- Check external playlists first (fast check)
- Only check Spotify if enabled (avoid extra API call)
- Fall back to regular Jellyfin playlists
- Maintains correct behavior for all playlist types
- Remove duplicate JellyfinUrl and ApiKey from SpotifyImportSettings
- Use existing JELLYFIN_URL and JELLYFIN_API_KEY settings
- Simplify configuration - no duplicate settings needed
- Update documentation and .env.example
- Check playlist name instead of creating virtual playlists
- Intercept playlist items request for Spotify playlists
- Fill empty playlists with matched tracks from providers
- Works with existing Jellyfin Spotify Import plugin playlists
Jellyfin sends ETag, Last-Modified, and Cache-Control headers that allow clients like Feishin to cache songs locally. Proxy now forwards these headers so clients don't re-download songs unnecessarily.
- Cache mode now registers downloaded songs in mappings
- Prevents duplicate downloads in cache mode
- Fixes wasted API calls and duplicate cache files
- Both cache and download modes benefit from deduplication
- Race condition fixed where multiple threads could download same song
- RegisterDownloadedSongAsync now called before lock release
- Second thread finds registered mapping and skips download
- Eliminates duplicate files with (1), (2) suffixes
- Move RegisterDownloadedSongAsync call before lock release
- Prevents race condition where multiple threads download same song
- Second thread now finds registered mapping before starting download
- Fixes duplicate files like 'Song (1).flac', 'Song (2).flac'
- Library scan and album download still happen after lock release
- Decode 6 endpoints at startup (triton, wolf, hund, maus, vogel, katze)
- Automatic fallback when endpoint fails
- All services cycle through endpoints on failure
- URLs stored as base64, decoded once in Program.cs
- Fixes search/download issues
- Decode 6 base64 URLs at startup (1 primary + 5 backups)
- Automatic fallback when endpoint fails
- All services try next endpoint on failure
- Metadata, Download, and Validator all support fallback
- Endpoints: triton.squid.wtf, wolf/hund/maus/vogel/katze.qqdl.site
- Logs which endpoint is being used
- Cycles through all endpoints before giving up
- Decode base64 URL once at startup in Program.cs
- Pass decoded URL as constructor parameter to all SquidWTF services
- Services receive plain URL, no decoding logic in service classes
- Cleaner separation: encoding/decoding only in Program.cs
- All service code remains unchanged, just receives decoded URL
- Changed from decoding in constructor to static readonly fields
- Decode happens once per class initialization, stored in memory
- Cleaner implementation, no repeated decoding per instance
- All three SquidWTF services updated (Metadata, Download, Validator)
