[rahul05ranjan]

Fixes #4283

Bug

Layout mode text extraction crashed with an internal TypeError when a content stream contained a Tm (or cm) operator with extra operands on the operand stack:

# content stream: BT /F1 12 Tf 72 720 1 0 0 1 100 600 Tm (TmText) Tj ET
reader.pages[0].extract_text(extraction_mode="layout")
# TypeError: TextStateManager.new_transform() takes from 0 to 6 positional arguments but 8 ...

Root cause

• TextStateManager.add_tm/add_trm pass operands through _complete_matrix, which pads a 2-operand Td-style set but does not trim operand sets longer than 6, so new_transform() receives more than its 6 matrix parameters.
• TextStateManager.add_cm splats *args straight into new_transform().

Per the PDF operand-stack model, a matrix operator consumes the 6 operands nearest to it; the extras are malformed-stream noise and should not overflow the transform API. This is the same class of operand hardening as #4232 (Tz without operand) and #4233 (TD with a missing/non-numeric ty).

Fix

_complete_matrix now returns only the last 6 operands, and add_cm splats only args[-6:], so Tm/cm consume the 6 operands nearest to the operator. Valid operand counts are unaffected.

Verification (stdlib-only harness on the Pi, Python 3.13.5, main @ c7d391c)

Hand-built one-page PDF (correct xref) with the following content streams, layout mode, before → after:

• BT /F1 12 Tf 72 720 1 0 0 1 100 600 Tm (TmText) Tj ET (8 operands): TypeError crash → 'TmText' ✓ (matrix taken from the nearest 6 operands: 1 0 0 1 100 600)
• BT /F1 12 Tf 9 1 0 0 1 100 600 Tm (TmText) Tj ET (7 operands): crash → 'TmText' ✓
• BT /F1 12 Tf 5 5 1 0 0 1 50 50 cm 72 720 Td (CmText) Tj ET (cm with 7 operands): crash → 'CmText' ✓
• Valid 6-operand Tm: 'TmText' before and after (unchanged) ✓
• Valid Td translation: 'Second' before and after (unchanged) ✓

Tests: pytest tests/text_extraction/ -q → 6 passed. pytest tests/test_text_extraction.py -q → 108 passed, 1 xpassed, 2 pre-existing failures (test_multi_language, FileNotFoundError: sample-files/015-arabic/habibi-rotated.pdf — the sample-files submodule is absent in a shallow clone, unrelated to this change).

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.55%. Comparing base (c7d391c) to head (62df42e).
⚠️ Report is 14 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4284      +/-   ##
==========================================
+ Coverage   98.41%   98.55%   +0.14%     
==========================================
  Files          59       59              
  Lines       11995    11992       -3     
  Branches     2270     2268       -2     
==========================================
+ Hits        11805    11819      +14     
+ Misses        103       93      -10     
+ Partials       87       80       -7     

☔ 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.

[stefan6419846]

Please add tests. Why are these changes part of #4286 as well?

[rahul05ranjan]

Thanks for the review! Tests added in 62df42e — two layout-mode cases covering extra operands before both Tm and cm, asserting the extracted text matches the equivalent clean 6-operand stream. Both reproduce the original TypeError when run against the pre-fix code.

On your question — why these changes are part of #4286 as well: the set_page_label branch (fix/page-label-start-none) was cut from this PR's branch, so it carries this Tm/cm commit (d70431ae) beneath its own commit. I'll rebase #4286 onto main so it contains only the set_page_label change, leaving this PR standalone.

[rahul05ranjan]

Thanks for merging #4284 (ROB: Tolerate extra operands before Tm/cm in layout mode)! 🎉

By the way, I'm currently open to new opportunities (Full Stack (PHP, NodeJs, Python) / Data Science / Applied AI / Devtools, based in India). If you or anyone around is hiring, I'd love to connect:
LinkedIn: https://www.linkedin.com/in/rahulswt7/ · GitHub: https://github.com/rahul05ranjan