[andrewfader]

Fixes #22370

The state of an XRef table/stream is kept, so that parsing can resume after a MissingDataException. However, when parsing failed for any other reason the state was left behind and then used for the next XRef table/stream.

In a linearized file where the linearization dictionary and the first-page XRef stream were overwritten, startxref points to the hint stream which is parsed as an XRef stream and fails, since it has no /W entry. The main XRef stream was then read with the hint stream's state and failed as well, which lost all of its compressed entries, hence "Invalid Root reference." even in recovery mode.

The new unit test reproduces this with issue22370.pdf, which is empty.pdf with its linearization dictionary and first-page XRef stream overwritten by spaces. It fails on master with "Invalid Root reference." and passes with this patch.

[Snuffleupagus]

Assuming that it passes all tests, this is probably a generally reasonable change.

[github-actions]

Viewer preview

🗑️ Viewer previews removed.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.58%. Comparing base (743102c) to head (4794c39).
⚠️ Report is 6 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #22375      +/-   ##
==========================================
- Coverage   90.58%   90.58%   -0.01%     
==========================================
  Files         278      278              
  Lines       68573    68586      +13     
==========================================
+ Hits        62117    62126       +9     
- Misses       6456     6460       +4     

Flag	Coverage Δ
browsertest	66.26% <100.00%> (-0.03%)	⬇️
fonttest	8.99% <ø> (ø)
integrationtest	69.59% <0.00%> (+0.04%)	⬆️
unittest	61.27% <100.00%> (+0.01%)	⬆️
unittestcli	59.00% <100.00%> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[codspeed]

Merging this PR will improve performance by 16.28%

⚡ 2 improved benchmarks
✅ 86 untouched benchmarks
⏩ 174 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	getUnicodeRangeFor (bug1473809.pdf)	33.4 ms	27.2 ms	+22.69%
⚡	PartialEvaluator.buildPath: lines (bug1743245.pdf)	18.4 ms	16.7 ms	+10.21%

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing andrewfader:xref-discard-corrupt-state (4794c39) with master (ab3819f)}

Footnotes

1. 174 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[github-actions]

Reference tests

Results for 4794c39 (run, references from mozilla/pdf.js.refs@46378a2139).

Platform	Status	Tests	Total runtime	Errors	Different	FBF	No reference	Report
	⚠️	1413	13m 1s	0	10	0	0	view differences
	⚠️	1413	16m 52s	0	11	0	0	view differences

The report is kept for 30 days and replaced by the next run.

[Snuffleupagus]

r=me, with merge conflicts fixed; thanks.

[Snuffleupagus]

The commits must be squashed (and merge commits aren't allowed in PRs).