Watchlist scanner: empty discography (no new releases in lookback) was
treated as API failure, causing "Failed to get artist discography" for
artists like Kendrick Lamar who simply had no recent releases. Now
distinguishes None (API failure → try next source) from [] (success,
no new tracks). Spotify backfill now uses the authenticated client
instance instead of creating a fresh unauthenticated one.
Wishlist nebula: album remove now sends album_name (API updated to
accept album_name as fallback alongside album_id). Track remove
re-renders the nebula after deletion. Toned down processing pulse
animation.
Updated test to verify fallback triggers on API failure (None), not
on empty results.
Replaced track-count-only release selection with deterministic scoring
across 6 factors: track count match (40pts), release status (10pts),
country preference with US/worldwide bias (10pts), format preference
favoring Digital/CD over Vinyl/Cassette (10pts), barcode presence (3pts),
and date completeness (2pts). Same inputs always produce the same release.
Also fixed critical bug: _embed_source_ids was missing the context
parameter, silently skipping ALL source ID tag embedding since the
MusicBrainz consistency commit. Now passes context from the caller.
Make discovery pool population respect provider priority while keeping Spotify strict, and reduce unnecessary request volume in the hot discovery paths.
- keep discovery fan-out source-priority aware
- preserve cache use where freshness is not required
- cap Spotify artist-album pagination in discovery and cache refresh paths
- keep incremental release checks to a single page, since they only need the newest releases
- add regression coverage for provider order, strict Spotify handling, and pagination caps
Rewrote the AcoustID scanner job to scan all library tracks (via DB file
paths resolved to disk) instead of only the Transfer folder. Checkpoints
by track ID for robust resume across restarts. Defaults changed to
enabled, 24h interval, batch size 200.
Added _fix_acoustid_mismatch handler with three actions:
- retag: update DB title/artist to match actual audio content
- redownload: add expected track to wishlist and delete wrong file
- delete: remove wrong file and DB record
This catches cases like a file tagged as "Dinosaur Bones" that is
actually "Helicopters" — the scanner fingerprints the audio, detects
the mismatch, and the user can fix it from Library Maintenance findings.
Resolve Spotify artist matching through the exact Spotify client only, so watchlist ID backfill cannot drift to fallback-provider results. Remove the remaining preemptive provider availability check from the backfill loop.
Allow cached Spotify search results to return even when Spotify is rate-limited or temporarily unavailable, and remove redundant rate-limit gating after auth checks.
Drop the old active-provider artist lookup helpers from watchlist_scanner now that the web scan flow resolves sources through the shared metadata priority.
Keep the Spotify-specific feature toggles in place for discovery and sync paths that still use them.
Move the web watchlist scan core onto the shared metadata source priority so primary provider settings are respected during artist, album, and image resolution.
Add coverage for primary-source-first discography lookup and fallback to later providers when the primary source has no albums.
Bring placeholder tracklist skipping back into the shared watchlist scan path, and centralize the DB-only artist image backfill helper so both web scan entrypoints reuse the same logic.
Drop the legacy watchlist scan entrypoints that are no longer used by the web scan flow, and keep the live refresh path pointed at the shared scanner helper.
Move the shared watchlist scan loop into core/watchlist_scanner.py so web_server.py only handles triggers, locks, progress, and post-scan orchestration.
Manual and scheduled watchlist scans now share the same scanner-side core, while the web entrypoints keep profile selection and automation progress updates.
Two-layer detection: (1) check the Qobuz API response for sample=True
before downloading, and (2) validate actual file duration with mutagen
after download — if under 35 seconds, delete and return None. Qobuz
returns valid audio files for previews (~2-5MB FLAC) that pass the
existing 100KB size check, so duration is the reliable signal.
Artists with an existing spotify_artist_id but NULL spotify_match_status
were fetched by the priority queue every ~3 seconds. _process_artist
returned early (preserving the ID) without marking the status, so the
same artist was re-queued indefinitely — burning CPU and inflating API
call counters. Now marks the artist as 'matched' on the early-return
path.
Respect the configured metadata source order when looking up album years, and re-check provider availability during the scan so Spotify can drop out cleanly if it becomes rate-limited.
Cover art lookup now honors an explicit prefer_source first,
falls back to the runtime primary metadata source when unset,
and uses the shared source priority for the remaining fallbacks.
Use the shared metadata source priority when resolving album IDs,
album searches, and tracklists in track number repair.
Keeps Deezer and iTunes ahead of Spotify where configured, while
still allowing the job to fall back through other supported sources.
Unknown artist resolution now uses the shared metadata source priority and only filters to the sources that can actually participate in this job. Deezer and iTunes remain direct lookup sources, while Hydrabase can now join the title-search path when it is the configured priority source.
- Pass playlist image_url to _run_sync_task from all source-specific sync
start handlers (Deezer, Tidal, Spotify public, YouTube, automation mirror)
— previously only the /api/sync/start endpoint passed it
- Fix plex_client.set_playlist_image: use uploadPoster(url=) instead of
uploadPoster(data=) which is not a valid PlexAPI argument
- deezer_client: use picture_xl > picture_big > picture_medium fallback
for better cover art resolution
- tidal_client: extract image_url in get_playlist() from JSON:API
relationships (was only extracted in metadata-only listing)
- parse_youtube_playlist: capture playlist thumbnail from yt-dlp result
- Add visible logging for image upload attempts and outcomes
Spotify was being called for album/artist data fetching across multiple
background workers and the Artists page search even when the user had
Deezer or iTunes set as their primary metadata source. Being authenticated
for playlist sync was treated as permission to use Spotify for everything.
- watchlist_scanner: add _spotify_is_primary_source() that checks both
auth and primary source config; use it for all album/artist data fetching
(discovery pool, recent album caching, playlist curation, similar artist
ID matching, proactive ID backfill). _spotify_available_for_run() is kept
for sync_spotify_library_cache which must run regardless of primary source
- repair_jobs/metadata_gap_filler: gate Spotify ISRC lookup on primary
source being 'spotify'; MusicBrainz lookup unaffected
- repair_jobs/unknown_artist_fixer: replace hardcoded spotify_client with
source-aware client selection — primary source ID tried first, each ID
matched to its correct client (fixes latent bug passing Deezer IDs to
Spotify)
- web_server.py /api/match/search: Artists page search was hardcoded to
spotify_client.search_artists(); now uses _get_metadata_fallback_client()
so results come from the configured primary source
Adds a new Last.fm Radio section to the Discover page that lets users
search a track on Last.fm, generate a similar-tracks playlist, and run
it through the existing discovery/download/sync pipeline. Also generates
playlists automatically from top listening history during watchlist scans
(max once per week).
- core/lastfm_client.py: Add get_similar_tracks() using track.getsimilar
- core/listenbrainz_manager.py: Add save_lastfm_radio_playlist() with
deterministic MBID (MD5 seed), cleanup limit of 5 for lastfm_radio type
- web_server.py: Add /api/lastfm/configured, /api/lastfm/search/tracks,
/api/lastfm/radio/generate, /api/discover/listenbrainz/lastfm-radio;
fix playlist['name'] KeyError in discovery worker that was resetting
phase back to 'fresh' after completion
- core/watchlist_scanner.py: Add _generate_lastfm_radio_playlists() with
weekly throttle, called at end of scan_all_watchlist_artists()
- webui/index.html: Add #lastfm-radio-section above ListenBrainz section,
hidden unless Last.fm API key is configured
- webui/static/script.js: Search/generation/card-load functions; fix
discovery modal labels (Last.fm Radio vs ListenBrainz), description
update on completion, belt-and-suspenders completion handling inside
updateYouTubeDiscoveryModal; fix album/duration display for tracks
without metadata; music note SVG placeholder for missing art
- webui/static/style.css: Styles for search bar, dropdown, result rows
Centralize the ordered metadata source list and source-priority helper so album completeness and the repair worker follow the same Deezer/iTunes-first fallback order. This also removes the last duplicate priority logic from the touched repair paths.
Album completeness and any other repair job now uses the centralized source/client helpers instead of a worker-local Spotify client or override plumbing
- This keeps source selection aligned with the configured primary provider and removes the last Spotify-only special case from the job path.
This change ultimately is a step towards further centralizing the Spotify client access and the associated `is_spotify_authenticated` check.
- Currently these look-ups are done all over the place in different feature implementations directly, but moving forward, any feature that uses `get_primary_client` or `get_client_for_source` to access the Spotify client, won't have to duplicate any rate-limiting or auth checks as long as these getters are used
New core/replaygain.py module uses FFmpeg's ebur128 filter (already a
project dependency) to analyze integrated loudness and true peak, then
writes ReplayGain 2.0 tags (-18 LUFS reference) to MP3 (TXXX frames),
FLAC/OGG/Opus (Vorbis comments), and M4A/MP4 (freeform atoms).
Three analysis modes in the enhanced library view:
- Per-track RG button: synchronous single-track analysis (~1-3 s)
- Album "ReplayGain" button: background job writing both track gain
and album gain (mean LUFS across all album tracks) to every file
- Bulk bar "ReplayGain" button: batch track-gain for selected tracks
read_file_tags() in tag_writer.py extended with four new optional keys
(replaygain_track_gain/_peak, replaygain_album_gain/_peak) so existing
RG values surface in the tag-preview diff view. Purely additive — no
existing endpoints or DB schema changed.
Hardcoded ports 8888/8889 conflict when SoulSync runs behind Gluetun or
other containers that claim those ports. Introduce SOULSYNC_SPOTIFY_CALLBACK_PORT
and SOULSYNC_TIDAL_CALLBACK_PORT env vars (defaulting to 8888/8889) so
users can remap without rebuilding the image.
docker-compose.yml exposes the vars with comments explaining how to keep
the port mappings in sync with the redirect URI in Settings → Connections.
Track per-quality-tier failure reasons across all failure paths (stream
error, empty manifest, download exception, stub file, MP4 extraction
failure) and include them in the exhausted-tiers log message so failures
are diagnosable from logs.
When HiRes is configured with no fallback and all tiers are exhausted,
log an actionable hint directing the user to enable Quality Fallback.
Surface Tidal-specific error messages in the UI task on retry
exhaustion: distinguishes HiRes-unavailable (with actionable guidance)
from general Tidal auth/quality failures, rather than showing the
generic Soulseek error string.
Album completeness and downstream repair flow now follow the configured
primary provider first, with Discogs and Hydrabase support added alongside
existing Spotify, iTunes, and Deezer paths.
Keep spotify_track_id for compatibility while preserving source-aware track
IDs for provider-neutral handling.
Builds a new Your Albums section on the Discover page that aggregates
saved/liked albums from all connected services, mirroring the Your Artists
pattern. Deezer works via both OAuth and ARL.
- tidal_client: add get_favorite_albums() with V2/V1 API fallback
- deezer_client: add get_user_favorite_albums() via OAuth (user/me/albums)
- deezer_download_client: add get_user_favorite_albums() via ARL session
- music_database: add liked_albums_pool table (deduped by artist::album
normalized key), upsert_liked_album, get_liked_albums,
get_liked_albums_last_fetch, clear_liked_albums
- web_server: GET /api/discover/your-albums (ownership-checked, paginated),
GET /api/discover/your-albums/sources, POST /api/discover/your-albums/refresh,
_fetch_liked_albums background worker (Spotify + Tidal + Deezer OAuth/ARL)
- frontend: Your Albums section with source selector cog, album grid reusing
spotify-library-card styles, search/filter/sort/pagination, download missing
button, auto-refresh poll on first load
Also fix: Deezer greyed out in Your Artists sources when using ARL — connection
check now accepts ARL auth (deezer_dl.is_authenticated()) in addition to OAuth,
and _fetch_and_match_liked_artists falls back to ARL client for artist fetching.
Add get_user_favorite_artists(limit=200) to DeezerDownloadClient to fetch a user's favorite artists via the public API using an ARL-authenticated session (paginated, error-handled, returns deezer_id, name, image_url).
Update web_server to treat Deezer as connected if either OAuth or ARL is authenticated, and to fetch favorite artists from OAuth client when available or from soulseek_client.deezer_dl (ARL) otherwise. Fetched artists are upserted into the database and appropriate log/console messages and counters are updated.
get_playlist_tracks() had no Limit or StartIndex params, so Jellyfin
defaulted to ~100 items per response. This caused the Server Playlists
comparison to show most tracks as missing even though they were present.
Now paginates in batches of 1000 until a partial batch signals the last page.
Spotify album art: replace the 4-char size segment after '0000ab67616d'
with '82c1' to request the original uploaded master (up to 2000px+).
Applied via _upgrade_spotify_image_url() in Track, Artist, and Album
dataclass constructors and as a catch-all in _download_cover_art.
Scoped to the ab67616d album art prefix only — artist images use a
different prefix (ab676161) where the trick does not apply.
iTunes/Apple Music: replace '100x100bb' with '3000x3000bb' in all
artworkUrl100 replacements across Track, Artist, Album, and the
get_album images arrays. Also applied as a catch-all in _download_cover_art.
Deezer already uses cover_xl at its maximum — no changes needed there.
The fallback test (used when no audio files exist in the library) sends
a dummy fingerprint to the AcoustID API. The API correctly rejects the
dummy fingerprint but this is not error code 4 (invalid key), so the
test was returning False instead of True. Any non-code-4 error from the
fallback means the API key was accepted — only code 4 means a bad key.
Deezer and iTunes defaulted to 50 albums max, silently truncating large
discographies. Deezer now paginates (100 per page) up to 200. iTunes
raised to 200 (single call). All callers in web_server.py updated to
use the new defaults instead of hardcoding limit=50.
Also adds diagnostic logging for allow_duplicates album comparison
to help debug inconsistent singles behavior.
Singles like "idol" weren't added when the same song existed on a
different album because check_track_exists used album-aware matching
that found the track via album name. Now skips the album hint when
allow_duplicates is on so matching is title+artist only, then compares
album names ourselves with a strict 0.85 threshold. Only affects users
with allow_duplicates enabled.
The setting only affected wishlist dedup but the watchlist scanner's
library check still skipped tracks by title+artist regardless. Now
when allow_duplicates is enabled, the scanner compares album names
and only skips if the same album matches. Same song on a different
album is allowed through to the wishlist.
Deezer's API returns a contributors array with all credited artists on
a track, but only the primary artist field was used. Now extracts all
contributor names into the artists array for feature/collab tracks.
Only affects the per-track ARTIST tag (TPE1) — album artist, folder
paths, and matching are unchanged. Falls back to single primary artist
when contributors field is absent (search results) or has only one
entry.
Spotify lists unreleased albums with placeholder names like "Track 1",
"Track 2" before the real tracklist is revealed. The scanner was trying
to download these, searching Soulseek for "Track 1" by artist which
matches random files. Now skips any album where more than half the
tracks match the placeholder pattern. Covers both the watchlist scan
and discovery pool paths.
The old subset check treated "Paradise" as matching "Club Paradise"
because {'paradise'} is a subset of {'club', 'paradise'}. Both got
the same +0.10 bonus, so the wrong album could be selected.
Now uses SequenceMatcher for full-string similarity between the wanted
album name and each path segment. Exact matches (>= 0.85) get +0.10,
partial matches (>= 0.60) get +0.03, no match gets +0.00. No penalty
applied — purely adjusts bonus sizing so the correct album ranks higher.
Plex API can return Tag objects mixed with playlists — these lack the
playlistType attribute, causing AttributeError. Use getattr with safe
default instead of direct attribute access.
Add 3-tier Unknown Artist guard in post-processing: checks track_info
artists, original search result, then re-fetches from metadata API
before building folder paths or embedding tags. Prevents files from
landing in Unknown Artist folders when the download context has
incomplete artist data.
The junk entity filter checked track name but not artist_name, allowing
tracks like "Woman Like You by Unknown Artist" to be cached. Now rejects
any track or album where artist_name matches the junk names list
(unknown, unknown artist, empty, null, etc). Prevents stale incomplete
data from persisting across retries.
- Stop active DB update work before tearing down executor pools.
- Short-circuit scan completion callbacks during shutdown so in-flight timer ticks don’t queue follow-up work.
- Prevent the download monitor from draining deferred/completed tasks after shutdown starts.
- Make listening stats startup stop-aware so it exits cleanly if teardown begins during warmup.
- If a metadata update is already running, it can now observe should_stop and exit cleanly instead of continuing after SIGTERM.
- Add interruptible stop events to background workers so shutdown
wakes out of long sleeps instead of waiting on fixed delays.
- Stop scan managers, repair worker, executors, and cleanup helpers
deterministically so process exit does not leave background threads
alive.
- Add startup warnings for stale SQLite WAL/SHM sidecars so unclean
shutdowns are easier to spot before init/migration errors cascade.
- Prevent forced kills from leaving SQLite sidecars behind, which
made rollbacks to older branches fail with malformed database
errors.
7-step full-screen wizard: Welcome, Metadata Source, Download Source,
Paths & Media Server, Add Artists, First Download, Done. All settings
save to DB identically to the Settings page. Supports all 6 download
sources with inline config and test buttons. First download goes through
the full matched download pipeline with metadata context.
Fixes:
- Download clients (YouTube/HiFi/Tidal/Qobuz/Deezer) now reload
download_path when settings change instead of caching from init
- watchlist_artists table migrations now include deezer_artist_id and
discogs_artist_id in all 3 table rebuild locations (was being dropped)
- CREATE TABLE for watchlist_artists includes all provider ID columns
- Serverless download sources (YouTube/HiFi/Qobuz) show green status
instead of red disconnected on sidebar and dashboard
- Suppress repeated slskd 401 errors — logs once then silences until
connection recovers
Stripped 4,200+ emoji characters from print(), logger calls across
39 Python files. Logs are now clean text — easier to grep, more
professional, no encoding issues on terminals without Unicode support.
Seasonal config icons preserved for UI display.
Navidrome provides musicBrainzId on tracks — now captured during
database updates so the MusicBrainz enrichment worker can skip
tracks that already have an MBID.
Uses COALESCE on UPDATE to never overwrite existing enrichment data
with NULL (safe for Plex/Jellyfin which don't provide this field).
Inspired by PR #279 — fixed data loss bug in the original where
unconditional UPDATE would erase existing MBIDs.
Tag preview and writer now use track_artist (per-track artist) for the
Artist tag when available, falling back to artist_name (album artist)
when NULL. Album Artist tag always uses artist_name.
Fixes#277 — DJ mix albums no longer overwrite per-track artists
(Technique, Gouryella) with the album artist (Tiësto).
Lidarr integration:
- New core/lidarr_download_client.py with full interface parity
(search, download, status, cancel — same as Qobuz/Tidal/HiFi)
- Registered in download orchestrator with source routing
- Settings: URL + API key on Downloads tab with connection test
- Available as standalone source or in Hybrid mode priority order
- API key encrypted at rest
- All streaming source checks updated to include 'lidarr'
Lidarr downloads full albums via Usenet/torrent — SoulSync imports
only the tracks it needs and discards the rest.
Music video path validation:
- Empty/unconfigured path returns clear error instead of silent failure
- Write permission test before starting download
- Default changed from './MusicVideos' to empty (must be configured)
Clients are for the most part being initialized per-request, which leads to a lot of redundant client initialization, as well as noise on the logs, since each client initialization emits a row on the logs, eg. 'Deezer client initialized'
Click any video card in Music Videos tab to download. Flow:
1. Search primary metadata source for clean artist/title
2. Fall back to YouTube title parsing if no match
3. Download video via yt-dlp (best quality MP4)
4. Save to configured Music Videos folder as Artist/Title-video.mp4
UI shows circular progress ring on the thumbnail during download,
green checkmark on completion, red X on error (clickable to retry).
Cards are non-interactive while downloading.
Backend: /api/music-video/download and /api/music-video/status endpoints
YouTube client: download_music_video() method keeps video format
New "Music Videos" pill tab alongside Spotify/Deezer/iTunes/Discogs
in both enhanced search and global search. Searches YouTube via yt-dlp
and displays results in a video card grid with 16:9 thumbnails, play
overlay, duration badge, channel name, and view count.
- Backend: /api/enhanced-search/source/youtube_videos endpoint with
search_videos() method on YouTubeClient returning YouTubeSearchResult
- Frontend: Video grid layout with responsive cards, YouTube red tab
color, proper section hiding when switching between metadata and
video tabs
- Global search: Full parity with enhanced search video rendering
- No download functionality yet — display only
The finding dedup check only looked for 'pending' and 'resolved' status,
missing 'dismissed'. Dismissed findings were recreated as new entries on
every scan. Now includes 'dismissed' in the dedup check.
Orphan file detector improvements:
- Increased path suffix matching depth from 3 to 4 segments (covers
Genre/Artist/Album/track.flac paths)
- Added filename-based fallback when Mutagen can't read file tags —
parses title from "NN - Title [Quality].ext" pattern and matches
against parent/grandparent folder names as artist
Prevents downloading tracks from completely wrong artists by adding
minimum artist score gates:
- Soulseek: artist_score < 0.25 → reject (catches Belvedere vs Periphery)
- YouTube: artist_score < 0.15 → reject (catches lizzylou06 vs Muse)
Fixes artist substring matching to use word boundaries instead of plain
containment — "muse" no longer matches "museum", "art" no longer matches
"heart". This was causing false positives where wrong artists passed with
artist_score=1.0 due to accidental substring containment.
Improves similarity fallback by comparing against individual path segments
instead of the full filename, so misspelled artist names (Radiohedd vs
Radiohead) still match correctly.
Adjusts YouTube weights from Title 70%/Artist 10% to Title 60%/Artist 20%
to give artist more influence in YouTube matching.
Addresses user reports of unreleased albums being downloaded with garbage
content from wrong artists on Soulseek and YouTube.
All callers of _create_fallback_client() and _get_configured_fallback_source()
now use get_primary_client() and get_primary_source() directly. No more
legacy alias usage anywhere in the codebase.
Seasonal discovery had 3 use_spotify checks using is_authenticated()
(always True) instead of deriving from the configured source. Search API
(tracks, albums, artists) also defaulted to Spotify when authenticated.
All now check configured primary source first via get_primary_source().
All metadata source decisions now flow through get_primary_source() and
get_primary_client() in core/metadata_service.py. Previously 6 different
files reimplemented this logic with inconsistent defaults ('itunes' vs
'deezer') and auth checks, causing bugs when any one was missed.
Changes:
- metadata_service.py: Added canonical get_primary_source/get_primary_client
- web_server.py: _get_metadata_fallback_source() and _get_active_discovery_source()
are now thin wrappers delegating to metadata_service
- seasonal_discovery.py: _get_source() delegates to metadata_service
- personalized_playlists.py: _get_active_source() delegates to metadata_service
- spotify_client.py: Fixed _fallback_source default from 'itunes' to 'deezer'
- watchlist_scanner.py: _get_fallback_metadata_client() delegates to metadata_service
Future changes to source selection only need to update one file.
New repair job that scans the library for tracks filed under "Unknown
Artist" and corrects them. Resolves correct metadata by:
1. Reading embedded file tags (if file has correct artist)
2. Looking up by source track ID (Spotify/Deezer/iTunes)
3. Searching by title as last resort
Dry run mode (default) creates findings for review. Live mode re-tags
the audio file, moves it to the correct folder structure, and updates
the database. Includes fix handler for applying individual findings.
Seasonal discovery, personalized playlists, and playlist explorer all
defaulted to Spotify when authenticated, ignoring the user's configured
primary source. Now they read from config first.
Spotify's related_artists API (no Deezer/iTunes equivalent) is preserved
as a fallback for all users in personalized playlists. Artist discography
endpoint intentionally unchanged — ID-based lookups need the source that
owns the ID.
Deezer's /artist/{id}/albums endpoint returns albums without an artist
field, causing 98% of cached Deezer albums to have empty artist_name.
Now injects the known artist before caching.
Also fixes get_track_details cache validation — was trusting search
result cache (which has isrc but no track_position), returning
track_number=0. Now only trusts cache entries with track_position.
Discovery workers now respect the user's configured primary metadata
source instead of always using Spotify when authenticated. This
completes the intent of commit 3c211ea.
The core fix addresses data loss in the discovery→sync→wishlist→download
pipeline: the Track dataclass strips album metadata to a plain string,
losing album ID, track_number, release_date, and images. Discovery
workers now enrich results via get_track_details() to recover this data.
Deezer's get_track_details() cache validation was incorrectly trusting
search-result cache (which lacks track_position), returning track_number=0.
Also fixes wishlist download processing where albums without IDs couldn't
map to artists, and the fallback read 'artist' (singular) instead of
'artists' (plural), always producing "Unknown Artist".
Includes a one-time migration to purge stale discovery cache entries.
Duplicate finding detail now shows each version as clickable — user
can choose which to keep instead of relying on auto-selection. Added
track_number as tiebreaker in auto-pick (higher track number wins
over 01, catching leftover duplicates from the playlist sync track
number bug). Track number displayed in the detail view for clarity.
New automation action that executes user scripts from a dedicated
scripts/ directory. Available as both a DO action and THEN action.
Scripts are selected from a dropdown populated by /api/scripts.
Security: only scripts in the scripts dir can run, path traversal
blocked, no shell=True, stdout/stderr capped, configurable timeout
(max 300s). Scripts receive SOULSYNC_EVENT, SOULSYNC_AUTOMATION,
and SOULSYNC_SCRIPTS_DIR environment variables.
Includes Dockerfile + docker-compose.yml changes for the scripts
volume mount, and three example scripts (hello_world.sh,
system_info.py, notify_ntfy.sh).
The scanner was creating a finding for every file that couldn't be
identified by AcoustID, flooding the findings list with non-actionable
entries. Users saw the scanner "stuck scanning the same files over
and over" because the no-match findings were dismissed but recreated
on every run. Now only genuine mismatches (AcoustID identifies a
different track) create findings. Errors are counted and shown in
the job log with actual error messages for debugging.
After a successful playlist sync, if the source playlist has cover
art (Spotify, Tidal, Deezer, etc.), the image is downloaded and
uploaded as the playlist poster on the media server. Plex uses
uploadPoster(), Jellyfin/Emby uses POST /Items/{id}/Images/Primary.
Navidrome skipped (no playlist image API). Failure is silent — sync
result unchanged. Automation-triggered syncs and playlists without
images are unaffected.
Sync rehydration: after loading Deezer ARL playlists, checks each
for active syncs via /api/sync/status and re-attaches polling with
live card updates. Download rehydration: rehydrateModal now handles
deezer_arl_ playlist IDs, and openDownloadMissingModal routes cache
misses to the correct ARL endpoint. Fix All now prompts for dead
file action.
Album data caching: get_playlist_tracks now checks the metadata
cache before fetching album release dates from the Deezer API.
Cache hits are instant, misses are fetched and stored for future
use across all playlists. Import fixed from core.metadata_cache
instead of web_server to avoid circular dependency.
Dead file fix now prompts with two options: Re-download (existing
behavior — adds to wishlist + deletes DB entry) or Remove from DB
(just deletes the dead track record without re-downloading). Works
for both single and bulk fix. Solves the issue where dismissing
dead files didn't remove the underlying track record, causing them
to reappear on every scan.
Albums with zero local tracks were flagged as incomplete but the
auto-fill fix failed because there were no existing tracks to
determine the album folder or quality standard from. Now skipped
during scan — they'll be detected once tracks are actually added.
Qobuz added reCAPTCHA to their login endpoint, blocking automated
email/password auth for new users. Token login lets users paste
their X-User-Auth-Token from the browser DevTools after logging in
manually. Added to both Connections and Downloads tabs with
instructions. Existing email/password flow completely unchanged.
Backend validates token via user/get API and saves the session
identically to email/password login.
Albums announced but not yet released have no real audio available,
causing Soulseek to match random tracks with similar names. Both
discography methods (Spotify and generic client) now filter out
albums with release dates in the future. Skipped albums are not
marked as processed — they will be picked up on the first scan
after their release date passes.
The _is_valid_guid method only accepted 32-char hex GUIDs (Jellyfin
format) but Emby uses plain integer IDs like "12345". All matched
tracks were rejected as "invalid/empty IDs" causing playlist creation
to fail with zero tracks. Now accepts both numeric strings (Emby)
and hex GUIDs (Jellyfin).
New "Deezer" tab on sync page shows authenticated user's playlists,
identical to the Spotify tab pattern — same card layout, details
modal with track list, sync button, and download missing tracks
flow. Existing URL import renamed to "Deezer Link" tab (unchanged).
Backend: get_user_playlists() and get_playlist_tracks() on the
download client fetch via public API with ARL session cookies.
Album release dates batch-fetched for $year template variable.
Three new endpoints: arl-status, arl-playlists, arl-playlist/<id>.
Frontend: cards use Spotify-identical HTML structure with live sync
status, progress indicators, and View Progress/Results buttons.
Downloads reuse openDownloadMissingModal with zero modifications.
Track data cached on first open, instant on subsequent clicks.
Streaming matching: add artist gate rejecting candidates with artist
similarity below 0.4, raise threshold to 0.60, block fallback to
Soulseek filename matcher for Tidal/Qobuz/HiFi/Deezer. Fix single-
char artist containment bug where normalize_string strips non-ASCII
(e.g. "B小町" → "b") causing "b" to match any artist containing
that letter. Fixed in both score_track_match and the Soulseek scorer.
YouTube and Soulseek matching behavior unchanged.
Global search: add registerSearchDownload() calls to _gsClickAlbum
and _gsClickTrack so downloads create bubble snapshots on dashboard
and search page, matching the enhanced search standard.
Global search escaping: add _escAttr() helper to handle newlines in
album/artist names that broke inline onclick string literals.
Changed ignore_cross_album default from True to False. Re-downloads of
the same song create separate album entries, so the detector was skipping
them. Users who want to keep compilations/greatest-hits intact can toggle
it back on. Updated help text to explain when to use this setting.
API Call Tracker:
- Save/load 24h minute-bucketed history + events to database/api_call_history.json
- Persists across server restarts via atexit + signal handler hooks
- New record_event() for rate limit bans (called from _set_global_rate_limit)
- New get_debug_summary() for Copy Debug Info — 24h totals, peak cpm with
timestamp, per-endpoint breakdown, and last 20 rate limit events
- Fixed race condition: events iteration now inside lock during save
Spotify Rate Limit Mitigation:
- Enrichment worker: max_pages=5 on get_artist_albums (was unlimited — artist
with 217 albums caused 22 paginated API calls, now capped at 5)
- Enrichment worker: inter_item_sleep raised from 0.5s to 1.5s
Spotify Re-Auth Fix:
- Both OAuth callbacks (port 8008 + 8888) now clear rate limit ban AND
post-ban cooldown after successful re-auth — Spotify usable immediately
instead of stuck on Deezer fallback for 5 minutes
- Auth cache invalidated on both global client and enrichment worker client
New repair job that scans track and album titles for live performances,
commentary, interviews, skits, and spoken word content. Creates findings
for user review — no auto-fix.
Configurable per content type (live, commentary, interviews, spoken word),
with optional album title scanning and tracks/albums scope toggle.
Fix action removes track from DB + deletes file, cleans up empty albums
and directories. Follows existing repair job pattern exactly.
Tidal, Qobuz, HiFi, and Deezer results were blindly taking the first
API result with minimal validation. Now all streaming sources use
score_track_match() — same 60% title / 30% artist / 10% duration
weighting as Soulseek, plus version detection penalties.
- web_server.py get_valid_candidates(): replaced loose title-sim check
with matching engine scoring, version penalty for live/remix/acoustic
- download_orchestrator.py: optional expected_track param enables
scoring in search_and_download_best (backward compatible)
- sync_service.py: passes spotify_track for validation
- Fixed wrong class name (MusicMatchingEngine not MatchingEngine)
YOUR ARTISTS (major feature):
- Aggregates liked/followed artists from Spotify, Tidal, Last.fm, Deezer
- Matches to ALL metadata sources (Spotify, iTunes, Deezer, Discogs)
- DB-first matching: library → watchlist → cache → API search (capped)
- Image backfill from Spotify API for artists missing artwork
- Carousel on Discover page with 20 random matched artists
- View All modal with search, source filters, sort, pagination
- Artist info modal: hero image, matched source badges, genres, bio,
listeners/plays from Last.fm, watchlist toggle, view discography
- Auto-refresh with loading state on first load, polls until ready
- Deduplication by normalized name across all services
DEEZER OAUTH:
- Full OAuth flow: /auth/deezer + /deezer/callback
- Settings UI on Connections tab (App ID, Secret, Redirect URI)
- Token stored encrypted, auto-included in API calls
- get_user_favorite_artists() for liked artists pool
SERVICE CLIENTS:
- Spotify: added user-follow-read scope + get_followed_artists()
- Tidal: get_favorite_artists() with V2/V1 fallback
- Last.fm: get_authenticated_username() + get_user_top_artists()
FAILED MB LOOKUPS MANAGER:
- Manage button on Cache Health modal
- Browse/filter/search all failed MusicBrainz lookups
- Search MusicBrainz directly and manually match entries
- Optimized cache health queries (11 → 4 consolidated)
- Dashboard cache stats now poll every 15s
EXPLORER IMPROVEMENTS:
- Discover button on undiscovered playlist cards
- Status badges: explored/wishlisted/downloaded/ready
- Auto-refresh during discovery via polling
- Redesigned controls: prominent Explore button, icons
BUG FIXES:
- Fix album artist splitting on collab albums (collab mode fed
album-level artists instead of per-track)
- Fix cover.jpg not moving during library reorganize (post-pass sweep)
- Fix cover.jpg missing when album detection takes fallback path
- Fix wishlist auto-processing toast spam (was firing every 2s)
- Fix media player collapsing on short viewports
- Fix watchlist rate limiting (~90% fewer API calls)
- Configurable spotify.min_api_interval setting
- Better Retry-After header extraction
- Encrypt Last.fm and Discogs credentials at rest
- Add $discnum template variable (unpadded disc number)
New feature: Failed MusicBrainz Lookups management modal accessible
from Cache Health. Browse all failed lookups with type filter tabs,
search bar, pagination. Click any entry to search MusicBrainz and
manually match — saves MBID at 100% confidence. Clear individual
entries or bulk clear all.
Backend: 4 new endpoints — failed-mb-lookups list, mb-entry delete,
musicbrainz/search (artist/release/recording), mb-match save.
Performance: Cache health stats consolidated from 11 queries to 4
using CASE expressions. Added partial index on musicbrainz_cache for
failed lookups. Dashboard cache stats now poll every 15s instead of
single fire-and-forget fetch. Failed MB type counts cached on frontend,
only re-fetched after mutations.
Also includes: library reorganize now moves cover.jpg via post-pass
sidecar sweep, and changelog updates.
Addresses all three points from community rate-limiting report:
1. Watchlist scans fetched ALL albums then filtered — 262 albums = 27
API calls per artist. Now determines upfront if full discography is
needed: subsequent scans and time-bounded lookbacks use max_pages=1
(1 API call). Only "full discography" global setting fetches all.
2. MIN_API_INTERVAL (350ms) now configurable via spotify.min_api_interval
setting. Users who get rate-limited frequently can increase the delay.
Floor at 100ms to prevent abuse.
3. Retry-After header extraction improved: added diagnostic logging when
headers exist but lack Retry-After key, plus regex fallback to parse
the value from the error message string.
- Add discogs_artist_id column to watchlist_artists table (migration)
- Add discogs_artist_id to WatchlistArtist dataclass
- Add to get_watchlist_artists optional_columns and constructor
- Add update_watchlist_discogs_id DB method
- Backfill loop includes Discogs when token is configured
- Add _match_to_discogs for cross-provider artist matching
- Backfill maps updated: id_attr, match_fn, update_fn all include discogs
- Add _extract_discogs_fields to metadata cache — handles Discogs field
names (title vs name, images array, Artist - Title format)
- Worker uses _fetch_and_cache_artist/_fetch_and_cache_album helpers
that cache raw data while returning it for enrichment
- All search/lookup methods cache results for repeat queries
- Cache browser: Discogs stat pill, source filter, clear button, badge
- Fixes albums showing as 'Unknown' and artists missing images in cache
- get_album and get_album_tracks now try /masters/{id} first, fall back
to /releases/{id} — artist discography returns master IDs which are
in a different namespace than release IDs
- Fixes wrong album showing in download modal (master ID 3664443 for
GNX was hitting /releases/3664443 which is a different album)
- Add Discogs source override to all 6 artist/album/track endpoints
- Add discogs_id to _resolve_db_album_id lookup
- Remove upfront master detail fetching (was 15+ API calls, 40+ seconds)
- Discography loads from releases list only (~1 second, 2 API calls)
- Track counts populate on-demand via get_album_tracks when clicking album
- New get_album_tracks method: tries /masters first, falls back to /releases,
returns Spotify-compatible format with proper disc/track numbering
- Album type defaults to 'album' for masters without format metadata —
Discogs limitation, singles only detectable from individual release format
- Search results return format as list ['Vinyl', 'LP'] while artist
releases return comma-separated string — handle both
- Fixes "'list' object has no attribute 'lower'" error
- Master releases now fetch /masters/{id} to get actual tracklist,
genres, styles, and images — fixes 0/0 track count display
- Album type re-evaluated with real track count: 1-3 = single,
4-6 = EP, 7+ = album
- Cover art from master detail used when search results have none
- Tested: Kendrick Lamar shows correct track counts, proper types,
and images for all albums
- Fetch artist name first, then compare against each release's primary
artist — skip releases where the artist is listed after Feat./Ft./&
- "Beyoncé Feat. Kendrick Lamar" → skipped (Kendrick is featured)
- "Kendrick Lamar Feat. Rihanna" → kept (Kendrick is primary)
- Fixes artist pages showing unrelated albums from other artists
- Add _normalize_name helper and re import to DiscogsClient
- Prefer master releases over individual pressings to avoid duplicates
(multiple pressings of same album showing separately)
- Individual releases only included if no master exists for that title
- Skip non-main roles (appearances, features, remixes by others)
- Better album type detection from format string: catches LP, Album,
EP, Single, Compilation from comma-separated format field
- Fetch more results (3x limit) to compensate for filtering
- Fix source name mapping so sidebar/dashboard shows 'Discogs' instead
of falling through to 'iTunes'
- Fix album type detection: parse format string from artist releases
endpoint (e.g. "File, FLAC, Single, 320") to correctly identify
singles, EPs, albums, compilations — was defaulting everything to
'single' because track count was 0
- Remove fake track search that returned albums as tracks — Discogs
has no track-level search API, so tracks section is empty (honest)
- Track data available via album tracklists instead
- SpotifyClient: add _discogs lazy-load property, route _fallback to
DiscogsClient when configured (requires token, falls back to iTunes)
- web_server: _get_metadata_fallback_client returns DiscogsClient when
selected and token present
- Enhanced search: Discogs added as source tab with NDJSON streaming,
only available when token configured
- Alternate sources list includes Discogs when token is set
- Frontend: source labels, tab styling, fetch list all include Discogs
- Consistent with iTunes/Deezer pattern — same interfaces, same routing
- New core/discogs_worker.py — background worker enriching artists and
albums with Discogs metadata following AudioDBWorker pattern exactly
- Artist enrichment: discogs_id, bio, members, URLs, image backfill,
genre backfill, summary backfill from bio
- Album enrichment: discogs_id, genres, styles (400+ taxonomy), label,
catalog number, country, community rating, image backfill
- DB migration: discogs columns on artists (id, match_status, bio,
members, urls) and albums (id, match_status, genres, styles, label,
catno, country, rating, rating_count)
- Worker initialization with pause/resume persistence
- Status/pause/resume API endpoints
- Integrated into enrichment status system, rate monitor, auto-pause
during downloads/scans, WebSocket status emission
- New Discogs section on Settings → Connections with personal token input
- Discogs added as fallback metadata source option alongside iTunes/Deezer
- Token saved to discogs.token config key
- Discogs added to API rate monitor gauges (60/min with auth)
- Help text links to discogs.com/settings/developers for token generation
- Full parity with iTunes/Deezer clients — same Track/Artist/Album
dataclasses, same method signatures (search_artists, search_albums,
search_tracks, get_artist, get_album, get_artist_albums)
- 25 req/min unauthenticated, 60 req/min with free personal token
- Rate limited via same decorator pattern with API call tracking
- Unique data: 400+ genre/style taxonomy, label info, catalog numbers,
community ratings, artist bios
- Smart "Artist - Title" parsing for search results
- Release deduplication (Discogs has many pressings of same album)
- Track search via release tracklist extraction
- Tested: artist/album/track search, artist detail with bio, album
detail with full tracklist + genres + styles + label
- Was only backfilling the active provider — artists added via Deezer
never got Spotify/iTunes IDs, and vice versa
- Now backfills iTunes (always), Deezer (always), and Spotify (if
authenticated) at the start of every scan
- Added _match_to_deezer() and update_watchlist_deezer_id() for
Deezer cross-provider matching
- Generalized backfill with provider→attribute/function maps
- New 'webhook' then-action: sends HTTP POST with JSON payload to any
user-configured URL (Gotify, Home Assistant, Slack, n8n, etc.)
- Config: URL, optional custom headers (Key: Value per line with
variable substitution), optional custom message
- Payload includes all event variables as JSON fields
- 15s timeout, errors on 400+ status codes
- Follows exact same pattern as Discord/Pushbullet/Telegram handlers
- Frontend: config fields, config reader, icon, help docs
- Updated changelogs with webhook, M3U fix, orchestrator hardening
- Each of the 6 download clients initializes independently via
_safe_init() — one failing client no longer kills the orchestrator
- All methods guarded against None clients with appropriate fallbacks
- Init failures logged at startup and tracked in _init_failures list
- Copy Debug Info shows "Download Client Failures" section when any
client failed to initialize, or "ALL" if orchestrator itself is dead
- Add rate limiting to all 4 Spotify pagination loops (get_artist_albums,
get_user_playlists, get_playlist_tracks, get_album_tracks) — these
called sp.next() bypassing the rate_limited decorator entirely, causing
unthrottled API calls that triggered 429 bans
- Track pagination calls in API rate monitor (separate endpoint names)
- Increase DELAY_BETWEEN_ARTISTS from 2s to 4s in watchlist scanner
- Abort watchlist scan immediately if Spotify rate limit detected mid-scan
instead of continuing to hammer the API
- New core/api_call_tracker.py — centralized tracker with rolling 60s
timestamps (speedometer) and 24h minute-bucketed history (charts)
- Instrument all 9 service client rate_limited decorators to record
actual API calls with per-endpoint tracking for Spotify
- 1-second WebSocket push loop for real-time gauge updates
- Modern radial arc gauges with service brand colors, glowing active
arc, endpoint dot, 0/max scale labels, smooth CSS transitions
- Click any gauge to open detail modal with 24h call history chart
(Canvas 2D, HiDPI, gradient fill, grid lines, danger zone band)
- Spotify modal shows per-endpoint history lines with color legend
and live per-endpoint breakdown bars
- Rate limited state indicator — blinking red badge with countdown
timer appears on gauge card when Spotify ban is active
- REST endpoint GET /api/rate-monitor/history/<service> for chart data
- Responsive grid layout (5 cols desktop, 3 tablet, 2 phone)
- New discovery_artist_blacklist table with NOCASE name matching
- Filter blacklisted artists from all 6 discovery pool queries, hero
endpoint, and recent releases via SQL subquery and Python set check
- Name-based filtering means one block covers all sources (Spotify/iTunes/Deezer)
- Hover any discovery track row → ✕ button to quick-block that artist
- 🚫 button on Discover hero opens management modal with search-to-add
(powered by enhanced search) and list of blocked artists with unblock
- CRUD API: GET/POST/DELETE /api/discover/artist-blacklist
- Updated changelogs
- Add MusicBrainz to Cache Browser: stats pill, source filter, dedicated
browse endpoint, cards with matched/failed status indicators
- Add Clear MusicBrainz and Clear Failed MB Only to cache clear dropdown
- Move MusicBrainz into Cache Health "By Source" bar chart alongside
Spotify/iTunes/Deezer instead of isolated metric row
- Rename ambiguous "Failed Lookups" to "Failed MB Lookups" in summary cards
- Add browse-musicbrainz and clear-musicbrainz API endpoints
- Add musicbrainz_total/musicbrainz_failed to cache stats response
- Add Global Search Bar and MusicBrainz cache to changelogs
Genius rate limits are undocumented but users hit 429s at ~40 req/min.
Bumping interval from 1.5s to 2s drops throughput to ~30 req/min which
stays under the threshold. No functional change — just slower enrichment.
The manual match modal for Genius only returned 0 or 1 artist result
because search_artist() searched songs (per_page=5), extracted the
primary artist, and returned the first match or None.
Added search_artists() that returns multiple unique artists extracted
from song results with broader search (per_page=20). The manual match
endpoint now shows up to 8 artist candidates and multiple track results
instead of one-or-nothing. Also shows the Genius URL as extra info.
The previous commit only added skip_cache=True to one call site in
web_server.py. The watchlist scanner in core/watchlist_scanner.py has
6 Spotify get_artist_albums calls that also need fresh data to detect
new releases. All now bypass cache. iTunes/Deezer calls are unaffected
(they don't have the skip_cache param, detected via hasattr check).
Five places in web_server.py called spotify_client.sp.search() directly,
bypassing the cached search_tracks()/search_artists() methods. Each
discovery worker (Tidal, YouTube, ListenBrainz, Beatport) was also
doubling API calls — sp.search() for raw data then search_tracks() for
Track objects.
Now all use cached methods only. Raw track data for album art is
retrieved from the metadata cache by track ID after matching. Also fixed
a pre-existing bug where Tidal discovery could pair stale Spotify raw
data with a newer iTunes match.
Bumped is_spotify_authenticated() probe cache TTL from 5 to 15 minutes
to reduce /v1/me calls (~288/day → ~96/day). Manual disconnect still
takes effect immediately via _invalidate_auth_cache().
The cache stores raw_data through _extract_fields which expects a dict
with a 'name' field. Storing a raw list caused silent AttributeError,
and storing a dict without 'name' triggered junk entity rejection
(empty string is in _JUNK_NAMES). Now wraps the albums list in a dict
with a valid name field so it passes validation and persists correctly.
The watchlist auto-scan needs fresh data from Spotify to detect new
releases, so it bypasses the cache added in the previous commit.
All other callers (UI browsing, completion badges, discography views)
continue to benefit from cached results.
get_artist_albums was making fresh API calls on every invocation with
no cache check, despite being one of the most called methods (discography
views, completion badges, watchlist scans). The method already cached
individual albums opportunistically but never checked for a cached result
before hitting the API.
Now follows the same check-then-fetch-then-cache pattern used by
get_album_tracks and get_artist. Cache key includes album_type param
so different queries (album,single vs compilation) are cached separately.
YouTube's auto-generated artist channels use the format "Artist - Topic"
as the channel name. This suffix was not being stripped during playlist
parsing, causing metadata discovery to fail (e.g., searching for
"Koven - Topic" instead of "Koven" on iTunes/Deezer).
Fixed in all three places where YouTube artist names are cleaned:
- web_server.py clean_youtube_artist() — playlist parsing
- ui/pages/sync.py clean_youtube_artist() — UI-side parsing
- core/youtube_client.py — search result fallback artist extraction
Large libraries (first import) can take longer than 10 seconds for
getArtists to respond. The short timeout caused the library fetch
to fail with 0 artists returned.
The dashboard status poll and hybrid connection check were pinging
slskd every 2 minutes regardless of download source, flooding logs
with connection errors when slskd wasn't running. Now only checks
slskd when the download mode is 'soulseek' or when 'soulseek' is
in the hybrid order. Hybrid mode also only checks sources in the
configured priority order instead of all six.
The metadata-only optimization broke two things:
1. Cards showed 0 tracks because tracks were no longer in the listing
2. Auto-mirror skipped all playlists because tracks array was empty
Fix: cards render instantly from metadata, then tracks are fetched
per-playlist in the background via /api/tidal/playlist/<id>. As each
playlist's tracks arrive, the card count updates and the playlist
is auto-mirrored. Also tried multiple V2 attribute names for track
count (numberOfTracks, numberOfItems, etc.) and fixed the card DOM
selector for count updates.
get_user_playlists_metadata_only() was fetching full track lists for
every playlist sequentially (1+ API calls per playlist with 1s sleep
between pagination pages). For 20+ playlists this took 30-60 seconds.
Now returns only metadata (name, ID, track count, image) from a
single V2 API call. Track count comes from the numberOfTracks
attribute. Tracks are fetched on-demand when the user selects a
specific playlist to sync/mirror via the existing get_playlist()
endpoint.
Fresh installs now default to hybrid download mode (HiFi → YouTube →
Soulseek) instead of Soulseek-only, and Deezer as the metadata
fallback source instead of iTunes. Existing users with saved settings
are unaffected — defaults only apply when config keys don't exist.
The original #221 fix only covered Genius and AudioDB. All other
workers (Spotify, iTunes, Last.fm, MusicBrainz, Deezer, Tidal,
Qobuz) had the same bug: enrichment overwrites manual match status
to not_found when name search fails. Each worker now checks for an
existing service ID before searching by name and returns early if
one exists, preserving the manual match.
When spotipy exhausts all retries on 429 errors, the actual
Retry-After value (often 10+ hours) is consumed internally by
spotipy and not passed in the exception. The default ban was
only 1 hour, causing an endless retry cycle. Increased to 4
hours to match the escalation max and give Spotify's ban time
to expire.
normalize_string() was running unidecode on all text, converting
Japanese kanji to Chinese pinyin gibberish (命の灯火 → "tvanimedei").
Now detects CJK characters (kanji, hiragana, katakana, hangul,
fullwidth forms) and skips unidecode for text containing them —
just lowercases instead. Non-CJK text (Latin accents, Cyrillic)
still goes through unidecode normally.
Plain (unsynced) lyrics were being saved with .lrc extension despite
having no timestamps, making them invalid for Plex and other players
that expect LRC format. Synced lyrics now write as .lrc, plain lyrics
write as .txt. Both types still get embedded in audio file tags.
Updated all file move/rename operations to handle .txt sidecars
alongside .lrc.
When a user manually matched an artist to a service ID then triggered
enrichment, the worker re-searched by name, failed to find a match,
and overwrote the status back to not_found — despite the ID being
valid. Now both Genius and AudioDB workers check for existing service
IDs before searching by name. If an ID exists (from manual match),
the worker uses it for a direct API lookup to enrich metadata while
preserving the matched status. Added AudioDB lookup-by-ID client
methods for artist, album, and track.
The tooltip only checked paused/authenticated/idle/running states.
When Spotify was rate limited or daily budget exhausted, the worker
thread was still alive (sleeping in guards) so it showed "Running"
with no current item and stale 0% progress.
Now checks rate_limited and daily_budget.exhausted before running:
- Rate limited: "Rate Limited — Waiting Xm for rate limit to clear"
- Budget exhausted: "Daily Limit Reached — Resets in Xh Xm"
- No current item: "Waiting for next item..." instead of blank
Also adds rate_limit info object to get_stats() response for the
countdown display.
Cache maintenance:
- Input validation rejects junk entities (Unknown Artist, empty names)
from being cached, with exemptions for synthetic entries (_features,
_tracks suffixes)
- CacheEvictorJob expanded to 4 phases: TTL eviction, junk cleanup,
orphaned search cleanup, MusicBrainz failed lookup cleanup
- MusicBrainz null results now expire after 30 days (was 90) so failed
lookups get retried sooner
Cache health UI:
- Polished modal accessible from Dashboard "Cache Health" button and
repair dashboard health bar
- Shows health status banner (healthy/fair/poor), stat cards, source
breakdown with colored progress bars, type pills, and metrics table
- Repair dashboard shows compact bar with health dot indicator
When the fingerprint score is >=0.95 but title/artist don't match
(e.g. English expected vs Japanese returned), SKIP instead of FAIL.
A 95%+ fingerprint means the audio IS the correct recording — the
metadata mismatch is just a language/script difference, not a wrong
file. Prevents Japanese, Chinese, Korean, and other non-Latin tracks
from being falsely quarantined.
Happy path unchanged — matching title/artist still returns PASS at
the earlier check before this code is reached.
When artist or title contains non-ASCII characters (Japanese, Chinese,
Korean, etc.), prepend the original un-romanized text as the first
search query. unidecode converts Japanese kanji to Chinese pinyin
(e.g. "藤澤慶昌" → "wu zhi zhuan sheng") which never matches on
Soulseek. The original characters match filenames directly.
Romanized fallback queries are still generated after for coverage.
Zero impact on ASCII-only tracks (isascii check skips them).
fix_finding() was using a potentially stale transfer_folder that was
only refreshed during scheduled job runs, not on manual fix attempts.
Now re-reads the transfer path from config before each fix, matching
the same logic used by _run_next_job().
Also surfaces fix failure reasons to the user — bulk fix now logs each
failure with finding ID and error, and the frontend toast shows the
actual error message instead of just "X failed".
Artist/album/track matching previously took the first Spotify search
result above the 0.80 similarity threshold. If Spotify returned a
near-match before the exact match (e.g. "Brother's Keeper" before
"Brothers Keepers"), the wrong entity would be selected.
Now scores all candidates and picks the highest, so an exact match
(1.0) always wins over a near-match (0.94). No change to threshold
or batch matching logic — strictly better or equal results.
The background enrichment worker now caps itself at 3,000 processed items
per calendar day. Counter resets at midnight automatically. When exhausted,
the worker sleeps and checks every 5 minutes for a new day.
This is scoped entirely to the enrichment worker — user-initiated Spotify
API calls (searches, playlist ops, album lookups, etc.) are completely
unaffected. Budget status is exposed in the worker's get_stats() response
for the dashboard widget.
The image update code only checked Spotify's images array format,
missing iTunes direct image_url. Also used spotify_artist_id for
DB lookup which missed Deezer-only artists. Now handles both image
formats and tries all provider IDs for the DB update.
Old Deezer watchlist artists missing images will get backfilled
on the next watchlist scan automatically.
OggOpus (Mutagen) is a separate class from OggVorbis but uses the
same VorbisComment tag API. All isinstance checks for (FLAC, OggVorbis)
missed OggOpus, so tags were cleared but never rewritten — resulting
in empty metadata and "Unknown Artist" on the media server.
Added _is_ogg_opus() helper and applied to all 15 format checks in
web_server.py and 2 in core/tag_writer.py. No change to FLAC/OGG/
MP3/M4A handling (short-circuit evaluation skips the new check).
The fix was failing silently — returning error results without
writing to any log. Added info/warning logs at each failure point
(missing paths, source not found, path escapes transfer, destination
conflict) so Docker path resolution issues can be diagnosed.
New setting in Settings → Library → File Organization: "Collaborative
Album Artist" — choose between first listed artist (default) or all
artists combined for $albumartist in folder paths and album_artist tag.
Per-source resolution:
- Spotify: artists array has separate objects — picks first directly
- Deezer: API already returns first artist only — no change needed
- iTunes: combined string ("Larry June, Curren$y & The Alchemist") —
resolves via artistId API lookup to get primary name ("Larry June").
Safe for "Tyler, the Creator" and "Simon & Garfunkel" because their
IDs resolve to the same combined name (no change).
Applied to both folder path ($albumartist template) and album_artist
metadata tag for consistency. Track artist tag always keeps all artists.
iTunes lookup only fires when source is iTunes (numeric ID + not Deezer).
When artist-specific track search fails, falls back to album-aware
matching: finds the album by title (any artist), then checks if the
track exists on it. Fixes daily re-downloads of collaborative albums
filed under a different artist (e.g., "Spiral Staircases" tagged
under "The Alchemist" but scanned from "Larry June's" watchlist).
- check_track_exists: new album parameter, album-aware fallback with
0.8 album title threshold + 0.7 track title threshold
- Watchlist scanner: passes album_data.get('name') to track checks
- Download modal: passes batch_album_context to fallback track search
- Wishlist callers (4 spots): extract and pass track album name
- Backwards compatible: album=None default, no change for callers
without album context (singles, playlists)
Root cause: search_albums cached album data without release_date or
track_position. get_album_metadata accepted this incomplete cache hit
(only checked for title), never called the full API. Result: year
empty, all tracks numbered 01.
Fix: get_album_metadata now requires release_date in cache to use it.
Search results without release_date trigger a fresh /album/{id} API
call which returns complete data. Also improved track_number fallback
in download context builder when Spotify isn't configured.
New "Scan Lookback" dropdown in the watchlist artist config modal.
Each artist can override the global lookback period (7d to entire
discography). Default is "Use Global Setting" (NULL in DB).
- Database: lookback_days INTEGER DEFAULT NULL on watchlist_artists,
auto-migrated on startup
- Scanner: checks per-artist lookback_days first, falls back to
global discovery_lookback_period if NULL
- Backend: GET/POST /api/watchlist/artist/<id>/config includes
lookback_days. Changing lookback clears last_scan_timestamp to
force a rescan with the new window
- Frontend: dropdown with 8 options in artist config modal
- Fully backwards compatible — existing artists unchanged
Root cause: two issues compounding.
1. extractor_args with player_client: ['android', 'web'] + skip: ['hls', 'dash']
stripped all real audio formats. Android client returns HLS/DASH streams,
skip removes them, leaving only storyboard thumbnails. bestaudio then fails
because there's nothing valid to select.
2. Browser cookies (cookiesfrombrowser) cause authenticated YouTube sessions
to return restricted format data for some videos. Same video works fine
without cookies.
Fix:
- Removed all hardcoded player_client and skip overrides from 4 locations
(download opts, connection check, search, retry). Let yt-dlp use its own
defaults which are updated with each release.
- Retry strategy: attempt 1 uses cookies (respects user setting), attempt 2
drops cookies (fixes auth-restricted formats), attempt 3 uses format 'best'
as last resort.
- Updated user_agent to Chrome 131.
Navidrome fix:
- createPlaylist and other write operations now use POST instead of
GET. Large playlists (161+ tracks) exceeded URL length limits when
all songId params were in the query string, causing silent truncation
(e.g., only 6 of 161 tracks added). POST sends params as form body
with no size limit.
- Write operation timeout bumped to 30s (was 10s)
- _WRITE_ENDPOINTS set defines which Subsonic endpoints use POST
UX fix:
- Mirrored playlist cards now show "161/161 discovered on Spotify"
instead of just "161/161 discovered" — clarifies that discovery
means metadata matching, not library ownership
Helper system phases 2-7:
- Setup Progress: onboarding checklist with progress ring, auto-detection
via /status, /api/settings, /api/library, /api/watchlist, /api/automations
- Quick Actions: accent pill buttons in popovers (service cards get
"Open Settings" and "View Docs" actions)
- Keyboard Shortcuts: full-screen overlay with key cap styling, grouped
by scope (Global, Player, Helper, Forms)
- Search: fuzzy search across 200+ help entries, 11 tours, and shortcuts
with cross-page navigation via _guessPageFromSelector()
- What's New: version-tagged highlights with "Show me" navigation,
red badge on ? button for unseen versions, older version cycling
- Troubleshoot: scans dashboard service cards for disconnected/error
states, shows fix steps with action buttons, "All Clear" when healthy
- Contextual menu: page-aware tour suggestion at top of menu
- Ctrl+K / Cmd+K opens helper search globally
- First-launch welcome tooltip with pulsing ? button
- Redesigned floating button (48px, accent gradient, glass effect)
- Redesigned menu (unified card panel, accent left-stripe on contextual)
Enrichment worker fixes:
- AcoustID: individual recording matches downgraded INFO→DEBUG to reduce
log noise (14 lines for one track → 1 summary line)
- Name normalization: strip " - Suffix" dash format (Spotify) same as
"(Suffix)" parens format across all 8 workers. Fixes false mismatch
on tracks like "Electric Eyes (Studio Brussels Remix)" vs
"Electric Eyes - Studio Brussels Remix" (was 0.54, now matches)
Two changes to the Soulseek quality filter:
1. Sort candidates by effective bitrate first, peer quality second.
Previously a 16-bit FLAC from a fast peer could beat a 24-bit FLAC
from a slower peer. Now highest audio quality always wins within the
same format tier, with peer speed as tiebreaker.
2. Enforce the FLAC bit depth UI preference (Any/16-bit/24-bit) that
was previously cosmetic-only. Uses effective bitrate threshold of
1450 kbps to distinguish 16-bit (<1411 kbps theoretical) from 24-bit
(>2116 kbps theoretical). Respects the bit_depth_fallback toggle —
when enabled, gracefully accepts any FLAC if preferred depth is
unavailable. When disabled, strictly rejects non-matching depth.
Default bit_depth="any" — zero behavior change for existing users.
The Deezer enrichment worker's 5 raw API methods (search_artist,
search_album, search_track, get_album_raw, get_track_raw) were
bypassing the metadata cache — every enrichment cycle hit the
Deezer API fresh for every item. Spotify and iTunes workers properly
cached all results.
Now all 5 methods store results via store_entity(). get_album_raw
and get_track_raw also check cache first (verified by presence of
full-detail fields like 'label' and 'bpm' to distinguish from
search stubs). Cache failures are silently ignored to never block
enrichment. This eliminates redundant API calls and automatically
populates genre data for the genre explorer.
Genre explorer and deep dive modal now combine data from all available
metadata sources (iTunes + Deezer always, Spotify when authenticated).
Artists are deduplicated by name across sources, preferring entries
with images. Source dots (green/red/purple) indicate data origin.
Deezer genre support:
- Extract genre_id from Deezer album search responses via ID-to-name
mapping table (26 Deezer genre categories)
- Extract full genre names from Deezer get_album responses
- One-time backfill updates existing cached albums from stored raw_json
- Propagate album genres to Deezer artist entities
Cross-source album routing:
- /api/discover/album endpoint uses source-specific client (iTunes or
Deezer) based on the item's source, not just the active fallback
- Spotify path falls back to active fallback when album not found
- Track clicks use album_id directly instead of name-based resolution
- resolve-cache-album adds partial match and live search fallback
Other fixes:
- Genre explorer positioned at top of Discover page (below hero)
- Genre explorer results cached 24hr in-memory for fast reload
- Related genres computed from all albums by matched artists
- Artist clicks open Artists page with discography (not library detail)
- Discovery pool genre queries restored to source-filtered (Browse by
Genre tabs stay source-isolated as designed)
The genre query filtered by active source (spotify/deezer/itunes)
but discovery pool entries keep their original source. Switching
metadata sources caused all genres to disappear. Removed the source
filter since artist genres are source-agnostic metadata.
Users who intentionally don't use Soulseek were getting spammed with
ERROR-level logs every second. These are expected when Soulseek isn't
configured as a source and don't need user attention.
Deezer client imports config_manager locally in __init__ and stores
it as self._config, unlike the other clients which import at module
level. The allow_fallback line was referencing the wrong name.
Each streaming source (Tidal, Qobuz, HiFi, Deezer) now has an "Allow
quality fallback" checkbox in Settings. When disabled, the source only
tries the exact quality selected — if unavailable, it skips and lets
the orchestrator try the next source. Default is ON (current behavior).
When missing_tracks was empty (API unavailable at scan time) or the
album was completed between scan and fix, the handler returned a 400
error. Now it re-fetches missing tracks from Spotify/iTunes/Deezer at
fix time and auto-resolves findings where the album is already complete.
Enhanced album reorganize now moves LRC, cover.jpg, folder.jpg and other
sidecar files alongside audio files. Added debug log to library reorganize
repair job showing which template is loaded from config vs default.
New algorithm: pick first track alphabetically from artist's library, search
both Deezer and iTunes APIs for 'artist track' to find the exact artist,
verify name matches, then use max(deezer_id, itunes_id) as the canonical
differentiator. Deterministic across instances — any SoulSync with the same
artist and at least one matching track will produce the same soul_id.
Fallback chain: canonical API ID → first album title → name only.
Migration clears all artist soul_ids on first v2.1 startup for regeneration.
Method exists on iTunesClient and DeezerClient but was missing from
HydrabaseClient, causing AttributeError when Hydrabase is active and
Build a Playlist tries to fetch artist images.
Before album tracks start post-processing, pre-populate the MusicBrainz
release cache with ONE verified release. All tracks then hit the cache
during per-track processing and get the same release MBID — no second pass,
no file lock contention, tags correct on first write.
Handles edition name mismatches (Spotify says 'Super Deluxe', MB says
just the base name) by stripping qualifiers and searching again.
Validates track count to pick the right edition. Three download paths
covered: download missing modal, enhanced album, and matched album.
New maintenance job scans albums for tracks with mismatched album names, album
artist names, or MusicBrainz release IDs. These inconsistencies cause Navidrome
and other media servers to split one album into multiple entries. The fix
normalizes outlier tracks to the majority value by rewriting file tags.
Event-triggered automations now receive playlist_id from the triggering event
when the action config doesn't have one set. Fixes silent 'No playlist specified'
failures in Discover/Sync chains. Added debug logging to trace event matching.
- All 9 enrichment workers: stop auto-retrying 'error' status items (was infinite loop)
Only 'not_found' items retry after configured days; errors require manual full refresh
- Cover art dedup: check both 'pending' AND 'resolved' findings to prevent recreation
- Cover art scanner: top-level Spotify rate limit check skips Spotify entirely when
banned, falls back to iTunes/Deezer only, logs once instead of spamming 429s
- Stats page: database storage donut chart with per-table breakdown and total size
- Discover page: 5 new sections mined from metadata cache (zero API calls):
Undiscovered Albums, New In Your Genres, From Your Labels, Deep Cuts, Genre Explorer
- Genre Deep Dive modal: artists (clickable → artist page), popular tracks,
albums with download flow, related genre pills, in-library badges
- All cache queries filtered by active metadata source (Spotify/iTunes/Deezer)
- Stale cache entries (404) gracefully fall back to name+artist resolution
- Album cards show "In Library" badge, artist avatars scaled by prominence
Integrates play history data into the discovery algorithm:
- Listening profile: _get_listening_profile() builds user's top artists,
genres, play counts, and listening velocity from the last 30 days
- Artist genre cache: pre-built from local DB for O(1) genre lookups
- Release Radar: +10 genre affinity, +15 artist familiarity, -10 overplay
penalty. Weights rebalanced to 45% recency + 25% popularity + bonuses
- Discovery Weekly: serendipity scoring within tiers — boosts unheard
artists in preferred genres, penalizes overplayed artists
- Recent Albums: adaptive time window (21-60 days) based on listening
velocity — heavy listeners get fresher content, casual listeners more
- New "Because You Listen To" sections: personalized carousels based on
user's top 3 played artists via similar artists + genre fallback
- New endpoint: /api/discover/because-you-listen-to with artist images
- Frontend: BYLT sections with artist photo headers on discover page
- All changes gracefully fall back when no listening data exists
Full stats dashboard that polls Plex/Jellyfin/Navidrome for play
history and presents it with Chart.js visualizations:
Backend:
- ListeningStatsWorker polls active server every 30 min
- listening_history DB table with dedup, play_count/last_played on tracks
- get_play_history() and get_track_play_counts() for all 3 servers
- Pre-computed cache for all time ranges (7d/30d/12m/all) rebuilt each sync
- Single cached endpoint serves all stats data instantly
- Stats query methods: top artists/albums/tracks, timeline, genres, health
Frontend:
- New Stats nav page with glassmorphic container matching dashboard style
- Overview cards (plays, time, artists, albums, tracks) with accent hover
- Listening timeline bar chart (Chart.js)
- Genre breakdown doughnut chart with legend
- Top artists visual bubbles with profile pictures + ranked list
- Top albums and tracks ranked lists with album art
- Library health: format breakdown bar, unplayed count, enrichment coverage
- Recently played timeline with relative timestamps
- Time range pills with instant switching via cache
- Sync Now button with spinner, last synced timestamp
- Clickable artist names navigate to library artist detail
- Last.fm global listeners shown alongside personal play counts
- SoulID badges on matched artists
- Empty state when no data synced yet
- Mobile responsive layout
DB migrations: listening_history table, play_count/last_played columns,
all with idempotent CREATE IF NOT EXISTS / PRAGMA checks.
Opus ffmpeg command now uses -map 0:a to extract only the audio stream,
matching the user's working command and preventing picture stream
interference with the encoder. After conversion, cover art is embedded
from the source FLAC using Mutagen: Opus gets METADATA_BLOCK_PICTURE
(base64-encoded per OGG spec), AAC gets MP4Cover. Applied to both
the post-download lossy copy and the repair job fix handler.
The lossy converter fix handler now reads codec and bitrate fresh
from the Settings page when converting, not from what was stored
in the finding details at scan time. Users can change their codec
preference after scanning and Fix All will use the new setting.
The lossy converter fix handler now reads delete_original from its
own job settings (repair.jobs.lossy_converter.settings.delete_original)
instead of the global lossy_copy.delete_original. Defaults to false.
Separate from the per-download Blasphemy Mode toggle in Settings.
New library maintenance job that scans for FLAC files missing a lossy
copy (MP3/Opus/AAC) and creates findings. Fix action converts via
ffmpeg using the configured codec/bitrate from Settings. Supports
Blasphemy Mode (delete original + update DB path). Finding details
store codec/bitrate from scan time for consistency. Disabled by
default, manual-run only, no auto-schedule.
- Add get_artist_top_tracks to Last.fm client (up to 100 tracks)
- Include lastfm_listeners, lastfm_playcount, lastfm_tags, lastfm_bio,
and soul_id in artist detail API response
- New endpoint: /api/artist/<id>/lastfm-top-tracks for lazy loading
- Hero layout: image (160px) | center (name, badges, genres, bio,
listener/play stats, progress bars) | right card (scrollable top
100 tracks from Last.fm)
- Badges 36px with hover lift, bio in subtle card with Read More
toggle, Last.fm tags merged with existing genres
- Numbers formatted: 1234567 → 1.2M
- Graceful degradation: sections hidden when Last.fm data unavailable
Orphan detector: add normalized tag matching that strips parentheticals
and brackets (feat. X, [FLAC 16bit], etc.) and tries first-artist-only
for comma-separated artists. Prevents false orphan flags for tracks
like "The Mountain (feat. Dennis Hopper...)" that exist in DB as
"The Mountain". All lookups remain O(1) set operations.
Orphan fix: replace auto-delete with user choice prompt. Single Fix
and Fix All both show modal asking "Move to Staging" or "Delete".
Move to Staging relocates file to import staging folder for proper
re-import with metadata matching. Fix action flows through API
endpoint → repair_worker.fix_finding → _fix_orphan_file handler.
Staging path uses docker_resolve_path for container compatibility.
When searching for "Playing the Angel (Deluxe)", MusicBrainz would
match the standard release because its higher popularity score
outweighed the lower title similarity. Now match_release extracts
version qualifiers (Deluxe, Remastered, etc.) from both the query
and each result, applying a +10 bonus for exact qualifier match,
+5 for partial, and -5 penalty when the query has a qualifier but
the result doesn't. This ensures deluxe downloads get the deluxe
MBID, preventing Navidrome from splitting albums.
SoulID worker generates deterministic soul IDs for all library entities:
- Artists: hash(name + debut_year) — searches iTunes + Deezer APIs,
verifies correct artist by matching discography against local DB
albums via MusicMatchingEngine, pools years from both sources and
picks the earliest. Falls back to hash(name) if no match found.
- Albums: hash(artist + album)
- Tracks: song ID hash(artist + track) + album ID hash(artist + album + track)
Dashboard button with trans2.png logo, rainbow spinner, hover tooltip.
Worker orb with rainbow effect. SoulSync badge on library artist cards.
DB migration adds soul_id columns with indexes to artists/albums/tracks.
Migration version flag auto-resets artist soul IDs when algorithm changes.
- Use correct server request types: 'tracks', 'albums', 'artists',
'artist.albums', 'album.tracks' (were singular, caused timeouts)
- Normalize artists to strings (server may send dicts)
- Use native plugin IDs (iTunes/Spotify) instead of soul_id for
album/artist/track IDs so downstream endpoints can resolve them
- Carry soul_id and plugin_id in external_urls for routing
- Pass plugin param from frontend to server for correct client routing
(iTunes vs Deezer vs Spotify) with isdigit() fallback
- Route source=hydrabase to iTunes client for artist images
- Include external_urls in enhanced search API response
- Reduce WebSocket timeout from 15s to 8s
- Remove stale hydrabase.enabled check, use is_connected() directly
- Add hydrabase to frontend alternate source fetch list
- Normalize Hydrabase artists to strings (server may send dicts),
fixing silent crashes that prevented albums/tracks from appearing
Implements get_track_details, get_album, get_artist, get_artist_albums,
get_track_features, is_authenticated, and reload_config to match the
iTunes/Deezer client interface. Each method tries a specific request
type first then falls back to search if the server doesn't support it.
Preserves existing get_album_tracks (List[Track]) for backward compat.
Post-processing now extracts release year from MusicBrainz, Deezer,
Tidal, Qobuz, and Spotify context (first source wins). Writes
ORIGINALDATE and DATE tags to file, and backfills the album year
in the DB if currently missing. Fixes Library Reorganize showing
blank years for Tidal-only downloads.
Also raises Library Reorganize API year lookup cap from 50 to 200.
- Add debouncedAutoSaveSettings() to moveHybridSource and toggleHybridSource
- Skip unconfigured sources at search time with is_configured() check
- Add get_source_status() to orchestrator, include in settings API response
- Auto-disable unconfigured sources in UI on settings load
- Remove redundant enable checkbox — fallback dropdown is the enable
- Hydrabase option only appears in dropdown when connected
- Connect/disconnect dynamically adds/removes dropdown option
- _is_hydrabase_active checks fallback_source == hydrabase (not config toggle)
- Fallback client returns hydrabase_client when selected, iTunes if disconnected
- Auto-reconnect respects fallback selection for dev_mode handling
- hydrabase added to settings save service list for persistence
- Status shows green Connected on page load when auto-connected
Worker checked self.client.sp (non-None even without Spotify auth due
to fallback) instead of is_spotify_authenticated(). Searched via
iTunes/Deezer fallback, got numeric IDs, rejected them all with
warnings. Now sleeps when Spotify isn't authenticated instead of
making pointless fallback searches.
Pending count queries included NULL-ID rows that _get_next_item filters
out, so pending stayed > 0 even when no processable items remained.
Workers reported running instead of idle, UI never turned green. Added
AND id IS NOT NULL to _count_pending_items across all 9 workers to
match the _get_next_item filter.
Workers would endlessly match the same track because UPDATE WHERE id =
NULL matches 0 rows in SQL. Added AND id IS NOT NULL to all enrichment
queries (individual, batch EXISTS, and batch fetch) across all 9
workers. Also added process-level guard for belt-and-suspenders safety.
Fix Deezer get_track → get_track_details method name mismatch.
New setting: Min Completion % (default 0 = disabled). Skips albums
where the user has fewer than N% of tracks — filters out playlist
imports where a single track exists but the full album was never
intended to be downloaded. Catches real failed downloads (6/12)
while ignoring incidental singles (1/12).
Skip refresh attempt if client_id/secret not configured — clears stale
tokens to stop retry loop permanently. Add 5-minute backoff after
failed refresh to prevent hammering Tidal API every 2 seconds.
Wishlist entries from auto-fill were missing album images, album ID,
track number, disc number, and total tracks. Downloads would have no
cover art and wrong file organization. Now includes full album context
matching the standard wishlist data format.
When suffix path matching fails, reads file tags (title+artist) and
checks against DB tracks. Prevents false orphan detection from path
mismatches in Docker where DB paths differ from filesystem paths.
Only runs for files that fail suffix matching — zero overhead in
normal cases.
- Only emit batch_complete automation event when successful_downloads > 0
- Remove int() cast on album_id in _fix_incomplete_album — Plex/Jellyfin
use hash/text IDs that SQLite stores as-is in INTEGER columns
- Use string comparison for album_id equality check
Finding was created with entity_id=None (file-based) but fix handler
required entity_id for DB update. Rewrote handler to work with file_path
as primary — writes corrected track number to file tags, renames file
if needed, updates DB path. Also stores total_tracks in finding details
for correct tag writing.
Replace fixed primary/secondary hybrid dropdowns with an ordered list
of all 5 download sources. Users enable/disable each source and reorder
with up/down arrows to set download priority. Sources are tried in
order until one returns results.
- New hybrid_order config field (backward compat with legacy primary/secondary)
- Download orchestrator loops ordered list with per-source error handling
- Sortable source list UI with icons, toggle switches, priority numbers
- Source-specific settings shown for all enabled hybrid sources
- Seamless migration from legacy 2-source to N-source format
- Add max_peer_queue setting to skip peers with long queues (soft filter
with fallback to unfiltered if all results removed)
- Add download_timeout setting replacing hardcoded 10-minute limit
- Include quality_score (peer health: upload speed, free slots, queue
length) in result ranking — was calculated but never used in sort key
- New UI controls in Soulseek settings section
Was Spotify-only — users without Spotify got zero results. Now queries
albums with any source ID (spotify_album_id, itunes_album_id, deezer_id)
and uses the matching API client for track count and missing track lookup.
Falls back gracefully across sources with client-type detection.
When >50% of files are flagged as orphans (likely a DB path mismatch),
findings are marked as warnings with mass_orphan flag. Fixing these
requires typing "witness me" to confirm — prevents nuking an entire
library from a false-positive orphan scan.
_sanitize_context_values passed empty strings through _sanitize_filename
which converts '' to '_' (empty-name fallback). Template $album ($year)
became Album (_) instead of Album () which the cleanup regex couldn't
match. Now preserves empty strings so the existing () cleanup works.
- Add _fix_path_mismatch handler so fixing Library Reorganize dry-run
findings actually moves files (was returning 400 with no handler)
- Add path_mismatch to fixable_types and _execute_fix handler map
- Add recent_releases and wishlist_tracks as year sources in
_load_album_years to cover more playlist-synced tracks missing years
- Add sys and json imports needed by new code
- Add single_album_redundant to fixable_types in bulk_fix_findings so
Fix All actually includes these findings (Fix Selected worked, Fix All
silently returned 0)
- Expand version keyword regex from 9 to 25 terms (remastered, deluxe,
unplugged, etc.) to reduce false positives in Single/Album Dedup
- Add word boundary anchors to prevent substring matches (e.g. "live"
inside "Alive", "edit" inside "Meditate")
- Cast similarity thresholds to float for config type safety
The SQL HAVING clause filtered on local track count instead of expected
track count, excluding albums with fewer than 3 local tracks from the
scan entirely. Now fetches all Spotify-matched albums and filters by
expected track count in the loop.
- Frontend was concatenating Track and Artist inputs into a single
query string, causing Spotify to return mixed results matching
either word in any field. Now sends track and artist as separate
params; backend builds field-filtered query (track:X artist:Y).
- Result limit was silently capped at 10 in spotify_client.search_tracks
via min(limit, 10). Raised to respect requested limit up to
Spotify's API max of 50.
- iTunes fallback endpoint updated with same field-specific params.
- Legacy ?query= param still supported for backward compatibility.
Fixes#194
EPs are no longer classified as singles — removing EP tracks would
break the release and conflict with album completeness checks.
Only actual singles (album_type='single' or unknown type with <=2
tracks) are flagged as redundant when the same song exists on an album.
Tracks on different albums (e.g., same song on a studio album and a
compilation) are no longer flagged as duplicates by default, keeping
albums complete. Setting is enabled by default and configurable via
the job settings UI.
Playlist-synced tracks often have no year in file tags or the DB albums
table. Now checks discovery_pool release_date as a second DB source,
then batch-lookups remaining missing years from the user's active
metadata source (Spotify/iTunes/Deezer) capped at 50 API calls.
Multiple code paths built wishlist album data as a bare string instead
of a dict with images/album_type/total_tracks, causing wishlist entries
to lose album name and artwork. Fixed in: Beatport fallback conversion,
sync normalization, _ensure_spotify_track_format, and wishlist service
track reconstruction. Now uses track name as album fallback and always
includes required album dict keys.
New dropdown in Soulseek settings lets users filter out slow peers at
search time (Any/1/2/3/4/5/10 Mbps). Passes minimumPeerUploadSpeed
to slskd API in bytes/sec.
Also fixes quality scoring tiers which were using wrong units — old
thresholds (5000, 1000, 500) treated bytes/sec values as kbps,
making speed scoring effectively meaningless. Now uses correct
bytes/sec thresholds based on real peer data.
Deep scan passed skip_existing_tracks=True which skipped calling
insert_or_update_media_track entirely for known tracks, so stale
file paths were never refreshed from the media server. Now always
calls insert_or_update (which safely uses UPDATE for existing tracks,
preserving enrichment data) so file paths stay in sync.
Normalization was stripping parenthetical content before comparison,
so 'title' and 'title (xxx remix)' both became 'title' and always
matched at 1.0 regardless of the user's threshold setting.
_get_source() hardcoded 'itunes' as fallback, so seasonal content was
stored with source='itunes' but API endpoints queried source='deezer' —
resulting in empty discover page seasons. Now reads the configured
fallback source from metadata_service.
Dead file Fix button now adds the track to wishlist for re-download
instead of just removing the DB entry. Builds full wishlist-compatible
track data from DB (artist, album, artwork, IDs) so the download
pipeline can process it like any other wishlist item.
- Add null guard for soulseek_client in test-connection endpoint (#190)
- Fix single_album_dedup query: use al.record_type and al.track_count
(al.album_type and al.total_tracks don't exist on main albums table)
- Fix library reorganize missing year: pre-load album years from DB as
fallback when file tags lack the date field (common with playlist syncs)
- Fix enrichment progress never updating: remove `continue` that skipped
progress_callback for successful tracks in enrich_chart_tracks
- Split chart/extract into two-step flow: extract raw tracks, then enrich
via polling endpoint with live progress overlay updates
- Move Beatport enrichment cache to persistent metadata cache system
- Fix metadata cache detail modal for Beatport (URL entity_ids with slashes)
- Add per-source Clear dropdown (Spotify/iTunes/Deezer/Beatport) to cache browser
- Remove debug logging from enrichment progress tracking
When the maintenance worker flags an incomplete album, users can now
click "Auto-Fill" to automatically locate missing tracks in the library,
move/copy them into the album folder, and apply full metadata enhancement
(MusicBrainz, Deezer, cover art, etc.). Singles are moved; tracks from
multi-track albums are copied. Quality gate prevents filling FLAC albums
with lossy files. Tracks not found in library are added to wishlist with
album context for auto-download.
Both clients have their own Track class separate from iTunes. The
album preference commit added album_type/total_tracks to from_*_track()
constructors but not to the class definitions, crashing all Deezer and
Spotify discovery searches.
Scans library for tracks on singles/EPs that also exist on a full album,
letting users clean up fragmented libraries. Includes version-tag safety
(live, acoustic, remix etc. won't cross-match) and verifies the album
version still exists before removing a single.
Add album_type and total_tracks fields to Track dataclass, populate from
Spotify/iTunes/Deezer API responses, and apply a small tiebreaker bonus
(+0.02 for albums, +0.01 for EPs) in all matching loops so album versions
win when confidence scores are otherwise equal.
Deezer search results (cached without release_date, track_position, isrc)
were being served by get_track_details as cache hits, preventing the actual
/track/{id} API call that returns complete data. This caused $year template
variable to remain empty even with the discovery worker fix in place.
trigger_library_scan and is_library_scanning were no-ops assuming
Navidrome auto-detects new files. Navidrome supports startScan and
getScanStatus via the Subsonic API, so use them like Plex and Jellyfin.
- DB path update uses suffix matching when exact match fails, so server
paths (/mnt/musicBackup/...) are found and updated after local moves
- Escape LIKE wildcards in artist/album names to prevent false matches
- Move album-level sidecars (cover.jpg, folder.jpg, etc.) alongside audio
- Post-pass sweep moves all remaining sidecars from emptied directories
- Multi-disc albums place album art in album root, not disc subfolders
Use config_manager for transfer path (same source as playback) instead of
separate DB read that silently falls back to ./Transfer under contention.
Abort scan if transfer folder doesn't exist to prevent mass false findings.
- personalized_playlists._get_active_source() now returns 'deezer' when
configured instead of always falling back to 'itunes'
- Add deezer_track_id to _build_track_dict() for discovery pool tracks
- Include album_deezer_id and artist_deezer_id in get_discovery_recent_albums()
response — fixes "No deezer album ID available" error when clicking cards
- Skip Spotify library section entirely when Spotify is not authenticated
Per-job and per-status filtering — respects active toolbar filters so
users can clear e.g. only track number findings or only dismissed items.
Confirmation uses the app's styled modal instead of browser confirm().
Previously only 502/503/504 triggered instance rotation. A 500 from one
instance (e.g. triton.squid.wtf choking on a specific query) would stop
the search entirely instead of trying the next instance.
Users can now choose between iTunes/Apple Music and Deezer as their free
metadata source in Settings. Spotify always takes priority when authenticated;
the fallback handles all lookups when it's not.
Core changes:
- DeezerClient: full metadata interface (search, albums, artists, tracks)
matching iTunesClient's API surface with identical dataclass return types
- SpotifyClient: configurable _fallback property switches between iTunes/Deezer
based on live config reads (no restart needed)
- MetadataService, web_server, watchlist_scanner, api/search, repair_worker,
seasonal_discovery, personalized_playlists: all direct iTunesClient imports
replaced with fallback-aware helpers
Database:
- deezer_artist_id on watchlist_artists and similar_artists tables
- deezer_track_id/album_id/artist_id on discovery_pool and discovery_cache
- Full CRUD for Deezer IDs: add, read, update, backfill, metadata enrichment
- Watchlist duplicate detection by artist name prevents re-adding across sources
- SimilarArtist dataclass and all query/insert methods handle Deezer columns
Bug fixes found during review:
- Similar artist backfill was writing Deezer IDs into iTunes columns
- Discover hero was storing resolved Deezer IDs in wrong column
- Status cache not invalidating on settings save (source name lag)
- Watchlist add allowing duplicates when switching metadata sources
Add minimum 60% title similarity gate to match_recording() — prevents
artist bonus + MB score from pushing unrelated titles past the confidence
threshold (e.g. "Sweet Surrender" matching "Answers" by same artist).
New MBID Mismatch Detector repair job reads embedded MusicBrainz recording
IDs from audio files, verifies them against the MusicBrainz API, and flags
tracks where the MBID points to a different song. Fix action strips the bad
MBID tag so media servers like Navidrome fall back to correct file tags.
New download mode alongside Soulseek, YouTube, Tidal, and Qobuz. Uses
community-run REST API instances (no auth required) that serve Tidal CDN
FLAC streams. Features quality fallback chain (hires→lossless→high→low),
automatic instance rotation on failure, and full hybrid mode support.
Also fixes 6 missing streaming source checks for HiFi and Qobuz in the
frontend that were blocking playback with "format not supported" errors.
Scrapes Spotify's embed endpoint to extract track data from any public
playlist or album URL. Full discovery flow with Deezer parity: parse →
card → discovery modal → live progress → sync → download missing.
- New scraper: core/spotify_public_scraper.py (embed endpoint parsing)
- 12 API endpoints mirroring Deezer's discovery/sync/download flow
- WebSocket live discovery updates via spotify_public_discovery_states
- Green-branded tab, cards, and input styling (#1DB954)
- Album vs playlist detection with distinct card icons (💿/🎵)
- Download persistence: card click reopens download modal after close
- Card phase reset on cancel/completion (closeDownloadMissingModal)
- Backend download linking for spotify_public_ and deezer_ prefixes
- Completion handlers (V2 + no-missing-tracks) for both platforms
- Source URLs stored on mirrored playlists for future auto-refresh
Southern hemisphere users now see correct seasons (e.g. March = Autumn,
December = Summer). Holidays (Halloween, Christmas, Valentine's) stay
calendar-fixed regardless of hemisphere.
- Hemisphere dropdown in Discovery Pool Settings
- GET/POST /api/discovery/hemisphere endpoints
- Season detection offsets months by 6 for southern hemisphere
- Stored in metadata table, defaults to northern
Repair jobs were making Spotify API calls without checking the global
rate limit ban, causing them to churn through every item getting rejected
one by one during a ban. Now all Spotify calls in repair jobs check
context.is_spotify_rate_limited() first and skip/fallback gracefully.
- Add is_spotify_rate_limited() helper to JobContext (base.py)
- Guard calls in track_number_repair, album_completeness,
missing_cover_art, and metadata_gap_filler
- Jobs fall through to iTunes/MusicBrainz fallbacks when rate-limited
- All 9 repair jobs now emit report_progress() for real-time card updates
(phase, log lines, per-item activity) via WebSocket repair:progress events
- Enrich finding details with album/artist thumb URLs across all repair jobs
(dead_file, duplicate, metadata_gap, album_completeness, missing_cover_art,
acoustid_scanner, track_number_repair, fake_lossless, orphan_file)
- Track number repair: return match_score from fuzzy matching, add suffix-based
DB lookup for album/artist art (handles cross-environment path mismatches)
- Fix Plex/Jellyfin relative thumb URLs in findings endpoint via fix_artist_image_url
- Labeled media cards in finding detail panels (album title + artist name under images)
- Dashboard tooltip shows current job name + per-job progress instead of stale stats
- Add whisoul.png to modal header with subtitle text and gradient background
- Responsive: smaller logo on mobile, hide subtitle
- Share _resolve_file_path in repair_worker for cross-environment path compat
- Use path resolution in orphan and duplicate file deletion
- Guard directory cleanup against removing the transfer folder itself
- Restore correct button label text on fix error recovery
- Add findings dashboard with summary stats and per-job clickable filter chips
- Redesign findings cards with expandable detail panels and per-type renderers
- Redesign history tab with status dots, stat pills, and full timestamps
- Fix dead file cleaner false positives by using suffix-based path resolution
- Fix orphan file detector false positives by matching via path suffixes
- Add help text modal for each repair job card
- Enlarge maintenance modal (1100px wide, 90vh tall)
Allow multiple users to share a single SoulSync instance with isolated personal data. Each profile gets its own watchlist, wishlist, discovery pool, similar artists, and bubble snapshots — while sharing the same music library, database, and service credentials.
- Netflix-style profile picker on startup when multiple profiles exist
- Optional PIN protection per profile; admin PIN required when >1 profiles
- Admin-only profile management (create, edit, rename, delete)
- Profile avatar images via URL with colored-initial fallback
- Zero-downtime SQLite migration — all existing data maps to auto-created
admin profile
- Single-user installs see no changes — profile system is invisible until
a second profile is created
- WebSocket count emitters scoped to profile rooms (watchlist/wishlist)
- Background scanners (watchlist, wishlist, discovery) iterate all profiles
Incremental database updates now detect when artists or albums have been removed from your media server (Plex, Jellyfin, or Navidrome) and automatically clean them up from SoulSync's database. Previously, deleted content would persist as ghost entries until you ran a full refresh. Removal counts are reported in the scan results. Includes safety checks to prevent accidental mass deletion if the server is unreachable or returns incomplete data.
Adds Tidal as a third download source alongside Soulseek and YouTube. Uses the tidalapi library with device-flow authentication to search and download tracks in configurable quality (Low/High/Lossless/HiRes) with automatic fallback. Integrates into the download orchestrator for all modes (Tidal Only, Hybrid with fallback chain), the transfer monitor, post-processing pipeline, and file finder. Frontend includes download settings with quality selector, device auth flow, and dynamic sidebar/dashboard labels that reflect the active download source. No breaking changes for existing users.
Filled all missing Hydrabase fallthrough gates across 6 endpoints (artist album tracks, iTunes album, discover album, Spotify track, both similar artists), added track_number/disc_number to Track dataclass, fixed get_album_tracks to send soul_id instead of text query, mapped soul_id field from Hydrabase responses across all search methods, updated 7 frontend call sites to pass album name/artist params, and fixed M3U export defaulting to enabled when users never turned it on.
Fix short artist names (e.g. "ano") getting buried by wildcard matches. Re-ranks Spotify and iTunes results so exact name matches appear first, without changing what's returned.
SoulSync was importing from all Navidrome libraries regardless of user access restrictions. Added a "Music Library" dropdown in Navidrome settings that lets users scope imports to a specific music folder. Uses the Subsonic musicFolderId parameter on artist, album, and search API calls. Selecting "All Libraries" reverts to the previous behavior.
AcoustID sometimes returns featuring info in square brackets like [W/ Barnes Blvd.] instead of parentheses. The normalizer only stripped parenthetical featuring tags, so these tracks failed verification and got quarantined despite being correct. Now strips [W/ ...], [with ...], [feat. ...], and [ft. ...] bracket patterns too.
Store album artists in the Spotify track data structure (handles both dict and object album forms) so downstream processing has access to album-level artist metadata. Also update missing-track processing to fall back to source_info['watchlist_artist_name'] when artist_name is absent, ensuring artist context is preserved for legacy/watchlist-sourced entries.
Raise the default Jellyfin API timeout for bulk operations from 30s to 120s to better handle slow servers and large database syncs. Updated core/jellyfin_client.py to set bulk_timeout to 120, webui/index.html to show 120s in the settings input, and webui/static/script.js to use 120 as the fallback when loading settings. Aligns UI and client behavior to reduce timeout errors during heavy syncs.
Add a configurable API timeout for Jellyfin bulk requests and improve fetch retry behavior. UI: adds an "API Timeout (seconds)" field (15–300s, default 30) in webui/index.html and persists it via webui/static/script.js (load/save). Client: jellyfin_client.py now reads api_timeout from config_manager, uses it as the bulk timeout and computes a sensible non-bulk timeout (max(5, bulk_timeout//6)). Fetch loops for tracks and albums were hardened: reducing batch size now resets consecutive failure counters, log messages were clarified, and the stopping/retry thresholds were adjusted to avoid premature aborts at minimum batch sizes.
Introduce a generic score_track_match(...) in core/matching_engine.py and make calculate_match_confidence(...) delegate to it. The new scorer is source-agnostic, consolidates artist/title/duration logic (core-title fast path, cleaned similarity, weighted 60/30/10 scoring) and improves artist matching.
In web_server.py add cache-validation (_validate_discovery_cache_artist) and a reusable _discovery_score_candidates(...) helper that calls the new scorer. Propagate per-match confidence through discovery flows (Tidal, YouTube, ListenBrainz, Beatport), increase Spotify/iTunes search limits, add an extended high-limit search strategy, tighten per-source thresholds, and save match confidence to the discovery cache. Overall this centralizes and standardizes matching logic and improves accuracy/validation for cached discovery results.
Expand track title normalization and refine album grouping logic.
- core/acoustid_verification.py: Broadened the parenthetical-suffix regex to strip year-based remasters and additional variants (e.g. "2025 Remaster", "single edit", "album edit") while still removing common extras like (Live), (Deluxe), (Radio Edit), and featuring tags.
- web_server.py: Restrict the smart album grouping to only run for singles/auto-detected albums; explicit album downloads now preserve the original Spotify album name to avoid mangling names (e.g. reworked/remastered vs deluxe). Added explicit logging for both smart grouping and skipped grouping paths. The verification post-processing worker now checks an is_album_download flag in context and skips re-grouping when true, with fallback logging on errors.
These changes prevent unintended renaming of explicit album downloads and improve normalization of common title suffixes.
- matching_engine.py: Add 'single edit' and 'album edit' tokens and clarify radio edit comment so edit/cut variants are recognized as different cuts rather than being silently normalized away.
- database/music_database.py: Fix SQL param ordering by appending server_source to params; add a pre-step to strip "(with ...)" / "[with ...]" only when used inside brackets (so titles like "Stay With Me" are preserved); stop removing edit/version tokens in the generic cleanup and document that radio/single/album edits are treated as distinct by the similarity scorer to avoid incorrect matches.
- web_server.py: Increase DB match confidence threshold from 0.70 to 0.80 and update the runtime check accordingly.
These changes prevent edit/cut variants from being conflated with original recordings, improve title normalization for "with" featuring syntax in brackets, and fix a params ordering bug and a too-low match threshold.
Use UUID nonces to correlate requests/responses and robustly handle interleaved stats/heartbeat messages from the Hydrabase server. Adds _extract_stats and _extract_results helpers, records last_peer_count and timestamp, and loops on recv() with timeouts to drain non-result messages until the matching response (or a results message without a nonce) arrives. Mirrors the same nonce/send-and-drain logic in HydrabaseWorker, adds necessary imports (time, uuid), and improves logging and timeout handling to avoid returning stale or misattributed data.
Improve Hydrabase response handling and add discography/album track helpers. core/hydrabase_client.py: extract peer counts from stats.connectedPeers, handle new "response" key and stats-only or unexpected response shapes (return empty instead of wrapping), and add search_discography() and get_album_tracks() to map Hydrabase results into Album/Track objects. web_server.py: avoid redundant Hydrabase round-trips by passing precomputed hydrabase_counts into the background comparison worker; prefer Hydrabase for artist discography and album track import when active (with Spotify fallback); and route album-context searches to Hydrabase when configured. These changes reduce duplicate network calls and improve robustness against varied Hydrabase payloads.
Add Hydrabase support as an optional/dev metadata source and comparison tool.
- Add core/hydrabase_client.py: synchronous Hydrabase WebSocket client that normalizes results to Track/Artist/Album types and exposes raw access.
- Update config/settings.py: add hydrabase settings (url, api_key, auto_connect) and getter.
- Update web_server.py: integrate HydrabaseClient, initialize client alongside the existing HydrabaseWorker, add auto-reconnect using saved config, persist credentials on connect/disconnect, add endpoints for status and stored comparisons, background comparison runner (Hydrabase vs Spotify vs iTunes), and adapt multiple search endpoints to optionally use Hydrabase as the primary metadata source with fallbacks.
- Update web UI (webui/index.html, webui/static/script.js, webui/static/style.css): add network stats and source comparison UI, pre-fill saved credentials, show peer count, load/display comparisons, update disconnect behavior to disable dev mode, and add Hydrabase badge styling.
Behavioral notes: when dev mode + Hydrabase are active, searches can be served from Hydrabase and comparisons to Spotify/iTunes are run in background; when Hydrabase fails the code falls back to Spotify/iTunes. Saved Hydrabase credentials are persisted for auto-reconnect; disconnect disables dev mode and auto_connect.
Files touched: config/settings.py, core/hydrabase_client.py, web_server.py, webui/index.html, webui/static/script.js, webui/static/style.css.
Increase Spotify client rate limit and reduce API contention during watchlist scans. Changes:
- core/spotify_client.py: Bumped MIN_API_INTERVAL from 0.2s to 0.35s (~171 calls/min) to stay safely under Spotify's ~180/min limit.
- web_server.py: In start_watchlist_scan and automatic scan flow, pause spotify_enrichment_worker and itunes_enrichment_worker before scanning (tracking with _enrichment_was_running/_itunes_enrichment_was_running) and resume them in finally blocks; added console prints for pause/resume. This prevents enrichment workers from contending for API quota during long scans.
- webui/static/script.js: Improved enrichment status tooltip logic to prioritize explicit currentType and then fall back to completion-based inference with explicit branches for artists, albums, and tracks for clearer progress text.
These changes aim to avoid API rate violations and make scan progress display more predictable.
When clearing the wishlist or changing the discovery lookback period, reset watchlist_artists.last_scan_timestamp to NULL so subsequent scans can re-discover older releases that were previously filtered by an earlier scan timestamp. clear_wishlist now deletes wishlist_tracks, updates last_scan_timestamp for all watchlist artists, and logs the number of tracks cleared and artists reset. set_discovery_lookback_period also resets last_scan_timestamp and reports how many artists were reset. Minor whitespace cleanups in watchlist_scanner and web_server included.
Add a release_date field to the Track dataclass for both iTunes and Spotify clients (iTunes: parsed from releaseDate, Spotify: from album.release_date). Propagate release_date into enhanced search results in web_server and into the client-side script so album objects include release_date when available. Also broaden playlistId matching in the missing-tracks process to include 'enhanced_search_track_'. Removed SQLite SHM/WAL files from the repo (cleanup of DB temporary files). These changes enable showing and using track release dates across the app.
Add duplicate-artist detection/merge and handle ratingKey (ID) migrations for artists and albums. Introduces MusicDatabase.merge_duplicate_artists that picks a canonical artist (most enrichment data), merges enrichment fields, migrates albums/tracks, and removes duplicates; DatabaseUpdateWorker now runs this merge during updates (even when no new content) and after orphan cleanup. insert_or_update_artist/album now detect same-name/title + server_source collisions (ratingKey changes), inserting a new record while preserving enrichment and migrating references, with safe deletion of old rows. Also deduplicate artist listing queries so results show a single canonical row per name+server_source while aggregating album/track counts across duplicates. Logging improved to report merge/migration outcomes.
Introduce deferred album repair scanning and robust tracklist resolution plus UI/presentation tweaks.
- core/repair_worker.py: Add lazy MusicBrainz and AudioDB client accessors, per-batch folder queues, and register_folder/process_batch to defer folder scans until a batch completes. Implement cascading tracklist resolution (_resolve_album_tracklist) using Spotify/iTunes IDs, Spotify track→album lookup, album search, MusicBrainz release lookup, and AudioDB→MusicBrainz fallback. Add helpers to read Spotify track IDs, MusicBrainz album IDs, album/artist tags from files, MB/AudioDB fetchers, placeholder-ID filtering, and rename associated .lrc files when renaming audio files. Cache/locking and background-threaded scanning included. Improves resilience to missing/placeholder IDs and avoids circular imports.
- web_server.py: Register album folders for repair in post-processing and trigger repair_worker.process_batch when batches complete (multiple completion paths) so scans run after downloads finish.
- webui/static/script.js: Reduce unnecessary background work by skipping many fetches when the tab is hidden, and refresh dashboard-specific data on visibility change to ensure UI updates after OAuth or tab switch.
- webui/static/style.css: Replace glassmorphic backdrop-filters with opaque dark gradients for sidebar and main content to improve GPU rendering and visual consistency.
Overall: Adds reliable post-download album repair scanning using multiple metadata sources and reduces unnecessary client polling and heavy CSS effects for better performance and robustness.
Add full-featured SpotifyWorker and iTunesWorker background workers to enrich artists, albums, and tracks with external metadata using batch cascading searches, fuzzy name matching, ID validation, and DB backfills. Update RepairWorker to re-read the transfer path from the database each scan, resolve host paths when running in Docker, and trigger immediate rescans when the transfer path changes; remove the static config_manager dependency. Also include supporting changes to the database layer and web UI/server (stats, controls, and styles) to integrate the new workers and reflect updated worker status.
Add _get_playlist_items_page to call the new playlists/{id}/items endpoint (Feb 2026 API migration) and fall back to spotipy.playlist_items (old /tracks) on 403/404. Update _get_playlist_tracks and web_server.get_playlist_tracks to use the new helper to avoid 403 errors for Development Mode apps while preserving compatibility with Extended Quota Mode.
Introduce a RepairWorker to scan the transfer folder and automatically detect/repair broken album track numbers (e.g. the "all tracks = 01" bug). The worker uses mutagen to read/write tags, fuzzy-matches titles against an album tracklist (Spotify/iTunes via a SpotifyClient), updates filenames and the tracks DB file_path when renamed, and caches album tracklists. It also adds DB schema support (repair_status, repair_last_checked, and an index).
Integrates the worker into the web server: initializes and starts the worker, and exposes /api/repair/status, /api/repair/pause and /api/repair/resume endpoints. Adds UI elements (button, tooltip), client-side JS to poll and control the worker, CSS for visuals/animations, and a new image asset (whisoul.png).
Add a new docs/api-response-shapes.md describing expected Spotify/iTunes dataclass and raw-dict response shapes and client behavior. Also update core/wishlist_service.py to include 'track_number' (default 1) and 'disc_number' (default 1) in each formatted track dict so consumers receive track ordering metadata.
Introduce a Hydrabase P2P mirror worker and integrate it into the web UI and server flows. Adds core/hydrabase_worker.py: a background thread with a capped queue (1000), enqueue API, rate limiting, basic stats (sent/dropped/errors), and logic to send JSON requests over a provided WebSocket (responses received and discarded). Integrates the worker into web_server.py (import, startup init, status/pause/resume endpoints, and enqueues queries from multiple search endpoints when dev mode is enabled). Adds UI elements, JavaScript polling/toggle logic, and CSS styling for a Hydrabase status button in webui (index.html, static/script.js, static/style.css) to display and control worker state.
Add Deezer as a third metadata enrichment source. Enriches tracks with BPM and explicit flags, albums with
record labels, explicit flags, and type classification (album/single/EP), and backfills artwork and genres across
all entities. Includes background worker with priority queue, rate-limited API client, database migration, server
endpoints, and UI button with purple-themed status tooltip.
Integrated TheAudioDB as a metadata enrichment source with a background worker that scans artists, albums, and tracks in priority order. Stores style, mood, and AudioDB IDs with automatic backfill of artwork and genres.
Includes artist ID cross-verification from album/track results to correct mismatches caused by same-name artists.
Integrated TheAudioDB as a metadata enrichment source. A background worker scans the library in priority order (artists → albums → tracks), matching entities via fuzzy name comparison and storing style, mood, and AudioDB IDs. Includes rate limiting, 30-day retry for not-found items, and a UI tooltip showing phase-based scan progress.
Add a UI button to disconnect Spotify and fall back to iTunes/Apple Music without restarting. Cache is_spotify_authenticated() with a 60s TTL to reduce redundant API calls (~46 call sites were each
triggering a live sp.current_user() call). Fix status endpoint calling the auth check twice per poll,
and ensure both OAuth callback handlers (port 8008 Flask route and port 8888 dedicated server)
invalidate the status cache so the UI updates immediately after authentication.
Spotify's @rate_limited decorator now retries on 429/5xx with exponential backoff (up to 5 retries) instead of sleeping once and raising. Watchlist scan delays scale dynamically based on lookback setting and artist count to prevent sustained API pressure. A circuit breaker pauses the scan after consecutive rate-limit failures.
Multi-disc albums (e.g., deluxe editions, double albums) now automatically organize
tracks into Disc 1/, Disc 2/ subfolders within the album folder. Detection uses the
disc_number field from Spotify's API — when an album has total_discs > 1, subfolders
are created. Single-disc albums are completely unaffected.
- Plumb disc_number through all download paths (enhanced, non-enhanced, download
missing modal, wishlist)
- Compute total_discs from album tracklist and store on album context
- Modify path builder to insert Disc N/ subfolder for multi-disc albums
- Preserve disc_number when tracks fail and get re-added to wishlist
- Preserve disc_number when adding tracks to wishlist from library page
- Add visual disc separators in Soulseek search result track lists
Summary: Navidrome incremental updates always found 0 new tracks because _get_recent_albums_navidrome() fetched all artists, sampled only the first 200, collected their albums, and sorted by created date — missing artists beyond the first 200 entirely. Replaced this with a single getAlbumList2?type=newest Subsonic API call that directly returns albums sorted by library addition date, matching how Jellyfin and Plex already use their native "recently added" endpoints.
Fixed issue where broken iTunes explicit album IDs were preferred over working clean versions during deduplication. Some iTunes explicit albums (e.g., "Mr. Morale & The Big Steppers" ID 1623854804) report track counts in metadata but return 0 tracks when queried.
Added validation in itunes_client.py get_artist_albums() to verify explicit albums actually have tracks before keeping them. If an explicit version has 0 tracks, it's skipped and the clean version is used instead.
This fixes:
- "No tracks found" error when clicking affected albums
- Incorrect track count mismatches (19/18) caused by broken API data
The validation only runs for explicit albums during deduplication, minimal performance impact.
Enrich downloaded audio files with external identifiers and improved genre metadata in a single post-processing write. During metadata enhancement, the app now looks up the MusicBrainz recording and artist MBIDs, retrieves the ISRC and MusicBrainz genres from a follow-up detail lookup, merges them with Spotify's artist-level genres (deduplicated, capped at 5), and embeds everything alongside the Spotify/iTunes track, artist, and album IDs. All MusicBrainz API calls are serialized through the existing global rate limiter, making concurrent download workers safe without needing to pause the background worker. Includes a database migration adding Spotify/iTunes ID columns to the library tables.
Add optional post-download audio fingerprint verification using AcoustID.
Downloads are verified against expected track/artist using fuzzy string
matching on AcoustID results. Mismatched files are quarantined and
automatically added to the wishlist for retry.
- AcoustID verification with title/artist fuzzy matching (not MBID comparison)
- Quarantine system with JSON metadata sidecars for failed verifications
- fpcalc binary auto-download for Windows, macOS (universal), and Linux
- MusicBrainz enrichment worker with live status UI and track badges
- Settings page AcoustID section with real-fingerprint connection test
- Source reuse for album downloads to keep tracks from same Soulseek user
- Enhanced search queries for better track matching
- Bug fixes: wishlist tracking, album splitting, regex & handling, log rotation
MusicBrainz library enrichment with real-time
status monitoring and manual control.
Features:
- Status icon button in dashboard header with glassmorphic design
- Animated loading spinner during active enrichment
- Hover tooltip showing:
- Worker status (Running/Paused/Idle)
- Currently processing item
- Artist matching progress with percentage
- Click-to-toggle pause/resume functionality
- Auto-polling every 2 seconds for live updates
Backend Changes:
- Added GET /api/musicbrainz/status endpoint
- Added POST /api/musicbrainz/pause endpoint
- Added POST /api/musicbrainz/resume endpoint
- Worker tracks current_item for UI display
- get_stats() returns enhanced status data
Frontend Changes:
- New MusicBrainz button component with tooltip
- Premium CSS styling with animations
- JavaScript polling and state management
- Positioned tooltip below button with centered arrow
Files Modified:
- web_server.py: API endpoints and worker initialization
- core/musicbrainz_worker.py: current_item tracking
- webui/index.html: Button and tooltip structure
- webui/static/style.css: Complete styling (240 lines)
- webui/static/script.js: Polling and interaction logic (115 lines)
- Single-worker mode for album batches (sequential downloads enable clean source reuse)
- Browse API integration to list files in a source's directory
- Failed source tracking per-batch to avoid retrying broken sources
- Graduated quality scoring for upload speed, queue length, and free slots
- Track number fallback fix (uses Spotify track number instead of hardcoded 1)
- Duplicate completion guard to prevent double-decrement of active worker count
- Dedicated logging for source reuse and post-processing diagnostics
When downloading an album/EP, perform a single album-level search on Soulseek to find a user with the complete album folder before falling back to per-track search. This improves album completion rates and ensures consistent quality across tracks.
ListenBrainzManager opens its own raw sqlite3 connection, bypassing MusicDatabase initialization. If it runs before MusicDatabase creates the tables, all queries fail with "no such table: listenbrainz_playlists". Added _ensure_tables() to ListenBrainzManager init that runs CREATE TABLE IF NOT EXISTS for both ListenBrainz tables — a no-op when they already exist, but creates them if MusicDatabase hasn't run yet.
- Expose suffix, bitRate, and path fields on NavidromeTrack from the Subsonic API response
- Add fallback in insert_or_update_media_track() to populate file_path and bitrate for Navidrome tracks, fixing the Quality Scanner
returning 0 results
- Increase ListenBrainz playlist cache limit from 4 to 25 per type
- Add sub-tab grouping in the Recommendations tab (Weekly Jams, Weekly Exploration, Top Discoveries, etc.)
Two fixes for iTunes integration:
1. iTunes failed tracks now properly added to wishlist
- Root cause: iTunes tracks had no 'album' field (unlike Spotify)
- Fix: Added album information to each track in get_album_tracks()
- Tracks now include: album id, name, images, release_date
2. Remove ' - Single' suffix from iTunes album names
- Root cause: iTunes API includes ' - Single' in collectionName
- Fix: Added _clean_album_name() helper method
- Strips ' - Single' and ' - EP' suffixes from all album names
- Applied to all 6 locations where collectionName is used
Both Spotify and iTunes sources now work identically for wishlist
auto-processing when tracks fail or are cancelled.
When Spotify is enabled after populating similar artists with only iTunes IDs, the freshness check now detects missing Spotify IDs and triggers a refetch. This fixes the Discover page not showing data when switching from iTunes-only mode.
- Add similar artists fetching to web UI scan loop
- Add database migration for UNIQUE constraint on similar_artists table - Add source-agnostic /api/discover/album endpoint for iTunes support
- Fix NOT NULL constraint on discovery_recent_albums blocking iTunes albums
- Add fallback to watchlist artists when no similar artists exist
- Add /api/discover/refresh and /api/discover/diagnose endpoints
- Add retry logic with exponential backoff for iTunes API calls
- Ensure cache_discovery_recent_albums runs even when pool population skips
Implement dual-source architecture where iTunes serves as always-available
primary source and Spotify as preferred source when authenticated.
- Make watchlist scans provider-aware (manual and auto paths)
- Update discovery pool population to process both sources
- Update recent albums caching for both sources
- Create source-specific curated playlists (Fresh Tape, Archives)
- Add on-the-fly iTunes ID resolution for similar artists
- Add iTunes ID check to similar artists freshness validation
- Fix sqlite3.Row compatibility in personalized playlists
- Fix iTunes ISO 8601 date format parsing
- Update API endpoints to serve source-appropriate data
This ensures the app remains fully functional if Spotify becomes
unavailable (rate limits, auth issues, bans) by seamlessly falling
back to iTunes data that has been building in parallel.
Implements lazy loading of artist images in search results, artist pages, and similar artist bubbles to improve performance and user experience. Updates the iTunes client to prefer explicit album versions and deduplicate albums accordingly. Adds a new API endpoint to fetch artist images, and updates frontend logic to asynchronously fetch and display images where missing.
Adds iTunes fallback to SpotifyClient for search and metadata when Spotify is not authenticated. Updates album type logic to distinguish EPs, singles, and albums more accurately. Refactors watchlist database methods to support both Spotify and iTunes artist IDs. Improves deduplication and normalization of album names from iTunes. Updates web server and frontend to use new album type logic and support both ID types. Adds artist bubble snapshot example data.
Refactored artist scanning logic to use the active metadata provider (Spotify or iTunes) for fetching artist data, discography, and album tracks. Introduced helper methods to select the correct client and artist ID based on the provider, and updated image and similar artist handling accordingly. This enables watchlist scanning to work with iTunes when Spotify is not authenticated, improving flexibility and provider support.
Updated get_album, get_album_tracks, and get_artist methods to return data structures compatible with Spotify's API format. This includes normalizing image arrays, artist and album fields, and adding synthetic URIs and external URLs for better interoperability.
Introduces iTunes artist ID support to WatchlistArtist and database schema, enabling proactive backfilling of missing provider IDs (Spotify/iTunes) for watchlist artists. Updates WatchlistScanner to use MetadataService for provider-agnostic scanning and ID matching, and modifies web_server to support scans with either provider. Includes new database migration and update methods for iTunes and Spotify artist IDs.
Introduces core/itunes_client.py implementing an iTunes Search API client for music metadata, providing search and lookup for tracks, albums, and artists with rate limiting. Adds METADATA-FALLBACK-IMPLEMENTATION.md, a comprehensive guide comparing fallback strategies for music metadata, including anonymous Spotify access and iTunes, and outlines integration approaches for seamless user experience without requiring Spotify credentials.
Updates the method for obtaining the next cursor during track pagination to use the approach from PR #113, retrieving it from 'links.meta.nextCursor' instead of 'meta.nextCursor'. This ensures correct pagination behavior.
Eliminated unnecessary filtering of playlists by owner or collaboration status, as the Spotify API already returns all accessible playlists. This simplifies the code and ensures all relevant playlists are processed.
Rewrites playlist and track fetching to use Tidal's JSON:API endpoints with cursor-based pagination and batch track hydration. Adds robust rate limiting with retry logic, and introduces ISO-8601 duration parsing. This improves reliability, performance, and compatibility with Tidal's latest API structure.
Added a call to self.soulseek._setup_client() in DownloadOrchestrator to ensure the Soulseek client configuration is reloaded when the orchestrator is initialized. This helps keep the client settings in sync with the latest configuration.
Introduces new filters for live versions, remixes, acoustic versions, and compilation albums to the watchlist artist configuration. Updates the database schema, backend API, and web UI to support these options, allowing users to customize which content types are included for each artist in their watchlist.
Introduces a reload_config method to SpotifyClient and refactors ConfigManager to support reloading configuration from a file. Updates web_server.py to use the new config loading mechanism, ensuring configuration is loaded into the existing singleton instance and SpotifyClient is properly re-initialized after settings changes.
Introduces an async method to search for a query and automatically download the best result based on quality preferences and source priority. The method handles empty results, filters by quality, and logs the download process.
Implemented clear_all_completed_downloads in YouTubeClient to remove completed, cancelled, errored, and aborted downloads from memory. Updated DownloadOrchestrator to call this method, ensuring both Soulseek and YouTube completed downloads are cleared.
Introduces a new priority 0 query that combines artist, album, and title for improved matching, especially for YouTube and hybrid download modes. This helps better match tracks where the album is significant, such as soundtracks, and only applies when the album is not a generic label like 'single' or 'greatest hits'.
Adjusts matching weights for YouTube sources to rely more on title and duration, adds a shutdown callback to the YouTube client to prevent new downloads during shutdown, and enhances post-processing to reliably resolve actual YouTube file paths. Improves error handling for file removal, ensures no new batch downloads start during shutdown, and refines download monitoring to trigger post-processing on completed YouTube downloads. Also increases YouTube download retries and improves logging for debugging.
Enhances streaming logic to better support YouTube as a download source, including improved filename handling, fuzzy file matching, and search query generation. Updates format checks in the frontend to skip them for YouTube (always MP3). Refactors backend to use a unified download status API for both Soulseek and YouTube, and improves service test messaging based on the active download mode.
Introduces a DownloadOrchestrator class to route downloads between Soulseek and YouTube based on user-configurable modes (Soulseek only, YouTube only, Hybrid with fallback). Updates web server and UI to support new download source settings, including hybrid mode options and YouTube confidence threshold. Refactors YouTube client for thread-safe download management and bot detection bypass. Ensures quality filtering is skipped for YouTube results and improves file matching and post-processing logic for YouTube downloads.
Updated the duration field in TrackResult to store duration in milliseconds instead of seconds, matching Spotify's expected format. This ensures consistency when integrating with services that use millisecond-based durations.
Implements an enhanced search endpoint in the backend that unifies Spotify and local database results, returning categorized artists, albums, and tracks. Updates the frontend with a new dropdown overlay for live search, debounced input, categorized result rendering, and direct integration with the main results area for album/track selection. Adds new CSS for the dropdown and result cards, and updates the Track dataclass to include image URLs for richer UI display.
Introduces 'search_timeout' and 'search_timeout_buffer' options to Soulseek settings in the config, backend, and web UI. The backend now uses these values to control search duration and polling, allowing users to fine-tune how long searches run and how long to wait for late results.
Enhanced the music matching engine to use stricter version handling, rebalance title/artist/duration weights, and raise confidence thresholds to reduce false positives. Updated string normalization to better handle separators and special characters. In the web UI, improved album ID sanitization and added a placeholder for missing album images in the wishlist view.
Refactored the sorting logic to handle cases where 'spotify_data' may be a JSON string or have varying structures. This ensures consistent sorting by artist and track name, even with malformed or unexpected data formats.
Introduces backend and frontend functionality to list and select Jellyfin music libraries. Adds API endpoints, updates the client logic, and provides a UI selector for users to choose their preferred Jellyfin music library.
Adds strict checks to ensure Soulseek track versions (live, remix, acoustic, instrumental) are only matched if the Spotify track title contains corresponding version indicators. This prevents mismatched versions from being accepted and applies a reduced penalty for matching non-original versions.
Redirect URIs for Spotify and Tidal OAuth are now configurable via the web UI and settings. Updated backend clients to use the configured redirect URI if provided, improving flexibility for deployments with custom callback URLs.
Wishlist tracks are now sorted by artist and track name for consistent display. The database query for wishlist tracks now orders by date_added instead of randomly. Added logic to remove completed tracks from the wishlist during failed track processing. Updated docker-compose.yml to mount the H: drive for transfer folders.
Added helper functions to resolve Windows paths for Docker containers and to extract filenames in a cross-platform manner. Updated usage throughout web_server.py and soulseek_client.py to ensure correct path mapping and filename handling. Modified docker-compose.yml to use named volumes and mount the E: drive for better Docker compatibility.
Updated Soulseek client, web server, and docker-compose.yml to better handle service URLs when running inside Docker containers. Localhost URLs are now resolved to host.docker.internal, and Docker-specific volume mounts and extra_hosts are configured for improved interoperability between container and host services.
Introduces web server routes and UI buttons for initiating Spotify and Tidal OAuth authentication flows, with dedicated callback servers for token exchange. Updates Docker ports for OAuth callbacks and refines PKCE handling for Tidal. Improves user experience by allowing authentication directly from the web UI.
Introduces Docker deployment files (.dockerignore, Dockerfile, docker-compose.yml, docker-setup.sh, requirements-webui.txt, and README-Docker.md) for SoulSync WebUI. Refactors core/database_update_worker.py and core/media_scan_manager.py to support headless operation without PyQt6, enabling signal/callback compatibility for both GUI and non-GUI environments. Removes logs/app.log file.
Introduces a global activity feed system with API endpoints for recent activities and toasts, and integrates activity tracking for key backend events (downloads, syncs, database updates, etc.). Updates the dashboard frontend to periodically fetch and display activity feed items and show toasts for recent actions. Improves error logging and reduces noise for expected 404s in Soulseek client. Adds related CSS for activity feed separators.
Introduces backend logic to automatically remove tracks from the wishlist if they already exist in the database after downloads or database updates. Adds a new API endpoint and frontend button for manual wishlist cleanup, allowing users to remove already-owned tracks from the wishlist. Enhances reliability and user experience by keeping the wishlist up-to-date.
Watchlist will allow the user to select artist to 'watch' and the app will automatically add new releases by those artists to the wishlist for automatic download.
included version information and modal
This is needed if users manually correct failed matches on plex when songs are scanned in. If a db sync is done after the scan but before the manual modification,
