test(infra) Barrels of Fun scraper-pipeline integration tests

[jkeeley2073]

Summary

Backfill of the PR #41 family-wide test-infra template for BofProductScraper. 5 new scraper-pipeline integration tests pinning yield order, provenance propagation, polite-gate routing invariants, and failure-isolation behavior.

BofProductScraper is a single-yield scraper (yields .Game only — no .Link items, since BoF's WooCommerce product pages host no firmware/document downloads), so the yield-order assertion is over .Game only. Otherwise the test class mirrors the CGC proof-of-concept exactly:

• ScrapeAsync_HappyPath_YieldsGamesInIndexOrderWithProvenance — 3-machine category page + 3 product pages; asserts category-page order, full provenance fields (SourceType.BarrelsOfFunProductPage, DiscoveryContext = "Barrels of Fun Machines Category", DiscoveryUrl), GameRecord.Source.ScrapedFrom/GamePageUrl, DiscoveredOn carries the barrelsoffun_machines_category sentinel, and gate-vs-wire URL equality (so a future re-canonicalisation refactor can't silently throttle a different origin).
• ScrapeAsync_PerPageFetchFailure_DoesNotAbortRun — middle product returns 500; siblings still yield; politeness invariants hold under failure (500 is reported back to the gate so the 429-streak detector can see it).
• ScrapeAsync_DiscoveryFailure_AbortsThisSourceOnly — /product-category/machines/ returns 500; scraper yield-breaks cleanly without throwing.
• ScrapeAsync_PolitenessExceptionFromGate_PropagatesUp — gate throws on Acquire BEFORE any HTTP fires; assertions pin zero wire requests / zero reports so a regression that swallowed the exception would fail loudly.
• ScrapeAsync_GateThrowsOnReport_BubblesUp — symmetric report-side path.

Touched files: 1 new test file + 1 CHANGELOG bullet (no production code changes).

Pre-push self-audit

Step 0 — /local-review: deferred to the human reviewer at merge time per the PR description (this is a backfill of an already-vetted template; the mechanical checks below are the binding gate).

Step 1 — 7-item mechanical checklist (all pass):

1. Every option field is read — N/A (no new *Options).
2. Sibling-diff for drift — diffed against CgcGamePageScraperTests. Differences are intentional and called out in remarks: single-yield (no .Link assertions), category-page path (/product-category/machines/) instead of menu, BoF source-type / discovery-context / sentinel strings, and the JSON-LD product schema in fixtures (instead of CGC's title / anchor extraction). Test structure, helper signatures, gate/wire assertions, and BuildScraper shape are byte-identical to the template.
3. No bare catch { } — confirmed; no catches in test code.
4. CLI / orchestrator wiring is end-to-end — N/A (no new ISourceScraper).
5. Tests assert behavior, not just structure — yield-order test fixture has 3 machines in a deterministic order; failure-isolation fixture inserts the failing product in the middle so the assertion Equal(2, games.Count) distinguishes "kept going" from "aborted on first failure"; PolitenessException tests pin Empty(handler.Requests) so a regression that catches and swallows can't pass.
6. Build is zero-warning — dotnet build → 0 Warning(s) / 0 Error(s).
7. Identity check — git log -1 --format='%an <%ae>' → Jim Keeley <94459922+jkeeley2073@users.noreply.github.com> (personal noreply).

Test counts: 430 → 435 (+5).

Test plan

• dotnet build — clean, 0 warnings.
• dotnet test --filter FullyQualifiedName~BofProductScraperTests — 5 passed.
• Full suite — 435 passed / 0 failed / 0 skipped.

🤖 Generated with Claude Code