Greptile Summary

This PR fixes \sout (strikethrough) so it works in text mode (e.g. \text{\sout{abc}}), which matches the behavior of LaTeX's ulem package where \sout is strictly a text-mode command. The change is backward compatible: math-mode usage (\sout{abc}) continues to work when strict is unset or false, while strict-mode users now get a proper warning/error indicating non-standard usage.

Key changes:

• \sout is extracted from the \cancel/\bcancel/\xcancel/\phase group into its own defineFunction block with allowedInText: true and a reportNonstrict call for math-mode usage under strict settings
• Existing build tests for \sout in math mode are updated to use nonstrictSettings, and a new test verifies that strict math-mode parsing fails
• Screenshotter fixtures and reference images updated to use \text{\sout{...}}
• Documentation and a11y tests updated to reflect the canonical text-mode usage
• The reportNonstrict guard uses an outer parser.settings.strict check that is slightly inconsistent with how other functions (accent.ts, kern.ts) call reportNonstrict directly without a guard — though functionally there is no difference

Confidence Score: 4/5

• Safe to merge — backward compatible, well-tested for the new strict-mode path, and consistent with LaTeX semantics.
• The core implementation is correct: \sout is allowed in text mode and continues working in non-strict math mode (no breaking change). The strict-mode failure path is unit-tested. The only notable gaps are the redundant strict guard style inconsistency (P2) and no dedicated parse/build unit test for the text-mode path specifically (covered visually by screenshotter and a11y tests). Neither of these blocks merge.
• src/functions/enclose.ts — minor guard style inconsistency on the reportNonstrict call

Important Files Changed

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["\\sout{...} encountered"] --> B{parser.mode?}
    B -->|text| C["handler: no strict check\nReturn enclose node"]
    B -->|math| D{parser.settings.strict?}
    D -->|false / falsy| E["Silently return enclose node\n(non-breaking backward compat)"]
    D -->|'warn'| F["reportNonstrict → console.warn\nReturn enclose node"]
    D -->|true / 'error'| G["reportNonstrict → throw ParseError\n(\\sout not valid in math mode)"]
    D -->|'ignore'| H["reportNonstrict → returns early\nReturn enclose node"]
    C --> I[htmlBuilder / mathmlBuilder\nshared with cancel group]
    E --> I
    F --> I
    H --> I

_{Reviews (2): Last reviewed commit: "Merge branch 'main' into sout" | Re-trigger Greptile}

Thanks! It looks like your changes are from an old commit though.

@edemaine, If you would prefer to put math mode behind a strict setting, let me know. I would prefer to avoid a breaking change.

I would prefer to match LaTeX's behavior, yes. So math mode should trigger strict, unless some LaTeX package adds it. I don't think it's breaking given that the default strict behavior is warn.

The latest commit writes an error message if \sout is invoked while in math mode. I've also written a couple of unit tests to verify the desired behavior.

I've edited the screenshotter input to put the \sout in text mode, so the letters should now be upright instead of italic.

Unfortunately, screenshotter is failing and I am also getting an error message that there are merge conflicts. Both failures are inscrutable to me. I think that I have taken this PR as far as I can.

OK, I can pick it up from here. Next time, synchronize your fork and pull the latest main before making a branch.

I did pull the latest main before running git branch sout. Before I did that, I did not check if a sout branch already existed. The message about sout being an old branch surprised me. Still confuses me.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.30%. Comparing base (314f856) to head (dd4c465).
⚠️ Report is 105 commits behind head on main.

@@            Coverage Diff             @@
##             main    #4173      +/-   ##
==========================================
+ Coverage   93.15%   93.30%   +0.14%     
==========================================
  Files          91       91              
  Lines        6795     6810      +15     
  Branches     1582     1589       +7     
==========================================
+ Hits         6330     6354      +24     
- Misses        429      454      +25     
+ Partials       36        2      -34     

Looks good!

@greptileai Another review?

By the way, I noticed that Greptile now edits the original review instead of posting a new one. Took me a while to find. 🙂

Thanks @ronkok for implementing this!

🎉 This PR is included in version 0.16.41 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀