fix: recognize structured rate claim windows

[karlwaldman]

Summary

• replace phrase-order alternations with a bounded semantic detector
• require a non-duration numeric count, an API request/call/query/hit/credit noun, and a cadence/window token within one sentence span
• choose the closest count to the action and cap both action-count distance and total claim span
• strip generated HTML tags for detection while preserving sentence/code-comment boundaries
• keep numeric claims in one semantic classifier and preserve the legacy classifier only for nonnumeric monthly station-limit wording
• report every distinct finding in one file instead of hiding later matches behind the first match

Red-green proof

Requested red: the three final orderings failed on the prior head: 50 request limit per day, 50-call limit per day, and 50 requests allowed daily. The earlier four duration/order cases and seven permutation cases were also captured red before their fixes.

Review red: 50 API requests per month produced two classifications before c6fdd4a; the regression now requires exactly one finding.

Green:

• 31 positive rate claims spanning arbitrary counts, action nouns, duration windows, order permutations, hyphens, and generated markup
• 10 negative boundaries covering versions, tests, page records, retry counts, historical-range labels, code-comment timeouts, and unrelated sentences
• 46 focused storefront tests
• 458 passed / 13 skipped, 62.26% coverage on the complete CI-equivalent unit gate
• Ruff, mypy across 46 source files, and 63-surface authored/generated scan: green
• exact 1.12.4 wheel clean install, pip check, recursive RECORD scan, and keyless production demo: green

No v1.12.4 tag or PyPI artifact exists. Release will bind only the final replacement merge SHA after hosted and post-main verification.

[coderabbitai]

📝 Walkthrough

Walkthrough

The storefront claim validator now detects broader fixed-rate claim wording and reports every occurrence. Parameterized tests cover additional valid phrases and negative cases.

Changes

Fixed-rate claim validation

Layer / File(s)	Summary
Broader fixed-rate patterns scripts/validate_storefront_claims.py, tests/test_storefront_claims.py	The validator recognizes more request terms, time units, frequencies, windows, scopes, limits, quotas, and number formats. Tests cover valid and invalid phrases.
Report every matching occurrence scripts/validate_storefront_claims.py	Claim scanning iterates over all matches for each pattern and reports each occurrence.

Estimated code review effort: 3 (Moderate) | ~20 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: recognizing structured rate claim windows.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/cover-duration-rate-claims

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[karlwaldman]

@coderabbitai review

[coderabbitai]

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@scripts/validate_storefront_claims.py`:
- Around line 41-58: Deduplicate claim matches before reporting them so
identical (path, match.start(), match.end()) spans produce only one failure,
including overlapping FIXED_RATE classifications for “50 API requests per
month.” Update the result-collection logic around line 178 to retain one
occurrence per span, and add a regression test asserting this phrase yields
exactly one failure.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 3102e9ee-f918-40a1-a8d5-93f41a57df9e

📥 Commits

Reviewing files that changed from the base of the PR and between 6e3e543 and 2777d2f.

📒 Files selected for processing (2)

• scripts/validate_storefront_claims.py
• tests/test_storefront_claims.py