wcag/ChunkParser: scale Type3 glyph advances by /FontMatrix, not hardcoded 1/1000

[augustovillar]

Problem

ChunkParser#parseString computes each glyph's text-space advance as:

width = width * getTextState().getTextFontSize() / 1000 * getTextState().getHorizontalScaling();

PDFont#getWidth(code) returns the raw /Widths value in the font's glyph space. For most fonts glyph space is 1/1000 text units, so dividing by 1000 is correct — but Type3 fonts define their own glyph space via /FontMatrix. For a Type3 font whose /FontMatrix is e.g. [.0004883 0 0 -.0004883 0 0] (= 1/2048), dividing by 1000 instead of by 2048 makes every advance ~2.048× too large, corrupting TextPiece end positions and the resulting text-chunk bounding boxes.

Impact

Consumers of these metrics mis-assemble text. In opendataloader-pdf (which bundles this module), real-world Type3 PDFs drop a duplicate adjacent glyph and gain a spurious space — 1.66E-2 is extracted as 1. 6E-2, Programme as Progra me. Downstream report: opendataloader-project/opendataloader-pdf#578

This follows up #722 (which refactored this method but kept the hardcoded / 1000); same spirit as veraPDF-parser #511 (CFF CID FontMatrix).

Fix

Add a getGlyphWidthToTextSpaceScale(PDFont) helper that returns the horizontal /FontMatrix component (fontMatrix[0]) for a PDType3Font (null/length/zero-guarded), and the unchanged 1/1000 for every other font; the factor is hoisted out of the per-code loop. The TJ-array displacement path (obj.getReal() / 1000, which is text-space thousandths and font-independent) is intentionally left unchanged.

Verification

• Compiles on the current integration branch (mvn -pl wcag-validation -am compile, BUILD SUCCESS).
• End-to-end through opendataloader-pdf 2.4.7 (which bundles wcag-validation 1.31.65) on a public real-world Type3 PDF (EPD epd28600): before → 1. 6E-2, 34 spacing artifacts; after → 1.66E-2, 0 artifacts. Same toolchain, only this patch toggled.
• Non-Type3 (Type1) PDFs produce byte-identical output — the instanceof PDType3Font guard leaves the default path untouched — and previously-correct values are unchanged.

Summary by CodeRabbit

• Bug Fixes
  • Improved text glyph width calculation accuracy in PDF validation, particularly for Type3 fonts, resulting in more precise text positioning and rendering measurements.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 051e88f5-da97-4e91-bddd-ed37aecae34a

📥 Commits

Reviewing files that changed from the base of the PR and between dc5bd35 and 6fae194.

📒 Files selected for processing (1)

• wcag-validation/src/main/java/org/verapdf/gf/model/factory/chunks/ChunkParser.java

---

📝 Walkthrough

Walkthrough

ChunkParser.parseString replaces the hardcoded 1/1000 glyph-width divisor with a computed glyphSpaceToTextSpace factor. A new private helper getGlyphWidthToTextSpaceScale(PDFont font) returns a scale derived from the Type3 font matrix when applicable, falling back to 1/1000 for other font types.

Changes

Type3 Font Glyph Width Scaling

Layer / File(s)	Summary
Type3 glyph width scaling helper and parseString integration wcag-validation/src/main/java/org/verapdf/gf/model/factory/chunks/ChunkParser.java	Adds PDFont and PDType3Font imports. Introduces getGlyphWidthToTextSpaceScale(PDFont font) to return a Type3 font matrix-derived scale or 1/1000 as fallback. Updates parseString to compute glyphSpaceToTextSpace via that helper and apply it when calculating each glyph's rendered width and text-piece advance.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• veraPDF/veraPDF-validation#722: Directly related — both PRs modify glyph width calculation logic in ChunkParser.parseString, with this PR extending that work to handle Type3 font matrix scaling.

Suggested reviewers

• MaximPlusov

Poem

🐇 A font named Type3 came knocking one day,
Its glyphs had their own special matrix to weigh.
No more fixed one-thousandth, a true scale we seek,
The rabbit derived it — peak-to-peak!
TextPieces now land where they rightfully play. ✨

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately and specifically describes the main change: fixing Type3 font glyph advance scaling by using /FontMatrix instead of hardcoded 1/1000.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[MaximPlusov]

@augustovillar Since the font matrix should also be used in other places, I created a separate PR that incorporates your code. Thank you for your contribution!