regex: accept literal [ in bracket expressions

[kevinburke]

POSIX allows '[' to represent itself inside a bracket expression unless it starts a character class, collating symbol, or equivalence class construct. The parser was consuming the following character after a literal '[', which made '[[]' look unterminated, and the Rust regex backend also rejects unescaped literal '[' in classes.

Reproduce the compatibility gap by comparing these commands:

printf 'x\n' | ./target/debug/sed -E 's/[[]/X/'

printf 'x\n' | gsed -E 's/[[]/X/'

printf 'x\n' | ./target/debug/sed -E 's/[^[]/X/'

printf 'x\n' | gsed -E 's/[^[]/X/'

Before this change, uutils sed rejected the scripts as unterminated or invalid regexes while GNU sed parsed them cleanly. After this change the outputs match GNU sed: x, X, x, and X for the reported cases.

Leave the following character available for normal class parsing, then escape literal '[' characters inside parsed classes before compiling with the regex backend. Add parser, compiler, and command-level regressions for the failing sed -E substitution cases.

Observed this while trying to install Python 3.14.5 with Pyenv and uutils sed on the PATH, which failed.

[codecov]

Codecov Report

❌ Patch coverage is 98.24561% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.50%. Comparing base (d842aa3) to head (4bb5fa0).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/sed/compiler.rs	97.77%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #439      +/-   ##
==========================================
+ Coverage   83.27%   83.50%   +0.23%     
==========================================
  Files          13       13              
  Lines        6994     7106     +112     
  Branches      397      406       +9     
==========================================
+ Hits         5824     5934     +110     
- Misses       1167     1169       +2     
  Partials        3        3              

Flag	Coverage Δ
macos_latest	84.15% <98.24%> (+0.23%)	⬆️
ubuntu_latest	84.34% <98.24%> (+0.22%)	⬆️
windows_latest	0.00% <0.00%> (ø)

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.

[codspeed-hq]

Merging this PR will degrade performance by 2.91%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 2 regressed benchmarks
✅ 9 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	access_log_subst	2.3 s	2.4 s	-3.24%
❌	number_fix	1.7 s	1.7 s	-2.59%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing kevinburke:fix-literal-open-bracket-class (4bb5fa0) with main (d842aa3)}

[sylvestre]

please add a rustdoc comment here :)

[kevinburke]

Done

[sylvestre]

Well it is a bit long now :/

[kevinburke]

Oh, sorry, ok.

[kevinburke]

How is this