- Dockerfile: replace single runner stage with web + scraper named targets
- web: Next.js standalone only — no playwright, tsx, or scripts
- scraper: scripts/lib/node_modules/playwright only — no Next.js output
- docker-compose.yml: each service pulls its dedicated image tag
- .gitea/workflows/deploy.yml: build both targets on push to main
- lib/db.ts: STALE_AFTER_MS reads PARK_HOURS_STALENESS_HOURS env var (default 72h)
- lib/park-meta.ts: COASTER_STALE_MS reads COASTER_STALENESS_HOURS env var (default 720h)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Imports must appear before other statements in ES modules.
Also drop the explicit .ts extension from import paths — Next.js
bundler resolves them without it, and the extension after const
was causing the module to silently fail in the app.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Extract matching logic into lib/coaster-match.ts (isCoasterMatch + normalizeForMatch)
so it can be imported by both the scraper and tests without duplication.
Add tests/coaster-matching.test.ts covering all known match/false-positive cases:
- Trademark symbols, leading THE, possessives, punctuation
- Subtitle variants in both directions (Apocalypse, New Revolution - Classic)
- Space-split brand words (BAT GIRL vs Batgirl)
- 4D subtitle extension (THE JOKER™ 4D Free Fly Coaster vs Joker)
- False positives: Joker y Harley Quinn, conjunction connectors
Run with: npm test
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Matching fixes:
- normalize() now strips all non-word/non-space chars via [^\w\s] instead of
a hand-rolled list, catching !, curly apostrophe (U+2019), and any future edge cases
- Add isCoaster() helper with prefix matching (min 5 chars) to handle subtitle
mismatches in either direction (e.g. "Apocalypse" vs "Apocalypse the Ride",
"The New Revolution - Classic" vs "New Revolution")
- Fix top-level rides loop which still used coasterNames.has(normalize()) instead
of isCoaster() — this was the recurring bug causing top-level rides to miss
UI:
- Dark neutral base (#111) replacing cold navy and muddy purple
- Neon accent palette: hot pink, electric green, vivid yellow, cyan
- Park page max-width 960→1280px, calendar cells 72→96px tall
- Scrollbar accent matches theme
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add lib/park-meta.ts to manage data/park-meta.json (rcdb_id + coaster lists)
- Add lib/scrapers/rcdb.ts to scrape operating coaster names from RCDB park pages
- discover.ts now seeds park-meta.json with skeleton entries for all parks
- scrape.ts now refreshes RCDB coaster lists (30-day staleness) for parks with rcdb_id set
- fetchLiveRides() accepts a coasterNames Set; isCoaster uses normalize() on both sides
to handle trademark symbols, 'THE ' prefixes, and punctuation differences between
Queue-Times and RCDB names — applies correctly to both land rides and top-level rides
- Commit park-meta.json so it ships in the Docker image (fresh volumes get it automatically)
- Update .gitignore / .dockerignore to exclude only *.db files, not all of data/
- Dockerfile copies park-meta.json into image before VOLUME declaration
- README: document coaster filter setup and correct staleness window (72h not 7d)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
