| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| langroid-0.67.8-py3-none-any.whl | 2026-09-12 | 486.9 kB | |
| langroid-0.67.8.tar.gz | 2026-09-12 | 434.6 kB | |
| 0.67.8 source code.tar.gz | 2026-09-12 | 59.2 MB | |
| 0.67.8 source code.zip | 2026-09-12 | 59.7 MB | |
| README.md | 2026-09-12 | 959 Bytes | |
| Totals: 5 Items | 119.9 MB | 0 | |
Task control: done_sequences patterns with commas and brackets (#1135)
Done-sequence string patterns were split on every comma, so a content-match
regex containing a comma — e.g. "T,A,C[done,\s*thanks]" — was broken into
fragments and could never match. The parser now splits only on top-level
commas, tracking bracket depth and honoring:
- escaped characters (
\],\,) insideC[...]andT[...]parameters - regex character classes (
C[[a-z,]+]) and literal closing brackets - newlines inside regex patterns
Any TaskConfig(done_sequences=[...]) that previously had to avoid commas
in regexes can now use them directly. Two new test modules cover the comma
and escaped-bracket cases.
Thanks to @yhz5613813 for reporting the regression and contributing the original fix (#1134), which this release supersedes with the additional bracket/escape handling.
Full changelog: https://github.com/langroid/langroid/compare/0.67.7...0.67.8