rtmx-ai / rtmx / 27980796614

83%

push

github

fix(from-pytest): capture req-ids with optional lowercase suffix (#127)

* feat(from-pytest): accept multiple test paths + path-qualified JUnit join (REQ-LANG-004)

`rtmx from-pytest` scanned and ran only a single test path (default `tests`), so
a multi-package layout — a top-level `tests/` plus `packages/*/tests/` — could
not be reconciled in one run: the other packages were silently never scored
(their requirements stayed MISSING even with passing tests).

- Accept one or more test paths (variadic args); scan and run all of them.
- Join results to markers by repo-relative FILE PATH first, falling back to the
  bare function name, so a passing test in one file can no longer be
  mis-attributed to a same-named test in another file (a real hazard once a
  multi-package tree is scanned).

Tests cover multi-file same-name disambiguation. Verified end-to-end against a
multi-package project: requirements in packages/*/tests now score with full
scope/technique dimensions.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(from-pytest): capture req-ids with optional lowercase suffix (REQ-VERIFY-011)

The pytest/multi-language marker scanners captured req-ids with an uppercase-only
char class (REQ-[A-Z0-9-]+), so an id with a trailing lowercase letter
(REQ-HW-STRUCT-002c, REQ-HW-RF-001b -- the decomposition-child convention) failed
the whole @pytest.mark.req(...) match and was SILENTLY DROPPED. verify already
honored the configurable req_id_pattern, but from-pytest capture did not.

- Broaden capture char-classes to REQ-[A-Za-z0-9-]+ across all language scanners
  (from_tests.go + from_tests_langs.go, 44 patterns).
- Accept an OPTIONAL trailing lowercase letter in the default validation grammar
  (markers/spec.go + results/schema.go): ...-[0-9]+[a-z]?$.
- Tests: lowercase-suffix capture regression + accept/reject cases.
- .gitignore: ignore local Python tooling artifacts (venv/, .coverage, .hypothesis/).

Manually validated (pre-commit tool absen... (continued)

44 of 44 new or added lines in 2 files covered. (100.0%)

16473 of 19866 relevant lines covered (82.92%)

29.79 hits per line