fix(map): recover missing route shapes

[soumajitgh]

Summary

• align map routes with the non-departed arrivals shown to riders
• try up to three upcoming trip shapes before falling back to route-level geometry
• cache successful shapes and briefly cache failures to avoid request amplification
• keep arrival polling from redrawing healthy routes and clean up fallback segments and vehicles
• add regression coverage for retries, fallback deduplication, caching, and redraw behavior

Testing

• npm test -- --run (1,799 tests passed)
• Prettier and ESLint checks on changed files
• npm run build

Closes #607

Summary by CodeRabbit

• Improvements
  • Map route displays now prioritize currently boardable arrivals and exclude trips that have already departed.
  • Route lines are more resilient when trip-specific geometry is unavailable, using alternate trips or route-level geometry when possible.
  • Multiple route segments can now be displayed for more complete coverage.
  • Map updates reuse previously loaded route geometry, improving redraw behavior.
  • Vehicle markers are cleaned up reliably when routes are removed from the map.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Team

Run ID: e6200e77-17ab-4758-b4fa-19fc51a9a0e1

📥 Commits

Reviewing files that changed from the base of the PR and between 2ef7b96 and e36916c.

📒 Files selected for processing (5)

• src/components/MapExperience.svelte
• src/components/map/StopRoutesLayer.svelte
• src/components/map/__tests__/StopRoutesLayer.test.js
• src/lib/__tests__/activeRoutes.test.js
• src/lib/activeRoutes.js

Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review.

---

📝 Walkthrough

Walkthrough

The change filters departed arrivals, exposes ordered trip candidates, and improves route shape rendering. Shape requests now use caching, alternate trip attempts, and route-level fallback segments. Rendering and teardown handle multiple polylines per route.

Changes

Route shape resilience

Layer / File(s)	Summary
Boardable arrival candidates src/lib/activeRoutes.js, src/lib/__tests__/activeRoutes.test.js, src/components/MapExperience.svelte	activeRoutesFromArrivals filters departed arrivals and returns ordered, deduplicated tripCandidates. MapExperience shares the boardable route set for selection and colors.
Cached shape retrieval and fallback src/components/map/StopRoutesLayer.svelte, src/components/map/__tests__/StopRoutesLayer.test.js	StopRoutesLayer tries up to three trip shapes, caches successes and temporary failures, and falls back to route-level polylines. Tests cover alternate candidates, fallback segments, and cache reuse.
Multi-segment rendering and teardown src/components/map/StopRoutesLayer.svelte, src/components/map/__tests__/StopRoutesLayer.test.js	Drawing, stop claiming, paint ordering, promotion, vehicle-marker cleanup, redraw behavior, and cache disposal now support multiple polylines per route.

Estimated code review effort: 4 (Complex) | ~45 minutes

Merge Risk: 🔵 Low · up to e3691

The change improves route recovery and is mergeable, but a route whose trip and fallback geometry all fail may remain missing for the current selection rather than retrying automatically after the failure-cache window.

Sequence Diagram(s)

sequenceDiagram
  participant StopRoutesLayer
  participant TransitAPI
  participant ShapeCache
  participant MapProvider
  StopRoutesLayer->>ShapeCache: request cached trip shape
  ShapeCache->>TransitAPI: fetch trip details and shape points
  TransitAPI-->>ShapeCache: return shape or failure
  StopRoutesLayer->>TransitAPI: fetch route fallback after trip failures
  TransitAPI-->>StopRoutesLayer: return fallback segments
  StopRoutesLayer->>MapProvider: create and reveal route polylines

Loading

Suggested reviewers: aaronbrethorst

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly describes the primary change: recovering missing route shapes in the map.
Linked Issues check	✅ Passed	The changes directly address issue #607 by retrying upcoming trip shapes, using route-level geometry as a fallback, and adding regression coverage for missing route rendering.
Out of Scope Changes check	✅ Passed	The changes remain within scope. Retry logic, caching, arrival filtering, cleanup, redraw behavior, and tests all support reliable map route rendering.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 1 functions across 3 files. (2 skipped: 2 …

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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.}

[coveralls]

Coverage is 93.41% — soumajitgh:fix/607-route-shape-fallback into OneBusAway:develop. No base build found for OneBusAway:develop.

[aaronbrethorst]

Code review

No issues found. Checked for bugs and CLAUDE.md compliance.

🤖 Generated with Claude Code

[aaronbrethorst]

This is a careful piece of work on a genuinely awkward problem. The escalation — try up to three boardable trips in arrival order, then fall back to route-level geometry, rather than dropping the route off the map — is the right shape, and I like that the fallback is honest about being degraded rather than pretending it's the real trip shape.

The parts I checked hardest and found sound:

• The polylinesByRouteId single-to-array conversion is complete. teardown(), reassertLinePaintOrder(), the promotion effect, and drawRoutes all handle the array form, and both providers already accept revealPolylines({only}) with multiple polylines and per-polyline removePolyline/setPolylineLayer, so nothing diverges between OSM and Google.
• polledRouteIds is the right fix for the actual leak: vehicle polling starts for every active route including ones whose shape never resolves, so teardown has to snapshot the union rather than just the drawn keys.
• The stale-token path inside the multi-segment loop correctly removes every segment the superseded route already attached, not just the one in flight.
• filterDeparted(arrivals, now) matches its real signature, and gating on Number.isFinite(now) keeps the no-options callers on the old unfiltered behavior.

One thing worth a follow-up, not a blocker. routeColors now derives from the departed-filtered boardableRoutes, but StopPane seeds its rows from the unfiltered entry on first render and only applies filterDeparted on subsequent polls. So a route whose only cold-load arrival has already departed renders a row with no assigned color and falls back to route?.color until the first poll corrects it. Cosmetic and self-healing, and the comment above routeColors about two consumers with different staleness semantics is now carrying a third axis it doesn't mention — worth a line when you're next in there.

Merging.