[KayhanB21]

A coverage_map shard written from its package directory now matches, and a coverage report that matches no scanned file says so. This is PR 2 of 0.3.10, based on master after #68.

The findings

These were reproduced on c1f64d4. The fixture is a monorepo with packages/a and packages/b, a [tool.riskratchet.coverage_map] table with one coverage file (a shard) per package, and each shard written by pytest --cov run from the package directory.

• The shards never match. A shard's keys are src/amod.py. MultiCoverageData.lookup picks the shard by prefix and then asks it for packages/a/src/amod.py. Every function scores 0% line coverage.
• Nothing says so. coverage_status reads present whenever a report loads. check prints No risk regressions detected., the PR comment shows 0% on every row, and only stderr has one warning per file. doctor reports the case as coverage-overlap, and no other command does.

Together these mean a monorepo that follows the README has a gate that ignores coverage: a PR that deletes every test moves no score.

The fix

• MultiCoverageData.lookup tries the repository-relative path first, then the path relative to the shard's prefix. The second lookup is exact (CoverageData.lookup_exact), so packages/alpha/core.py can't pick up an unrelated vendored/core.py by basename.
• RiskReport gains coverage_checked_files and coverage_unmatched_files: the Python files with at least one function that the engine looked up in a present report, and how many had no entry. DiffReport carries the resulting note, because the check renderers get the diff and never the report.
• When no file matched, scan, baseline, check, diff, explain, and the pytest plugin print one warning. The scan summary line, the markdown report's **Coverage:** line, and the Baseline: line of every check format carry coverage matched 0 of N scanned files.
• A partial match prints nothing, so no existing snapshot moves.

What changes for an adopter

Scores drop in a repository whose shards matched nothing: coverage_gap and branch_gap fall to their real values. check reports those functions as improved, so no gate fails. The CHANGELOG states this under Changed and says to regenerate the baseline after upgrading.

The following don't change: the gate verdict, exit codes, SCORING_MODEL_VERSION, and all nine JSON schemas. coverage_status stays present, and the two new fields aren't serialized. The note is a count, so it's safe under redaction.

Tests

18 new tests in tests/test_coverage.py, tests/test_coverage_overlap.py, and tests/test_pytest_plugin.py. Against the v0.3.9 src/, 5 fail directly and the 10 in tests/test_coverage_overlap.py fail at import. The other 3 guard behavior that must stay the same: the repository-relative key wins, a path outside every prefix doesn't match, and an empty prefix has no second spelling. The suite has 1391 passed and 2 skipped, with 93.28% coverage.

Baseline bump rationale

regenerate-baseline.yml regenerated the baseline in CI (run 37102003419, commit fdb19bd). It goes from 567 to 574 entries, and the scoring block doesn't change. 29 of the 567 existing entries change score. The largest rise is +3.25, under fail_regression_above = 5.

• New (7 functions): all score 5.8 or less. The highest is models.py::coverage_overlap_note at 5.8.
• Churn only (23 entries): this PR shifts spans in cli.py, engine.py, models.py, and summary.py.
• Edited here (6 functions): MultiCoverageData.lookup +3.25 (one more branch for the second lookup), engine.py::analyze +3.12 (the two counters), pytest_sessionfinish +2.39, diff +1.25, _summary_line +0.32, and baseline_line +0.19.
• This repository uses one coverage file, so the lookup change moves none of its own scores.

🤖 Generated with Claude Code

[github-actions]

riskratchet

Regressions: 2 · New above threshold: 0 · Regressed: 0 · Existing above threshold: 0 · Component regressed: 2 · Above threshold: 0
Since the baseline: Changed: 2 · New: 7 · Ambiguous renames: 0 · Improved: 0 · Moved: 0 · Removed: 0
Baseline: 567 entries · 567 compared · 0 not seen this run

Kind	Function	Before	After	Delta	Reason
component_regressed	src/riskratchet/coverage.py::MultiCoverageData.lookup	7.2	10.5	+3.2	churn grew by +20.0 (from 10.0 to 30.0); component tolerance is +15.0
component_regressed	src/riskratchet/coverage.py::MultiCoverageData.from_map	1.2	3.2	+2.0	churn grew by +20.0 (from 0.0 to 20.0); component tolerance is +15.0

New functions (7)

Status	Function	Before	After	Delta	Reason
new	src/riskratchet/models.py::coverage_overlap_note	n/a	5.8	n/a	function is absent from baseline with score 5.8
new	src/riskratchet/coverage.py::_prefix_relative_hit	n/a	3.5	n/a	function is absent from baseline with score 3.5
new	src/riskratchet/cli.py::_warned_about_unmatched_coverage	n/a	2.2	n/a	function is absent from baseline with score 2.2
new	src/riskratchet/models.py::RiskReport.coverage_warning	n/a	2.2	n/a	function is absent from baseline with score 2.2
new	src/riskratchet/reporting/markdown.py::_coverage_note_suffix	n/a	2.2	n/a	function is absent from baseline with score 2.2
new	src/riskratchet/coverage.py::CoverageData.lookup_exact	n/a	1.0	n/a	function is absent from baseline with score 1.0
new	src/riskratchet/models.py::RiskReport.coverage_note	n/a	1.0	n/a	function is absent from baseline with score 1.0

Unchanged functions (565)

Status	Function	Before	After	Delta	Reason
unchanged	src/riskratchet/cli.py::diff	37.5	37.5	+0.0	risk unchanged at 37.5
unchanged	src/riskratchet/typescript.py::_anon_class_name	33.8	33.8	+0.0	risk unchanged at 33.8
unchanged	src/riskratchet/cli.py::scan	32.5	32.5	+0.0	risk unchanged at 32.5
unchanged	src/riskratchet/pytest_plugin.py::_guarded_typescript_identity	31.5	31.5	+0.0	risk unchanged at 31.5
unchanged	src/riskratchet/cli.py::explain	27.5	27.5	+0.0	risk unchanged at 27.5
unchanged	src/riskratchet/diagnostics.py::Diagnostics.to_lines	26.8	26.8	+0.0	risk unchanged at 26.8
unchanged	src/riskratchet/config.py::_resolve_coverage	23.8	23.8	+0.0	risk unchanged at 23.8
unchanged	src/riskratchet/reporting/markdown.py::render_report_pr_comment	23.8	23.8	+0.0	risk unchanged at 23.8
unchanged	src/riskratchet/config.py::_load_config_strict	23.5	23.5	+0.0	risk unchanged at 23.5
unchanged	src/riskratchet/cli.py::baseline	23.2	23.2	+0.0	risk unchanged at 23.2
unchanged	src/riskratchet/pytest_plugin.py::_typescript_override	23.0	23.0	+0.0	risk unchanged at 23.0
unchanged	src/riskratchet/pytest_plugin.py::_report_or_fail	22.8	22.8	+0.0	risk unchanged at 22.8
unchanged	src/riskratchet/git.py::_normalize_diff_path	22.5	22.5	+0.0	risk unchanged at 22.5
unchanged	src/riskratchet/typescript_engine.py::_count_lines	22.5	22.5	+0.0	risk unchanged at 22.5
unchanged	src/riskratchet/cli.py::_should_run_baseline	22.2	22.2	+0.0	risk unchanged at 22.2
unchanged	src/riskratchet/baseline/io.py::runtime_scoring_block	21.0	21.0	+0.0	risk unchanged at 21.0
unchanged	src/riskratchet/baseline/io.py::_weight_reasons	20.8	20.8	+0.0	risk unchanged at 20.8
unchanged	src/riskratchet/baseline/regressions.py::regressions_from_diff	20.6	20.6	+0.0	risk unchanged at 20.6
unchanged	src/riskratchet/baseline/compare.py::compare	20.1	20.1	+0.0	risk unchanged at 20.1
unchanged	src/riskratchet/cli.py::config_show	20.0	20.0	+0.0	risk unchanged at 20.0
... 545 more hidden.