden, remove_hidden};
+ let objs: Vec> = vec![
+ // 1 catalog, with a structure tree
+ b"<< /Type /Catalog /Pages 2 0 R /MarkInfo << /Marked true >> /StructTreeRoot 8 0 R >>".to_vec(),
+ b"<< /Type /Pages /Kids [3 0 R] /Count 1 >>".to_vec(),
+ // 3 page: a tagged paragraph (MCID 0), a link and a comment, both tagged
+ b"<< /Type /Page /Parent 2 0 R /MediaBox [0 0 300 300] /Contents 4 0 R /Resources << /Font << /F1 5 0 R >> >> /Annots [6 0 R 7 0 R] /StructParents 0 >>".to_vec(),
+ stream("", b"/P << /MCID 0 >> BDC BT /F1 10 Tf 10 230 Td (See the site) Tj ET EMC"),
+ FONT.replace("95 0 R", "13 0 R").into_bytes(),
+ // 6 link, 7 comment
+ b"<< /Type /Annot /Subtype /Link /Rect [10 228 80 240] /StructParent 1 /A << /S /URI /URI (https://secret.example/token=ZX81) >> >>".to_vec(),
+ b"<< /Type /Annot /Subtype /Text /Rect [200 200 220 220] /StructParent 2 /Contents (COMMENTSECRETNOTE) >>".to_vec(),
+ // 8 struct tree root, 9 document element, 10 paragraph, 11 link element (direct OBJR),
+ // 12 annotation element (OBJR by reference, as K's only kid)
+ b"<< /Type /StructTreeRoot /K 9 0 R /ParentTree << /Nums [0 [10 0 R] 1 11 0 R 2 12 0 R] >> /ParentTreeNextKey 3 >>".to_vec(),
+ b"<< /Type /StructElem /S /Document /P 8 0 R /K [10 0 R 11 0 R 12 0 R] >>".to_vec(),
+ b"<< /Type /StructElem /S /P /P 9 0 R /Pg 3 0 R /K 0 >>".to_vec(),
+ // the link element's alternate text repeats the URL
+ b"<< /Type /StructElem /S /Link /P 9 0 R /Pg 3 0 R /Alt (secret.example link) /K [<< /Type /OBJR /Obj 6 0 R /Pg 3 0 R >>] >>".to_vec(),
+ b"<< /Type /StructElem /S /Annot /P 9 0 R /Pg 3 0 R /K 14 0 R >>".to_vec(),
+ widths(),
+ b"<< /Type /OBJR /Obj 7 0 R /Pg 3 0 R >>".to_vec(),
+ ];
+ let mut doc = pdf(objs);
+ let done = remove_hidden(&mut doc, &[Hidden::LinksActionsScripts, Hidden::Comments]).unwrap();
+ assert!(done.contains(&(Hidden::LinksActionsScripts, 1)) && done.contains(&(Hidden::Comments, 1)), "{done:?}");
+ // Without object streams, whose compression would hide the strings from the search.
+ let bytes = write_full(&doc, &SaveOptions { object_streams: false, ..SaveOptions::default() }).unwrap();
+ let has = |needle: &[u8]| bytes.windows(needle.len()).any(|w| w == needle);
+ assert!(has(b"(See the site)"), "the search sees the saved objects");
+ assert!(!has(b"secret.example"), "the removed link's URL survived the save");
+ assert!(!has(b"COMMENTSECRETNOTE"), "the removed comment's text survived the save");
+ // The tagged paragraph and its content stay; the elements lose only the references.
+ let saved = Document::open(Arc::new(bytes)).unwrap();
+ assert!(content(&saved, 0).contains("(See the site)"));
+ let objrs = saved.object_numbers().into_iter().filter(|&n| {
+ let o = saved.get(ObjRef::new(n, saved.generation(n)));
+ let direct = o.as_dict().is_some_and(|d| d.name(b"Type") == Some(b"OBJR"));
+ let nested = o.as_dict().and_then(|d| d.get(b"K")).is_some_and(|k| match &*saved.resolve(k) {
+ Object::Array(a) => a.iter().any(|x| saved.resolve(x).as_dict().is_some_and(|d| d.name(b"Type") == Some(b"OBJR"))),
+ _ => false,
+ });
+ direct || nested
+ });
+ assert_eq!(objrs.count(), 0, "no object reference to a removed annotation is left");
+ let root = saved.get(saved.root().unwrap()).as_dict().cloned().unwrap();
+ assert!(root.contains(b"StructTreeRoot"), "the structure tree itself stays");
+}
+
+/// #2532 review: a tagged form's widgets, flattened when form fields are removed, leave the
+/// structure tree too (their field value with them), and a pop-up listed on another page than
+/// its removed comment goes with it.
+#[test]
+fn flattened_widgets_and_far_pop_ups_leave_the_structure_tree() {
+ use crate::sanitize::{Hidden, remove_hidden};
+ let objs: Vec> = vec![
+ // 1 catalog with a form and a structure tree
+ b"<< /Type /Catalog /Pages 2 0 R /AcroForm << /Fields [6 0 R] >> /StructTreeRoot 8 0 R >>".to_vec(),
+ b"<< /Type /Pages /Kids [3 0 R 13 0 R] /Count 2 >>".to_vec(),
+ // 3 page 1: a text field's widget and a comment
+ b"<< /Type /Page /Parent 2 0 R /MediaBox [0 0 300 300] /Contents 4 0 R /Resources << /Font << /F1 5 0 R >> >> /Annots [6 0 R 7 0 R] /StructParents 0 >>".to_vec(),
+ stream("", b"BT /F1 10 Tf 10 230 Td (Name:) Tj ET"),
+ FONT.replace("95 0 R", "12 0 R").into_bytes(),
+ // 6 the field and its widget, with an appearance
+ b"<< /Type /Annot /Subtype /Widget /FT /Tx /T (name) /V (FIELDSECRETVALUE) /Rect [50 225 200 240] /StructParent 1 /AP << /N 11 0 R >> >>".to_vec(),
+ // 7 comment, its pop-up (15) listed on page 2
+ b"<< /Type /Annot /Subtype /Text /Rect [200 200 220 220] /Contents (note) /Popup 15 0 R >>".to_vec(),
+ // 8 struct tree root, 9 document, 10 form element tagging the widget
+ b"<< /Type /StructTreeRoot /K 9 0 R /ParentTree << /Nums [1 10 0 R] >> /ParentTreeNextKey 2 >>".to_vec(),
+ b"<< /Type /StructElem /S /Document /P 8 0 R /K [10 0 R] >>".to_vec(),
+ b"<< /Type /StructElem /S /Form /P 9 0 R /Pg 3 0 R /K [<< /Type /OBJR /Obj 6 0 R /Pg 3 0 R >>] >>".to_vec(),
+ stream("/Type /XObject /Subtype /Form /BBox [0 0 150 15] /Resources << /Font << /F1 5 0 R >> >>", b"BT /F1 10 Tf 2 3 Td (FIELDSECRETVALUE) Tj ET"),
+ widths(),
+ // 13 page 2, listing the pop-up of page 1's comment
+ b"<< /Type /Page /Parent 2 0 R /MediaBox [0 0 300 300] /Annots [15 0 R 16 0 R] >>".to_vec(),
+ b"<< >>".to_vec(),
+ b"<< /Type /Annot /Subtype /Popup /Rect [0 0 100 50] /Parent 7 0 R >>".to_vec(),
+ // 16 an unrelated link on page 2 stays
+ b"<< /Type /Annot /Subtype /Link /Rect [0 0 10 10] >>".to_vec(),
+ ];
+ let mut doc = pdf(objs);
+ remove_hidden(&mut doc, &[Hidden::FormFields, Hidden::Comments]).unwrap();
+ let page2 = pdfcraft_model::pages(&doc)[1].dict.clone();
+ let left: Vec = page2.get(b"Annots").map(|a| doc.resolve(a).as_array().map(|a| a.to_vec()).unwrap_or_default()).unwrap_or_default();
+ assert_eq!(left, [Object::Ref(ObjRef::new(16, 0))], "the far pop-up went with its comment");
+ let form = doc.get(ObjRef::new(10, 0)).as_dict().cloned().unwrap();
+ let kids = form.get(b"K").map(|k| doc.resolve(k).as_array().map(Vec::len).unwrap_or(1));
+ assert!(kids.unwrap_or(0) == 0, "the form element no longer references the flattened widget");
+ let bytes = write_full(&doc, &SaveOptions { object_streams: false, ..SaveOptions::default() }).unwrap();
+ let saved = Document::open(Arc::new(bytes)).unwrap();
+ let field_left = saved.object_numbers().into_iter().any(|n| {
+ saved.get(ObjRef::new(n, saved.generation(n))).as_dict().is_some_and(|d| d.name(b"Subtype") == Some(b"Widget") || d.contains(b"FT"))
+ });
+ assert!(!field_left, "the flattened field didn't survive the save through the structure tree");
+}
+
+/// #2532 review: a structure tree larger than the walk's limit can't be searched in full, so a
+/// reference the walk never reached could keep a removed annotation. Then the whole tree goes
+/// (with `/MarkInfo`), rather than reporting success with references left behind; under the
+/// limit only the reference goes.
+#[test]
+fn a_structure_tree_too_large_to_walk_is_removed_whole() {
+ let objs: Vec> = vec![
+ b"<< /Type /Catalog /Pages 2 0 R /MarkInfo << /Marked true >> /StructTreeRoot 5 0 R >>".to_vec(),
+ b"<< /Type /Pages /Kids [3 0 R] /Count 1 >>".to_vec(),
+ b"<< /Type /Page /Parent 2 0 R /MediaBox [0 0 300 300] >>".to_vec(),
+ // 4: the removed link
+ b"<< /Type /Annot /Subtype /Link /Rect [0 0 10 10] /A << /S /URI /URI (https://secret.example) >> >>".to_vec(),
+ // 5 root, 6 document, 7 section, 8 link element with the OBJR
+ b"<< /Type /StructTreeRoot /K 6 0 R >>".to_vec(),
+ b"<< /Type /StructElem /S /Document /P 5 0 R /K [7 0 R] >>".to_vec(),
+ b"<< /Type /StructElem /S /Sect /P 6 0 R /K [8 0 R] >>".to_vec(),
+ b"<< /Type /StructElem /S /Link /P 7 0 R /K [<< /Type /OBJR /Obj 4 0 R >>] >>".to_vec(),
+ ];
+ let gone: std::collections::HashSet = [ObjRef::new(4, 0)].into();
+ let catalog = |doc: &Document| doc.get(doc.root().unwrap()).as_dict().cloned().unwrap();
+
+ let mut doc = pdf(objs.clone());
+ assert_eq!(crate::tags::drop_object_refs_within(&mut doc, &gone, 100).unwrap(), 1);
+ assert!(catalog(&doc).contains(b"StructTreeRoot") && catalog(&doc).contains(b"MarkInfo"), "under the limit the tree stays");
+
+ let mut doc = pdf(objs);
+ crate::tags::drop_object_refs_within(&mut doc, &gone, 2).unwrap();
+ assert!(!catalog(&doc).contains(b"StructTreeRoot"), "a tree the walk couldn't finish goes whole");
+ assert!(!catalog(&doc).contains(b"MarkInfo"));
+ let bytes = write_full(&doc, &SaveOptions { object_streams: false, ..SaveOptions::default() }).unwrap();
+ assert!(!bytes.windows(14).any(|w| w == b"secret.example"), "the removed link can't be reached any more");
+}
+
/// Issue #1325: removing comments keeps the annotations that are not comments: the multimedia
/// and print-production ones (Watermark, PrinterMark, TrapNet, Screen, Movie, RichMedia, 3D).
#[test]
diff --git a/crates/render/src/inspect.rs b/crates/render/src/inspect.rs
index 3c3988b7b5..0979f3d38a 100644
--- a/crates/render/src/inspect.rs
+++ b/crates/render/src/inspect.rs
@@ -56,6 +56,9 @@ pub struct DocInfo {
pub keywords: Option,
pub creator: Option,
pub producer: Option,
+ /// Custom document properties: `/Info` entries with a key the spec does not name and a text
+ /// value, as (name, text), in the dictionary's order (at most 256).
+ pub custom: Vec<(String, String)>,
pub pdf_version: String,
pub file_size: usize,
pub encrypted: bool,
@@ -616,6 +619,23 @@ impl<'a> Inspector<'a> {
info.keywords = self.text(d, b"Keywords");
info.creator = self.text(d, b"Creator");
info.producer = self.text(d, b"Producer");
+ const STANDARD: [&[u8]; 9] =
+ [b"Title", b"Author", b"Subject", b"Keywords", b"Creator", b"Producer", b"CreationDate", b"ModDate", b"Trapped"];
+ // The same rule as the engine's list after an edit (`pdfcraft_organize::custom_info`,
+ // trimmed and without empty values): text strings only, so a name value is not a
+ // custom property, and the cap counts listed entries.
+ info.custom = d
+ .iter()
+ .filter(|(k, _)| !STANDARD.contains(&k.as_slice()))
+ .filter_map(|(k, v)| match self.resolve(v) {
+ Object::String(bytes, _) => {
+ let v = text_string(bytes).trim_matches(|c: char| c == '\0' || c.is_whitespace()).to_string();
+ (!v.is_empty()).then(|| (String::from_utf8_lossy(k).into_owned(), v))
+ }
+ _ => None,
+ })
+ .take(256)
+ .collect();
}
let Ok(catalog) = self.doc.catalog() else { return };
info.tagged = catalog
@@ -678,6 +698,12 @@ impl<'a> Inspector<'a> {
}
info.has_javascript |= names.get(b"JavaScript").is_ok();
}
+ if let Ok(action) = catalog.get(b"OpenAction")
+ && let Some(dict) = self.dict(action)
+ && dict.get(b"S").ok().and_then(|o| self.resolve(o).as_name().ok()) == Some(&b"JavaScript"[..])
+ {
+ info.has_javascript = true;
+ }
}
// ── object helpers ──────────────────────────────────────────────────────────────────────
@@ -2302,6 +2328,27 @@ trailer << /Root 1 0 R >>
);
}
+ const CUSTOM_INFO: &[u8] = b"%PDF-1.7
+1 0 obj << /Type /Catalog /Pages 2 0 R >> endobj
+2 0 obj << /Type /Pages /Kids [3 0 R] /Count 1 /MediaBox [0 0 200 200] >> endobj
+3 0 obj << /Type /Page /Parent 2 0 R >> endobj
+4 0 obj << /Title (Report) /Producer (Test) /Project (Apollo) /Revision 7 /Dept (R&D) /Trapped /False /Note >> endobj
+trailer << /Root 1 0 R /Info 4 0 R >>
+%%EOF";
+
+ #[test]
+ fn custom_document_properties_are_listed_apart_from_the_standard_ones() {
+ let info = inspect(Arc::new(CUSTOM_INFO.to_vec()), None).expect("opens");
+ assert_eq!((info.title.as_deref(), info.producer.as_deref()), (Some("Report"), Some("Test")));
+ let mut custom = info.custom.clone();
+ custom.sort();
+ // Revision is a number, not text, so it is not listed; Trapped is a standard entry.
+ assert_eq!(
+ custom,
+ [("Dept".to_string(), "R&D".to_string()), ("Note".to_string(), "部".to_string()), ("Project".to_string(), "Apollo".to_string())]
+ );
+ }
+
#[test]
fn radio_button_layer_groups_are_read() {
let info = inspect(Arc::new(LAYERS.to_vec()), None).expect("opens");
@@ -2982,3 +3029,49 @@ trailer << /Size 7 /Root 1 0 R >>
assert_eq!((f.substitute.as_deref(), f.substitution.as_deref()), (None, None), "the document's own glyphs are shown");
}
}
+
+#[cfg(test)]
+mod open_action_javascript_tests {
+ use std::sync::Arc;
+
+ use super::inspect;
+
+ /// One-page PDF whose catalog points its /OpenAction at object 4 (written by the caller).
+ fn pdf_with_open_action(action: &[u8]) -> Vec {
+ let objs: Vec<&[u8]> = vec![
+ b"<< /Type /Catalog /Pages 2 0 R /OpenAction 4 0 R >>",
+ b"<< /Type /Pages /Kids [3 0 R] /Count 1 >>",
+ b"<< /Type /Page /Parent 2 0 R /MediaBox [0 0 612 792] >>",
+ action,
+ ];
+ let mut out = b"%PDF-1.7\n".to_vec();
+ let mut offsets = Vec::new();
+ for (i, o) in objs.iter().enumerate() {
+ offsets.push(out.len());
+ out.extend_from_slice(format!("{} 0 obj\n", i + 1).as_bytes());
+ out.extend_from_slice(o);
+ out.extend_from_slice(b"\nendobj\n");
+ }
+ let xref = out.len();
+ out.extend_from_slice(format!("xref\n0 {}\n0000000000 65535 f \n", objs.len() + 1).as_bytes());
+ for o in offsets {
+ out.extend_from_slice(format!("{o:010} 00000 n \n").as_bytes());
+ }
+ out.extend_from_slice(format!("trailer\n<< /Size {} /Root 1 0 R >>\nstartxref\n{xref}\n%%EOF\n", objs.len() + 1).as_bytes());
+ out
+ }
+
+ #[test]
+ fn open_action_javascript_sets_has_javascript() {
+ let bytes = pdf_with_open_action(b"<< /S /JavaScript /JS (app.alert('hi')) >>");
+ let info = inspect(Arc::new(bytes), None).expect("inspects");
+ assert!(info.has_javascript, "an /OpenAction /JavaScript is JavaScript");
+ }
+
+ #[test]
+ fn open_action_goto_does_not_set_has_javascript() {
+ let bytes = pdf_with_open_action(b"<< /S /GoTo /D [3 0 R /Fit] >>");
+ let info = inspect(Arc::new(bytes), None).expect("inspects");
+ assert!(!info.has_javascript, "a /GoTo /OpenAction is not JavaScript");
+ }
+}
diff --git a/crates/render/src/raster.rs b/crates/render/src/raster.rs
index b2127894fd..1e4747ef18 100644
--- a/crates/render/src/raster.rs
+++ b/crates/render/src/raster.rs
@@ -3631,6 +3631,46 @@ endstream endobj
}
}
+ /// #3308: a fax image whose `/Rows` says fewer rows than its `/Height` (a plotter driver wrote
+ /// `/Rows` equal to `/Columns`) is decoded to its height: only the first `/Rows` rows were
+ /// decoded and the rest drawn black, as MuPDF and Acrobat don't. `/BlackIs1` turns the fixture
+ /// white, so rows left undecoded show as black. Also through an `/Indexed` colour space, the
+ /// filter's other output path.
+ #[test]
+ fn group4_fax_with_rows_below_its_height_decodes_to_the_image_height() {
+ const ALL_BLACK_8X4: [u8; 6] = [0x26, 0xa2, 0xfe, 0x00, 0x20, 0x02];
+ for (space, rows) in [("/DeviceGray", "/Rows 2"), ("[/Indexed /DeviceRGB 1 <000000FFFFFF>]", "/Rows 2"), ("/DeviceGray", "/Rows 4")] {
+ let content = "q 20 0 0 20 5 5 cm /Im1 Do Q";
+ let mut pdf = format!(
+ "%PDF-1.7
+1 0 obj << /Type /Catalog /Pages 2 0 R >> endobj
+2 0 obj << /Type /Pages /Kids [3 0 R] /Count 1 >> endobj
+3 0 obj << /Type /Page /Parent 2 0 R /MediaBox [0 0 40 40] /Contents 4 0 R /Resources << /XObject << /Im1 5 0 R >> >> >> endobj
+4 0 obj << /Length {} >> stream
+{content}
+endstream endobj
+5 0 obj << /Type /XObject /Subtype /Image /Width 8 /Height 4 /ColorSpace {space} /BitsPerComponent 1 /Filter /CCITTFaxDecode /DecodeParms << /K -1 /Columns 8 {rows} /BlackIs1 true >> /Length {} >> stream
+",
+ content.len(),
+ ALL_BLACK_8X4.len()
+ )
+ .into_bytes();
+ pdf.extend_from_slice(&ALL_BLACK_8X4);
+ pdf.extend_from_slice(
+ b"
+endstream endobj
+trailer << /Root 1 0 R >>
+%%EOF",
+ );
+ let mut renderer = PageRenderer::new(Arc::new(pdf), RenderConfig::default());
+ let page = renderer.render(RenderRequest { page: 0, kind: RequestKind::Pixels, tile: None, scale: 1.0, tag: 0, ..Default::default() });
+ assert!(page.error.is_none(), "{space} {rows}: {:?}", page.error);
+ for y in [17, 22, 27, 32] {
+ assert_eq!(&page.rgba[((y * 40 + 15) * 4)..][..4], &[255, 255, 255, 255], "{space} {rows}: image row at y {y} is decoded");
+ }
+ }
+ }
+
#[test]
fn tiff_predictor_packed_rgb_rows_keep_their_byte_boundaries() {
use hayro_syntax::object::{FromBytes, Object};
@@ -6852,4 +6892,36 @@ trailer << /Root 1 0 R >>
assert!(render(glyph, 3).as_chunks::<4>().0.iter().all(|p| p[..3] == [255, 255, 255]), "{glyph}: 3 Tr hides the glyph");
}
}
+
+ /// #3109 / #3116: clipping-only text (`7 Tr`) installs a clip for subsequent painting
+ /// but draws no pixels itself.
+ #[test]
+ fn test_clip_only_text_raster_preserves_clipping_without_painting() {
+ let pdf_bytes = |content: &str| {
+ Arc::new(
+ format!(
+ "%PDF-1.7\n1 0 obj << /Type /Catalog /Pages 2 0 R >> endobj\n2 0 obj << /Type /Pages /Kids [3 0 R] /Count 1 >> endobj\n3 0 obj << /Type /Page /Parent 2 0 R /MediaBox [0 0 200 100] /Contents 4 0 R /Resources << /Font << /F1 5 0 R >> >> >> endobj\n4 0 obj << /Length {} >> stream\n{content}\nendstream endobj\n5 0 obj << /Type /Font /Subtype /Type1 /BaseFont /Helvetica /Encoding /WinAnsiEncoding >> endobj\ntrailer << /Root 1 0 R >>\n%%EOF",
+ content.len()
+ )
+ .into_bytes(),
+ )
+ };
+ let render = |content: &str| {
+ let mut r = PageRenderer::new(pdf_bytes(content), RenderConfig::default());
+ let p = r.render(RenderRequest { page: 0, kind: RequestKind::Pixels, tile: None, scale: 1.0, tag: 0, ..Default::default() });
+ assert!(p.error.is_none(), "{content}: {:?}", p.error);
+ p
+ };
+ let white = [255, 255, 255, 255];
+
+ // The text alone paints nothing.
+ let p = render("BT /F1 40 Tf 7 Tr 20 60 Td (Hi) Tj ET");
+ assert!(p.rgba.as_chunks::<4>().0.iter().all(|c| *c == white), "7 Tr alone leaves the page blank");
+
+ // A fill afterwards shows only through the text-shaped clip.
+ let p = render("q BT /F1 40 Tf 7 Tr 20 60 Td (Hi) Tj ET 0 0 1 rg 0 0 200 100 re f Q");
+ let pixels = p.rgba.as_chunks::<4>().0;
+ assert!(pixels.contains(&[0, 0, 255, 255]), "fill shows through the 7 Tr clip");
+ assert!(pixels.contains(&white), "the clip is text-shaped, not the whole page");
+ }
}
diff --git a/crates/ui-egui/src/canvas.rs b/crates/ui-egui/src/canvas.rs
index 855361f6f3..ad74834656 100644
--- a/crates/ui-egui/src/canvas.rs
+++ b/crates/ui-egui/src/canvas.rs
@@ -446,6 +446,9 @@ pub struct DocView {
pub(crate) text_failed: HashSet,
pub find: Option,
selection: Option,
+ /// The selection whose action bar was closed by its Copy button: the bar stays hidden
+ /// until the selection changes, as it does once a markup button consumes the selection.
+ actions_dismissed: Option,
/// The drag in progress started a text selection: the wheel scrolls the page during it
/// (#721), which egui's scroll area refuses while anything is dragged.
selection_drag: bool,
@@ -479,6 +482,9 @@ pub struct DocView {
/// (#759). The window outlasts a single frame so egui's internal easing tail is suppressed as
/// it decays, instead of adding on top of the 1:1 motion.
precise_scroll_until: f64,
+ /// Per-axis line-wheel state; horizontal input cannot extend a vertical burst.
+ wheel_boost_x: crate::wheel_boost::WheelBoost,
+ wheel_boost_y: crate::wheel_boost::WheelBoost,
/// Touchpad momentum after the fingers lift (#531).
fling: crate::fling::Fling,
pub(crate) auto_scroll: crate::autoscroll::AutoScroll,
@@ -735,6 +741,7 @@ impl DocView {
text_failed: HashSet::new(),
find: None,
selection: None,
+ actions_dismissed: None,
selection_drag: false,
clicks: 0,
last_press_time: 0.0,
@@ -752,6 +759,8 @@ impl DocView {
shown: false,
wheel: Default::default(),
precise_scroll_until: 0.0,
+ wheel_boost_x: Default::default(),
+ wheel_boost_y: Default::default(),
fling: Default::default(),
auto_scroll: Default::default(),
selected: BTreeSet::new(),
@@ -831,6 +840,7 @@ impl DocView {
self.flash = None;
// A thumbnail drag holds page indexes from before the change.
self.panel_drag = None;
+ self.reset_coarse_state();
}
/// The document was replaced by a newer version of its file (#431). The reader's place
@@ -1048,6 +1058,7 @@ impl DocView {
pub(crate) fn clear_selection(&mut self) {
self.selection = None;
+ self.actions_dismissed = None;
}
/// Select text on `page` from glyph `from` to glyph `to` (tests and automation).
@@ -1148,6 +1159,7 @@ impl DocView {
.unwrap_or(0.0);
self.goto = Some((page, frac));
self.current = page;
+ self.reset_coarse_state();
}
/// Screen position of the centre of glyph `glyph` on `page`, if that page is on screen and its
@@ -1169,6 +1181,14 @@ impl DocView {
self.thumbs.get(&page).map(|t| &t.tex)
}
+ /// Drops wheel state after a discontinuous jump so the destination cannot
+ /// inherit momentum from the previous viewport.
+ fn reset_coarse_state(&mut self) {
+ self.wheel.clear_motion();
+ self.wheel_boost_x.reset();
+ self.wheel_boost_y.reset();
+ }
+
pub fn go_to_page(&mut self, page: usize) {
let page = page.min(self.page_count.saturating_sub(1));
if page != self.current {
@@ -1182,7 +1202,7 @@ impl DocView {
self.forward.clear();
}
// Partial wheel motion belongs to the old place.
- self.wheel.clear_motion();
+ self.reset_coarse_state();
self.goto = Some((page, 0.0));
self.current = page;
}
@@ -1317,6 +1337,7 @@ impl DocView {
let page = page.min(self.page_count.saturating_sub(1));
self.goto = Some((page, 0.0));
self.current = page;
+ self.reset_coarse_state();
true
}
@@ -1341,12 +1362,14 @@ impl DocView {
pub fn rotate_view(&mut self, clockwise: bool) {
self.rotation = (self.rotation + if clockwise { 90 } else { 270 }) % 360;
self.goto = Some((self.current, 0.0));
+ self.reset_coarse_state();
}
/// Switch the zoom mode, keeping the current page in view.
pub fn set_fit(&mut self, fit: Fit) {
self.fit = fit;
self.goto = Some((self.current, 0.0));
+ self.reset_coarse_state();
}
/// Switch the base page display (Single or Two-page), leaving continuous scrolling as it is.
@@ -1358,6 +1381,7 @@ impl DocView {
}
self.layout = layout;
self.goto = Some((self.current, 0.0));
+ self.reset_coarse_state();
}
/// Whether the view scrolls continuously (as opposed to one page or spread at a time).
@@ -1371,6 +1395,7 @@ impl DocView {
if self.continuous != continuous {
self.continuous = continuous;
self.goto = Some((self.current, 0.0));
+ self.reset_coarse_state();
}
}
@@ -1386,6 +1411,7 @@ impl DocView {
if self.cover != cover {
self.cover = cover;
self.goto = Some((self.current, 0.0));
+ self.reset_coarse_state();
}
}
@@ -2190,11 +2216,20 @@ pub fn shortcuts(view: &mut DocView, ctx: &egui::Context, speed: crate::scroll_s
if ctx.input_mut(|i| i.consume_shortcut(&KeyboardShortcut::new(Modifiers::COMMAND, Key::F))) {
view.open_find();
}
+ let cmd = |k| KeyboardShortcut::new(Modifiers::COMMAND, k);
+ let pressed = |s: KeyboardShortcut| ctx.input_mut(|i| i.consume_shortcut(&s));
+ // ⌘G / ⇧⌘G step through the matches from the find field too, as in Acrobat and Preview
+ // (#2669): ahead of the text-input check, but only while there is a search to step.
+ if view.find.is_some() {
+ if pressed(KeyboardShortcut::new(Modifiers::COMMAND | Modifiers::SHIFT, Key::G)) {
+ view.find_step(false);
+ } else if pressed(cmd(Key::G)) {
+ view.find_step(true);
+ }
+ }
if ctx.egui_wants_keyboard_input() {
return;
}
- let cmd = |k| KeyboardShortcut::new(Modifiers::COMMAND, k);
- let pressed = |s: KeyboardShortcut| ctx.input_mut(|i| i.consume_shortcut(&s));
// Rotation first: egui matches ⌘+ loosely with respect to Shift.
if pressed(KeyboardShortcut::new(Modifiers::COMMAND | Modifiers::SHIFT, Key::Plus))
|| pressed(KeyboardShortcut::new(Modifiers::COMMAND | Modifiers::SHIFT, Key::Equals))
@@ -2235,11 +2270,6 @@ pub fn shortcuts(view: &mut DocView, ctx: &egui::Context, speed: crate::scroll_s
view.fit = Fit::Width;
view.goto = Some((view.current, 0.0));
}
- if pressed(KeyboardShortcut::new(Modifiers::COMMAND | Modifiers::SHIFT, Key::G)) {
- view.find_step(false);
- } else if pressed(cmd(Key::G)) {
- view.find_step(true);
- }
if pressed(cmd(Key::OpenBracket)) {
view.view_history(false);
}
@@ -2309,6 +2339,12 @@ pub fn shortcuts(view: &mut DocView, ctx: &egui::Context, speed: crate::scroll_s
/// How far ↓ / ↑ scroll at Normal speed (Scroll speed halves or doubles it). Page Down / Page Up
/// keep this much of the previous screen in view.
const KEY_SCROLL_LINE: f32 = 40.0;
+/// Suppression window for egui's eased copy of point input.
+///
+/// egui 0.36.2 drains `WheelState::after_events` tail deltas by 90% per 0.1 s.
+const SUPPRESS_SECS: f64 = 0.25;
+/// Short rolling extension while egui still has an unprocessed wheel tail.
+const SUPPRESS_TAIL_SECS: f64 = 0.05;
pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
let t = Tokens::get(ui.ctx());
@@ -2446,9 +2482,10 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
// When not scrolling continuously, only one row is shown at a time: the current page, or the
// current two-page spread (with the cover rule). These indices drive paging, which pages are
// drawn, and the content height (#752). Empty for a document with no pages.
- let row: Vec = if rects.is_empty() {
- Vec::new()
- } else {
+ let shown_row = |view: &DocView| -> Vec {
+ if rects.is_empty() {
+ return Vec::new();
+ }
let last = rects.len() - 1;
let cur = view.current.min(last);
match view.layout {
@@ -2460,6 +2497,7 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
}
}
};
+ let row = shown_row(view);
let middle_gesture = view.auto_scroll.blocks_input();
// Not scrolling: when the whole page or spread fits, the wheel would do nothing, so it turns
// pages instead; zoomed in far enough to pan, it pans, and at the edge it turns the page on
@@ -2496,36 +2534,25 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
ui.input_mut(|i| i.smooth_scroll_delta.y = 0.0);
}
}
- // Touchpad scrolling 1:1 (#759). A precision touchpad sends pixel-precise `MouseWheel`
- // events; on Windows they carry no `TouchPhase::Start`, so egui never takes its no-smoothing
- // path and eases deltas of 8 px or more, leaving the page a frame behind the finger. Apply
- // the precise delta ourselves this frame through the un-animated scroll path, and suppress
- // egui's eased copy while it decays so it neither doubles nor trails the motion. A coarse
- // mouse wheel keeps egui's easing (pleasant for chunky notches).
+ // Point input bypasses egui smoothing for 1:1 motion. Line input keeps egui's configured,
+ // smoothed base and adds only the accelerated surplus. While point suppression is active,
+ // reconstruct line and page input here so current coarse events survive without exposing an
+ // older eased tail.
let now = ui.input(|i| i.time);
- let mut precise_scroll = Vec2::ZERO;
- if ui.rect_contains_pointer(avail) {
- let mut coarse_wheel = false;
- ui.input(|i| {
- for e in &i.events {
- if let egui::Event::MouseWheel { unit, delta, modifiers, .. } = e {
- if modifiers.command || modifiers.ctrl {
- continue; // Zooming, not scrolling.
- }
- match unit {
- // Shift scrolls sideways, matching egui's own wheel handling.
- egui::MouseWheelUnit::Point if modifiers.shift => precise_scroll.x += delta.x + delta.y,
- egui::MouseWheelUnit::Point => precise_scroll += *delta,
- egui::MouseWheelUnit::Line | egui::MouseWheelUnit::Page => coarse_wheel = true,
- }
- }
- }
- });
- if coarse_wheel {
- view.precise_scroll_until = 0.0;
- } else if precise_scroll != Vec2::ZERO {
- view.precise_scroll_until = now + 0.25;
- }
+ let input = if ui.rect_contains_pointer(avail) {
+ // Match the line distance used by egui's other scroll areas.
+ let live = ui.ctx().options(|o| o.input_options.line_scroll_speed);
+ let line_points = if live.is_finite() && live > 0.0 { live } else { crate::scroll_speed::EGUI_LINE_POINTS };
+ ui.input(|i| crate::wheel_boost::accumulate(&i.events, line_points, single_paging))
+ } else {
+ crate::wheel_boost::FrameInput::default()
+ };
+ let mut precise_scroll = input.precise;
+ let mut line_base = input.line;
+ let mut boost_extra = crate::wheel_boost::boost_extra(input, &mut view.wheel_boost_x, &mut view.wheel_boost_y, now);
+ let has_page = input.page != Vec2::ZERO;
+ if now.is_finite() && precise_scroll != Vec2::ZERO && !has_page {
+ view.precise_scroll_until = now + SUPPRESS_SECS;
}
if edge_turned {
// The turn leaves egui's eased copy of the notch to decay for a few frames; it must not
@@ -2538,11 +2565,40 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
let paging_owns_vertical = single_paging || (!view.continuous && view.wheel.gesture_turned());
if paging_owns_vertical {
precise_scroll.y = 0.0;
+ line_base.y = 0.0;
+ boost_extra.y = 0.0;
}
if now < view.precise_scroll_until {
+ if ui.input(|i| i.is_scrolling()) {
+ // Keep hiding egui's unprocessed copy until it has drained. The short rolling grace
+ // avoids a delayed duplicate without prolonging suppression after the tail is gone.
+ view.precise_scroll_until = view.precise_scroll_until.max(now + SUPPRESS_TAIL_SECS);
+ }
ui.input_mut(|i| i.smooth_scroll_delta = Vec2::ZERO);
+ // The point path suppresses every eased wheel delta. Restore the line base through the
+ // same un-eased path as its surplus; paging-owned axes were cleared above.
+ boost_extra = crate::wheel_boost::cap_frame(boost_extra + line_base);
+ if has_page {
+ // Page units normally stay on egui's viewport-sized path. During suppression only,
+ // reproduce the current event directly so an older hidden tail stays hidden.
+ let page_scale = ui.input(|i| i.viewport_rect().height());
+ let mut page_delta =
+ if page_scale.is_finite() && page_scale > 0.0 { crate::wheel_boost::cap_frame(input.page * page_scale) } else { Vec2::ZERO };
+ if paging_owns_vertical {
+ page_delta.y = 0.0;
+ }
+ boost_extra = crate::wheel_boost::cap_frame(boost_extra + page_delta);
+ }
} else {
precise_scroll = Vec2::ZERO;
+ if line_base != Vec2::ZERO || has_page {
+ // Selection handling below uses zero to distinguish an inactive window.
+ view.precise_scroll_until = 0.0;
+ }
+ }
+ if has_page && paging_owns_vertical {
+ // Page input belongs to egui on the free axis; the pager still owns vertical motion.
+ ui.input_mut(|i| i.smooth_scroll_delta.y = 0.0);
}
// Touchpad momentum (#531): a swipe over the page lifted at speed glides on and slows to a
// stop, through the same un-eased path. Never while the pager owns the wheel, a gesture or
@@ -2560,6 +2616,9 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
if view.fling.gliding() {
ui.ctx().request_repaint();
}
+ // The wheel may have turned the page above: draw the row it turned to. Measured from the old
+ // row, the jump to the new page scrolled both out of view, a blank frame on every turn forward.
+ let row = shown_row(view);
let visible_pages: Vec = if view.continuous { (0..rects.len()).collect() } else { row.clone() };
let (y_shift, content_h) = if view.continuous {
(0.0, rects.iter().map(|r| r.bottom()).reduce(f32::max).map(|bottom| bottom + MARGIN).unwrap_or(0.0))
@@ -2618,13 +2677,26 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
let r = rects[page.min(rects.len() - 1)];
let point = pos2(r.left() + fx * r.width(), r.top() - y_shift + fy * r.height());
scroll = scroll.scroll_offset(vec2((point.x - rel.x).max(0.0), (point.y - rel.y).max(0.0)));
- } else if edge_turned {
- // The row above was laid out for the page before the turn: place the new one next frame.
- ui.ctx().request_repaint();
} else if let Some((page, frac)) = view.goto.take() {
let page = page.min(rects.len() - 1);
let r = rects[page];
- scroll = scroll.vertical_scroll_offset((r.top() - y_shift - GAP + frac * r.height()).max(0.0));
+ // Clamped here: egui clamps only after drawing, so a page that fits was drawn MARGIN - GAP
+ // too high for a frame and then hopped into place. (`avail` is the viewport, as the
+ // scroll bars float.)
+ let max_y = (content_h - avail.height()).max(0.0);
+ scroll = scroll.vertical_scroll_offset((r.top() - y_shift - GAP + frac * r.height()).max(0.0).min(max_y));
+ // Horizontally too, as Acrobat does (#2665): beside a much wider page the others sit off
+ // to one side. A page already shown whole, or one filling the window's width, keeps the
+ // scroll, so paging through a zoomed-in document stays on the same column. Otherwise a
+ // page that fits is centred, and a wider one scrolled just far enough to fill the window.
+ // The page counts with its side gutters, which keep its edge clear of the floating
+ // quick-action bar as the layout does.
+ let (x, w) = (view.last_scroll.x, avail.width());
+ let (left, right) = (r.left() - SIDE, r.right() + SIDE);
+ let target = if right - left <= w { if left >= x && right <= x + w { x } else { r.center().x - w / 2.0 } } else { x.clamp(left, right - w) };
+ if (target - x).abs() > 0.5 {
+ scroll = scroll.horizontal_scroll_offset(target.clamp(0.0, (content_w - w).max(0.0)));
+ }
}
let ppp = ui.ctx().pixels_per_point();
let scale = view.render_scale(ppp);
@@ -2739,6 +2811,7 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
scroll = scroll.vertical_scroll_offset((r.top() - y_shift + fy * r.height() - rel.y).clamp(0.0, max_y));
}
}
+ let selection_wheel = view.selection_drag && navigation_enabled && unobstructed && !single_paging && ui.rect_contains_pointer(avail);
let out = scroll.show_viewport(ui, |ui, viewport| {
// egui's drag responses accept every pointer button. Keep a wheel gesture from also
// selecting text, drawing a mark, or panning with the Hand tool.
@@ -2749,10 +2822,13 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
ui.set_opacity(opacity);
}
// Middle-button auto-scroll, the keyboard, precise touchpad input (#759) and its glide
- // (#531), and a drawing's edge scroll; positive y moves the content down. All scroll
- // without easing, unlike egui's wheel default.
- let delta = auto_delta + edge_delta - vec2(0.0, key_scroll) + precise_scroll + glide;
- if delta != Vec2::ZERO {
+ // (#531), accelerated wheel surplus (#1072), and a drawing's edge scroll; positive y
+ // moves the content down. All scroll without easing, unlike egui's wheel default.
+ // ScrollArea ignores wheel input during a widget drag. Defer the wheel contribution
+ // until after the text widgets when a selection owns that drag.
+ let wheel_delta = if selection_wheel { Vec2::ZERO } else { precise_scroll + boost_extra };
+ let delta = auto_delta + edge_delta - vec2(0.0, key_scroll) + wheel_delta + glide;
+ if delta != Vec2::ZERO && delta.is_finite() {
ui.scroll_with_delta_animation(delta, egui::style::ScrollAnimation::none());
}
let (resp_rect, resp) = ui.allocate_exact_size(vec2(content_w, content_h), Sense::click_and_drag());
@@ -3250,6 +3326,9 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
}
}
if resp.drag_started() && press_here && !over_link {
+ // A drag replaces the selection without passing through none, so a
+ // re-drag over text whose bar Copy closed would keep the bar hidden.
+ view.actions_dismissed = None;
view.selection_drag = true;
let origin = ui.input(|inp| inp.pointer.press_origin()).unwrap_or(p);
let (ox, oy) = xf.screen_to_view(origin);
@@ -3334,6 +3413,7 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
if tool == QuickTool::Select && !consumed {
crate::image_pick::page_input(&resp, &xf, i, info, doc, pointer, view);
}
+ crate::image_pick::drop_stale(doc, view);
crate::image_pick::paint(painter, &xf, i, info, view);
comments::page_after_text(&resp, &pcx, view);
if tool == QuickTool::Redact && can_modify {
@@ -3481,17 +3561,13 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
// ScrollArea ignores the wheel while any widget is dragged, so a text-selection drag
// scrolls the page itself (#721). After the page loop, so it keeps scrolling once the
// pointer leaves the selection's page for the gap or the next page; over the whole
- // document area, like #759. A Line/Page event clears precise_scroll_until; a Point
- // gesture's eased tail after its suppression deadline is not picked up.
- if view.selection_drag
- && navigation_enabled
- && unobstructed
- && !single_paging
- && view.precise_scroll_until == 0.0
- && ui.rect_contains_pointer(avail)
- {
- let delta = ui.input_mut(|input| std::mem::take(&mut input.smooth_scroll_delta));
- if delta != Vec2::ZERO {
+ // document area, like #759. Combine direct and smoothed motion in one request so mixed
+ // Point and Line events retain their full contribution.
+ if selection_wheel {
+ let smooth =
+ if view.precise_scroll_until == 0.0 { ui.input_mut(|input| std::mem::take(&mut input.smooth_scroll_delta)) } else { Vec2::ZERO };
+ let delta = precise_scroll + boost_extra + smooth;
+ if delta != Vec2::ZERO && delta.is_finite() {
ui.scroll_with_delta_animation(delta, egui::style::ScrollAnimation::none());
}
}
@@ -3666,6 +3742,26 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
view.last_scroll = scroll_offset;
let pan_held = now_settle < view.pan_hold_until;
let (mut wanted, visible_now) = out.inner;
+ // Single-page and two-page views draw only the current row, so the next one was first asked
+ // for on the turn and showed blank paper until it rendered. Prefetch the rows either side,
+ // as continuous scrolling does for the pages just off screen.
+ if !view.continuous
+ && let (Some(&first), Some(&last)) = (visible_pages.first(), visible_pages.last())
+ {
+ let n = if view.layout == PageLayout::Single { 1 } else { 2 };
+ let max_side = ui.ctx().input(|inp| inp.max_texture_side) as f32;
+ for i in (last.saturating_add(1)..=last.saturating_add(n)).chain(first.saturating_sub(n)..first) {
+ let Some(p) = info.pages.get(i).filter(|_| !view.errors.contains_key(&i)) else { continue };
+ // As drawn: the whole page at this scale, or the backdrop of a tiled page.
+ let side = p.width.max(p.height).max(1.0);
+ wanted.push(if side * scale > TILE_THRESHOLD.min(max_side) {
+ let bs = BASE_SIDE.min(max_side) / side;
+ (i, bs, scale_tag(bs), None)
+ } else {
+ (i, scale, tag, None)
+ });
+ }
+ }
let cur = view.current;
wanted.sort_by_key(|w| (!visible_now.contains(&w.0), w.0.abs_diff(cur), w.3.is_some()));
// Tiles of far-away pages are useless: free them. Tiles of an earlier zoom stay for a page
@@ -3956,7 +4052,10 @@ pub fn document_area(app: &mut PdfCraftApp, index: usize, ui: &mut egui::Ui) {
if copy_selected {
app.execute("edit.copy");
}
- quick_bar(app, avail, ui);
+ // Read mode hides the tools (#2646).
+ if app.mode != crate::Mode::Read {
+ quick_bar(app, avail, ui);
+ }
if let Some((source, file, destination)) = remote_open {
app.open_remote_link(source, file, destination);
}
@@ -3983,7 +4082,13 @@ fn selection_actions(
if ui.input(|input| input.pointer.is_decidedly_dragging()) {
return false;
}
- let Some(selection) = view.selection else { return false };
+ let Some(selection) = view.selection else {
+ view.actions_dismissed = None;
+ return false;
+ };
+ if view.actions_dismissed == Some(selection) {
+ return false;
+ }
let Some(text) = view.texts.get(&selection.page) else { return false };
let Some(xf) = view.page_xform(selection.page) else { return false };
let Some(last) = selection.rects(text).last().copied() else { return false };
@@ -4020,6 +4125,8 @@ fn selection_actions(
});
let Some(action) = chosen else { return false };
if let SelectionAction::Copy = action {
+ // Copying closes the bar but keeps the words selected, for ⌘C or a second action.
+ view.actions_dismissed = Some(selection);
return true;
}
let Some((page, quads)) = view.selection_quads(info) else { return false };
@@ -4800,6 +4907,15 @@ pub(crate) fn drop_gap(cells: &[(usize, Rect)], p: Pos2) -> Option {
Some(if p.x < r.center().x { *i } else { i + 1 })
}
+/// Ctrl (Option on macOS) held when a page drag drops copies the dragged pages into the gap
+/// instead of moving them, as Acrobat's Organize Pages does.
+///
+/// The OS comes from the context, not `cfg!`: the wasm build runs on macOS too, where
+/// `cfg!(target_os = "macos")` is false, so Option has to copy there as well (#3092 review).
+fn copy_modifier(ctx: &egui::Context, m: &egui::Modifiers) -> bool {
+ if crate::commands::mac_shortcuts(ctx) { m.alt } else { m.ctrl }
+}
+
/// A "+" on a gap of the page grid: insert files there. Returns whether it was clicked.
fn gap_button(ui: &mut egui::Ui, gap: usize, at: Pos2, height: f32, label: String, t: &Tokens) -> bool {
let resp = ui.interact(Rect::from_center_size(at, Vec2::splat(22.0)), ui.id().with(("org-gap", gap)), Sense::click());
@@ -5031,6 +5147,8 @@ fn organize_grid(view: &mut DocView, info: &DocInfo, editable: bool, dirty: bool
}
// While dragging: the gap the pages would go to, drawn as a bar.
if let (Some(pages), Some(p)) = (view.org_drag.clone(), ui.input(|i| i.pointer.hover_pos())) {
+ // Ctrl (Option on macOS) held over the drop copies the pages; see #2997.
+ let copying = copy_modifier(ui.ctx(), &ui.input(|i| i.modifiers));
if let Some(gap) = drop_gap(&cells, p) {
let x = match cells.iter().find(|(i, _)| *i == gap) {
Some((_, r)) => r.left() + 3.0,
@@ -5038,8 +5156,18 @@ fn organize_grid(view: &mut DocView, info: &DocInfo, editable: bool, dirty: bool
};
let row = cells.iter().find(|(i, _)| *i == gap).or(cells.last()).map(|(_, r)| r.y_range()).unwrap_or(egui::Rangef::new(0.0, 0.0));
ui.painter().line_segment([pos2(x, row.min + 10.0), pos2(x, row.max - 10.0)], Stroke::new(3.0, t.accent));
+ let mut text_at = p + vec2(14.0, 14.0);
+ if copying {
+ // A "+" by the pointer: this drop copies the pages, it doesn't move them.
+ let badge = p + vec2(12.0, 12.0);
+ ui.painter().circle(badge, 9.0, t.accent, Stroke::new(1.0, t.accent));
+ let ink = Stroke::new(1.5, Color32::WHITE);
+ ui.painter().line_segment([badge - vec2(4.0, 0.0), badge + vec2(4.0, 0.0)], ink);
+ ui.painter().line_segment([badge - vec2(0.0, 4.0), badge + vec2(0.0, 4.0)], ink);
+ text_at.x += 18.0;
+ }
ui.painter().text(
- p + vec2(14.0, 14.0),
+ text_at,
Align2::LEFT_TOP,
if pages.len() == 1 { tl!("1 page").to_string() } else { crate::i18n::fmt(tl!("{n} pages"), &[("n", &pages.len().to_string())]) },
theme::medium(12.0),
@@ -5049,13 +5177,19 @@ fn organize_grid(view: &mut DocView, info: &DocInfo, editable: bool, dirty: bool
if drop {
view.org_drag = None;
if let Some(gap) = drop_gap(&cells, p) {
- // `to` counts positions without the moving pages.
- let to = gap - pages.iter().filter(|x| **x < gap).count();
- let first = pages[0];
- let contiguous = pages.windows(2).all(|w| w[1] == w[0] + 1);
- if !(contiguous && to == first) {
- view.pending_edit = Some(Edit::MovePages { pages: pages.clone(), to });
- view.selected = (to..to + pages.len()).collect();
+ if copying {
+ // The copies land at the gap as it is: nothing leaves its place.
+ view.pending_edit = Some(Edit::DuplicatePages { pages: pages.clone(), at: Some(gap) });
+ view.selected = (gap..gap + pages.len()).collect();
+ } else {
+ // `to` counts positions without the moving pages.
+ let to = gap - pages.iter().filter(|x| **x < gap).count();
+ let first = pages[0];
+ let contiguous = pages.windows(2).all(|w| w[1] == w[0] + 1);
+ if !(contiguous && to == first) {
+ view.pending_edit = Some(Edit::MovePages { pages: pages.clone(), to });
+ view.selected = (to..to + pages.len()).collect();
+ }
}
}
}
diff --git a/crates/ui-egui/src/commands.rs b/crates/ui-egui/src/commands.rs
index 832ce7cfbd..0d230982eb 100644
--- a/crates/ui-egui/src/commands.rs
+++ b/crates/ui-egui/src/commands.rs
@@ -627,7 +627,7 @@ impl PdfCraftApp {
"create.clipboard" => self.create_from_clipboard(),
"optimize.reduce" => self.reduce_file_size(),
"page.duplicate" => {
- self.apply_edit(Edit::DuplicatePages { pages: targets });
+ self.apply_edit(Edit::DuplicatePages { pages: targets, at: None });
}
// Two or more selected pages reverse among themselves; otherwise the whole document.
"page.reverse" => {
diff --git a/crates/ui-egui/src/control.rs b/crates/ui-egui/src/control.rs
index 630d66190c..a4ab35d969 100644
--- a/crates/ui-egui/src/control.rs
+++ b/crates/ui-egui/src/control.rs
@@ -410,12 +410,23 @@ impl Control {
let steps = p.get("steps").and_then(Value::as_u64).unwrap_or(8).clamp(1, 200) as usize;
let modifiers = modifiers(p.get("modifiers"))?;
let which = pointer_button(p, "ui.drag")?;
+ // The events carry the modifiers and so does the input state: gestures that read
+ // `ui.input(|i| i.modifiers)` (a Ctrl-drag copy, an Alt-drag column select) see them.
+ let held = |mut events: Vec| {
+ if modifiers != egui::Modifiers::NONE {
+ events.insert(0, egui::Event::ModifiersChanged(modifiers));
+ }
+ events
+ };
let button = |pos, pressed| egui::Event::PointerButton { pos, button: which, pressed, modifiers };
- let mut frames = vec![vec![egui::Event::PointerMoved(from)], vec![button(from, true)]];
+ let mut frames = vec![held(vec![egui::Event::PointerMoved(from)]), held(vec![button(from, true)])];
for k in 1..=steps {
- frames.push(vec![egui::Event::PointerMoved(from + (to - from) * (k as f32 / steps as f32))]);
+ frames.push(held(vec![egui::Event::PointerMoved(from + (to - from) * (k as f32 / steps as f32))]));
+ }
+ frames.push(held(vec![button(to, false)]));
+ if modifiers != egui::Modifiers::NONE {
+ frames.push(vec![egui::Event::ModifiersChanged(egui::Modifiers::NONE)]);
}
- frames.push(vec![button(to, false)]);
let n = self.inject(frames);
Ok(Handled::AfterFrames(n, json!({ "dragged": [[from.x, from.y], [to.x, to.y]] })))
}
diff --git a/crates/ui-egui/src/dialogs.rs b/crates/ui-egui/src/dialogs.rs
index 0de1c6a7ea..981db214b2 100644
--- a/crates/ui-egui/src/dialogs.rs
+++ b/crates/ui-egui/src/dialogs.rs
@@ -354,17 +354,20 @@ fn dialogs(app: &mut PdfCraftApp, ctx: &egui::Context) {
let ends = doc.revision_ends();
if ends.len() > 1 {
ui.label(egui::RichText::new(tl!("Revisions")).color(t.text_muted));
- ui.horizontal_wrapped(|ui| {
- ui.label(ends.len().to_string());
- for n in (1..ends.len()).rev().take(12) {
- if ui
- .small_button(crate::i18n::fmt(tl!("View revision {n}"), &[("n", &n.to_string())]))
- .on_hover_text(tl!("Open the file as it was saved then"))
- .clicked()
- {
- open_revision = Some(n);
+ // Reserve the full wrapped height before the grid starts the next row.
+ ui.vertical(|ui| {
+ ui.horizontal_wrapped(|ui| {
+ ui.label(ends.len().to_string());
+ for n in (1..ends.len()).rev().take(12) {
+ if ui
+ .small_button(crate::i18n::fmt(tl!("View revision {n}"), &[("n", &n.to_string())]))
+ .on_hover_text(tl!("Open the file as it was saved then"))
+ .clicked()
+ {
+ open_revision = Some(n);
+ }
}
- }
+ });
});
ui.end_row();
}
diff --git a/crates/ui-egui/src/editing.rs b/crates/ui-egui/src/editing.rs
index 0e5307f0e9..e554d7a1b0 100644
--- a/crates/ui-egui/src/editing.rs
+++ b/crates/ui-egui/src/editing.rs
@@ -53,6 +53,12 @@ impl PdfCraftApp {
let Some(doc) = self.session.get(id) else { return true };
let info = &doc.info;
let view = &mut self.views[i];
+ // A picture picked with the Select tool may be the one this edit removed or moved:
+ // drop its highlight now, not on the next frame (which only comes with the next input).
+ view.picked_image = None;
+ if let Some(ctx) = &self.ctx {
+ ctx.request_repaint();
+ }
if matches!(edit, Edit::AddAnnotation(_)) {
// A released drawing's raster can come now: it will have the annotation.
view.comments.drawings_committed();
@@ -127,6 +133,11 @@ impl PdfCraftApp {
let to = to.min(info.pages.len() - n);
view.select_pages(&(to..to + n).collect::>());
}
+ Edit::DuplicatePages { pages, at: Some(at) } => {
+ let n = pages.iter().collect::>().len();
+ let at = at.min(info.pages.len().saturating_sub(n));
+ view.select_pages(&(at..at + n).collect::>());
+ }
Edit::InsertBlankPage { at, .. } => view.select_pages(&[at.min(info.pages.len() - 1)]),
Edit::DeletePages { .. } => view.select_pages(&[]),
// A stroke drawn with the pen stays unselected, so the selection box and
@@ -990,13 +1001,15 @@ fn is_form_edit(edit: &Edit) -> bool {
/// classic Outlook (`Temporary Internet Files\Content.Outlook\` on older Windows), and the
/// `Microsoft.OutlookForWindows` package folder for the new Outlook. Saving there either fails,
/// because Outlook can still hold the file, or appears to work while the edits stay out of the
-/// email and are deleted with the cache. Matched on either separator and in any case, so it can
-/// be tested on every platform.
+/// email and are deleted with the cache. Apple Mail opens one from `Mail Downloads//` in
+/// its sandbox container, where other apps can't write, so saving there fails (#2751). Matched on
+/// either separator and in any case, so it can be tested on every platform.
pub(crate) fn is_mail_temp_copy(path: &str) -> bool {
let p = path.replace('\\', "/").to_ascii_lowercase();
p.contains("/inetcache/content.outlook/")
|| p.contains("/temporary internet files/content.outlook/")
|| p.contains("/microsoft.outlookforwindows")
+ || p.contains("/library/containers/com.apple.mail/data/library/mail downloads/")
}
/// Write via a temporary file in the same directory and rename over the target, so a crash or
@@ -1244,8 +1257,8 @@ mod tests {
let _ = std::fs::remove_dir_all(&dir);
}
- /// Windows refuses to replace a read-only file, so the rename fails: the staging file must not
- /// be left behind.
+ /// A read-only target is refused before anything is staged, so no staging file is left
+ /// behind.
#[cfg(windows)]
#[test]
fn a_failed_save_removes_the_staging_file() {
@@ -1256,7 +1269,7 @@ mod tests {
perms.set_readonly(true);
std::fs::set_permissions(&target, perms.clone()).unwrap();
let e = write_atomically_with(&target.to_string_lossy(), b"NEW", [7]).unwrap_err();
- assert_eq!(e.kind(), std::io::ErrorKind::PermissionDenied, "the rename failed: {e}");
+ assert_eq!(e.kind(), std::io::ErrorKind::PermissionDenied, "a read-only target is refused: {e}");
assert_eq!(read(&target), "OLD");
assert_eq!(listing(&dir), ["out.pdf"], "the staging file was removed");
#[allow(clippy::permissions_set_readonly_false)]
@@ -1314,6 +1327,16 @@ mod tests {
}
}
+ #[test]
+ fn apple_mail_attachment_copies_are_recognised() {
+ for p in [
+ "/Users/me/Library/Containers/com.apple.mail/Data/Library/Mail Downloads/6F1C2A9E-3B4D-4E5F-8A7B-9C0D1E2F3A4B/booking.pdf",
+ "/Users/me/library/containers/COM.APPLE.MAIL/data/library/mail downloads/6F1C2A9E/Booking (2).pdf",
+ ] {
+ assert!(is_mail_temp_copy(p), "{p}");
+ }
+ }
+
#[test]
fn ordinary_paths_are_not_mail_copies() {
for p in [
@@ -1321,6 +1344,8 @@ mod tests {
r"C:\Users\me\OneDrive - Contoso\Content.Outlook notes\Quote.pdf",
r"C:\Users\me\AppData\Local\Microsoft\Windows\INetCache\IE\Quote.pdf",
"/home/me/Documents/Quote.pdf",
+ "/Users/me/Downloads/Mail Downloads/booking.pdf",
+ "/Users/me/Library/Containers/com.apple.Notes/Data/Library/Mail Downloads/booking.pdf",
"Quote.pdf",
] {
assert!(!is_mail_temp_copy(p), "{p}");
diff --git a/crates/ui-egui/src/image_pick.rs b/crates/ui-egui/src/image_pick.rs
index a05aefe2d2..53e4bdf158 100644
--- a/crates/ui-egui/src/image_pick.rs
+++ b/crates/ui-egui/src/image_pick.rs
@@ -9,12 +9,14 @@ use crate::PdfCraftApp;
use crate::canvas::{DocView, PageXform};
use pdfcraft_render::DocInfo;
-/// The selected picture: page, index in `page_images` and its box in user space.
+/// The selected picture: page, index in `page_images` and its box in user space, and the
+/// document's edit generation when it was picked (any later edit may have removed or renumbered it).
#[derive(Clone, Copy, Debug, PartialEq)]
pub struct PickedImage {
pub page: usize,
pub index: usize,
pub rect: [f64; 4],
+ pub generation: u64,
}
const BLUE: Color32 = Color32::from_rgb(0x3A, 0x7B, 0xF0);
@@ -43,8 +45,18 @@ pub(crate) fn page_input(
let (vx, vy) = xf.screen_to_view(p);
let u = info_page.view_to_user(vx, vy);
let at = [f64::from(u[0]), f64::from(u[1])];
+ let generation = doc.edit_generation();
view.picked_image =
- doc.page_image_at(page, at).and_then(|index| doc.page_images(page).get(index).map(|i| PickedImage { page, index, rect: i.rect }));
+ doc.page_image_at(page, at).and_then(|index| doc.page_images(page).get(index).map(|i| PickedImage { page, index, rect: i.rect, generation }));
+}
+
+/// Forget the selected picture once the document has been edited since it was picked: Edit text &
+/// images (or Undo) may have deleted or moved it, and its highlight would stay behind on an empty
+/// spot of the page.
+pub(crate) fn drop_stale(doc: &pdfcraft_engine::Document, view: &mut DocView) {
+ if view.picked_image.is_some_and(|p| p.generation != doc.edit_generation()) {
+ view.picked_image = None;
+ }
}
/// The selected picture's highlight on page `page`.
diff --git a/crates/ui-egui/src/lib.rs b/crates/ui-egui/src/lib.rs
index 725bcad4b0..7b421263d3 100644
--- a/crates/ui-egui/src/lib.rs
+++ b/crates/ui-egui/src/lib.rs
@@ -139,9 +139,12 @@ mod tab_context;
pub mod theme;
pub mod ui_scale;
pub mod updates;
+mod wheel_boost;
mod wheel_pager;
mod widgets;
pub mod window_controls;
+#[cfg(not(target_arch = "wasm32"))]
+pub mod window_state;
#[cfg(all(test, not(target_arch = "wasm32")))]
mod open_worker_tests;
@@ -744,6 +747,9 @@ pub struct PdfCraftApp {
dialog_epoch: u64,
/// The egui context, for commands that change window or theme state.
ctx: Option,
+ /// The window's size and place kept across a minimized close (desktop).
+ #[cfg(not(target_arch = "wasm32"))]
+ pub window_state: window_state::WindowState,
styled: bool,
fonts_ready: bool,
/// The CJK fallback order the installed fonts follow (see `theme::installed_font_definitions`).
@@ -1033,6 +1039,8 @@ impl PdfCraftApp {
dialog_seen: None,
dialog_epoch: 0,
ctx: None,
+ #[cfg(not(target_arch = "wasm32"))]
+ window_state: Default::default(),
styled: false,
fonts_ready: false,
fonts_cjk: theme::CjkPreference::Default,
@@ -2162,6 +2170,11 @@ impl PdfCraftApp {
} else if matches!(self.quick_tool, QuickTool::EditText | QuickTool::AddText) {
self.put_down_edit_text();
}
+ // Read mode hides the quick toolbar, so no other tool may stay active there. Opening a
+ // document with Read as the default mode comes here without `switch_mode`.
+ if mode == Mode::Read {
+ self.quick_tool = QuickTool::Select;
+ }
}
/// Back to the Select tool from an editing tool. A paragraph still open in the editor is
@@ -2179,6 +2192,7 @@ impl PdfCraftApp {
/// Switches workspace because the user chose to (mode bar, Read mode command). Entering Read
/// mode closes the right panel, as Acrobat's Read mode hides its panes, and leaving it reopens
/// that panel. The rail stays in Read mode, and its buttons open panels there too (#932).
+ /// Read mode hides the quick toolbar, so entering it also puts the current tool down.
pub(crate) fn switch_mode(&mut self, mode: Mode, select_tools: bool) {
let was_read = self.mode == Mode::Read;
if select_tools {
@@ -2188,6 +2202,12 @@ impl PdfCraftApp {
}
if mode == Mode::Read && !was_read {
self.right_before_read = self.right.take();
+ // An editing tool is put down so its open paragraph is applied, as `select_mode` does.
+ if matches!(self.quick_tool, QuickTool::EditText | QuickTool::AddText) {
+ self.put_down_edit_text();
+ } else {
+ self.quick_tool = QuickTool::Select;
+ }
} else if mode != Mode::Read && was_read {
let before = self.right_before_read.take();
if self.right.is_none() {
@@ -2920,11 +2940,19 @@ impl eframe::App for PdfCraftApp {
}
fn save(&mut self, storage: &mut dyn eframe::Storage) {
+ // eframe has just stored the window as it is now: not while it is minimized.
+ #[cfg(not(target_arch = "wasm32"))]
+ {
+ let minimized = self.ctx.as_ref().is_some_and(|c| c.input(|i| i.viewport().minimized == Some(true)));
+ self.window_state.keep(storage, minimized);
+ }
storage.set_string("pdfcraft", self.persist());
}
fn logic(&mut self, ctx: &egui::Context, _frame: &mut eframe::Frame) {
self.ctx = Some(ctx.clone());
+ #[cfg(not(target_arch = "wasm32"))]
+ self.window_state.repair(ctx);
self.dialog_epoch();
// Notices raised outside `ui` (opened files, OS events, the control channel) translate too.
let lang = i18n::Lang::from_pref(&self.language);
diff --git a/crates/ui-egui/src/marks_ui.rs b/crates/ui-egui/src/marks_ui.rs
index 4a60c40a6e..d91a69f937 100644
--- a/crates/ui-egui/src/marks_ui.rs
+++ b/crates/ui-egui/src/marks_ui.rs
@@ -28,8 +28,10 @@ pub struct PageRange {
}
impl Default for PageRange {
+ /// From the first page to the last: `to` starts past any page count, and the range field
+ /// shows (and clamps it to) the document's last page.
fn default() -> Self {
- Self { all: true, from: 1, to: 1, subset: Subset::All }
+ Self { all: true, from: 1, to: usize::MAX, subset: Subset::All }
}
}
@@ -53,9 +55,6 @@ impl PageRange {
widgets::radio_value(ui, &mut self.all, false, tl!("Pages from"));
ui.add_enabled(!self.all, egui::DragValue::new(&mut self.from).range(1..=count.max(1)));
ui.label(tl!("to"));
- if self.to == 1 && self.from == 1 {
- self.to = count.max(1);
- }
ui.add_enabled(!self.all, egui::DragValue::new(&mut self.to).range(1..=count.max(1)));
ui.label(tl!("Subset"));
let label = match self.subset {
@@ -519,6 +518,11 @@ mod tests {
assert_eq!(r.pages(6), [1, 3, 5]);
let r = PageRange { all: true, subset: Subset::Odd, ..PageRange::default() };
assert_eq!(r.pages(5), [0, 2, 4]);
+ // Not all pages: by default up to the last one; a single page is a range too.
+ let r = PageRange { all: false, ..PageRange::default() };
+ assert_eq!(r.pages(4), [0, 1, 2, 3]);
+ let r = PageRange { all: false, from: 1, to: 1, subset: Subset::All };
+ assert_eq!(r.pages(38), [0]);
}
#[test]
diff --git a/crates/ui-egui/src/prepare.rs b/crates/ui-egui/src/prepare.rs
index 386c42e872..a46611274d 100644
--- a/crates/ui-egui/src/prepare.rs
+++ b/crates/ui-egui/src/prepare.rs
@@ -734,6 +734,7 @@ pub struct FieldDraft {
pub tab: FieldTab,
pub name: String,
pub tooltip: String,
+ pub caption: String,
pub read_only: bool,
pub required: bool,
pub multiline: bool,
@@ -805,6 +806,7 @@ impl FieldDraft {
tab: FieldTab::General,
name: f.name.rsplit('.').next().unwrap_or(&f.name).to_string(),
tooltip: f.tooltip.clone().unwrap_or_default(),
+ caption: f.widgets.get(widget).and_then(|w| w.caption.clone()).unwrap_or_default(),
read_only: f.read_only(),
required: f.has(pdfcraft_engine::field_flags::REQUIRED),
multiline: f.has(pdfcraft_engine::field_flags::MULTILINE),
@@ -848,7 +850,7 @@ impl FieldDraft {
fn tabs(&self) -> Vec<(FieldTab, &'static str)> {
let mut t = vec![(FieldTab::General, "General"), (FieldTab::Appearance, "Appearance"), (FieldTab::Position, "Position")];
- if !matches!(self.kind, FormFieldKind::PushButton | FormFieldKind::Signature) {
+ if self.kind != FormFieldKind::Signature {
t.push((FieldTab::Options, "Options"));
}
if matches!(self.kind, FormFieldKind::Text | FormFieldKind::Combo) {
@@ -865,6 +867,7 @@ impl FieldDraft {
let p = FieldProps {
name: (self.name.trim() != o.name).then(|| self.name.trim().to_string()),
tooltip: (self.tooltip != o.tooltip).then(|| self.tooltip.clone()),
+ caption: (self.caption != o.caption).then(|| (self.widget, self.caption.clone())),
read_only: ch(self.read_only, o.read_only),
required: ch(self.required, o.required),
multiline: ch(self.multiline, o.multiline),
@@ -1087,6 +1090,12 @@ pub(crate) fn body(ui: &mut egui::Ui, d: &mut FieldDraft, t: &crate::theme::Toke
FieldTab::Calculate => calculate_tab(ui, d, t),
FieldTab::Actions => actions_tab(ui, d, t),
FieldTab::Options => match d.kind {
+ FormFieldKind::PushButton => {
+ ui.horizontal(|ui| {
+ let l = ui.label(tl!("Label"));
+ ui.add(crate::widgets::line(&mut d.caption).desired_width(340.0)).labelled_by(l.id);
+ });
+ }
FormFieldKind::Text => {
use pdfcraft_engine::field_flags as ff;
egui::Grid::new("text-options").num_columns(2).spacing([12.0, 8.0]).show(ui, |ui| {
diff --git a/crates/ui-egui/src/recovery.rs b/crates/ui-egui/src/recovery.rs
index 4f1f1d8e8e..3d5f78ad1b 100644
--- a/crates/ui-egui/src/recovery.rs
+++ b/crates/ui-egui/src/recovery.rs
@@ -9,6 +9,9 @@
//! Encrypted documents are stored encrypted (the working file is), so recovery never writes
//! plaintext of a protected document to disk. The web build has no recovery store yet.
//!
+//! On Unix the folder is owner-only (`0o700`), so other accounts on the machine can't read the
+//! snapshots of documents that aren't encrypted.
+//!
//! A running instance holds an exclusively locked `instance--.lock` in the folder and
//! names it in the snapshots it writes, so a second instance can tell live autosaves from a
//! crash's leftovers and neither offers nor discards them (#2166). The OS releases the lock
@@ -75,12 +78,31 @@ impl RecoveryStore {
Self { dir: dir.into(), lock: Arc::default() }
}
+ /// Create the folder if it is missing. On Unix only the owner may open it: snapshots are
+ /// unsaved copies of open documents, and with the default mode other accounts on the
+ /// machine could read them. A folder an earlier version made is tightened too. A volume
+ /// that can't change modes (FAT, some network shares) works as before, with one warning in
+ /// the log.
+ fn create_dir(&self) -> std::io::Result<()> {
+ std::fs::create_dir_all(&self.dir)?;
+ #[cfg(unix)]
+ {
+ use std::os::unix::fs::PermissionsExt;
+ let open_to_others = std::fs::metadata(&self.dir).is_ok_and(|m| m.permissions().mode() & 0o077 != 0);
+ if open_to_others && let Err(e) = std::fs::set_permissions(&self.dir, std::fs::Permissions::from_mode(0o700)) {
+ static WARNED: std::sync::Once = std::sync::Once::new();
+ WARNED.call_once(|| log::warn!("recovery folder {} can't be made owner-only: {e}", self.dir.display()));
+ }
+ }
+ Ok(())
+ }
+
/// Mark the folder as in use by a running instance (see the module doc). Without a lock
/// (the folder can't be written, or the platform has no file locks) snapshots are written
/// and offered as before.
pub fn claim(&self) {
self.lock.get_or_init(|| {
- if let Err(e) = std::fs::create_dir_all(&self.dir) {
+ if let Err(e) = self.create_dir() {
log::warn!("recovery folder {} can't be made: {e}", self.dir.display());
return None;
}
@@ -166,7 +188,7 @@ impl RecoveryStore {
/// Write (or replace) an entry. The PDF is written before its metadata, both atomically, so
/// a listed entry always has complete bytes.
pub fn write(&self, meta: &RecoveryMeta, bytes: &[u8]) -> std::io::Result<()> {
- std::fs::create_dir_all(&self.dir)?;
+ self.create_dir()?;
crate::editing::write_atomically(&self.dir.join(format!("{}.pdf", meta.key)).to_string_lossy(), bytes)?;
let json = serde_json::to_vec_pretty(meta).map_err(std::io::Error::other)?;
crate::editing::write_atomically(&self.dir.join(format!("{}.json", meta.key)).to_string_lossy(), &json)
diff --git a/crates/ui-egui/src/scroll_speed.rs b/crates/ui-egui/src/scroll_speed.rs
index a2d857c741..cc3a7e86e5 100644
--- a/crates/ui-egui/src/scroll_speed.rs
+++ b/crates/ui-egui/src/scroll_speed.rs
@@ -18,7 +18,7 @@ pub enum ScrollSpeed {
}
/// egui's own `line_scroll_speed`, which its wheel zoom rate is tuned against.
-const EGUI_LINE_POINTS: f32 = if cfg!(target_arch = "wasm32") { 8.0 } else { 40.0 };
+pub(crate) const EGUI_LINE_POINTS: f32 = if cfg!(target_arch = "wasm32") { 8.0 } else { 40.0 };
/// Points one wheel line scrolls at Normal speed.
const NORMAL_LINE_POINTS: f32 = if cfg!(target_arch = "wasm32") {
diff --git a/crates/ui-egui/src/tab_context.rs b/crates/ui-egui/src/tab_context.rs
index aa16ef7105..7cc502722b 100644
--- a/crates/ui-egui/src/tab_context.rs
+++ b/crates/ui-egui/src/tab_context.rs
@@ -55,9 +55,8 @@ fn reveal(path: &std::path::Path) -> Result<(), String> {
return Err("The file no longer exists".into());
}
let platform = std::env::consts::OS;
- let run = |program: &str, args: &[std::ffi::OsString]| {
- std::process::Command::new(program).args(args).output().map(|o| o.status.success()).map_err(|e| e.to_string())
- };
+ let run =
+ |program: &str, args: &[std::ffi::OsString]| reveal_process(program, args).output().map(|o| o.status.success()).map_err(|e| e.to_string());
if matches!(platform, "linux" | "freebsd") {
// `dbus-send --print-reply` waits up to 2 s for the file manager (then xdg-open may run),
// so it runs off the UI thread; a failure there is only logged.
@@ -71,6 +70,24 @@ fn reveal(path: &std::path::Path) -> Result<(), String> {
reveal_via(platform, &path, run)
}
+/// The process that runs `program` with `args`. Windows: `reveal_command` already quotes what
+/// Explorer needs quoted, and Rust's own quoting of an argument with a space breaks `/select,`
+/// (#2579), so the arguments go on the command line as written.
+#[cfg(not(target_arch = "wasm32"))]
+fn reveal_process(program: &str, args: &[std::ffi::OsString]) -> std::process::Command {
+ let mut command = std::process::Command::new(program);
+ #[cfg(windows)]
+ {
+ use std::os::windows::process::CommandExt as _;
+ for arg in args {
+ command.raw_arg(arg);
+ }
+ }
+ #[cfg(not(windows))]
+ command.args(args);
+ command
+}
+
#[cfg(not(target_arch = "wasm32"))]
fn reveal_via(
platform: &str,
@@ -95,9 +112,16 @@ fn reveal_via(
fn reveal_command(platform: &str, path: &std::path::Path) -> Result<(&'static str, Vec), String> {
match platform {
"windows" => {
- // Explorer uses /select,, rather than a separate option value.
- let mut selection = std::ffi::OsString::from("/select,");
+ // Explorer reads `/select,""`: the path quoted after the comma (#2579). Quoting
+ // the whole argument instead ("/select,C:\a b.pdf", what an ordinary argument with a
+ // space becomes) opens Documents. A Windows path can't contain `"`, so the quotes
+ // can't be closed early; one that does is refused.
+ if path.as_os_str().as_encoded_bytes().contains(&b'"') {
+ return Err("The file name can't be shown in File Explorer".into());
+ }
+ let mut selection = std::ffi::OsString::from("/select,\"");
selection.push(path);
+ selection.push("\"");
Ok(("explorer.exe", vec![selection]))
}
"macos" => Ok(("open", vec!["-R".into(), path.as_os_str().into()])),
@@ -221,6 +245,19 @@ mod tests {
assert!(reveal_via("linux", path, |_, _| Err("command missing".into())).is_err());
}
+ /// #2579 (review of #2625): the `/select,"…"` argument reaches the process exactly as written.
+ /// `cmd /c echo` prints its command line back; with Rust's own quoting (`args`) it would come
+ /// back wrapped in quotes with the inner ones escaped, which Explorer reads as no selection.
+ #[cfg(windows)]
+ #[test]
+ fn the_reveal_argument_reaches_the_process_as_written() {
+ let (_, args) = reveal_command("windows", std::path::Path::new(r"C:\Reports\annual report.pdf")).unwrap();
+ let mut echo: Vec = ["/d", "/c", "echo"].map(std::ffi::OsString::from).to_vec();
+ echo.extend(args);
+ let out = reveal_process("cmd", &echo).output().unwrap();
+ assert_eq!(String::from_utf8_lossy(&out.stdout).trim_end(), r#"/select,"C:\Reports\annual report.pdf""#);
+ }
+
#[test]
fn reveal_arguments_preserve_spaces_and_unicode() {
let path = std::path::Path::new("/tmp/年度 report.pdf");
@@ -229,10 +266,13 @@ mod tests {
assert_eq!(args, [std::ffi::OsString::from("-R"), path.as_os_str().into()]);
let (exe, args) = reveal_command("windows", path).unwrap();
assert_eq!(exe, "explorer.exe");
- assert_eq!(args, [std::ffi::OsString::from("/select,/tmp/年度 report.pdf")]);
+ assert_eq!(args, [std::ffi::OsString::from("/select,\"/tmp/年度 report.pdf\"")]);
+ // #2579: the path is quoted after the comma, never the whole argument, which Explorer
+ // reads as no selection and answers by opening Documents.
let special = std::path::Path::new(r"C:\Reports\年度 report, #1 & (draft).pdf");
let (_, windows) = reveal_command("windows", special).unwrap();
- assert_eq!(windows, [std::ffi::OsString::from(r"/select,C:\Reports\年度 report, #1 & (draft).pdf")]);
+ assert_eq!(windows, [std::ffi::OsString::from(r#"/select,"C:\Reports\年度 report, #1 & (draft).pdf""#)]);
+ assert!(reveal_command("windows", std::path::Path::new(r#"C:\a" b.pdf"#)).is_err(), "a quote can't end the path early");
let (exe, args) = reveal_command("linux", path).unwrap();
assert_eq!(exe, "dbus-send");
assert_eq!(args[6], "array:string:file:///tmp/%E5%B9%B4%E5%BA%A6%20report.pdf");
diff --git a/crates/ui-egui/src/wheel_boost.rs b/crates/ui-egui/src/wheel_boost.rs
new file mode 100644
index 0000000000..b9d474565e
--- /dev/null
+++ b/crates/ui-egui/src/wheel_boost.rs
@@ -0,0 +1,395 @@
+//! Bounded acceleration for coarse wheel input in the document viewport.
+//!
+//! Line-unit events describe notches, not motion over time. A fixed distance per
+//! notch makes a fast spin feel slow, so this module derives a bounded multiplier
+//! from arrival time and direction. The state is constant-size and leaves isolated
+//! notches unchanged.
+
+use egui::Vec2;
+
+/// Idle time after which a burst is forgotten, in seconds.
+const RESET_GAP_SECS: f64 = 0.3;
+/// Notch spacing at or below which input counts as a continuous spin, in seconds.
+const FAST_GAP_SECS: f64 = 0.08;
+/// Multiplier gained per fast consecutive notch.
+const BOOST_PER_TICK: f32 = 0.4;
+/// Hard limit for accelerated motion.
+const MAX_BOOST: f32 = 6.0;
+/// Ticks that affect the output.
+const MAX_TICKS: u32 = 13;
+/// Extra ticks retained beyond the ceiling. They absorb a few slow events
+/// before the visible multiplier starts to fall.
+const HYSTERESIS_TICKS: u32 = 4;
+/// Maximum scroll applied from one frame of coarse input, in points per axis.
+/// Each event is clamped first; this caps the sum.
+pub(crate) const MAX_FRAME_POINTS: f32 = 8192.0;
+/// Per-event clamp, in native units (lines or pages). Keeps one hostile event
+/// bounded before points conversion.
+pub(crate) const MAX_EVENT_UNITS: f32 = 10.0;
+
+/// Constant-size arrival-rate state for line wheel events.
+#[derive(Debug, Default)]
+pub(crate) struct WheelBoost {
+ last: f64,
+ burst: u32,
+ dir: i8,
+}
+
+impl WheelBoost {
+ /// Returns the multiplier for one bucket of motion carrying `motion`.
+ ///
+ /// `motion` is signed in native units; only its sign matters. Zero means
+ /// no motion. Non-finite time returns 1.0 and leaves state unchanged;
+ /// backwards time resets the burst and returns 1.0.
+ pub(crate) fn multiplier(&mut self, now: f64, motion: f32) -> f32 {
+ if !now.is_finite() || !motion.is_finite() || motion == 0.0 {
+ return 1.0;
+ }
+ let dir = if motion > 0.0 { 1 } else { -1 };
+ let idle = now - self.last;
+ if !idle.is_finite() || idle < 0.0 {
+ self.last = now;
+ self.burst = 0;
+ self.dir = dir;
+ return 1.0;
+ }
+ if idle > RESET_GAP_SECS || dir != self.dir {
+ self.burst = 0;
+ } else if idle <= FAST_GAP_SECS {
+ self.burst = self.burst.saturating_add(1).min(MAX_TICKS + HYSTERESIS_TICKS);
+ } else {
+ // Slow input in the same direction decays rather than extending the burst.
+ self.burst = self.burst.saturating_sub(1);
+ }
+ self.last = now;
+ self.dir = dir;
+ let ticks = self.burst.min(MAX_TICKS);
+ // MAX_TICKS is well within f32's exact integer range.
+ (1.0 + ticks as f32 * BOOST_PER_TICK).min(MAX_BOOST)
+ }
+
+ /// Drops burst state. Call after a programmatic scroll jump.
+ pub(crate) fn reset(&mut self) {
+ *self = Self::default();
+ }
+}
+
+/// Caps one frame of accumulated coarse scroll, per axis. A non-finite axis
+/// maps to zero; the other axis is preserved.
+pub(crate) fn cap_frame(delta: Vec2) -> Vec2 {
+ Vec2::new(
+ if delta.x.is_finite() { delta.x.clamp(-MAX_FRAME_POINTS, MAX_FRAME_POINTS) } else { 0.0 },
+ if delta.y.is_finite() { delta.y.clamp(-MAX_FRAME_POINTS, MAX_FRAME_POINTS) } else { 0.0 },
+ )
+}
+
+/// One frame of wheel input, normalized in a single pass. Line motion is in
+/// points; page motion remains in native units until the viewport is known.
+#[derive(Clone, Copy, Debug, Default, PartialEq)]
+pub(crate) struct FrameInput {
+ pub precise: Vec2,
+ pub line: Vec2,
+ pub page: Vec2,
+}
+
+/// Folds raw wheel events into frame totals without consulting time or UI state.
+/// Invalid scales contribute no line motion.
+pub(crate) fn accumulate(events: &[egui::Event], line_points: f32, single_paging: bool) -> FrameInput {
+ let line_points = sanitize_scale(line_points);
+ let mut out = FrameInput::default();
+ for event in events {
+ let egui::Event::MouseWheel { unit, delta, modifiers, .. } = event else { continue };
+ if modifiers.command || modifiers.ctrl {
+ continue; // Zooming, not scrolling.
+ }
+ if !delta.is_finite() {
+ continue;
+ }
+ match unit {
+ egui::MouseWheelUnit::Point if modifiers.shift => {
+ // Shift scrolls sideways, matching egui's own wheel handling.
+ let sum = delta.x + delta.y;
+ if sum.is_finite() {
+ out.precise.x += sum;
+ }
+ }
+ egui::MouseWheelUnit::Point => out.precise += *delta,
+ egui::MouseWheelUnit::Line if modifiers.shift => {
+ let sum = delta.x + delta.y;
+ if sum.is_finite() {
+ out.line.x += sum.clamp(-MAX_EVENT_UNITS, MAX_EVENT_UNITS) * line_points;
+ }
+ }
+ egui::MouseWheelUnit::Line => {
+ out.line.x += delta.x.clamp(-MAX_EVENT_UNITS, MAX_EVENT_UNITS) * line_points;
+ if !single_paging {
+ out.line.y += delta.y.clamp(-MAX_EVENT_UNITS, MAX_EVENT_UNITS) * line_points;
+ }
+ }
+ egui::MouseWheelUnit::Page if modifiers.shift => {
+ let sum = delta.x + delta.y;
+ if sum.is_finite() {
+ out.page.x += sum.clamp(-MAX_EVENT_UNITS, MAX_EVENT_UNITS);
+ }
+ }
+ egui::MouseWheelUnit::Page => {
+ out.page.x += delta.x.clamp(-MAX_EVENT_UNITS, MAX_EVENT_UNITS);
+ out.page.y += delta.y.clamp(-MAX_EVENT_UNITS, MAX_EVENT_UNITS);
+ }
+ }
+ }
+ // Bound batched events before acceleration is applied.
+ out.precise = cap_frame(out.precise);
+ out.line = cap_frame(out.line);
+ out.page = cap_frame(out.page);
+ out
+}
+
+/// Converts the frame total to the extra scroll delta beyond egui's share.
+///
+/// egui applies the base distance through its own smoothing. This function
+/// returns only the accelerated surplus, so an isolated notch adds nothing.
+/// Each axis advances once per frame; events batched into one frame cannot
+/// accelerate one another. Output is frame-capped.
+pub(crate) fn boost_extra(input: FrameInput, boost_x: &mut WheelBoost, boost_y: &mut WheelBoost, now: f64) -> Vec2 {
+ let mut extra = Vec2::ZERO;
+ if input.line.x != 0.0 && input.line.x.is_finite() {
+ extra.x += input.line.x * (boost_x.multiplier(now, input.line.x) - 1.0);
+ }
+ if input.line.y != 0.0 && input.line.y.is_finite() {
+ extra.y += input.line.y * (boost_y.multiplier(now, input.line.y) - 1.0);
+ }
+ cap_frame(extra)
+}
+
+/// Maps a non-finite or non-positive points-per-unit scale to zero.
+fn sanitize_scale(points: f32) -> f32 {
+ if points.is_finite() && points > 0.0 { points } else { 0.0 }
+}
+
+#[cfg(test)]
+mod tests {
+ use super::*;
+
+ #[test]
+ fn single_and_slow_notches_stay_unboosted() {
+ let mut b = WheelBoost::default();
+ assert_eq!(b.multiplier(0.0, -1.0), 1.0);
+ assert_eq!(b.multiplier(0.5, -1.0), 1.0);
+ assert_eq!(b.multiplier(1.0, -1.0), 1.0);
+ }
+
+ #[test]
+ fn fast_spin_accelerates_then_caps() {
+ let mut b = WheelBoost::default();
+ assert_eq!(b.multiplier(1.0, -1.0), 1.0);
+ let mut last = 1.0;
+ for (i, t) in (1..=20).map(|i| 1.0 + f64::from(i) * 0.02).enumerate() {
+ last = b.multiplier(t, -1.0);
+ if i < 2 {
+ assert!(last >= 1.0, "early burst must not jump, got {last}");
+ }
+ }
+ assert!(last > 2.0, "a sustained fast spin must accelerate, got {last}");
+ assert!(last <= MAX_BOOST, "boost must be capped, got {last}");
+ assert_eq!(last, MAX_BOOST, "a long spin should reach the ceiling");
+ for i in 1..=20 {
+ let m = b.multiplier(1.5 + f64::from(i) * 0.02, -1.0);
+ assert!(m <= MAX_BOOST, "boost must stay capped, got {m}");
+ }
+ }
+
+ #[test]
+ fn reversal_and_idle_reset_the_burst() {
+ let mut b = WheelBoost::default();
+ b.multiplier(1.0, -1.0);
+ for i in 1..=6 {
+ b.multiplier(1.0 + f64::from(i) * 0.02, -1.0);
+ }
+ let boosted = b.multiplier(1.0 + 7.0 * 0.02, -1.0);
+ assert!(boosted > 1.0, "precondition: burst is boosted, got {boosted}");
+ assert_eq!(b.multiplier(1.0 + 8.0 * 0.02, 1.0), 1.0);
+ for i in 9..=14 {
+ b.multiplier(1.0 + f64::from(i) * 0.02, 1.0);
+ }
+ assert_eq!(b.multiplier(5.0, 1.0), 1.0);
+ }
+
+ #[test]
+ fn invalid_input_never_boosts_or_corrupts() {
+ let mut b = WheelBoost::default();
+ for bad_now in [f64::NAN, f64::INFINITY, f64::NEG_INFINITY] {
+ assert_eq!(b.multiplier(bad_now, -1.0), 1.0);
+ }
+ for bad_motion in [0.0, -0.0, f32::NAN, f32::INFINITY, f32::NEG_INFINITY] {
+ assert_eq!(b.multiplier(1.0, bad_motion), 1.0);
+ }
+ assert_eq!(b.multiplier(2.0, -1.0), 1.0);
+ b.reset();
+ assert_eq!(b.multiplier(3.0, -1.0), 1.0);
+ }
+
+ #[test]
+ fn same_timestamp_events_count_as_a_burst() {
+ let mut b = WheelBoost::default();
+ assert_eq!(b.multiplier(1.0, -1.0), 1.0);
+ for _ in 0..6 {
+ b.multiplier(1.0, -1.0);
+ }
+ assert!(b.multiplier(1.0, -1.0) > 1.0);
+ }
+
+ #[test]
+ fn fractional_notches_share_the_burst() {
+ let mut b = WheelBoost::default();
+ b.multiplier(1.0, -0.25);
+ for i in 1..=8 {
+ b.multiplier(1.0 + f64::from(i) * 0.02, -0.25);
+ }
+ assert!(b.multiplier(1.0 + 9.0 * 0.02, -0.25) > 1.0);
+ }
+
+ #[test]
+ fn sleep_wake_resets_the_burst() {
+ let mut b = WheelBoost::default();
+ b.multiplier(1.0, -1.0);
+ for i in 1..=6 {
+ b.multiplier(1.0 + f64::from(i) * 0.02, -1.0);
+ }
+ assert_eq!(b.multiplier(3600.0, -1.0), 1.0);
+ }
+
+ #[test]
+ fn backwards_time_resets_without_corrupting() {
+ let mut b = WheelBoost::default();
+ b.multiplier(1.0, -1.0);
+ b.multiplier(1.02, -1.0);
+ assert_eq!(b.multiplier(0.5, -1.0), 1.0);
+ assert_eq!(b.multiplier(0.52, -1.0), 1.4);
+ }
+
+ #[test]
+ fn unhurried_spin_decays() {
+ let mut b = WheelBoost::default();
+ b.multiplier(1.0, -1.0);
+ for i in 1..=8 {
+ b.multiplier(1.0 + f64::from(i) * 0.02, -1.0);
+ }
+ let peak = b.multiplier(1.0 + 9.0 * 0.02, -1.0);
+ assert!(peak > 1.0);
+ // Same direction but slower than the fast gap: each event sheds one tick.
+ let mut prev = peak;
+ for i in 10..=13 {
+ let m = b.multiplier(1.0 + f64::from(i) * 0.15, -1.0);
+ assert!(m <= prev, "decay must not grow: {m} > {prev}");
+ prev = m;
+ }
+ }
+
+ #[test]
+ fn hysteresis_holds_the_ceiling_briefly() {
+ let mut b = WheelBoost::default();
+ let mut t = 1.0;
+ b.multiplier(t, -1.0);
+ for _ in 0..20 {
+ t += 0.02;
+ b.multiplier(t, -1.0);
+ }
+ assert_eq!(b.multiplier(t + 0.02, -1.0), MAX_BOOST);
+ // A couple of unhurried events consume hysteresis before output falls.
+ t += 0.15;
+ assert_eq!(b.multiplier(t, -1.0), MAX_BOOST);
+ t += 0.15;
+ assert_eq!(b.multiplier(t, -1.0), MAX_BOOST);
+ }
+
+ #[test]
+ fn frame_cap_bounds_each_axis() {
+ use egui::vec2;
+ assert_eq!(cap_frame(vec2(40.0, -40.0)), vec2(40.0, -40.0));
+ assert_eq!(cap_frame(vec2(1e9, -1e9)), vec2(MAX_FRAME_POINTS, -MAX_FRAME_POINTS));
+ assert_eq!(cap_frame(vec2(f32::NAN, 40.0)), vec2(0.0, 40.0));
+ assert_eq!(cap_frame(vec2(40.0, f32::INFINITY)), vec2(40.0, 0.0));
+ }
+
+ fn wheel(unit: egui::MouseWheelUnit, dx: f32, dy: f32) -> egui::Event {
+ egui::Event::MouseWheel { unit, delta: egui::vec2(dx, dy), phase: egui::TouchPhase::Move, modifiers: egui::Modifiers::NONE }
+ }
+
+ fn shifted(unit: egui::MouseWheelUnit, dx: f32, dy: f32) -> egui::Event {
+ egui::Event::MouseWheel { unit, delta: egui::vec2(dx, dy), phase: egui::TouchPhase::Move, modifiers: egui::Modifiers::SHIFT }
+ }
+
+ #[test]
+ fn isolated_notch_adds_no_extra() {
+ use egui::MouseWheelUnit::Line;
+ for base in [50.0, 100.0, 200.0, 40.0, 8.0] {
+ let input = accumulate(&[wheel(Line, 0.0, -1.0)], base, false);
+ assert_eq!(input.line, egui::vec2(0.0, -base));
+ let mut bx = WheelBoost::default();
+ let mut by = WheelBoost::default();
+ assert_eq!(boost_extra(input, &mut bx, &mut by, 1.0), egui::Vec2::ZERO);
+ }
+ }
+
+ #[test]
+ fn fast_spin_adds_a_bounded_extra() {
+ use egui::MouseWheelUnit::Line;
+ let mut bx = WheelBoost::default();
+ let mut by = WheelBoost::default();
+ let mut peak = 0.0f32;
+ for i in 0..20 {
+ let input = accumulate(&[wheel(Line, 0.0, -1.0)], 100.0, false);
+ let extra = boost_extra(input, &mut bx, &mut by, 1.0 + f64::from(i) * 0.02);
+ assert!(extra.y <= 0.0, "extra follows the spin direction, got {extra:?}");
+ peak = peak.min(extra.y);
+ }
+ assert!(peak < 0.0, "a sustained spin must add an extra");
+ assert!(peak >= -MAX_FRAME_POINTS, "extra stays frame-capped, got {peak}");
+ }
+
+ #[test]
+ fn point_and_page_events_stay_out_of_the_line_boost() {
+ use egui::{MouseWheelUnit::Line, MouseWheelUnit::Page, MouseWheelUnit::Point};
+ let input = accumulate(&[wheel(Page, 0.0, -1.0), wheel(Point, 0.0, -30.0)], 100.0, false);
+ assert_eq!(input.precise, egui::vec2(0.0, -30.0));
+ assert_eq!(input.page, egui::vec2(0.0, -1.0));
+ assert_eq!(input.line, egui::Vec2::ZERO);
+ let input = accumulate(&[wheel(Line, 0.0, -1.0), wheel(Page, 0.0, -1.0)], 100.0, false);
+ assert_eq!(input.line, egui::vec2(0.0, -100.0));
+ assert_eq!(input.page, egui::vec2(0.0, -1.0));
+ let mut bx = WheelBoost::default();
+ let mut by = WheelBoost::default();
+ assert_eq!(boost_extra(input, &mut bx, &mut by, 1.0), egui::Vec2::ZERO);
+ }
+
+ #[test]
+ fn single_paging_drops_vertical_keeps_horizontal() {
+ use egui::MouseWheelUnit::Line;
+ let input = accumulate(&[wheel(Line, 5.0, -1.0)], 100.0, true);
+ assert_eq!(input.line, egui::vec2(500.0, 0.0));
+ }
+
+ #[test]
+ fn shift_merges_into_horizontal() {
+ use egui::MouseWheelUnit::{Line, Page};
+ let input = accumulate(&[shifted(Line, 1.0, -2.0)], 100.0, false);
+ assert_eq!(input.line, egui::vec2(-100.0, 0.0));
+ let input = accumulate(&[shifted(Page, 1.0, -2.0)], 100.0, false);
+ assert_eq!(input.page, egui::vec2(-1.0, 0.0));
+ }
+
+ #[test]
+ fn invalid_scales_and_events_contribute_nothing() {
+ use egui::MouseWheelUnit::Line;
+ for bad in [f32::NAN, 0.0, -10.0, f32::INFINITY] {
+ let input = accumulate(&[wheel(Line, 0.0, -1.0)], bad, false);
+ assert_eq!(input.line, egui::Vec2::ZERO, "line scale {bad}");
+ }
+ let zoom =
+ egui::Event::MouseWheel { unit: Line, delta: egui::vec2(0.0, -1.0), phase: egui::TouchPhase::Move, modifiers: egui::Modifiers::COMMAND };
+ let key = egui::Event::Key { key: egui::Key::Escape, physical_key: None, pressed: true, repeat: false, modifiers: egui::Modifiers::NONE };
+ let input = accumulate(&[zoom, key, wheel(Line, 0.0, f32::NAN)], 100.0, false);
+ assert_eq!(input, FrameInput::default());
+ }
+}
diff --git a/crates/ui-egui/src/window_state.rs b/crates/ui-egui/src/window_state.rs
new file mode 100644
index 0000000000..0475604a48
--- /dev/null
+++ b/crates/ui-egui/src/window_state.rs
@@ -0,0 +1,137 @@
+//! The window's saved size and place across starts. eframe stores them just before
+//! `App::save`, from the window as it is at that moment. A minimized window has neither: Windows
+//! reports it as 0 × 0 at (-32000, -32000), and the next start then opened at egui's 64-point
+//! floor in the screen's corner. So while the window is minimized the last settings it was shown
+//! with are put back, and a window that opens at that floor is given a usable size.
+
+/// eframe's storage key for the window settings.
+const WINDOW_KEY: &str = "window";
+/// The size egui opens a stored size below (its floor for "ridiculously small" windows), in points.
+const EGUI_FLOOR: f32 = 64.0;
+/// The size a window that opened at [`EGUI_FLOOR`] is given (apps/pdfcraft opens a first start at
+/// this size too), at most nine tenths of the monitor.
+const USABLE: egui::Vec2 = egui::vec2(1440.0, 920.0);
+
+#[derive(Debug, Default)]
+pub struct WindowState {
+ /// The settings eframe last stored for a shown (not minimized) window.
+ shown: Option,
+ /// The size at start has been checked ([`Self::repair`]).
+ checked: bool,
+}
+
+impl WindowState {
+ /// The settings the window started with: what to put back if it is minimized before eframe
+ /// stores a shown one.
+ pub fn started_with(storage: Option<&dyn eframe::Storage>) -> Self {
+ Self { shown: storage.and_then(|s| s.get_string(WINDOW_KEY)), checked: false }
+ }
+
+ /// After eframe has stored the window: remember it while the window is shown, and while it
+ /// is minimized put the last shown settings back (or none, so the next start opens at the
+ /// default size).
+ pub fn keep(&mut self, storage: &mut dyn eframe::Storage, minimized: bool) {
+ if !minimized {
+ self.shown = storage.get_string(WINDOW_KEY);
+ } else if let Some(shown) = &self.shown {
+ storage.set_string(WINDOW_KEY, shown.clone());
+ } else {
+ storage.remove_string(WINDOW_KEY);
+ }
+ }
+
+ /// Once the window's size is known: a shown window at egui's floor in both directions came
+ /// from settings stored while it was minimized (before this fix), so give it a usable size.
+ pub fn repair(&mut self, ctx: &egui::Context) {
+ if self.checked {
+ return;
+ }
+ let (rect, minimized, monitor) = ctx.input(|i| (i.viewport().inner_rect, i.viewport().minimized, i.viewport().monitor_size));
+ let Some(rect) = rect else { return };
+ if minimized == Some(true) {
+ return;
+ }
+ self.checked = true;
+ if rect.width() <= EGUI_FLOOR + 0.5 && rect.height() <= EGUI_FLOOR + 0.5 {
+ // The stored settings are the broken ones: never put them back.
+ self.shown = None;
+ let size = monitor.map_or(USABLE, |m| USABLE.min(m * 0.9));
+ ctx.send_viewport_cmd(egui::ViewportCommand::InnerSize(size));
+ }
+ }
+}
+
+#[cfg(test)]
+mod tests {
+ use super::*;
+ use eframe::Storage as _;
+
+ #[derive(Default)]
+ struct Store(std::collections::HashMap);
+
+ impl eframe::Storage for Store {
+ fn get_string(&self, key: &str) -> Option {
+ self.0.get(key).cloned()
+ }
+ fn set_string(&mut self, key: &str, value: String) {
+ self.0.insert(key.to_string(), value);
+ }
+ fn remove_string(&mut self, key: &str) {
+ self.0.remove(key);
+ }
+ fn flush(&mut self) {}
+ }
+
+ const SHOWN: &str = "(outer_position_pixels:Some((x:200.0,y:200.0)),inner_size_points:Some((x:1320.0,y:940.0)))";
+ const MINIMIZED: &str = "(outer_position_pixels:Some((x:-32000.0,y:-32000.0)),inner_size_points:Some((x:0.0,y:0.0)))";
+
+ /// Closing PdfCraft while it was minimized stored a 0 × 0 window off screen.
+ #[test]
+ fn a_minimized_window_keeps_the_last_shown_settings() {
+ let mut store = Store::default();
+ store.set_string(WINDOW_KEY, SHOWN.into());
+ let mut state = WindowState::started_with(Some(&store));
+ // eframe stores the minimized window, then the app saves.
+ store.set_string(WINDOW_KEY, MINIMIZED.into());
+ state.keep(&mut store, true);
+ assert_eq!(store.get_string(WINDOW_KEY).as_deref(), Some(SHOWN));
+ // Shown again and resized: that is what is kept from then on.
+ let moved = SHOWN.replace("200.0", "300.0");
+ store.set_string(WINDOW_KEY, moved.clone());
+ state.keep(&mut store, false);
+ store.set_string(WINDOW_KEY, MINIMIZED.into());
+ state.keep(&mut store, true);
+ assert_eq!(store.get_string(WINDOW_KEY), Some(moved));
+ }
+
+ #[test]
+ fn with_nothing_shown_yet_a_minimized_save_stores_no_window() {
+ let mut store = Store::default();
+ let mut state = WindowState::started_with(Some(&store));
+ store.set_string(WINDOW_KEY, MINIMIZED.into());
+ state.keep(&mut store, true);
+ assert_eq!(store.get_string(WINDOW_KEY), None, "the next start opens at the default size");
+ }
+
+ fn repaired(size: egui::Vec2, minimized: Option) -> Vec {
+ let ctx = egui::Context::default();
+ let mut input = egui::RawInput::default();
+ let info = input.viewports.entry(egui::ViewportId::ROOT).or_default();
+ info.inner_rect = Some(egui::Rect::from_min_size(egui::Pos2::ZERO, size));
+ info.minimized = minimized;
+ info.monitor_size = Some(egui::vec2(2674.0, 1504.0));
+ let mut state = WindowState { shown: Some(MINIMIZED.into()), checked: false };
+ let out = ctx.run_logic(&input, |ctx| state.repair(ctx));
+ let commands = out.viewport_commands.get(&egui::ViewportId::ROOT).cloned().unwrap_or_default();
+ // Repaired: the settings it started with are not put back by a minimized close.
+ assert_eq!(state.shown.is_none(), !commands.is_empty());
+ commands
+ }
+
+ #[test]
+ fn a_window_at_eguis_floor_gets_a_usable_size() {
+ assert_eq!(repaired(egui::vec2(64.0, 64.0), Some(false)), [egui::ViewportCommand::InnerSize(USABLE)]);
+ assert!(repaired(egui::vec2(820.0, 520.0), Some(false)).is_empty(), "a small window someone chose stays");
+ assert!(repaired(egui::vec2(64.0, 64.0), Some(true)).is_empty(), "not while minimized");
+ }
+}
diff --git a/crates/ui-egui/tests/comments.rs b/crates/ui-egui/tests/comments.rs
index 9a3716c3ee..3d53f1e3bb 100644
--- a/crates/ui-egui/tests/comments.rs
+++ b/crates/ui-egui/tests/comments.rs
@@ -366,6 +366,40 @@ fn selected_text_shows_actions_and_underlines_without_changing_tools() {
assert_eq!(h.state().quick_tool, QuickTool::Select);
}
+#[test]
+fn selected_text_copy_action_closes_the_bar_and_keeps_the_selection() {
+ let mut h = harness(|_| {});
+ h.state_mut().views[0].select_text(0, 4, 8);
+ h.run_steps(3);
+ h.get_by_label("Copy").click();
+ h.run_steps(3);
+ assert!(h.query_by_label("Add comment").is_none(), "the action bar closes after Copy");
+ assert!(h.state().views[0].selected_text().is_some_and(|t| !t.is_empty()), "the words stay selected");
+ // A new selection brings the bar back.
+ h.state_mut().views[0].select_text(0, 2, 6);
+ h.run_steps(3);
+ h.get_by_label("Add comment");
+}
+
+#[test]
+fn dragging_over_copied_text_again_brings_the_action_bar_back() {
+ let mut h = harness(|_| {});
+ let (a, b) = {
+ let v = &h.state().views[0];
+ (v.glyph_screen_pos(0, 4).expect("text layer"), v.glyph_screen_pos(0, 8).expect("glyph"))
+ };
+ drag(&mut h, a, b);
+ let copied = h.state().views[0].selected_text();
+ assert!(copied.is_some(), "the drag selects text");
+ h.get_by_label("Copy").click();
+ h.run_steps(3);
+ assert!(h.query_by_label("Add comment").is_none(), "the action bar closes after Copy");
+ // The same drag again selects the same words, which is a new selection: the bar is back.
+ drag(&mut h, a, b);
+ assert_eq!(h.state().views[0].selected_text(), copied, "the drag selects the same words");
+ assert!(h.query_by_label("Add comment").is_some(), "the action bar is back after a new drag");
+}
+
#[test]
fn selected_text_replace_action_opens_replacement_composer() {
let mut h = harness(|_| {});
diff --git a/crates/ui-egui/tests/control.rs b/crates/ui-egui/tests/control.rs
index 88ed1c0a90..e6af166d77 100644
--- a/crates/ui-egui/tests/control.rs
+++ b/crates/ui-egui/tests/control.rs
@@ -5,10 +5,23 @@ use std::sync::{Arc, Mutex};
use egui_kittest::Harness;
use egui_kittest::kittest::{NodeT, Queryable};
+use pdfcraft_render::{PageRenderer, RenderRequest, RequestKind};
use pdfcraft_ui_egui::PdfCraftApp;
use pdfcraft_ui_egui::control::{ControlClient, Reply};
use serde_json::{Value, json};
+/// The text each page shows, in order, as the app's session holds it.
+fn page_texts(app: &PdfCraftApp) -> Vec {
+ let doc = app.session.get(app.views[0].id).unwrap();
+ let mut r = PageRenderer::new(doc.bytes.clone(), Default::default());
+ (0..r.page_count())
+ .map(|p| {
+ let out = r.render(RenderRequest { page: p, kind: RequestKind::Text, scale: 1.0, ..Default::default() });
+ out.text.map(|t| t.plain_text().trim().to_string()).unwrap_or_default()
+ })
+ .collect()
+}
+
fn fixture(n: usize) -> Vec {
let mut objs: Vec = vec!["<< /Type /Catalog /Pages 2 0 R >>".into()];
let kids: Vec = (0..n).map(|i| format!("{} 0 R", 4 + 2 * i)).collect();
@@ -1034,6 +1047,45 @@ fn state_reports_the_selected_pages() {
assert_eq!(ok(&mut h, &c, "ui.state", json!({}))["active"]["selected_pages"], json!([]));
}
+#[test]
+fn ctrl_dragging_a_thumbnail_through_the_control_channel_copies_the_page() {
+ let (mut h, c) = harness_pages(3);
+ ok(&mut h, &c, "ui.set", json!({ "key": "tool", "value": "organize" }));
+ h.run_steps(3);
+ // The grid cell of a page: the node labelled exactly "Page N" (the checkbox says "Select page N").
+ let cell = |h: &mut Harness<'static, PdfCraftApp>, c: &ControlClient, page: usize| -> [f64; 4] {
+ let nodes = ok(h, c, "ui.inspect", json!({ "query": &format!("page {page}") }));
+ let want = format!("Page {page}");
+ let r = nodes["widgets"]
+ .as_array()
+ .unwrap()
+ .iter()
+ .find(|w| w["label"].as_str() == Some(want.as_str()))
+ .and_then(|w| w["rect"].as_array().cloned())
+ .expect("the page's grid cell");
+ let f = |i: usize| r[i].as_f64().unwrap();
+ [f(0), f(1), f(2), f(3)]
+ };
+ let (a, b) = (cell(&mut h, &c, 1), cell(&mut h, &c, 3));
+ // The gesture reads the OS the context reports: Option on macOS, Ctrl elsewhere.
+ let copy = if h.ctx.os() == egui::os::OperatingSystem::Mac { "alt" } else { "ctrl" };
+ ok(
+ &mut h,
+ &c,
+ "ui.drag",
+ json!({
+ "from": [(a[0] + a[2]) / 2.0, (a[1] + a[3]) / 2.0],
+ "to": [b[2] - 4.0, (b[1] + b[3]) / 2.0],
+ "modifiers": [copy],
+ }),
+ );
+ h.run_steps(3);
+ let st = ok(&mut h, &c, "ui.state", json!({}));
+ assert_eq!(st["documents"][0]["pages"], 4, "the drag copied the page instead of moving it: {st}");
+ assert_eq!(st["active"]["selected_pages"], json!([4]), "{st}");
+ assert_eq!(page_texts(h.state()), ["Page 1", "Page 2", "Page 3", "Page 1"], "the copy landed at the drop gap, after page 3");
+}
+
#[test]
fn organize_overview_is_reachable_through_the_control_channel() {
let (mut h, c) = harness();
diff --git a/crates/ui-egui/tests/editing.rs b/crates/ui-egui/tests/editing.rs
index 0c7f7196c8..2ef52a4403 100644
--- a/crates/ui-egui/tests/editing.rs
+++ b/crates/ui-egui/tests/editing.rs
@@ -2390,6 +2390,81 @@ fn dragging_thumbnails_reorders_pages() {
assert_eq!(sel, vec![2], "the moved page stays selected");
}
+/// A page drag with the copy modifier held — Ctrl (Option on macOS) — instead of none.
+/// Modifiers are input state: `ModifiersChanged` once, as the column-select drag does.
+fn copy_drag(h: &mut Harness<'static, PdfCraftApp>, from: egui::Pos2, to: egui::Pos2) {
+ page_drag(h, from, to, copy_modifiers(h));
+}
+
+/// The copy modifier of the OS the app reports: Option on macOS, Ctrl elsewhere. That is the OS
+/// the gesture itself reads, so a Mac browser — a wasm build, where `cfg!(target_os = "macos")` is
+/// false — is covered by asking the context rather than the target.
+fn copy_modifiers(h: &Harness<'static, PdfCraftApp>) -> Modifiers {
+ if h.ctx.os() == egui::os::OperatingSystem::Mac { Modifiers::ALT } else { Modifiers::CTRL }
+}
+
+/// Drag the page under `from` to `to` with `modifiers` held for the whole gesture.
+fn page_drag(h: &mut Harness<'static, PdfCraftApp>, from: egui::Pos2, to: egui::Pos2, modifiers: Modifiers) {
+ h.event(egui::Event::ModifiersChanged(modifiers));
+ h.run_steps(1);
+ h.hover_at(from);
+ h.run_steps(1);
+ h.event(egui::Event::PointerButton { pos: from, button: egui::PointerButton::Primary, pressed: true, modifiers });
+ h.run_steps(1);
+ for k in 1..=5 {
+ h.hover_at(from + (to - from) * (k as f32 / 5.0));
+ h.run_steps(1);
+ }
+ h.event(egui::Event::PointerButton { pos: to, button: egui::PointerButton::Primary, pressed: false, modifiers });
+ h.run_steps(4);
+ h.event(egui::Event::ModifiersChanged(Modifiers::NONE));
+ h.run_steps(1);
+}
+
+#[test]
+fn dragging_with_the_copy_modifier_copies_the_pages_into_the_gap() {
+ let mut h = organize(3);
+ let before = page_texts(h.state());
+ let grab = |h: &Harness<'static, PdfCraftApp>, label: &str| h.get_by_label(label).rect();
+ let (from, to) = (grab(&h, "Page 1").center(), grab(&h, "Page 3").right_center() - egui::vec2(10.0, 0.0));
+ copy_drag(&mut h, from, to);
+ // The original stays first; a copy of page 1 lands after page 3 (#2997).
+ let after = page_texts(h.state());
+ assert_eq!(after, vec![before[0].clone(), before[1].clone(), before[2].clone(), before[0].clone()], "{after:?}");
+ assert_eq!(h.state().session.get(h.state().views[0].id).unwrap().can_undo(), Some("Duplicate page"));
+ let sel: Vec = h.state().views[0].selected.iter().copied().collect();
+ assert_eq!(sel, vec![3], "the copy is selected");
+}
+
+/// macOS is not only `cfg!(target_os = "macos")`: the wasm build runs in a Mac browser, where the
+/// cfg is false but the context says Mac. The gesture must follow the context, or Option-drag
+/// there moves the pages instead of copying them (#3092 review).
+#[test]
+fn option_copies_and_ctrl_moves_when_the_context_reports_macos() {
+ let grab = |h: &Harness<'static, PdfCraftApp>, label: &str| h.get_by_label(label).rect();
+ let points = |h: &Harness<'static, PdfCraftApp>| (grab(h, "Page 1").center(), grab(h, "Page 3").right_center() - egui::vec2(10.0, 0.0));
+
+ // Ctrl is not the copy modifier on macOS, so that drag moves the page.
+ let mut h = organize(3);
+ h.ctx.set_os(egui::os::OperatingSystem::Mac);
+ let before = page_texts(h.state());
+ let (from, to) = points(&h);
+ page_drag(&mut h, from, to, Modifiers::CTRL);
+ assert_eq!(page_texts(h.state()), vec![before[1].clone(), before[2].clone(), before[0].clone()], "Ctrl moves on macOS");
+ assert_eq!(h.state().session.get(h.state().views[0].id).unwrap().can_undo(), Some("Move page"));
+
+ // Option is, so the same drag drops a copy at the gap.
+ let mut h = organize(3);
+ h.ctx.set_os(egui::os::OperatingSystem::Mac);
+ let before = page_texts(h.state());
+ let (from, to) = points(&h);
+ page_drag(&mut h, from, to, Modifiers::ALT);
+ assert_eq!(page_texts(h.state()), vec![before[0].clone(), before[1].clone(), before[2].clone(), before[0].clone()], "Option copies on macOS");
+ assert_eq!(h.state().session.get(h.state().views[0].id).unwrap().can_undo(), Some("Duplicate page"));
+ let sel: Vec = h.state().views[0].selected.iter().copied().collect();
+ assert_eq!(sel, vec![3], "the copy is selected");
+}
+
#[test]
fn copying_cutting_and_pasting_pages() {
let mut h = organize(4);
@@ -4777,3 +4852,32 @@ fn leaving_organize_goes_back_to_the_workspace_panel() {
h.run_steps(3);
assert_eq!(h.state().left, LeftPanel::AllTools);
}
+
+/// The Read mode command (not the mode bar) applies a paragraph still open in the editor, as
+/// putting the Edit text tool down does; left open, it would stay drawn in Read mode with no tool.
+#[test]
+fn read_mode_shortcut_applies_the_open_paragraph() {
+ use pdfcraft_ui_egui::{Mode, QuickTool};
+ let mut h = harness(1, |_| {});
+ assert!(h.state_mut().execute("edit.edit_text"));
+ h.run_steps(2);
+ // The line "Page 1" sits at (20, 150) on a 200 × 300 page, 24 pt.
+ let r = h.state().views[0].page_screen_rect(0).expect("on screen");
+ let at = egui::pos2(r.left() + 40.0 / 200.0 * r.width(), r.top() + (300.0 - 158.0) / 300.0 * r.height());
+ h.hover_at(at);
+ h.run_steps(1);
+ h.drag_at(at);
+ h.run_steps(1);
+ h.drop_at(at);
+ h.run_steps(3);
+ assert!(h.state().views[0].line_editor.is_some(), "the line opens for editing");
+ h.state_mut().views[0].line_editor.as_mut().unwrap().text = "Chapter One".into();
+ h.run_steps(1);
+ h.state_mut().run_command("view.read_mode");
+ h.run_steps(3);
+ let s = h.state();
+ assert_eq!(s.mode, Mode::Read);
+ assert_eq!(s.quick_tool, QuickTool::Select);
+ assert!(s.views[0].line_editor.is_none(), "no editor stays open in Read mode");
+ assert_eq!(s.session.get(s.views[0].id).unwrap().text_lines(0)[0].text, "Chapter One", "the edit is applied");
+}
diff --git a/crates/ui-egui/tests/prepare.rs b/crates/ui-egui/tests/prepare.rs
index eb25e42a03..d63df5232c 100644
--- a/crates/ui-egui/tests/prepare.rs
+++ b/crates/ui-egui/tests/prepare.rs
@@ -250,6 +250,59 @@ fn options_tab_sets_the_check_box_style() {
assert_eq!(s.session.get(s.views[0].id).unwrap().field_check_style("agree"), Some(CheckStyle::Cross));
}
+#[test]
+fn button_label_is_editable_without_renaming_the_field() {
+ use pdfcraft_engine::{Edit, NewField};
+ let mut h = harness();
+ h.state_mut().apply_edit(Edit::AddField {
+ page: 0,
+ rect: [40.0, 300.0, 190.0, 330.0],
+ kind: NewField::Button { caption: String::new() },
+ name: Some("clear_button".into()),
+ });
+ h.state_mut().open_field_props("clear_button", 0);
+ h.run_steps(2);
+ h.get_by_label("Options").click();
+ h.run_steps(2);
+ h.get_by_role_and_label(egui::accesskit::Role::TextInput, "Label").click();
+ h.run_steps(1);
+ h.get_by_role_and_label(egui::accesskit::Role::TextInput, "Label").type_text("Clear Form");
+ h.run_steps(2);
+ h.get_by_label("OK").click();
+ h.run_steps(3);
+
+ let app = h.state();
+ let id = app.views[0].id;
+ let field = app.session.get(id).unwrap().form.iter().find(|f| f.name == "clear_button").unwrap();
+ let bytes = app.session.save_bytes(id).unwrap();
+ let doc = pdfcraft_cos::Document::open(bytes).unwrap();
+ let widget = doc.get(field.widgets[0].obj);
+ let mk = doc.resolve(widget.as_dict().unwrap().get(b"MK").unwrap());
+ assert_eq!(doc.resolve(mk.as_dict().unwrap().get(b"CA").unwrap()).as_string().unwrap().to_text(), "Clear Form");
+ assert_eq!(app.session.get(id).unwrap().can_undo(), Some("Change field properties"));
+
+ // Read the existing label back, and leave both Cancel and an unchanged OK harmless.
+ let before = app.session.save_bytes(id).unwrap();
+ h.state_mut().open_field_props("clear_button", 0);
+ h.run_steps(2);
+ h.get_by_label("Options").click();
+ h.run_steps(2);
+ assert_eq!(h.state().field_props.as_ref().unwrap().caption, "Clear Form");
+ h.get_by_role_and_label(egui::accesskit::Role::TextInput, "Label").click();
+ h.run_steps(1);
+ h.key_press_modifiers(egui::Modifiers::COMMAND, egui::Key::A);
+ h.get_by_role_and_label(egui::accesskit::Role::TextInput, "Label").type_text("Discard this");
+ h.run_steps(2);
+ h.get_by_label("Cancel").click();
+ h.run_steps(2);
+ assert_eq!(h.state().session.save_bytes(id).unwrap(), before);
+ h.state_mut().open_field_props("clear_button", 0);
+ h.run_steps(2);
+ h.get_by_label("OK").click();
+ h.run_steps(2);
+ assert_eq!(h.state().session.save_bytes(id).unwrap(), before);
+}
+
#[test]
fn dropdown_alignment_is_editable_and_preserved() {
choice_alignment(pdfcraft_engine::NewField::Combo { options: vec!["One".into(), "Two".into()], editable: false }, "Center", 1);
diff --git a/crates/ui-egui/tests/recovery.rs b/crates/ui-egui/tests/recovery.rs
index 807261c8c6..a259ca3ad9 100644
--- a/crates/ui-egui/tests/recovery.rs
+++ b/crates/ui-egui/tests/recovery.rs
@@ -411,3 +411,23 @@ fn a_running_instances_autosaves_are_not_offered_to_a_second_instance() {
let fourth = harness(instance());
assert_eq!(fourth.state().recoverable.len(), 1, "a name outside the folder is not looked at, and the snapshot is offered");
}
+
+/// Snapshots are unsaved copies of open documents, so on Unix only the owner may open the
+/// folder: a new one is made that way, and one an earlier version left with the default mode is
+/// tightened when an instance starts.
+#[cfg(unix)]
+#[test]
+fn the_recovery_folder_is_owner_only() {
+ use std::os::unix::fs::PermissionsExt;
+ let mode = |s: &RecoveryStore| std::fs::metadata(s.dir()).unwrap().permissions().mode() & 0o777;
+ let fresh = store("owner-only");
+ crashed_session(&fresh, fixture(2), None, None);
+ assert_eq!(files_in(&fresh), 2);
+ assert_eq!(mode(&fresh), 0o700);
+
+ let earlier = store("owner-only-earlier");
+ std::fs::create_dir_all(earlier.dir()).unwrap();
+ std::fs::set_permissions(earlier.dir(), std::fs::Permissions::from_mode(0o755)).unwrap();
+ earlier.claim();
+ assert_eq!(mode(&earlier), 0o700);
+}
diff --git a/crates/ui-egui/tests/ui.rs b/crates/ui-egui/tests/ui.rs
index 0082e62fd8..b86f7699f0 100644
--- a/crates/ui-egui/tests/ui.rs
+++ b/crates/ui-egui/tests/ui.rs
@@ -246,6 +246,39 @@ fn read_mode_closes_the_panes_and_leaving_it_brings_them_back() {
assert_eq!(h.state().right, Some(RightPanel::Pages));
}
+/// Read mode hides the quick toolbar too, and puts down the tool that was in use (#2646).
+#[test]
+fn read_mode_hides_the_quick_toolbar() {
+ use pdfcraft_ui_egui::{Mode, QuickTool};
+ let mut h = harness(|app| app.open_bytes("fixture.pdf", None, FIXTURE.to_vec()).expect("fixture opens"));
+ h.get_by_label("Hand (H)").click();
+ h.run_steps(2);
+ assert_eq!(h.state().quick_tool, QuickTool::Hand);
+ h.get_by_label("Read").click();
+ h.run_steps(3);
+ assert!(h.query_by_label("Select (V)").is_none(), "no quick toolbar in Read mode");
+ assert_eq!(h.state().quick_tool, QuickTool::Select);
+ h.state_mut().run_command("view.read_mode");
+ h.run_steps(3);
+ assert_eq!(h.state().mode, Mode::AllTools);
+ h.get_by_label("Select (V)");
+}
+
+/// Opening a document with Read as the default mode puts down the tool in use, as entering Read
+/// mode from the mode bar does: the hidden toolbar would leave no way to see or change it.
+#[test]
+fn opening_into_read_mode_puts_the_tool_down() {
+ use pdfcraft_ui_egui::{Mode, QuickTool};
+ let mut h = harness(|app| app.open_bytes("fixture.pdf", None, FIXTURE.to_vec()).expect("fixture opens"));
+ assert_eq!(h.state().mode, Mode::AllTools);
+ h.state_mut().set_option("quick", "hand").unwrap();
+ h.state_mut().set_option("default-mode", "read").unwrap();
+ h.state_mut().open_bytes("again.pdf", None, FIXTURE.to_vec()).unwrap();
+ h.run_steps(3);
+ assert_eq!(h.state().mode, Mode::Read, "the new document opens in the default mode");
+ assert_eq!(h.state().quick_tool, QuickTool::Select, "no tool stays active behind the hidden toolbar");
+}
+
/// A pane closed in Read mode stays closed on the way out: leaving Read mode doesn't bring back the
/// pane saved on the way in, and closing Comments keeps its close preference.
#[test]
diff --git a/crates/ui-egui/tests/ui_group/marks.rs b/crates/ui-egui/tests/ui_group/marks.rs
index 0a0e45b152..1f1515ad60 100644
--- a/crates/ui-egui/tests/ui_group/marks.rs
+++ b/crates/ui-egui/tests/ui_group/marks.rs
@@ -96,3 +96,37 @@ fn backgrounds_and_watermarks_from_a_file() {
let s = h.state();
assert_eq!(s.session.get(s.views[0].id).unwrap().can_undo(), Some("Add background"));
}
+
+/// Pages from 1 to 1 stays one page: the dialog used to turn "to" back into the last page on
+/// every frame while both were 1, so only page 1 could not be chosen alone. The field starts on
+/// the last page.
+#[test]
+fn page_range_from_one_to_one_selects_only_the_first_page() {
+ let mut h = harness();
+ h.state_mut().execute("edit.background");
+ h.run_steps(2);
+ assert_eq!(h.state().marks_draft.range.pages(3), [0, 1, 2]);
+ h.get_by_label("Pages from").click();
+ h.run_steps(2);
+ let range = h.state().marks_draft.range;
+ assert_eq!((range.from, range.to), (1, 3), "from the first page to the last");
+ h.state_mut().marks_draft.range.to = 1;
+ h.run_steps(3);
+ let range = h.state().marks_draft.range;
+ assert_eq!((range.from, range.to), (1, 1), "the range stays as typed");
+ h.get_by_label("Applies to 1 page");
+ h.get_by_label("OK").click();
+ h.run_steps(3);
+ let s = h.state();
+ let doc = s.session.get(s.views[0].id).unwrap();
+ assert_eq!(doc.marks(), [MarkKind::Background]);
+ // Rendered: page 1 has the (default cream) background, page 2 is still blank.
+ let mut r = pdfcraft_render::PageRenderer::new(doc.bytes.clone(), pdfcraft_render::RenderConfig::default());
+ let corner = |r: &mut pdfcraft_render::PageRenderer, page: usize| {
+ let p = r.render(pdfcraft_render::RenderRequest { page, scale: 1.0, ..Default::default() });
+ assert!(p.error.is_none(), "{:?}", p.error);
+ p.rgba[..4].to_vec()
+ };
+ assert_eq!(corner(&mut r, 0), vec![255, 255, 217, 255], "page 1 has the background");
+ assert_eq!(corner(&mut r, 1), vec![255, 255, 255, 255], "page 2 does not");
+}
diff --git a/crates/ui-egui/tests/view.rs b/crates/ui-egui/tests/view.rs
index c30f081e4a..9b6d57bf67 100644
--- a/crates/ui-egui/tests/view.rs
+++ b/crates/ui-egui/tests/view.rs
@@ -185,6 +185,53 @@ fn single_page_layout_shows_one_page_at_a_time() {
assert!(rect(&h, 0).is_none() && rect(&h, 1).is_some(), "only page 2");
}
+#[test]
+fn single_page_view_prefetches_the_next_page() {
+ // Only the shown page is drawn, but the next one renders ahead of the turn: otherwise
+ // turning forward shows blank paper until it arrives.
+ let mut h = harness(&[("layout", "single"), ("zoom", "50")]);
+ let deadline = std::time::Instant::now() + std::time::Duration::from_secs(10);
+ while h.state().render_pending() {
+ assert!(std::time::Instant::now() < deadline, "renders settle");
+ h.run_steps(2);
+ std::thread::sleep(std::time::Duration::from_millis(5));
+ }
+ assert!(rect(&h, 1).is_none(), "page 2 is not shown");
+ assert!(h.state().views[0].raster_memory().page_count >= 2, "page 2 is rendered ahead of the turn");
+}
+
+#[test]
+fn single_page_wheel_turn_draws_the_new_page_at_once() {
+ // The frame that turns the page draws the new one. Its jump used to be measured from the
+ // old page's place, which scrolled both out of view: a blank frame on every turn forward.
+ let mut h = harness_stepping(1.0 / 60.0, &[("layout", "single")]);
+ h.key_press_modifiers(Modifiers::COMMAND, Key::Num0); // fit page
+ h.run_steps(4);
+ let vp = h.state().views[0].viewport_rect();
+ h.hover_at(vp.center());
+ h.run_steps(2);
+ h.event(wheel(MouseWheelUnit::Line, -1.0, TouchPhase::Move));
+ h.step();
+ assert_eq!(h.state().views[0].current, 1, "one notch turns one page");
+ let r = rect(&h, 1).expect("page 2 drawn in the frame that turns to it");
+ assert!(r.min.y >= vp.min.y && r.max.y <= vp.max.y, "page 2 fits on screen: {r:?} in {vp:?}");
+}
+
+#[test]
+fn single_page_jump_draws_a_fitting_page_where_it_stays() {
+ // A page that fits has nothing to scroll, but the jump to it asked for the gap above it:
+ // drawn that much too high for a frame, it then hopped into place.
+ let mut h = harness_stepping(1.0 / 60.0, &[("layout", "single")]);
+ h.key_press_modifiers(Modifiers::COMMAND, Key::Num0); // fit page
+ h.run_steps(4);
+ h.state_mut().set_option("page", "2").unwrap();
+ h.step();
+ let first = rect(&h, 1).expect("page 2 drawn in the frame after the jump");
+ h.run_steps(4);
+ let settled = rect(&h, 1).expect("page 2");
+ assert!((first.min.y - settled.min.y).abs() < 0.5, "page 2 drawn where it settles: {first:?} then {settled:?}");
+}
+
#[test]
fn single_page_wheel_pans_instead_when_zoomed_in() {
// The zoomed-in guard: with room to pan, the wheel pans and must not turn pages.
@@ -345,10 +392,78 @@ fn precise_touchpad_wheel_scrolls_one_to_one() {
assert!(drift.abs() < 2.0, "no easing tail after the swipe: drifted {drift}");
}
+#[test]
+fn rapid_wheel_notches_accelerate_and_follow_touchpad_input() {
+ use pdfcraft_ui_egui::scroll_speed::ScrollSpeed;
+
+ let mut h = harness_stepping(1.0 / 60.0, &[("layout", "continuous"), ("zoom", "300")]);
+ h.state_mut().set_option("animation_effects", "off").unwrap();
+ let at = h.state().views[0].viewport_rect().center();
+ h.hover_at(at);
+ h.run_steps(2);
+
+ h.event(wheel(MouseWheelUnit::Line, -1.0, TouchPhase::Move));
+ let single = scroll_over(&mut h, 2);
+ let base = ScrollSpeed::default().line_points();
+ assert!((single - base).abs() < 1.0, "one notch keeps the configured base: moved {single}, base {base}");
+
+ // Let the isolated notch expire, then spin five notches one frame apart. The burst must move
+ // farther than five base steps without adding work on idle frames.
+ h.run_steps(30);
+ let mut burst = 0.0;
+ for _ in 0..5 {
+ h.event(wheel(MouseWheelUnit::Line, -1.0, TouchPhase::Move));
+ burst += scroll_over(&mut h, 2);
+ }
+ assert!(burst > 7.0 * base, "a rapid five-notch spin accelerates: moved {burst}, base {base}");
+
+ // A line notch inside the precise path's suppression window keeps its base share (#1072):
+ // suppressing egui's touchpad tail must not swallow mixed-device input.
+ h.run_steps(30);
+ h.event(wheel(MouseWheelUnit::Point, -20.0, TouchPhase::Move));
+ scroll_over(&mut h, 2);
+ // Arrive near the old suppression deadline: the line's new easing tail must extend it.
+ h.run_steps(11);
+ h.event(wheel(MouseWheelUnit::Line, -1.0, TouchPhase::Move));
+ let after_touchpad = scroll_over(&mut h, 2);
+ assert!(after_touchpad >= base, "a notch after point input is not dropped: moved {after_touchpad}, base {base}");
+ let delayed = scroll_over(&mut h, 20);
+ assert!(delayed.abs() < 2.0, "the reconstructed notch has no delayed easing copy: moved {delayed}");
+
+ h.event(wheel(MouseWheelUnit::Point, -20.0, TouchPhase::Move));
+ scroll_over(&mut h, 2);
+ h.event(wheel(MouseWheelUnit::Page, -1.0, TouchPhase::Move));
+ let page_after_touchpad = scroll_over(&mut h, 2);
+ assert!(page_after_touchpad > base, "a page event after point input is preserved: moved {page_after_touchpad}");
+ let delayed_page = scroll_over(&mut h, 20);
+ assert!(delayed_page.abs() < 2.0, "the reconstructed page has no delayed easing copy: moved {delayed_page}");
+}
+
+#[test]
+fn changing_continuous_layout_resets_wheel_acceleration() {
+ use pdfcraft_ui_egui::scroll_speed::ScrollSpeed;
+
+ let mut h = harness_stepping(1.0 / 60.0, &[("layout", "continuous"), ("zoom", "300")]);
+ h.state_mut().set_option("animation_effects", "off").unwrap();
+ let at = h.state().views[0].viewport_rect().center();
+ h.hover_at(at);
+ h.run_steps(2);
+ for _ in 0..5 {
+ h.event(wheel(MouseWheelUnit::Line, -1.0, TouchPhase::Move));
+ scroll_over(&mut h, 2);
+ }
+
+ h.state_mut().views[0].set_continuous(false);
+ h.run_steps(3);
+ h.event(wheel(MouseWheelUnit::Line, -1.0, TouchPhase::Move));
+ let moved = scroll_over(&mut h, 2);
+ let base = ScrollSpeed::default().line_points();
+ assert!((moved - base).abs() < 1.0, "a layout jump clears the old burst: moved {moved}, base {base}");
+}
+
/// Steps `frames` frames and returns how far the document scrolled down over them, from the
/// pages on screen in each pair of consecutive frames (a page can leave the screen on the way).
-/// Only the non-macOS glide tests below use it, hence the matching gate.
-#[cfg(not(target_os = "macos"))]
+/// Used by wheel and non-macOS glide tests below.
fn scroll_over(h: &mut Harness<'static, PdfCraftApp>, frames: usize) -> f32 {
let tops = |h: &Harness<'static, PdfCraftApp>| (0..3).map(|p| rect(h, p).map(|r| r.min.y)).collect::>();
let mut moved = 0.0;
diff --git a/crates/ui-egui/tests/viewer.rs b/crates/ui-egui/tests/viewer.rs
index 31530bf235..b154694854 100644
--- a/crates/ui-egui/tests/viewer.rs
+++ b/crates/ui-egui/tests/viewer.rs
@@ -175,6 +175,36 @@ fn find_shortcuts_advance_and_retreat_matches() {
assert_eq!(h.state().views[0].find.as_ref().unwrap().current, Some((start + n - 1) % n));
}
+/// #2669: ⌘G and ⇧⌘G step through the matches while the find field still has the focus.
+#[test]
+fn find_shortcuts_work_from_the_find_field() {
+ let mut h = harness();
+ h.state_mut().active = Some(0);
+ h.run_steps(2);
+ h.state_mut().views[0].open_find();
+ h.state_mut().views[0].find.as_mut().unwrap().query = "page".into();
+ h.state_mut().views[0].rerun_find();
+ let deadline = std::time::Instant::now() + std::time::Duration::from_secs(20);
+ while std::time::Instant::now() < deadline {
+ h.run_steps(2);
+ std::thread::sleep(std::time::Duration::from_millis(5));
+ if h.state().views[0].find.as_ref().unwrap().matches.len() >= 15 {
+ break;
+ }
+ }
+ let n = h.state().views[0].find.as_ref().unwrap().matches.len();
+ assert_eq!(n, 15);
+ assert!(h.ctx.egui_wants_keyboard_input(), "the find field has the focus");
+ let start = h.state().views[0].find.as_ref().unwrap().current.unwrap();
+ h.key_press_modifiers(egui::Modifiers::COMMAND, egui::Key::G);
+ h.run_steps(2);
+ assert_eq!(h.state().views[0].find.as_ref().unwrap().current, Some((start + 1) % n));
+ h.key_press_modifiers(egui::Modifiers::COMMAND | egui::Modifiers::SHIFT, egui::Key::G);
+ h.run_steps(2);
+ assert_eq!(h.state().views[0].find.as_ref().unwrap().current, Some(start));
+ assert!(h.ctx.egui_wants_keyboard_input(), "and keeps it");
+}
+
#[test]
fn marking_while_searching_does_not_jitter_or_reset_search_state() {
let mut h = harness();
@@ -902,6 +932,70 @@ fn an_earlier_revision_opens_from_document_properties() {
);
}
+#[test]
+fn wrapped_revision_buttons_stay_above_binding_and_open_the_saved_revision() {
+ use pdfcraft_engine::{Edit, Session};
+ let mut bytes = std::sync::Arc::new(fixture(1));
+ // Reopen each save so the next edit appends to the preceding saved revision.
+ for revision in 2..=13 {
+ let mut source = Session::new().with_clock(|| 1_700_000_000);
+ let id = source.open("revisions.pdf", None, bytes, None).unwrap();
+ source.apply(id, Edit::SetInfo { key: "Title".into(), value: format!("Revision {revision}") }).unwrap();
+ bytes = source.save_bytes(id).unwrap();
+ }
+ for width in [1400.0, 720.0] {
+ let pdf = bytes.clone();
+ let mut h = Harness::builder().with_size(egui::vec2(width, 900.0)).build_eframe(move |_cc| {
+ let mut app = PdfCraftApp::new();
+ app.set_option("language", "en").unwrap();
+ app.open_bytes("revisions.pdf", None, pdf.to_vec()).unwrap();
+ app
+ });
+ let s = h.state();
+ let doc = s.session.get(s.views[s.active.unwrap()].id).unwrap();
+ assert_eq!(doc.revision_ends().len(), 13, "the fixture contains thirteen saved revisions");
+ h.state_mut().dialog = Some(Dialog::Properties(pdfcraft_ui_egui::PropsTab::Advanced));
+ // Let the grid and the centered dialog settle before measuring or clicking.
+ h.run_steps(6);
+ let buttons: Vec<_> = (1..13).map(|n| (n, h.get_by_label(&format!("View revision {n}")).rect())).collect();
+ let first_row = h.get_by_label("View revision 12").rect();
+ let lower_revision = buttons
+ .iter()
+ .find(|(n, rect)| *n > 1 && rect.top() > first_row.bottom())
+ .map(|(n, _)| *n)
+ .expect("revision buttons wrap to a lower row");
+ let binding = h.get_by_label("Binding").rect();
+ let left = h.get_by_label("Left Edge").rect();
+ let right = h.get_by_label("Right Edge").rect();
+ // Properties requests a fixed width. Measure its card in each viewport.
+ let card = h.ctx.data(|d| d.get_temp::(egui::Id::new("dialog").with("card"))).expect("Properties is shown");
+ for (n, rect) in &buttons {
+ for (label, below) in [("Binding", binding), ("Left Edge", left), ("Right Edge", right)] {
+ assert!(rect.bottom() <= below.top(), "revision {n} overlaps {label} at viewport width {width}: {rect:?}, {below:?}");
+ }
+ }
+ for (n, rect) in &buttons {
+ assert!(
+ rect.left() >= card.left() && rect.right() <= card.right(),
+ "revision {n} escapes Properties horizontally at viewport width {width}: {rect:?}, measured card {card:?}"
+ );
+ }
+ h.get_by_label(&format!("View revision {lower_revision}")).scroll_to_me();
+ h.run_steps(3);
+ h.get_by_label(&format!("View revision {lower_revision}")).click();
+ h.run_steps(3);
+ let s = h.state();
+ assert!(s.dialog.is_none(), "opening a revision closes Properties");
+ let view = &s.views[s.active.unwrap()];
+ let doc = s.session.get(view.id).unwrap();
+ assert_eq!(doc.name, format!("revisions (revision {lower_revision}).pdf"));
+ assert_eq!(doc.info_value("Title"), Some(format!("Revision {lower_revision}")));
+ let snapshot = view.snapshot.as_ref().expect("the saved revision opens as a snapshot");
+ assert_eq!(snapshot.kind, pdfcraft_ui_egui::canvas::SnapshotKind::Revision);
+ assert_eq!(snapshot.revision, lower_revision);
+ }
+}
+
#[test]
fn advanced_search_lists_results_with_context() {
let mut h = harness();
@@ -1448,6 +1542,55 @@ fn a_clicked_picture_is_selected_and_copied_in_its_original_format() {
assert_eq!(h.state().views[i].picked_image, None);
}
+/// A picture picked with the Select tool, then deleted in Edit text & images, leaves no highlight
+/// behind on the empty spot where it was.
+#[test]
+fn a_picked_picture_that_is_deleted_in_edit_mode_loses_its_highlight() {
+ let (_, mut h) = picture_harness();
+ let i = h.state().active.unwrap();
+ let id = h.state().views[i].id;
+ let at = h.state().views[i].page_screen_rect(0).expect("on screen").center();
+ h.hover_at(at);
+ h.run_steps(1);
+ h.event(egui::Event::PointerButton { pos: at, button: egui::PointerButton::Primary, pressed: true, modifiers: Default::default() });
+ h.event(egui::Event::PointerButton { pos: at, button: egui::PointerButton::Primary, pressed: false, modifiers: Default::default() });
+ h.run_steps(2);
+ assert!(h.state().views[i].picked_image.is_some(), "picked with the Select tool");
+ // Edit text & images: click the picture, delete it with the Delete key.
+ assert!(h.state_mut().execute("edit.edit_text"));
+ h.run_steps(2);
+ h.hover_at(at);
+ h.run_steps(1);
+ h.drag_at(at);
+ h.run_steps(1);
+ h.drop_at(at);
+ h.run_steps(2);
+ h.key_press(egui::Key::Delete);
+ // One frame, no more input: the highlight goes at once, it does not wait for a click.
+ h.run_steps(1);
+ assert!(h.state().session.get(id).unwrap().page_images(0).is_empty(), "the picture is gone");
+ assert_eq!(h.state().views[i].picked_image, None, "and so is its highlight");
+}
+
+/// The highlight goes in the very edit that removes the picture, not on a later frame: the window
+/// only redraws on the next input, so waiting for one leaves the highlight on screen until a click.
+#[test]
+fn an_edit_clears_the_picked_picture_at_once() {
+ let (_, mut h) = picture_harness();
+ let i = h.state().active.unwrap();
+ let at = h.state().views[i].page_screen_rect(0).expect("on screen").center();
+ h.hover_at(at);
+ h.run_steps(1);
+ h.event(egui::Event::PointerButton { pos: at, button: egui::PointerButton::Primary, pressed: true, modifiers: Default::default() });
+ h.event(egui::Event::PointerButton { pos: at, button: egui::PointerButton::Primary, pressed: false, modifiers: Default::default() });
+ h.run_steps(2);
+ assert!(h.state().views[i].picked_image.is_some(), "picked with the Select tool");
+ let delete = pdfcraft_engine::Edit::EditPageImage { page: 0, index: 0, change: pdfcraft_engine::ImageEdit::Delete };
+ assert!(h.state_mut().apply_edit(delete));
+ // No frame has run since the edit.
+ assert_eq!(h.state().views[i].picked_image, None);
+}
+
#[test]
fn right_click_on_a_picture_copies_it_at_full_size() {
let (px, mut h) = picture_harness();
@@ -1845,3 +1988,27 @@ fn arrow_keys_follow_the_scroll_speed_preference() {
let (slow, normal, fast) = (lines_to_page_2("slow"), lines_to_page_2("normal"), lines_to_page_2("fast"));
assert!(fast * 3 < normal * 2 && normal * 3 < slow * 2, "presses to page 2: slow {slow}, normal {normal}, fast {fast}");
}
+
+/// #2665: next to a very wide page the normal pages sit far to its right, outside the window.
+/// Going to one changed the page counter but kept the horizontal scroll, so it stayed out of
+/// sight; as in Acrobat, it is brought into view.
+#[test]
+fn go_to_page_brings_the_page_into_view_horizontally() {
+ use pdfcraft_ui_egui::canvas::Fit;
+ let mut h = Harness::builder().with_size(egui::vec2(1400.0, 900.0)).build_eframe(|_cc| {
+ let mut app = PdfCraftApp::new();
+ app.set_option("language", "en").unwrap();
+ app.open_bytes("wide.pdf", None, sized_fixture(&[(595, 842), (595, 842), (595, 842), (6000, 2000)])).unwrap();
+ app.views[0].fit = Fit::None;
+ app.views[0].zoom = 1.0;
+ app
+ });
+ h.run_steps(4);
+ assert!(h.state().views[0].page_screen_rect(0).is_none(), "page 1 starts off-screen, right of the window");
+ h.state_mut().views[0].go_to_page(2);
+ h.run_steps(4);
+ let v = &h.state().views[0];
+ assert_eq!(v.current, 2);
+ let (r, vp) = (v.page_screen_rect(2).expect("page 3 on screen"), v.viewport_rect());
+ assert!(r.left() >= vp.left() && r.right() <= vp.right(), "page 3 at {r:?}, viewport {vp:?}");
+}
diff --git a/crates/xfa/src/layout.rs b/crates/xfa/src/layout.rs
index a491d3f493..62d7beedf1 100644
--- a/crates/xfa/src/layout.rs
+++ b/crates/xfa/src/layout.rs
@@ -232,6 +232,7 @@ pub fn layout(tpl: &Template, data: Option<&DataNode>) -> Result
content: vec![Rect::new(36.0, 36.0, 540.0, 720.0)],
items: Vec::new(),
occur_max: None,
+ occur_min: 0,
}],
};
let mut values = HashMap::new();
@@ -261,6 +262,7 @@ pub fn layout(tpl: &Template, data: Option<&DataNode>) -> Result
if tpl.root.common.presence.occupies() {
l.flow_subform(&root_node, None, &ctx, 0, 0)?;
}
+ l.satisfy_page_area_minimums()?;
l.check_measures()?;
let total = l.pages.len();
let count = total.to_string();
@@ -497,11 +499,28 @@ impl Layouter<'_> {
// ───────────────────────────────────────────────────────────────────────── pages
fn new_page(&mut self) -> Result<(), XfaError> {
+ let idx =
+ (0..self.areas.len()).find(|&i| self.areas[i].occur_max.is_none_or(|m| self.uses[i] < m)).unwrap_or(self.areas.len().saturating_sub(1));
+ self.new_page_of(idx)
+ }
+
+ /// Once the body has been laid out, give every page area that has been used fewer times than
+ /// its `occur min` the pages it still needs, in template order, after the body's pages.
+ fn satisfy_page_area_minimums(&mut self) -> Result<(), XfaError> {
+ let areas = Rc::clone(&self.areas);
+ for (idx, area) in areas.iter().enumerate() {
+ while self.uses.get(idx).is_some_and(|u| *u < area.occur_min) {
+ self.new_page_of(idx)?;
+ }
+ }
+ Ok(())
+ }
+
+ /// Start a page from page area `idx`.
+ fn new_page_of(&mut self, idx: usize) -> Result<(), XfaError> {
if self.pages.len() >= MAX_PAGES {
return Err(XfaError::TooLarge(format!("the form lays out to more than {MAX_PAGES} pages")));
}
- let idx =
- (0..self.areas.len()).find(|&i| self.areas[i].occur_max.is_none_or(|m| self.uses[i] < m)).unwrap_or(self.areas.len().saturating_sub(1));
let areas = Rc::clone(&self.areas);
let Some(area) = areas.get(idx) else {
return Err(XfaError::Malformed("the template has no page area".into()));
diff --git a/crates/xfa/src/model.rs b/crates/xfa/src/model.rs
index f4ef3fabce..572675f958 100644
--- a/crates/xfa/src/model.rs
+++ b/crates/xfa/src/model.rs
@@ -680,6 +680,9 @@ pub struct PageArea {
pub items: Vec,
/// How many pages may use this master (`None`: unbounded).
pub occur_max: Option,
+ /// How many pages must use this master: a layout that ends with fewer adds pages for it.
+ /// Zero without an ``.
+ pub occur_min: usize,
}
#[derive(Clone, Debug, Default, PartialEq)]
diff --git a/crates/xfa/src/parse.rs b/crates/xfa/src/parse.rs
index f6c69789b4..4582d59b22 100644
--- a/crates/xfa/src/parse.rs
+++ b/crates/xfa/src/parse.rs
@@ -529,8 +529,11 @@ fn parse_page_set(n: XmlNode, warnings: &mut Vec, depth: usize) -> PageS
content.push(Rect::new(0.0, 0.0, width, height));
}
let items = pa.children().filter(|c| c.is_element()).filter_map(|c| parse_node(c, warnings, depth + 1)).collect();
- let occur = parse_occur(child(pa, "occur"));
- areas.push(PageArea { name: pa.attribute("name").map(str::to_string), width, height, content, items, occur_max: occur.max });
+ // A page area with no `` has no limit and no minimum (a subform's default of one
+ // would exhaust it, and demand a page for every master).
+ let occur = child(pa, "occur").map(|o| parse_occur(Some(o)));
+ let (occur_min, occur_max) = occur.map_or((0, None), |o| (o.min, o.max));
+ areas.push(PageArea { name: pa.attribute("name").map(str::to_string), width, height, content, items, occur_max, occur_min });
}
// Nested page sets: take their areas too.
for ps in children(n, "pageSet") {
diff --git a/crates/xfa/src/tests.rs b/crates/xfa/src/tests.rs
index 1e8730586d..03eb2c67ad 100644
--- a/crates/xfa/src/tests.rs
+++ b/crates/xfa/src/tests.rs
@@ -1191,6 +1191,75 @@ fn repeating_subforms_inside_positioned_content_get_their_instances() {
assert!(((after.rect.y - rows[2].rect.y) - step).abs() < 0.01, "after: {:?}, last row: {:?}", after.rect, rows[2].rect);
}
+/// A page area without an `` repeats for as long as the body needs pages. Before, it was
+/// treated like a subform (at most once), so once the first body page was out the layout fell
+/// back to the last page area, here one limited to a single use, and drew its content over every
+/// body page (and listed a copy of its fields per page).
+#[test]
+fn page_area_without_occur_serves_every_body_page() {
+ let section = |n: u32| {
+ format!(
+ r#"
+
+
+
+
+ {}{}{}{}
+
+
+
+"##,
+ section(1),
+ section(2),
+ section(3),
+ section(4)
+ );
+ let form = layout_xml(&xdp).unwrap();
+ let soms: Vec> = form.pages.iter().map(|p| widgets(p).iter().map(|w| w.som.clone()).collect()).collect();
+ let has_button = |page: &Vec| page.iter().any(|s| s.contains("PrintButton"));
+ // Four body pages, then the once-only page: its button is on that page alone, not over the body.
+ assert_eq!(form.pages.len(), 5, "{soms:?}");
+ assert!(soms[..4].iter().all(|p| !has_button(p)), "the single-use page area is not drawn over body pages: {soms:?}");
+ assert_eq!(soms[4].iter().filter(|s| s.contains("PrintButton")).count(), 1, "the required page is laid out once: {soms:?}");
+}
+
+/// A page area's `occur min` is a page it must have even when the body never reaches it, and a
+/// body that already used it does not get another. An area with no `` needs none.
+#[test]
+fn page_area_minimum_adds_pages_after_the_body_only_when_unused() {
+ let area = |name: &str, occur: &str| {
+ format!(
+ r#"{occur} "#
+ )
+ };
+ let pages = |areas: String| {
+ let xdp = format!(
+ r##"
+
+{areas} ` and `min="0"` need no page.
+ assert_eq!(pages(area("A", "") + &area("B", r#">().join(","),
@@ -1044,7 +1046,7 @@ pub fn merge_csv(files: &[Vec<(String, Vec)>]) -> String {
}
}
}
- let q = |s: &str| if s.contains([',', '"', '\n', '\r']) { format!("\"{}\"", s.replace('"', "\"\"")) } else { s.to_string() };
+ let q = |s: &str| if s.contains([',', '"', '\n', '\r', '\t']) { format!("\"{}\"", s.replace('"', "\"\"")) } else { s.to_string() };
let mut out = columns.iter().map(|c| q(c)).collect::>().join(",");
out.push('\n');
for f in files {
@@ -1104,6 +1106,31 @@ fn import_fdf(doc: &mut Document, bytes: &[u8]) -> Result {
Ok(report)
}
+// Inspect the first logical record: tabs within quoted CSV cells are data. A quote
+// in the middle of an unquoted TSV name stays literal, as in the TSV reader.
+fn has_tab_delimiter(text: &str) -> bool {
+ let mut quoted = false;
+ let mut cell_start = true;
+ let mut chars = text.chars().peekable();
+ while let Some(c) = chars.next() {
+ match c {
+ '"' if quoted && chars.peek() == Some(&'"') => {
+ chars.next();
+ }
+ '"' if quoted || cell_start => {
+ quoted = !quoted;
+ cell_start = false;
+ }
+ '\t' if !quoted => return true,
+ '\r' | '\n' if !quoted => return false,
+ ',' if !quoted => cell_start = true,
+ _ => cell_start = false,
+ }
+ }
+ // An unclosed quote can be a literal TSV field-name character.
+ quoted && text.lines().next().is_some_and(|line| line.contains('\t'))
+}
+
/// Form data from XML, CSV or tab-delimited text.
fn import_table(doc: &mut Document, text: &str) -> Result {
let mut report = Report::default();
@@ -1121,7 +1148,7 @@ fn import_table(doc: &mut Document, text: &str) -> Result {
(name, vec![c.children().filter(|node| node.is_text()).filter_map(|node| node.text()).collect()])
})
.collect()
- } else if t.lines().next().is_some_and(|l| l.contains('\t')) {
+ } else if has_tab_delimiter(t) {
tab_delimited(t)?
} else {
// CSV with quotes: a quoted value may hold commas, doubled quotes and line breaks.
diff --git a/crates/xfdf/src/tests.rs b/crates/xfdf/src/tests.rs
index 47952e642b..a4f6abb2d3 100644
--- a/crates/xfdf/src/tests.rs
+++ b/crates/xfdf/src/tests.rs
@@ -1,6 +1,6 @@
use std::sync::Arc;
-use pdfcraft_annot::{Markup, Meta, NewAnnotation, Shape, Style, add_annotation, add_reply, summaries};
+use pdfcraft_annot::{Markup, Meta, NewAnnotation, Shape, Style, add_annotation, add_reply, add_text_replacement, summaries};
use pdfcraft_cos::{SaveOptions, write_full, write_incremental};
use pdfcraft_forms::{NewField, add_field, fields, set_value};
@@ -142,6 +142,48 @@ fn values(doc: &Document) -> Vec<(String, Vec)> {
fields(doc).into_iter().map(|f| (f.name, f.value)).collect()
}
+/// #2666: Replace Text writes a StrikeOut and a Caret that replies to it. XFDF names a reply by
+/// the `/NM` of what it replies to, so when both carried one id the caret's `inreplyto` was its
+/// own `name`: importing it found a comment replying to itself and the pair was lost.
+#[test]
+fn a_replace_text_pair_survives_an_xfdf_round_trip() {
+ let mut src = blank();
+ let quads = [[100.0, 720.0, 200.0, 720.0, 100.0, 706.0, 200.0, 706.0]];
+ let style = Style { color: [0.89, 0.13, 0.13], opacity: 1.0, width: 1.0, fill: None };
+ let meta = |id: &str| Meta { date: Some("D:20261001120000Z".into()), id: id.into() };
+ add_text_replacement(&mut src, 0, &quads, "new words", "Ada", &style, &style, &meta("strike"), &meta("caret")).unwrap();
+ assert_eq!(summaries(&src).len(), 2, "a strikeout and a caret");
+
+ let xfdf = export_xfdf(&src, true, false, "a.pdf");
+ assert!(xfdf.contains("name=\"strike\"") && xfdf.contains("name=\"caret\""), "each gets its own name: {xfdf}");
+ assert!(xfdf.contains("inreplyto=\"strike\""), "the caret replies to the strikeout: {xfdf}");
+ assert!(!xfdf.contains("inreplyto=\"caret\""), "never a reply to itself");
+
+ let mut dst = blank();
+ let r = import(&mut dst, xfdf.as_bytes()).unwrap();
+ assert_eq!(r.comments, 2, "both comments import: {r:?}");
+ let found = summaries(&dst);
+ let kinds: Vec<&str> = found.iter().map(|s| s.subtype.as_str()).collect();
+ assert!(kinds.contains(&"StrikeOut") && kinds.contains(&"Caret"), "{kinds:?}");
+ assert_eq!(comment_view(&dst), comment_view(&src), "the pair and its thread survive");
+}
+
+#[test]
+fn xml_exports_drop_xml_noncharacters() {
+ let mut src = filled();
+ set_value(&mut src, "Name", &FieldValue::Text("a\u{FFFE}b\u{FFFF}c".into())).unwrap();
+ let xfdf = export_xfdf(&src, false, true, "f.pdf");
+ let doc = roxmltree::Document::parse(&xfdf).unwrap_or_else(|e| panic!("XFDF is not well-formed: {e}: {xfdf:?}"));
+ let value = doc
+ .descendants()
+ .find(|n| n.has_tag_name("field") && n.attribute("name") == Some("Name"))
+ .and_then(|n| n.children().find(|c| c.has_tag_name("value")))
+ .and_then(|n| n.text());
+ assert_eq!(value, Some("abc"));
+ let data = export_data(&src, Format::Xml);
+ roxmltree::Document::parse(&data).unwrap_or_else(|e| panic!("XML is not well-formed: {e}: {data:?}"));
+}
+
#[test]
fn xml_exports_preserve_whitespace_in_field_names() {
let names = ["Line\nName", "Line\tName", "Line\rName", "Line\r\nName", "\t Leading\nTrailing \t", "Line Name"];
@@ -1144,3 +1186,49 @@ fn xml_form_import_preserves_text_around_comments_and_instructions() {
assert_eq!(fields(&original).into_iter().find(|f| f.name == "Name").unwrap().value, vec!["replace me"]);
}
}
+
+#[test]
+fn csv_headers_with_tabs_survive_export_import_and_save() {
+ for name in ["First\tName", "\tName", "Name\t", "Name\t\"quoted\"", "Name\nwith\ta tab"] {
+ let mut source = one_field();
+ add_field(&mut source, 0, [50.0, 650.0, 250.0, 670.0], &NewField::Text { multiline: false }, Some("Other")).unwrap();
+ set_value(&mut source, "Other", &FieldValue::Text("other field".into())).unwrap();
+ let field = fields(&source).into_iter().find(|f| f.name == "Name").unwrap().obj;
+ source.update_dict(field, |dict| dict.set(b"T".to_vec(), PdfString::text(name))).unwrap();
+ set_value(&mut source, name, &FieldValue::Text("original, \"value\"".into())).unwrap();
+ let bytes = write_full(&source, &SaveOptions::default()).unwrap();
+ let source = Document::open(Arc::new(bytes.clone())).unwrap();
+ let row = data_values(&bytes).unwrap();
+ for csv in [export_data(&source, Format::Csv), merge_csv(&[row])] {
+ let mut doc = Document::open(Arc::new(bytes.clone())).unwrap();
+ set_value(&mut doc, name, &FieldValue::Text("replacement".into())).unwrap();
+ set_value(&mut doc, "Other", &FieldValue::Text("other replacement".into())).unwrap();
+ let report = import(&mut doc, csv.as_bytes()).unwrap();
+ assert_eq!(report.fields, 2);
+ assert!(report.rejected.is_empty(), "{:?}", report.rejected);
+ for saved in [write_full(&doc, &SaveOptions::default()).unwrap(), write_incremental(&doc, &SaveOptions::default()).unwrap()] {
+ let reopened = Document::open(Arc::new(saved)).unwrap();
+ assert_eq!(fields(&reopened).into_iter().find(|f| f.name == name).unwrap().value, vec!["original, \"value\""], "{csv:?}");
+ assert_eq!(values(&reopened), values(&source), "{csv:?}");
+ }
+ }
+ }
+}
+
+#[test]
+fn tabular_import_keeps_tsv_quotes_literal_and_csv_value_tabs() {
+ for name in ["Name", "Na\"me", "\"Name\"", "\"Name"] {
+ let mut doc = one_field();
+ add_field(&mut doc, 0, [50.0, 650.0, 250.0, 670.0], &NewField::Text { multiline: false }, Some("Other")).unwrap();
+ let field = fields(&doc).into_iter().find(|f| f.name == "Name").unwrap().obj;
+ doc.update_dict(field, |dict| dict.set(b"T".to_vec(), PdfString::text(name))).unwrap();
+ let text = format!("{name}\tOther\r\nAda\tother field\r\n");
+ let report = import(&mut doc, text.as_bytes()).unwrap();
+ assert_eq!(report.fields, 2, "{text:?}");
+ assert_eq!(fields(&doc).into_iter().find(|f| f.name == name).unwrap().value, vec!["Ada"]);
+ }
+ let mut doc = one_field();
+ add_field(&mut doc, 0, [50.0, 650.0, 250.0, 670.0], &NewField::Text { multiline: false }, Some("Other")).unwrap();
+ import(&mut doc, b"Name,Other\r\n\"Ada\tLovelace\",other field\r\n").unwrap();
+ assert_eq!(fields(&doc).into_iter().find(|f| f.name == "Name").unwrap().value, vec!["Ada\tLovelace"]);
+}
diff --git a/docs/roadmap.md b/docs/roadmap.md
index b8c0403e33..696e0c5d73 100644
--- a/docs/roadmap.md
+++ b/docs/roadmap.md
@@ -3,12 +3,16 @@
> **2026-10-18:** M2 / #3116: clipping-only glyphs stay extractable with the image filter; invisible OCR remains excluded. Estimates unchanged.
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M14/#2383 GPU-limit candidate; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M7: the Edit text caret is as tall as the row it is on; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
+> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M6: CSV tab-containing headers; estimates unchanged)
+> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-18 · **Change:** minor (M4, #1919: repeated page indices accumulate rotations; M4: indirect page rotations refreshed upstream; M13: CLI check visits each directory once through aliases and cycles; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-18 · **Last updated:** 2026-10-18 · **Change:** minor (M3, #3149: discard render demand after page geometry changes and check the GPU texture limit before upload; estimates unchanged)
> **Last reviewed:** 2026-10-18 · **Last updated:** 2026-10-18 · **Change:** minor (M8, #1350: named replacement descriptions retired during Sanitize and across shared resources; estimates unchanged)
+> **Last reviewed:** 2026-10-17 · **Last updated:** 2026-10-17 · **Change:** minor (M4 #2997: Ctrl-drag (Option on macOS) in the Organize grid copies the pages into the drop gap; `page_duplicate` takes `at`; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M6: AcroForm XML text; estimates unchanged)
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M6: preserve XML/XFDF attribute whitespace; refreshed upstream; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-15 · **Change:** minor (M1: raw-deflate fallback preserves decoding-limit errors; refreshed upstream; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-13 · **Last updated:** 2026-10-14 · **Change:** minor (M14: the Italian catalog has every entry the other catalogs had and the code looks up, 2,528 entries; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
+> **Last reviewed:** 2026-10-16 | **Last updated:** 2026-10-16 | **Change:** minor (M10 #1201 partial: lossless image Decode; estimates unchanged) | **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-12 · **Last updated:** 2026-10-13 · **Change:** minor (M2/M6: spot-colour images convert once per pixel value; widget placement and the Fields panel scale with the field count; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-14 · **Last updated:** 2026-10-14 · **Change:** minor (M3 candidate: coordinate control clicks on slow frames and coalesced text selection; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-14 · **Last updated:** 2026-10-15 · **Change:** minor (M2: sampled functions index a flat table, one image decode serves every zoom level, patch meshes copy less; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
@@ -41,6 +45,7 @@
> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-11 · **Change:** minor (M13: reject malformed CLI --root options before dispatch; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-11 · **Change:** minor (Traditional Chinese accessibility semantics and Taiwan form/table terminology; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
> **Last reviewed:** 2026-10-12 · **Last updated:** 2026-10-12 · **Change:** patch (M1 #1337: exhausted xref hints are repaired; allocation avoids overflow; estimates unchanged) · **Target:** Adobe Acrobat Pro (Acrobat DC, continuous track 26.002.21931, macOS)
+> **Last reviewed:** 2026-10-18 - **Last updated:** 2026-10-18 - **Change:** minor (M6 #2892: Push-button normal labels are editable per widget through Options and form_set_props, independently of field names; rendering, undo/redo and incremental/full save/reopen are covered. Estimates unchanged)
Forward-looking: the milestones, the current focus and what comes next. The parity numbers behind
it are in [target-app-parity.md](target-app-parity.md), the ranked work list in [gaps.md](gaps.md),
@@ -105,11 +110,12 @@ total matches the parity estimate.
| M | Milestone | Plan size (h) | Done | Remaining (h, 2026-10-10) | Notes |
|---|---|---|---|---|---|
| M0 | Skeleton: workspace, xtask gates, CI | 15–30 | 98% | 1–2 | GitHub workflow, `deny.toml` (licence audit of every dependency), parity checklist (826 features, `xtask parity`) done. Missing: remaining crate stubs, testkit/oracle crates |
-| M1 | COS: filters, crypt, parser, xref, writer | 120–200 | 80% | 20–40 | Done:` argument (pass `None` for true colours); the upstream-only `[dev-dependencies.log]` entry is removed from `Cargo.toml` (log is a normal dependency since the logging patch), so `cargo test -p hayro` runs this crate's unit tests; (14) a shrunk, interpolated, axis-aligned image is snapped to whole device pixels: shrunk once to its rounded device rectangle and drawn 1:1 at the rounded corner, instead of a ceil-sized raster resampled again at a fractional scale and offset (`snap_minified_image`; scans looked soft next to Acrobat); (15) a soft mask of another size than its image is filtered when shrunk, whatever its `/Interpolate` (#1649): such a mask carries the shape on its own (Word saves text as a tiny black image with a glyph-bitmap mask), and nearest-neighbour dropped every stroke thinner than the shrink factor; (16) under a soft mask or a blend mode, axial and radial shadings are sampled instead of drawn as native gradients (#2128): vello_cpu 0.0.8 copies an opaque gradient over fully covered tiles without the mask or blend mode, so PowerPoint's faded gradient fills came out solid; (17) the renderer keeps annotation paints out of the reading-colours mapping between `begin_annotations` and `end_annotations` (#1548); (18) the page renders as one transparency group with its background painted inside it as the initial backdrop, so a Multiply highlight blends with a dark reading-colours background instead of staying an opaque block (identity on white pages) | (1) `raster::tests::tiles_match_full_render`; (2) `raster::tests::absurd_image_dimensions_are_skipped`; (3) `raster::tests::tiling_cells_stay_small_whatever_the_step`; (4) `raster::tests::absurd_predictor_columns_and_line_widths_render`; (5) `raster::tests::dash_patterns_that_cut_a_stroke_into_billions_of_pieces_render`; (6) `raster::tests::invalid_dash_patterns_terminate_and_are_drawn_solid`; (7) `raster::tests::pixels_are_a_plain_renders_bytes_in_a_buffer_a_gui_can_take_over`; (8) `raster::tests::image_resampling_rejects_anisotropic_target_growth`, `raster::tests::image_resampling_bounds_intermediate_before_planning`, `raster::tests::image_resampling_rejects_invalid_sources_and_scales`, `raster::tests::image_resampling_keeps_valid_sizes_and_exact_limits`, `raster::tests::resampling_fallback_preserves_original_geometry_and_pixels`, `raster::tests::image_resampling_wide_sources_shrink_before_backend_side_limits`, `raster::tests::image_resampling_wide_transparent_sources_still_shrink`, `raster::tests::image_resampling_padded_backend_dimensions_stay_checked`, `raster::tests::image_resampling_mismatched_alpha_mask_keeps_pixels_and_placement`; (9) `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel`; (10) `raster::tests::axial_and_radial_shadings_draw_as_native_gradients`; (11) `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel`; (12) `raster::tests::preview_renders_skip_tiny_images_but_finals_draw_everything`; (13) `color_replacement_tests::{black_maps_to_text_colour, white_maps_to_background, midgrey_interpolates, alpha_copied_verbatim, only_black_threshold, line_art_false_skips_paths_but_not_text}`, `color_replacement_render_tests::{color_replacement_maps_text_and_line_art, color_replacement_maps_stencil_mask_ink}`; (14) `raster::tests::interpolated_minified_image_edges_snap_to_whole_pixels` (`cargo test -p hayro`); (15) `raster::tests::shrunk_soft_mask_larger_than_its_image_keeps_thin_strokes`; (16) `raster::tests::soft_mask_applies_to_a_shading_pattern_fill`; (17), (18) `raster::tests::reading_colors_keep_highlight_annotations_readable` |
| hayro | 0.7.1 | Apache-2.0 OR MIT | (1) `RenderSettings::{x_offset, y_offset}` to render one tile of a large page; (2) images over `MAX_IMAGE_PIXELS` (2^28) are skipped (a fuzzed `/W 4294967295` hung in resampling); (3) `tiling_cell_scale` keeps a tiling pattern's cell pixmap within 3000 px a side whatever its `/XStep`/`/YStep` (a fuzzed step far beyond `/BBox` allocated 2 GB); (4) stroke widths clamped to 4 × (canvas width + height) device pixels (a fuzzed `/LW 9223372036854775807` allocated 10 GB in stroke expansion); (5) `usable_dash_pattern`: a stroke whose dash pattern could cut it into more than `MAX_DASHES_PER_STROKE` (250,000) pieces, counting kurbo's restart of the pattern at every subpath and `ClosePath`, is drawn solid (a fuzzed `/D [[11] 0]` on a line from x = 92234775807 allocated over 5 GB, `[0.000001] 0 d` on a 20 pt line over 10 GB); (6) a dash pattern with a negative or non-finite entry, or a period that isn't positive, is drawn solid (`[-1 -1] 0 d` looped for ever in kurbo); (7) `render_size` and `render_into`: render into a buffer the caller allocates (PdfCraft renders into pixel-aligned memory that the GUI takes over as texture data); `render` is `render_into` on a new pixmap; (8) checked image source/target/RGBA, alpha-premultiplication scratch and intermediate sizes before resampling, finite transforms and exact source/mask buffers; sources wider than u16 may shrink before backend limits, with a 2^20 source-axis cap and conservative 128 MiB filter-planning estimate; an over-limit or failed resize keeps the original image resolution and geometry when the source fits the backend, otherwise skips it, with a logged warning; staging, resampling output/scratch, RGBA conversion and padding buffers use fallible reserves; (9) `may_paint_canvas`: a fill or stroke whose control-point box, grown by how far its paint reaches past the path (half the stroke width times the larger of the miter limit and √2) plus 0.5 user units and 2 pixels, misses the canvas is not drawn, so a tile no longer strokes and fills every path of its page (item 5 of #623); `set_skip_offscreen_paths` turns this off for the regression test; (10) `/Interpolate` governs the pre-resize too (#624): a minified image that does not request interpolation is pre-resized with nearest-neighbour, planned without the Catmull-Rom filter limits (`image_resampling_size` takes the flag); a Type 3 glyph image is sampled bicubically only when interpolated; a minified image with a matching soft mask is nearest-sampled straight into premultiplied RGBA; (11) `may_paint_rect`: an image whose stream-size box, mapped to the device, misses the canvas is skipped before decoding (decoded data can only shrink against that box), so a tile of an image-heavy page no longer decodes, resamples and rasterizes every image on it; (12) preview renders (`InterpreterSettings::preview`: backdrops, thumbnails, grid cells, signature-drag backgrounds) skip images under `PREVIEW_MIN_IMAGE_AREA` device pixels, while final renders always draw them; (13) reading colours (#897): `RenderSettings::color_replacement` takes a `ColorReplacement` (text colour, background, `only_black`, `line_art`) that remaps solid text and line-art paint while rendering, glyph outlines counting as text and stencil-mask ink as line art; images, image alpha masks, soft masks and pattern cells keep their colours; `render_pdf` gains a `color_replacement: Option` argument (pass `None` for true colours); the upstream-only `[dev-dependencies.log]` entry is removed from `Cargo.toml` (log is a normal dependency since the logging patch), so `cargo test -p hayro` runs this crate's unit tests; (14) a shrunk, interpolated, axis-aligned image is snapped to whole device pixels: shrunk once to its rounded device rectangle and drawn 1:1 at the rounded corner, instead of a ceil-sized raster resampled again at a fractional scale and offset (`snap_minified_image`; scans looked soft next to Acrobat); (15) a soft mask of another size than its image is filtered when shrunk, whatever its `/Interpolate` (#1649): such a mask carries the shape on its own (Word saves text as a tiny black image with a glyph-bitmap mask), and nearest-neighbour dropped every stroke thinner than the shrink factor; (16) under a soft mask or a blend mode, axial and radial shadings are sampled instead of drawn as native gradients (#2128): vello_cpu 0.0.8 copies an opaque gradient over fully covered tiles without the mask or blend mode, so PowerPoint's faded gradient fills came out solid; (17) the renderer keeps annotation paints out of the reading-colours mapping between `begin_annotations` and `end_annotations` (#1548); (18) the page renders as one transparency group with its background painted inside it as the initial backdrop, so a Multiply highlight blends with a dark reading-colours background instead of staying an opaque block (identity on white pages) | (1) `raster::tests::tiles_match_full_render`; (2) `raster::tests::absurd_image_dimensions_are_skipped`; (3) `raster::tests::tiling_cells_stay_small_whatever_the_step`; (4) `raster::tests::absurd_predictor_columns_and_line_widths_render`; (5) `raster::tests::dash_patterns_that_cut_a_stroke_into_billions_of_pieces_render`; (6) `raster::tests::invalid_dash_patterns_terminate_and_are_drawn_solid`; (7) `raster::tests::pixels_are_a_plain_renders_bytes_in_a_buffer_a_gui_can_take_over`; (8) `raster::tests::image_resampling_rejects_anisotropic_target_growth`, `raster::tests::image_resampling_bounds_intermediate_before_planning`, `raster::tests::image_resampling_rejects_invalid_sources_and_scales`, `raster::tests::image_resampling_keeps_valid_sizes_and_exact_limits`, `raster::tests::resampling_fallback_preserves_original_geometry_and_pixels`, `raster::tests::image_resampling_wide_sources_shrink_before_backend_side_limits`, `raster::tests::image_resampling_wide_transparent_sources_still_shrink`, `raster::tests::image_resampling_padded_backend_dimensions_stay_checked`, `raster::tests::image_resampling_mismatched_alpha_mask_keeps_pixels_and_placement`; (9) `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel`; (10) `raster::tests::non_interpolated_images_are_not_smoothed_when_minified`, `raster::tests::non_interpolated_gray_images_are_not_smoothed_when_minified`, `raster::tests::non_integer_minification_samples_the_same_columns`, `raster::tests::non_integer_minification_samples_colour_and_mask_from_the_same_columns`, `raster::tests::non_interpolated_alpha_images_keep_their_colours_and_mask_when_minified`, `raster::tests::non_interpolated_type3_glyph_images_are_not_smoothed_when_minified`, `raster::tests::image_resampling_nearest_accepts_wide_sources_the_filter_refuses`, `raster::tests::wide_non_interpolated_sources_draw_after_shrinking`; (11) `raster::tests::skipping_paths_off_the_canvas_changes_no_pixel`; (12) `raster::tests::preview_renders_skip_tiny_images_but_finals_draw_everything`; (13) `color_replacement_tests::{black_maps_to_text_colour, white_maps_to_background, midgrey_interpolates, alpha_copied_verbatim, only_black_threshold, line_art_false_skips_paths_but_not_text}`, `color_replacement_render_tests::{color_replacement_maps_text_and_line_art, color_replacement_maps_stencil_mask_ink}`; (14) `raster::tests::interpolated_minified_image_edges_snap_to_whole_pixels` (`cargo test -p hayro`); (15) `raster::tests::shrunk_soft_mask_larger_than_its_image_keeps_thin_strokes`; (16) `raster::tests::soft_mask_applies_to_a_shading_pattern_fill`; (17), (18) `raster::tests::reading_colors_keep_highlight_annotations_readable` |
| hayro-jbig2 | 0.3.0 | Apache-2.0 OR MIT | (1) at most 2^28 pixels per bitmap, page bitmap, halftone grid and grayscale image (each side could be 65535: 4.3 billion pixels decoded one by one); (2) at most 2^26 symbol pixels per symbol dictionary (a fuzzed dictionary decoded for minutes); (3) a symbol dictionary's height classes are capped at one per new symbol and its export runs at `2 × symbols + 1`, each plus `EXTRA_EMPTY_STEPS` (1024): an empty height class or a zero-length export run doesn't advance, and past the end of its data the arithmetic decoder can decode them for ever (one changed byte hung the renderer) | (1) `raster::tests::jbig2_regions_of_billions_of_pixels_are_refused`; (2) the nightly `cargo xtask fuzz` finding, which now finishes in about a second; (3) `raster::tests::jbig2_symbol_dictionaries_that_never_advance_terminate` |
| lopdf | 0.45.0 | MIT | (1) PNG predictor rows longer than the data are refused before their buffers are allocated (a fuzzed `/Columns 4294967295` allocated two 4 GiB rows), and empty frames return without row buffers; (2) predictor row widths use saturating arithmetic or reject overflow; (3) sub-byte TIFF Predictor 2 scratch is bounded by the samples present in the input instead of the declared `/Colors`, aligned with [lopdf#589](https://github.com/J-F-Liu/lopdf/pull/589) | (1) `inspect::tests::png_predictor_rows_longer_than_the_data_are_refused_before_allocating`, `inspect::tests::empty_png_predictor_frames_do_not_allocate_rows`; (2) `inspect::tests::tiff_subbyte_predictor_row_width_overflow_is_refused`; (3) `inspect::tests::tiff_subbyte_predictor_scratch_is_bounded_by_available_samples`, `inspect::tests::tiff_subbyte_predictors_keep_components_rows_and_padding` |
-| hayro-syntax | 0.7.2 | Apache-2.0 OR MIT | (1) Page-tree cycle guard in `resolve_pages` (visited set + depth cap): a `/Kids` loop among compressed objects overflowed the stack and aborted the process; (2) `ccitt_size_ok`: CCITT images wider than 2^20 columns or over 2^28 pixels are not decoded (a fuzzed `/Columns 4294967295` allocated 4 GiB); (3) Flate/LZW predictor row widths are saturating `usize`, a zero width is refused and a row longer than the data is not allocated (a fuzzed `/Columns 9223372036854775807` wrapped to a 2^61-byte allocation and aborted); (4) the inline-image end-of-data check (`EI`) doesn't nest inside its own look-ahead and examines at most 256 candidates and steps (50 images with "EI" in their data ran for over ten minutes); (5) a stream whose `/Length` is wrong is recovered only when `stream` is the next token — scanning ahead to a later object's `stream` made an indirect appearance-state dictionary (Quartz checkboxes: `/AP /N` → `<< /Yes … /Off … >>`) render as a non-form, so the widget was skipped; (6) JPX rescaling uses an exact integer sample range for 1–32-bit components (32 shifted past u32, and floating-point maxima at 25–31 bits wrapped to zero), and refuses zero-sized rows; (7) `decode_limit`: Flate (through flate2 and the fallback inflater), LZW and RunLength never write past `MAX_DECODED_STREAM` (256 MiB), or, for the last filter of an image of known size, past twice its pixel bytes (a predictor tag byte per byte), at most 1 GiB; `Stream::decoded_within` for budgeted callers; a page's `/Contents` array shares one `MAX_DECODED_STREAM` (a 2 MB page whose content stream or 8 × 8 image inflated to 2 GB took the renderer past 3 GB in under a second; naming the stream twenty times in `/Contents`, 5 GB); (8) high-precision numbers (jsPDF MediaBox coordinates with over 19 digits) parse without overflowing the u64 mantissa: fractional digits past it are dropped, integer digits past it scale the value (a real, kept finite); (9) large dictionary indexes are built on first use and shared across repeated reads within each PDF (#307); the validated-byte cache is bounded, reader contexts remain separate, repairs and inline/content dictionaries bypass sharing, and small dictionaries stay eager; (10) decoded object-stream slots publish once under contention, cold loading rejects recursive dependencies, and xref repair serializes without holding recursive read guards; this permits one parser per native render pool with private worker caches; (11) `sync.rs`: a poisoned lock is recovered instead of unwrapped (a worker panic no longer cascades into every reader of the same PDF), and xref repair takes a blocking write lock instead of asserting a non-blocking one; (12) `SegmentList` (`util.rs`) returns `None` for an out-of-range index instead of panicking; (13) a page's `/UserUnit` scales `render_dimensions` and `initial_transform`, so the page and everything on it are sized in points, as Acrobat shows them; it is read as Acrobat reads it (the page's own entry, not inherited; a number from 1 up, at most 75,000; anything else counts as 1), and a page without it keeps exactly the transform it had; (14) revision 2–4 passwords are tried as given and then in PDFDocEncoding (`legacy_passwords` in `crypto/mod.rs`, which resolves the upstream `TODO: Convert to PDFDocEncoding`): passwords arrive as UTF-8 text, so a password with characters outside ASCII (`Çok güzel`, `ılık€`) written in PDFDocEncoding, as ISO 32000-2 §7.6.4.3.2 asks and qpdf and Acrobat write it, was always wrong; (15) page content is read a piece at a time (`page::ContentReader`, `Page::content_reader`): plain streams and Flate streams without a predictor are decoded as they are read, with a budget of 64 times their compressed size and never less than `MAX_DECODED_STREAM` (a stream that expands past it is cut and reported); other streams are decoded whole, as before, and share `MAX_DECODED_STREAM`. The content tokenizer reads a window at a time (`UntypedIter::new_window`, `TypedIter::needs_more`, `TypedIter::committed`): an instruction that runs past the window is not decided until more data follows, which `Reader` records by flagging reads and `at_end` past the end of the data, including through the clones the parser makes. The Flate decoder keeps its own 32 KiB history, so it pauses at any output size. `ContentReader::new` reuses raw/decrypted stream bytes for budget accounting and piece creation, and `decoded_raw_within` applies the existing Whole-stream filters and limits to those bytes without decrypting twice; (16) revision 5–6 (AES-256) passwords are tried in their SASLprep form (RFC 4013, then UTF-8, at most 127 bytes, as ISO 32000-2 §7.6.4.3.3 Algorithm 2.A asks and pdfcraft-crypt writes them) and then, if that differs, as given (`aes256_passwords` in `crypto/mod.rs`, which resolves the upstream `TODO: Do the above.`; `stringprep`, already in the tree through pdfcraft-crypt, is an optional dependency enabled by `std`, and without `std` passwords are used as given): the renderer used the password as typed, so one that SASLprep changes (a decomposed `café`, a no-break space, the fullwidth `ａｂ` a Japanese IME types in full-width mode) was wrong, although pdfcraft-cos accepted it: protecting a document with such an open password was rolled back ("the password is incorrect"), and a file with such a password did not open with it.; (17) CCITT `/Rows 0` means the height is not predetermined (ISO 32000-2 Table 11), not zero rows: scanned Group 4 pages written that way rendered blank | (1) `cos/tests/writer.rs::page_tree_loops_in_object_streams_do_not_crash_readers`; (2) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (3) `raster::tests::absurd_predictor_columns_and_line_widths_render`; (4) `raster::tests::inline_images_with_ei_in_their_data_parse_in_linear_time`; (5) `raster::tests::indirect_appearance_state_dict_is_not_another_stream`; (6) `raster::tests::jpx_stream_bit_depth_boundaries_do_not_panic`, `raster::tests::jpx_32_bit_image_does_not_prevent_other_page_content_from_rendering`, `raster::tests::jpx_gray_rgb_and_alpha_keep_eight_bit_samples`, `raster::tests::jpx_streams_reject_out_of_range_bit_depths`; (7) `raster::tests::decompression_bombs_stop_at_the_stream_limit`, `raster::tests::page_contents_and_nested_forms_share_one_budget`; (8) `inspect::tests::jspdf_high_precision_mediabox_does_not_overflow_dimensions`, `object::number::tests::high_precision_float_does_not_overflow_mantissa` (hayro-syntax); (9) `object::dict::lazy_index_tests`, `raster::tests::large_resource_indexes_preserve_pixels_and_text`; (10) `data::tests`, `xref::tests::concurrent_cold_object_stream_reads_all_succeed`, `raster::tests::shared_parser_is_lazy_and_initializes_once_under_contention`; (11) `data::tests::panicking_initializer_wakes_waiters_and_prevents_retries`, `xref::tests::concurrent_cold_object_stream_reads_all_succeed`; (12) `util::tests::segment_list_rejects_out_of_range_indices`; (13) `raster::tests::user_unit_scales_the_page_and_everything_on_it`; (14) `crypto::pdfcraft_password_tests`, `pdfcraft-engine` `tests::legacy_passwords_with_turkish_characters_open`; (15) `content::tests::windows_keep_every_instruction_at_every_split`, `content::tests::windows_keep_every_instruction_in_pieces_of_any_size`, `content::tests::windows_match_the_slice_on_random_content`, `content::tests::windows_drop_blanks_and_keep_a_comment_that_may_go_on`, `page::content_reader_tests::content_past_the_cap_is_read_to_the_end`, `page::content_reader_tests::a_stream_that_expands_past_the_budget_is_cut_there`, `page::content_reader_tests::a_small_expansion_under_the_cap_is_not_cut`, `raster::tests::page_content_past_the_old_cap_renders`, `raster::tests::content_reader_returns_the_same_bytes_in_any_pieces`, `raster::tests::content_bombs_stop_at_the_budget_and_say_so`, `page::content_reader_tests::plain_and_whole_pieces_keep_their_bytes`, `page::content_reader_tests::repeated_content_reference_counts_compressed_bytes_once`, `page::content_reader_tests::whole_cap_still_counts_later_unique_streams`; (16) `crypto::pdfcraft_password_tests::aes256_passwords_are_tried_in_their_saslprep_form_then_as_given`, `crypto::pdfcraft_password_tests::aes256_passwords_open_as_typed_and_in_their_saslprep_form`, `pdfcraft-engine` `tests::aes256_open_passwords_that_saslprep_changes_protect_and_reopen_as_typed`, `tests::aes256_files_whose_passwords_saslprep_changes_open_as_typed`; (17) `raster::tests::group4_fax_with_rows_zero_decodes_to_the_image_height`
+| hayro-syntax | 0.7.2 | Apache-2.0 OR MIT | (1) Page-tree cycle guard in `resolve_pages` (visited set + depth cap): a `/Kids` loop among compressed objects overflowed the stack and aborted the process; (2) `ccitt_size_ok`: CCITT images wider than 2^20 columns or over 2^28 pixels are not decoded (a fuzzed `/Columns 4294967295` allocated 4 GiB); (3) Flate/LZW predictor row widths are saturating `usize`, a zero width is refused and a row longer than the data is not allocated (a fuzzed `/Columns 9223372036854775807` wrapped to a 2^61-byte allocation and aborted); (4) the inline-image end-of-data check (`EI`) doesn't nest inside its own look-ahead and examines at most 256 candidates and steps (50 images with "EI" in their data ran for over ten minutes); (5) a stream whose `/Length` is wrong is recovered only when `stream` is the next token — scanning ahead to a later object's `stream` made an indirect appearance-state dictionary (Quartz checkboxes: `/AP /N` → `<< /Yes … /Off … >>`) render as a non-form, so the widget was skipped; (6) JPX rescaling uses an exact integer sample range for 1–32-bit components (32 shifted past u32, and floating-point maxima at 25–31 bits wrapped to zero), and refuses zero-sized rows; (7) `decode_limit`: Flate (through flate2 and the fallback inflater), LZW and RunLength never write past `MAX_DECODED_STREAM` (256 MiB), or, for the last filter of an image of known size, past twice its pixel bytes (a predictor tag byte per byte), at most 1 GiB; `Stream::decoded_within` for budgeted callers; a page's `/Contents` array shares one `MAX_DECODED_STREAM` (a 2 MB page whose content stream or 8 × 8 image inflated to 2 GB took the renderer past 3 GB in under a second; naming the stream twenty times in `/Contents`, 5 GB); (8) high-precision numbers (jsPDF MediaBox coordinates with over 19 digits) parse without overflowing the u64 mantissa: fractional digits past it are dropped, integer digits past it scale the value (a real, kept finite); (9) large dictionary indexes are built on first use and shared across repeated reads within each PDF (#307); the validated-byte cache is bounded, reader contexts remain separate, repairs and inline/content dictionaries bypass sharing, and small dictionaries stay eager; (10) decoded object-stream slots publish once under contention, cold loading rejects recursive dependencies, and xref repair serializes without holding recursive read guards; this permits one parser per native render pool with private worker caches; (11) `sync.rs`: a poisoned lock is recovered instead of unwrapped (a worker panic no longer cascades into every reader of the same PDF), and xref repair takes a blocking write lock instead of asserting a non-blocking one; (12) `SegmentList` (`util.rs`) returns `None` for an out-of-range index instead of panicking; (13) a page's `/UserUnit` scales `render_dimensions` and `initial_transform`, so the page and everything on it are sized in points, as Acrobat shows them; it is read as Acrobat reads it (the page's own entry, not inherited; a number from 1 up, at most 75,000; anything else counts as 1), and a page without it keeps exactly the transform it had; (14) revision 2–4 passwords are tried as given and then in PDFDocEncoding (`legacy_passwords` in `crypto/mod.rs`, which resolves the upstream `TODO: Convert to PDFDocEncoding`): passwords arrive as UTF-8 text, so a password with characters outside ASCII (`Çok güzel`, `ılık€`) written in PDFDocEncoding, as ISO 32000-2 §7.6.4.3.2 asks and qpdf and Acrobat write it, was always wrong; (15) page content is read a piece at a time (`page::ContentReader`, `Page::content_reader`): plain streams and Flate streams without a predictor are decoded as they are read, with a budget of 64 times their compressed size and never less than `MAX_DECODED_STREAM` (a stream that expands past it is cut and reported); other streams are decoded whole, as before, and share `MAX_DECODED_STREAM`. The content tokenizer reads a window at a time (`UntypedIter::new_window`, `TypedIter::needs_more`, `TypedIter::committed`): an instruction that runs past the window is not decided until more data follows, which `Reader` records by flagging reads and `at_end` past the end of the data, including through the clones the parser makes. The Flate decoder keeps its own 32 KiB history, so it pauses at any output size. `ContentReader::new` reuses raw/decrypted stream bytes for budget accounting and piece creation, and `decoded_raw_within` applies the existing Whole-stream filters and limits to those bytes without decrypting twice; (16) revision 5–6 (AES-256) passwords are tried in their SASLprep form (RFC 4013, then UTF-8, at most 127 bytes, as ISO 32000-2 §7.6.4.3.3 Algorithm 2.A asks and pdfcraft-crypt writes them) and then, if that differs, as given (`aes256_passwords` in `crypto/mod.rs`, which resolves the upstream `TODO: Do the above.`; `stringprep`, already in the tree through pdfcraft-crypt, is an optional dependency enabled by `std`, and without `std` passwords are used as given): the renderer used the password as typed, so one that SASLprep changes (a decomposed `café`, a no-break space, the fullwidth `ａｂ` a Japanese IME types in full-width mode) was wrong, although pdfcraft-cos accepted it: protecting a document with such an open password was rolled back ("the password is incorrect"), and a file with such a password did not open with it.; (17) CCITT `/Rows 0` means the height is not predetermined (ISO 32000-2 Table 11), not zero rows: scanned Group 4 pages written that way rendered blank; (18) The streaming Flate decoder (`lzw_flate::flate::fallback`) copies a back-reference in runs of up to its distance (`extend_from_within`), not byte by byte. The copy reads only bytes already in the output, and a copy whose source does not exist still writes nothing, so the bytes are those a byte-by-byte copy writes. Inflating the A1 sample's content stream (278.6 MB) takes about 14% less time (0.66 s to 0.56 s); the output is identical; (19) CCITT `/Rows` below the image's `/Height` decodes to the height: a plotter driver wrote `/Rows` equal to `/Columns`, so only the first rows were decoded and the rest drawn black (#3308) | (1) `cos/tests/writer.rs::page_tree_loops_in_object_streams_do_not_crash_readers`; (2) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (3) `raster::tests::absurd_predictor_columns_and_line_widths_render`; (4) `raster::tests::inline_images_with_ei_in_their_data_parse_in_linear_time`; (5) `raster::tests::indirect_appearance_state_dict_is_not_another_stream`; (6) `raster::tests::jpx_stream_bit_depth_boundaries_do_not_panic`, `raster::tests::jpx_32_bit_image_does_not_prevent_other_page_content_from_rendering`, `raster::tests::jpx_gray_rgb_and_alpha_keep_eight_bit_samples`, `raster::tests::jpx_streams_reject_out_of_range_bit_depths`; (7) `raster::tests::decompression_bombs_stop_at_the_stream_limit`, `raster::tests::page_contents_and_nested_forms_share_one_budget`; (8) `inspect::tests::jspdf_high_precision_mediabox_does_not_overflow_dimensions`, `object::number::tests::high_precision_float_does_not_overflow_mantissa` (hayro-syntax); (9) `object::dict::lazy_index_tests`, `raster::tests::large_resource_indexes_preserve_pixels_and_text`; (10) `data::tests`, `xref::tests::concurrent_cold_object_stream_reads_all_succeed`, `raster::tests::shared_parser_is_lazy_and_initializes_once_under_contention`; (11) `data::tests::panicking_initializer_wakes_waiters_and_prevents_retries`, `xref::tests::concurrent_cold_object_stream_reads_all_succeed`; (12) `util::tests::segment_list_rejects_out_of_range_indices`; (13) `raster::tests::user_unit_scales_the_page_and_everything_on_it`; (14) `crypto::pdfcraft_password_tests`, `pdfcraft-engine` `tests::legacy_passwords_with_turkish_characters_open`; (15) `content::tests::windows_keep_every_instruction_at_every_split`, `content::tests::windows_keep_every_instruction_in_pieces_of_any_size`, `content::tests::windows_match_the_slice_on_random_content`, `content::tests::windows_drop_blanks_and_keep_a_comment_that_may_go_on`, `page::content_reader_tests::content_past_the_cap_is_read_to_the_end`, `page::content_reader_tests::a_stream_that_expands_past_the_budget_is_cut_there`, `page::content_reader_tests::a_small_expansion_under_the_cap_is_not_cut`, `raster::tests::page_content_past_the_old_cap_renders`, `raster::tests::content_reader_returns_the_same_bytes_in_any_pieces`, `raster::tests::content_bombs_stop_at_the_budget_and_say_so`, `page::content_reader_tests::plain_and_whole_pieces_keep_their_bytes`, `page::content_reader_tests::repeated_content_reference_counts_compressed_bytes_once`, `page::content_reader_tests::whole_cap_still_counts_later_unique_streams`; (16) `crypto::pdfcraft_password_tests::aes256_passwords_are_tried_in_their_saslprep_form_then_as_given`, `crypto::pdfcraft_password_tests::aes256_passwords_open_as_typed_and_in_their_saslprep_form`, `pdfcraft-engine` `tests::aes256_open_passwords_that_saslprep_changes_protect_and_reopen_as_typed`, `tests::aes256_files_whose_passwords_saslprep_changes_open_as_typed`; (17) `raster::tests::group4_fax_with_rows_zero_decodes_to_the_image_height`; (18) `filter::lzw_flate::tests::back_reference_longer_than_its_distance_repeats_its_source`, `back_reference_cut_by_the_limit_resumes_in_the_next_call` and `back_reference_reads_bytes_written_by_an_earlier_call`; each fails when the copy returns the wrong remaining count; (19) `raster::tests::group4_fax_with_rows_below_its_height_decodes_to_the_image_height`
| hayro-syntax | 0.7.2 | Apache-2.0 OR MIT | (1) Page-tree cycle guard in `resolve_pages` (visited set + depth cap): a `/Kids` loop among compressed objects overflowed the stack and aborted the process; (2) `ccitt_size_ok`: CCITT images wider than 2^20 columns or over 2^28 pixels are not decoded (a fuzzed `/Columns 4294967295` allocated 4 GiB); (3) Flate/LZW predictor row widths are saturating `usize`, a zero width is refused and a row longer than the data is not allocated (a fuzzed `/Columns 9223372036854775807` wrapped to a 2^61-byte allocation and aborted); (4) the inline-image end-of-data check (`EI`) doesn't nest inside its own look-ahead and examines at most 256 candidates and steps (50 images with "EI" in their data ran for over ten minutes); (5) a stream whose `/Length` is wrong is recovered only when `stream` is the next token — scanning ahead to a later object's `stream` made an indirect appearance-state dictionary (Quartz checkboxes: `/AP /N` → `<< /Yes … /Off … >>`) render as a non-form, so the widget was skipped; (6) JPX rescaling uses an exact integer sample range for 1–32-bit components (32 shifted past u32, and floating-point maxima at 25–31 bits wrapped to zero), and refuses zero-sized rows; (7) `decode_limit`: Flate (through flate2 and the fallback inflater), LZW, RunLength and ASCII85 never write past `MAX_DECODED_STREAM` (256 MiB), or, for the last filter of an image of known size, past twice its pixel bytes (a predictor tag byte per byte), at most 1 GiB; `Stream::decoded_within` for budgeted callers; a page's `/Contents` array shares one `MAX_DECODED_STREAM` (a 2 MB page whose content stream or 8 × 8 image inflated to 2 GB took the renderer past 3 GB in under a second; naming the stream twenty times in `/Contents`, 5 GB); (8) high-precision numbers (jsPDF MediaBox coordinates with over 19 digits) parse without overflowing the u64 mantissa: fractional digits past it are dropped, integer digits past it scale the value (a real, kept finite); (9) large dictionary indexes are built on first use and shared across repeated reads within each PDF (#307); the validated-byte cache is bounded, reader contexts remain separate, repairs and inline/content dictionaries bypass sharing, and small dictionaries stay eager; (10) decoded object-stream slots publish once under contention, cold loading rejects recursive dependencies, and xref repair serializes without holding recursive read guards; this permits one parser per native render pool with private worker caches; (11) `sync.rs`: a poisoned lock is recovered instead of unwrapped (a worker panic no longer cascades into every reader of the same PDF), and xref repair takes a blocking write lock instead of asserting a non-blocking one; (12) `SegmentList` (`util.rs`) returns `None` for an out-of-range index instead of panicking; (13) a page's `/UserUnit` scales `render_dimensions` and `initial_transform`, so the page and everything on it are sized in points, as Acrobat shows them; it is read as Acrobat reads it (the page's own entry, not inherited; a number from 1 up, at most 75,000; anything else counts as 1), and a page without it keeps exactly the transform it had; (14) revision 2–4 passwords are tried as given and then in PDFDocEncoding (`legacy_passwords` in `crypto/mod.rs`, which resolves the upstream `TODO: Convert to PDFDocEncoding`): passwords arrive as UTF-8 text, so a password with characters outside ASCII (`Çok güzel`, `ılık€`) written in PDFDocEncoding, as ISO 32000-2 §7.6.4.3.2 asks and qpdf and Acrobat write it, was always wrong; (15) page content is read a piece at a time (`page::ContentReader`, `Page::content_reader`): plain streams and Flate streams without a predictor are decoded as they are read, with a budget of 64 times their compressed size and never less than `MAX_DECODED_STREAM` (a stream that expands past it is cut and reported); other streams are decoded whole, as before, and share `MAX_DECODED_STREAM`. The content tokenizer reads a window at a time (`UntypedIter::new_window`, `TypedIter::needs_more`, `TypedIter::committed`): an instruction that runs past the window is not decided until more data follows, which `Reader` records by flagging reads and `at_end` past the end of the data, including through the clones the parser makes. The Flate decoder keeps its own 32 KiB history, so it pauses at any output size. `ContentReader::new` reuses raw/decrypted stream bytes for budget accounting and piece creation, and `decoded_raw_within` applies the existing Whole-stream filters and limits to those bytes without decrypting twice; (16) revision 5–6 (AES-256) passwords are tried in their SASLprep form (RFC 4013, then UTF-8, at most 127 bytes, as ISO 32000-2 §7.6.4.3.3 Algorithm 2.A asks and pdfcraft-crypt writes them) and then, if that differs, as given (`aes256_passwords` in `crypto/mod.rs`, which resolves the upstream `TODO: Do the above.`; `stringprep`, already in the tree through pdfcraft-crypt, is an optional dependency enabled by `std`, and without `std` passwords are used as given): the renderer used the password as typed, so one that SASLprep changes (a decomposed `café`, a no-break space, the fullwidth `ａｂ` a Japanese IME types in full-width mode) was wrong, although pdfcraft-cos accepted it: protecting a document with such an open password was rolled back ("the password is incorrect"), and a file with such a password did not open with it. | (1) `cos/tests/writer.rs::page_tree_loops_in_object_streams_do_not_crash_readers`; (2) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (3) `raster::tests::absurd_predictor_columns_and_line_widths_render`; (4) `raster::tests::inline_images_with_ei_in_their_data_parse_in_linear_time`; (5) `raster::tests::indirect_appearance_state_dict_is_not_another_stream`; (6) `raster::tests::jpx_stream_bit_depth_boundaries_do_not_panic`, `raster::tests::jpx_32_bit_image_does_not_prevent_other_page_content_from_rendering`, `raster::tests::jpx_gray_rgb_and_alpha_keep_eight_bit_samples`, `raster::tests::jpx_streams_reject_out_of_range_bit_depths`; (7) `raster::tests::decompression_bombs_stop_at_the_stream_limit`, `raster::tests::page_contents_and_nested_forms_share_one_budget`; (8) `inspect::tests::jspdf_high_precision_mediabox_does_not_overflow_dimensions`, `object::number::tests::high_precision_float_does_not_overflow_mantissa` (hayro-syntax); (9) `object::dict::lazy_index_tests`, `raster::tests::large_resource_indexes_preserve_pixels_and_text`; (10) `data::tests`, `xref::tests::concurrent_cold_object_stream_reads_all_succeed`, `raster::tests::shared_parser_is_lazy_and_initializes_once_under_contention`; (11) `data::tests::panicking_initializer_wakes_waiters_and_prevents_retries`, `xref::tests::concurrent_cold_object_stream_reads_all_succeed`; (12) `util::tests::segment_list_rejects_out_of_range_indices`; (13) `raster::tests::user_unit_scales_the_page_and_everything_on_it`; (14) `crypto::pdfcraft_password_tests`, `pdfcraft-engine` `tests::legacy_passwords_with_turkish_characters_open`; (15) `content::tests::windows_keep_every_instruction_at_every_split`, `content::tests::windows_keep_every_instruction_in_pieces_of_any_size`, `content::tests::windows_match_the_slice_on_random_content`, `content::tests::windows_drop_blanks_and_keep_a_comment_that_may_go_on`, `page::content_reader_tests::content_past_the_cap_is_read_to_the_end`, `page::content_reader_tests::a_stream_that_expands_past_the_budget_is_cut_there`, `page::content_reader_tests::a_small_expansion_under_the_cap_is_not_cut`, `raster::tests::page_content_past_the_old_cap_renders`, `raster::tests::content_reader_returns_the_same_bytes_in_any_pieces`, `raster::tests::content_bombs_stop_at_the_budget_and_say_so`, `page::content_reader_tests::plain_and_whole_pieces_keep_their_bytes`, `page::content_reader_tests::repeated_content_reference_counts_compressed_bytes_once`, `page::content_reader_tests::whole_cap_still_counts_later_unique_streams`; (16) `crypto::pdfcraft_password_tests::aes256_passwords_are_tried_in_their_saslprep_form_then_as_given`, `crypto::pdfcraft_password_tests::aes256_passwords_open_as_typed_and_in_their_saslprep_form`, `pdfcraft-engine` `tests::aes256_open_passwords_that_saslprep_changes_protect_and_reopen_as_typed`, `tests::aes256_files_whose_passwords_saslprep_changes_open_as_typed`
| lopdf | 0.45.0 | MIT | (1) PNG predictor rows longer than the data are refused before their buffers are allocated (a fuzzed `/Columns 4294967295` allocated two 4 GiB rows), and empty frames return without row buffers; (2) predictor row widths use saturating arithmetic or reject overflow; (3) sub-byte TIFF Predictor 2 scratch is bounded by the samples present in the input instead of the declared `/Colors`, aligned with [lopdf#589](https://github.com/J-F-Liu/lopdf/pull/589); (4) the object lookup keeps its cycle history through compressed objects, requires an ordinary in-file container, and caps the chain of indirect lookups at 16 (a chain of new objects, each a `/Length` compressed in a container whose own `/Length` names the next, or a stream whose `/Length` names the next stream, overflowed the stack: a dev build's 2 MiB thread within 800 hops); (5) object-stream members are parsed within their own extent (up to the next distinct offset), an entry whose offset repeats an earlier entry's is skipped (so a repeated object number can resolve to its earlier entry), and offsets are processed in index order | (1) `inspect::tests::png_predictor_rows_longer_than_the_data_are_refused_before_allocating`, `inspect::tests::empty_png_predictor_frames_do_not_allocate_rows`; (2) `inspect::tests::tiff_subbyte_predictor_row_width_overflow_is_refused`; (3) `inspect::tests::tiff_subbyte_predictor_scratch_is_bounded_by_available_samples`, `inspect::tests::tiff_subbyte_predictors_keep_components_rows_and_padding`; (4) `inspect::tests::compressed_object_self_container_does_not_recurse`, `inspect::tests::compressed_object_mutual_containers_do_not_recurse`, `inspect::tests::object_stream_length_cycle_does_not_recurse`, `inspect::tests::compressed_object_container_chain_does_not_recurse`, `inspect::tests::compressed_length_chain_does_not_recurse`, `inspect::tests::plain_stream_length_chain_does_not_recurse`, `inspect::tests::compressed_objects_and_indirect_stream_length_still_load`, `inspect::tests::compressed_length_in_a_container_with_an_indirect_length_still_loads`; (5) `inspect::tests::object_stream_member_extents_bound_owned_strings`, `inspect::tests::object_stream_member_extents_preserve_index_order_and_whitespace` |
| hayro-syntax | 0.7.2 | Apache-2.0 OR MIT | (1) Page-tree cycle guard in `resolve_pages` (visited set + depth cap): a `/Kids` loop among compressed objects overflowed the stack and aborted the process; (2) `ccitt_size_ok`: CCITT images wider than 2^20 columns or over 2^28 pixels are not decoded (a fuzzed `/Columns 4294967295` allocated 4 GiB); (3) Flate/LZW predictor row widths are saturating `usize`, a zero width is refused and a row longer than the data is not allocated (a fuzzed `/Columns 9223372036854775807` wrapped to a 2^61-byte allocation and aborted); (4) the inline-image end-of-data check (`EI`) doesn't nest inside its own look-ahead and examines at most 256 candidates and steps (50 images with "EI" in their data ran for over ten minutes); (5) a stream whose `/Length` is wrong is recovered only when `stream` is the next token — scanning ahead to a later object's `stream` made an indirect appearance-state dictionary (Quartz checkboxes: `/AP /N` → `<< /Yes … /Off … >>`) render as a non-form, so the widget was skipped; (6) JPX rescaling uses an exact integer sample range for 1–32-bit components (32 shifted past u32, and floating-point maxima at 25–31 bits wrapped to zero), and refuses zero-sized rows; (7) `decode_limit`: Flate (through flate2 and the fallback inflater), LZW and RunLength never write past `MAX_DECODED_STREAM` (256 MiB), or, for the last filter of an image of known size, past twice its pixel bytes (a predictor tag byte per byte), at most 1 GiB; `Stream::decoded_within` for budgeted callers; a page's `/Contents` array shares one `MAX_DECODED_STREAM` (a 2 MB page whose content stream or 8 × 8 image inflated to 2 GB took the renderer past 3 GB in under a second; naming the stream twenty times in `/Contents`, 5 GB); (8) high-precision numbers (jsPDF MediaBox coordinates with over 19 digits) parse without overflowing the u64 mantissa: fractional digits past it are dropped, integer digits past it scale the value (a real, kept finite); (9) large dictionary indexes are built on first use and shared across repeated reads within each PDF (#307); the validated-byte cache is bounded, reader contexts remain separate, repairs and inline/content dictionaries bypass sharing, and small dictionaries stay eager; (10) decoded object-stream slots publish once under contention, cold loading rejects recursive dependencies, and xref repair serializes without holding recursive read guards; this permits one parser per native render pool with private worker caches; (11) `sync.rs`: a poisoned lock is recovered instead of unwrapped (a worker panic no longer cascades into every reader of the same PDF), and xref repair takes a blocking write lock instead of asserting a non-blocking one; (12) `SegmentList` (`util.rs`) returns `None` for an out-of-range index instead of panicking; (13) a page's `/UserUnit` scales `render_dimensions` and `initial_transform`, so the page and everything on it are sized in points, as Acrobat shows them; it is read as Acrobat reads it (the page's own entry, not inherited; a number from 1 up, at most 75,000; anything else counts as 1), and a page without it keeps exactly the transform it had | (1) `cos/tests/writer.rs::page_tree_loops_in_object_streams_do_not_crash_readers`; (2) `raster::tests::absurd_mask_and_fax_sizes_are_refused_before_decoding`; (3) `raster::tests::absurd_predictor_columns_and_line_widths_render`; (4) `raster::tests::inline_images_with_ei_in_their_data_parse_in_linear_time`; (5) `raster::tests::indirect_appearance_state_dict_is_not_another_stream`; (6) `raster::tests::jpx_stream_bit_depth_boundaries_do_not_panic`, `raster::tests::jpx_32_bit_image_does_not_prevent_other_page_content_from_rendering`, `raster::tests::jpx_gray_rgb_and_alpha_keep_eight_bit_samples`, `raster::tests::jpx_streams_reject_out_of_range_bit_depths`; (7) `raster::tests::decompression_bombs_stop_at_the_stream_limit`, `raster::tests::page_contents_and_nested_forms_share_one_budget`; (8) `inspect::tests::jspdf_high_precision_mediabox_does_not_overflow_dimensions`, `object::number::tests::high_precision_float_does_not_overflow_mantissa` (hayro-syntax); (9) `object::dict::lazy_index_tests`, `raster::tests::large_resource_indexes_preserve_pixels_and_text`; (10) `data::tests`, `xref::tests::concurrent_cold_object_stream_reads_all_succeed`, `raster::tests::shared_parser_is_lazy_and_initializes_once_under_contention`; (11) `data::tests::panicking_initializer_wakes_waiters_and_prevents_retries`, `xref::tests::concurrent_cold_object_stream_reads_all_succeed`; (12) `util::tests::segment_list_rejects_out_of_range_indices`; (13) `raster::tests::user_unit_scales_the_page_and_everything_on_it` |
diff --git a/vendor/hayro-syntax/src/filter/ccitt.rs b/vendor/hayro-syntax/src/filter/ccitt.rs
index 687d087149..685d680d2f 100644
--- a/vendor/hayro-syntax/src/filter/ccitt.rs
+++ b/vendor/hayro-syntax/src/filter/ccitt.rs
@@ -17,10 +17,13 @@ pub(crate) fn decode(
// PdfCraft patch: `/Rows 0` means the height is not predetermined (ISO 32000-2 Table 11, and
// the default), not zero rows. Scanners write it on Group 4 pages, which rendered blank.
+ // PdfCraft patch (19): fewer `/Rows` than the image's `/Height` doesn't cut the image short: a
+ // plotter driver wrote `/Rows` equal to `/Columns`, and the rows after it were drawn black
+ // (#3308). The decoder still stops where the data ends.
let rows = params
.get::
