fix(refactor): exclude Storybook-only evidence from production liveness

[azizur100389]

Linked issue

Closes #902.

What & why

A component used only by its Storybook story currently appears live. Classify CSF story filenames through the shared parser file classifier and exclude story-only calls, imports, references, TESTED_BY evidence and class-member calls from dead-code liveness. Path checks cover previously built graphs too. Preserve genuine production callers, including the bare-call fallback, and ordinary test evidence.

How it was tested

Applied the same regression/affected-module/full-baseline methodology used in my previous PRs (#274, #276, #279, #353 and #354), against unchanged staging 6b12d11625cbec3b6773e076cb3d136464fa90e5.

• Final regression module: 11 passed. Bugs reproduce on unchanged staging; compatibility guards cover existing behavior.
• Affected modules: 255 passed, 1 existing Windows failure. Failure identities are a subset of the unchanged-staging affected baseline (1,169 passed, 15 Windows failures); no new failures.
• Ruff clean; Bandit zero findings; git diff --check clean.
• Mypy: the same three pre-existing Windows fcntl LOCK_EX/LOCK_NB/flock errors in daemon.py as staging; no new errors.
• Schema remains version 13.

python -m pytest tests/test_storybook_dead_code.py tests/test_refactor.py tests/test_test_file_detection.py -q --tb=short
$env:GIT_CEILING_DIRECTORIES = $env:TEMP.Replace('\\', '/') + ';' + (Split-Path $PWD).Replace('\\', '/')
python -m pytest -q --tb=short -m 'not browser and not upgrade' --cov=code_review_graph --cov-report=term --cov-fail-under=65
ruff check code_review_graph tests/test_storybook_dead_code.py
mypy code_review_graph --ignore-missing-imports --no-strict-optional
bandit -r code_review_graph -c pyproject.toml

GitHub CI: 17 applicable checks passed on the final commit, including Python 3.10–3.13, browser tests, native Windows tests, and the three MCP-client platforms. Python 3.13: 4,279 passed, 804 skipped, 46 deselected, 2 xfailed, 2 xpassed; 85.87% coverage (65% required). The manual upgrade job is skipped by design.

Full local Windows suite: 4,153 passed, 111 failed, 819 skipped, 46 deselected, 2 xfailed, 2 xpassed; 86% coverage. Unchanged staging full baseline: 4,143 passed, 110 failed. The extra daemon-status assertion rejects the word dead anywhere in its printed output, including the temporary directory name crg-fix-storybook-dead-code-retry. The reported watcher is alive. The same test fails on unchanged staging under a path containing dead; it passes 3/3 neutral-path reruns on this branch and 3/3 neutral-path reruns on unchanged staging. This is a path-sensitive existing test assertion, not changed daemon behavior. An earlier parallel local attempt ended without a final report in the watch tests. This complete rerun used per-test diagnostics and short temporary paths; the earlier partial log is retained. PATH/PYTHONPATH select the isolated checkout, and Git ceilings keep fixture repositories away from the enclosing home Git repository. Browser tests passed in CI; browser and opt-in upgrade tests were excluded locally.

All ten issue fixes were additionally tested together: 283 regression checks passed, and the combined affected suite had 1,142 passed / 25 unchanged-staging Windows failures, with no new failure identities. The tested combination is preserved in this integration branch.

Merge coordination: this PR and #1082 both change find_dead_code. If #1082 lands first, retain its single streaming edge scan and exclude Storybook rows before adding them to the evidence maps. The integration branch contains the tested resolution (the shared regex import plus if _STORYBOOK_FILE_RE.search(file_path): continue in that scan). Both the Storybook and constant-query-count regressions pass together.

[github-actions]

code-review-graph review

Overall risk: 0.55 (MEDIUM) — 6 changed function(s)/class(es), 47 affected flow(s), 0 test gap(s)

Risk-scored changes

Risk	Level	Symbol	Location	Tested
0.55	medium	code_review_graph/refactor.py::find_dead_code	code_review_graph/refactor.py:236	yes
0.05	low	tests/test_storybook_dead_code.py::test_story_filename_is_nonproduction	tests/test_storybook_dead_code.py:13	(test)
0.05	low	tests/test_storybook_dead_code.py::test_story_like_production_paths_remain_production	tests/test_storybook_dead_code.py:21	(test)
0.05	low	tests/test_storybook_dead_code.py::test_only_story_callers_do_not_keep_component_alive	tests/test_storybook_dead_code.py:26	(test)
0.05	low	tests/test_storybook_dead_code.py::test_story_call_does_not_hide_bare_production_caller_fallback	tests/test_storybook_dead_code.py:43	(test)
0.05	low	tests/test_storybook_dead_code.py::test_story_member_calls_do_not_keep_class_alive	tests/test_storybook_dead_code.py:55	(test)

Affected execution flows

• process — criticality 0.74, 538 node(s) across 14 file(s)
• _run — criticality 0.74, 565 node(s) across 16 file(s)
• build_or_update_graph_tool — criticality 0.73, 617 node(s) across 24 file(s)
• main — criticality 0.72, 923 node(s) across 40 file(s)
• run_token_benchmark — criticality 0.70, 38 node(s) across 5 file(s)
• ...and 42 more affected flow(s)

---

Powered by code-review-graph — local-first analysis; no code leaves the CI runner.