[torrent] match RuTracker album bundle files

[ramonskie]

DRAFT:
this is a draft. as i did not test the whole worflow.
i just want to start a discussion. as rutracker is not working with lidarr either.
and i want to know if this is the right approach.

Summary

Fix torrent/usenet album-bundle staging for RuTracker-style releases.

RuTracker music torrents often download files with release-style names instead of clean tags, e.g.:

• Nosferatu_-_Beaver_Cleaver.flac
• 01 - 4 Tune Fairytales - Take Me 2 Wonderland (Extended Mix).flac

The album-bundle flow already downloads and stages these files correctly, but try_staging_match() compared raw filename stems against clean Spotify track titles. That caused staged album files to be missed, especially for compilations and artist-prefixed filenames.

Changes

• Parse common release filename shapes during private album-bundle staging:
  • Artist_-_Title
  • Artist - Title
  • NN - Artist - Title
  • NN - Title
• Use parsed filename title/artist as additional match candidates.
• Use explicit track numbers as a guard/boost for private album-bundle matches.
• Skip stale staged-file cache entries when the file was already claimed and removed.
• Add integration coverage using real TorrentDownloadPlugin.download_album_to_staging() with mocked Prowlarr/torrent-client boundaries.
• Add RuTracker source examples as regression fixtures:
  • https://rutracker.org/forum/viewtopic.php?t=3503447
  • https://rutracker.org/forum/viewtopic.php?t=1599889

Why

This fixes album downloads where the torrent completes successfully but tracks are not claimed from private staging due to filename/tag mismatch.

Tests

• python -m pytest tests/downloads/test_album_bundle_rutracker_integration.py tests/downloads/test_downloads_staging.py -q
• python -m ruff check core/downloads/staging.py tests/downloads/test_album_bundle_rutracker_integration.py tests/downloads/test_downloads_staging.py
• python -m pytest

[ramonskie]

Torrent/usenet album-bundle downloads are already release-level: the selected torrent/NZB downloads the whole album/release. The per-track part happens only after download, during SoulSync’s existing import/post-processing handoff.

Current flow:

1. Pick one torrent/usenet release for the album.
2. Downloader fetches the whole release.
3. SoulSync copies all audio files into private album-bundle staging.
4. Existing per-track workers run for each expected album track.
5. Each worker calls try_staging_match() to claim the staged file that belongs to that track.
6. The matched file is post-processed/tagged/moved using that track’s metadata.

This PR does not make torrents download per-track. It fixes the current per-track staging handoff by teaching it to understand RuTracker-style release filenames like Artist_-_Title and NN - Artist - Title.

Longer term, the better architecture would be an album-level manifest matcher:

• build a manifest from all downloaded files
• compare the whole manifest once against the expected album tracklist
• assign file → track mappings globally
• let per-track post-processing consume those explicit assignments

That would better handle weak filenames, duplicate compilation tracks, durations, disc numbers, and partial/extra files. This PR intentionally avoids that larger redesign and keeps scope limited to making the existing post-download handoff work for common RuTracker album releases.

[Zombiehamser]

Sorry, my question is a bit off-topic. However, I can’t help but ask: how do you use rutracker? Prowlarr can’t bypass Cloudflare protection, even with additional plugins.

[ramonskie]

rutracker works perfectly fine with flaresolver.
here is some documentation https://github.com/ramonskie/groovearr/blob/main/docs/prowlarr-integration.md

but soulsync can never really work. as at its core it does per track searching.
i wrote my own app. that does make a distinction between full albums and per track.
and with fallback mechanisms. that perfectly integrates with rutracker