IMDb Watchlist to Radarr RSS Worker

This project is a Cloudflare Worker plus Cloudflare Pages proxy that turns public IMDb lists and watchlists into a Radarr RSS feed and a Sonarr Custom List feed.

Flow:

1. Open the site on imdbwatcharr.pages.dev
2. Paste a public IMDb watchlist or list URL
3. Get back deterministic movie and TV list URLs derived from the IMDb identifier
4. Use the movie URL in Radarr's RSS List and the TV URL in Sonarr's Custom List

Examples:

• Radarr movie feed: https://www.imdb.com/user/p.kdbeq6dtmzzpiin4k7t4fnunf4/watchlist/ becomes https://imdbwatcharr.pages.dev/radarr/p/p.kdbeq6dtmzzpiin4k7t4fnunf4
• Sonarr custom list: https://www.imdb.com/user/p.kdbeq6dtmzzpiin4k7t4fnunf4/watchlist/ becomes https://imdbwatcharr.pages.dev/sonarr/p/p.kdbeq6dtmzzpiin4k7t4fnunf4
• Radarr movie feed: https://www.imdb.com/list/ls006123300/ becomes https://imdbwatcharr.pages.dev/radarr/l/ls006123300
• Sonarr custom list: https://www.imdb.com/list/ls006123300/ becomes https://imdbwatcharr.pages.dev/sonarr/l/ls006123300

Architecture

• Cloudflare Worker serves the UI, API, and RSS generation logic
• Cloudflare Pages exposes the public pages.dev hostname and proxies requests to the Worker
• D1 stores feed metadata plus the last successful snapshot of all IMDb items, any resolved TVDB IDs, and cached Radarr/Sonarr payloads
• Browser Rendering fetches IMDb pages and lets the parser extract data from stable page payloads such as #__NEXT_DATA__
• The Worker tries a direct IMDb fetch first and only falls back to Browser Rendering when needed
• The Worker fingerprints the stable IMDb payload block instead of the whole HTML page, so unchanged lists can reuse the cached Radarr RSS or Sonarr JSON response without rebuilding it

Routes

• GET / simple input form
• POST /api/create normalize an IMDb URL, create the feed record if needed, refresh it, resolve TVDB IDs for shows when possible, and return both public URLs
• GET /radarr/p/:profileId dynamically serve the Radarr movie feed for a watchlist
• GET /radarr/l/:listId dynamically serve the Radarr movie feed for a list
• GET /sonarr/p/:profileId dynamically serve the Sonarr Custom List JSON for a watchlist
• GET /sonarr/l/:listId dynamically serve the Sonarr Custom List JSON for a list
• GET /radarr/f/:imdbKey optional generic Radarr route that infers the source from values like ls..., p...., or ur...
• GET /sonarr/f/:imdbKey optional generic Sonarr route with the same inference rules
• GET /p/:profileId, GET /l/:listId, and GET /f/:imdbKey are legacy shortcuts that redirect to /radarr/...
• GET /f/:slug.xml legacy slug route that redirects to the deterministic path
• GET /api/feeds/:slug internal metadata lookup for stored feed records

Local development

npm install
npm run check
npm run dev

For remote D1 migrations after the database exists:

npm run db:migrate:remote

Deployment

This repo includes:

• ci.yml for fixture-based parser checks
• deploy-worker.yml for Worker deployment
• deploy-pages-proxy.yml for the public pages.dev proxy

Required GitHub repo secrets:

• CLOUDFLARE_API_TOKEN
• CLOUDFLARE_ACCOUNT_ID

Important note

IMDb access is the hardest part of the system. The Worker now tries a direct HTTP fetch first because it is lighter and avoids Browser Rendering rate limits in the common case. If IMDb returns a challenge page, the Worker falls back to Browser Rendering. If both fail, the route serves the last stored error or snapshot until a later refresh succeeds.

The /sonarr/... route is now designed for Sonarr's Custom List provider, which expects JSON entries with TvdbId. TVDB IDs are resolved from IMDb IDs through TVMaze when possible, so some IMDb TV entries may be skipped if no TVDB mapping is available.

When IMDb does respond, the Worker hashes the stable list payload extracted from the page. If that fingerprint has not changed, it keeps serving the cached list body with ETag headers so Radarr, Sonarr, and any upstream cache can avoid re-downloading the full payload unnecessarily.