test(import): cover landing stubs, api/docs rescue, and staging layout

[cursor]

🛖 Resolves
—

🧰 Changes

Adds regression tests for recent import fixes that shipped without coverage. Production behavior is unchanged aside from exposing existing helpers on import.js's __test__ hook.

Risky behavior now covered

• Section landing-page injection — real common-prefix landings are imported; client-redirect stubs (meta refresh, canonical-to-child, HTTP redirect-to-child), import-base probes, single-child categories, 404s, and fetch failures are skipped.
• /api/docs vs API reference — guides under /api/docs/* stay docs; /api/docs/api|reference|api-reference and /api/reference/* stay reference. skipApiReference staging rescues the docs subtree instead of dropping it.
• Parent page staging — parents-with-children write index.md inside the folder (not a sibling <slug>.md); nested category folders that share a basename are namespaced and backed by a titled index.md; YAML-unsafe slugs are quoted in _order.yaml.
• URL / title classification — API-reference and changelog path sweeps, auth-chrome URL pruning, discoverable-link filtering, and sidebar title cleanup (&, zero-width chars).
• llms.txt (section) markers — trailing (section) is stripped from H2/H3 titles so it never becomes a sidebar category name.

Test files added/updated

• src/commands/import.landing.test.js (added)
• src/commands/import.classify.test.js (added)
• src/commands/import.staging.test.js (added)
• src/utils/llms.test.js (updated)
• src/commands/import.js (__test__ exports only)

Why these tests materially reduce regression risk

These paths decide what gets imported, how pages are classified (docs vs reference vs changelog vs auth chrome), and how the sidebar is written to disk. The recent fixes for empty parent pages (#32), /api/docs misclassification (#28), client-redirect stub landings (#29), and (section) titles (#25) had no tests; a future organize/staging change could silently drop guides, invent stub section pages, or write contentless sidebar parents.

🧬 QA & Testing

npm test

All 154 tests passed, including the new import/llms coverage. Tests are deterministic (mocked fetch, temp staging dirs) and do not depend on live sites.

  

[greptile-apps]

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains.

_{Reviews (2): Last reviewed commit: "Merge branch 'main' into erunion/missing..." | Re-trigger Greptile}