The CAA branch of _upgrade_art_url capped art at the /front-1200 thumbnail —
a deliberate flakiness trade-off, but the policy had rotted into inconsistency:
iTunes art already shipped at 3000x3000, and bare /front URLs (release-group
lookups — exactly what the Re-tag flow produces) bypassed the cap entirely,
which is how Sokhi observed retag delivering full-res while downloads got 1200.
CAA URLs now upgrade to the bare /front ORIGINAL (native res, frequently
3000px+). The flakiness concern that motivated the old cap is handled where it
belongs, in the fetch: _fetch_art_bytes now walks an attempt chain — original
-> /front-1200 midpoint -> the original sized thumbnail — so a flaky
archive.org degrades to the old 1200px behavior, never below it.
Tests updated to the new contract (+3 chain tests: native-first, flaky
degrades to 1200 not 250, full chain ends at the thumbnail). 623 metadata +
1267 art-path tests pass.
Caught live by the new lookup timing ("Genius track lookup took 242.4s"):
the 429 handler slept the backoff (30/60/120s) in the CALLING thread and then
re-raised anyway — the import pipeline waited 2x120s per track for lookups
that still failed. Worse, the pre-flight backoff wait also slept while
HOLDING the global Genius API lock, so every other Genius caller queued
serially behind the nap.
Now the backoff is a gate: a 429 opens a 30s->60s->120s window and re-raises
immediately; any call inside the window raises GeniusRateLimitedError on the
spot. The error subclasses requests.RequestException, so every existing
caller (the import's source lookups catch RequestException and skip; the
worker's per-item guards) already handles it as a one-line skip — lyrics and
Genius tags are garnish, nothing is allowed to WAIT for them.
Tests: backoff window fails fast (<0.5s vs the old full-window sleep), a 429
opens and escalates the gate without sleeping, the error is a
RequestException (the no-call-site-changes hinge), success decays the gate.
Measured during a live album download: ~4m15s per track in post-processing
(normal is ~20s), with the time vanishing silently inside embed_source_ids —
up to 5 MusicBrainz calls per track crawling against a degraded musicbrainz.org
while the MB enrichment worker kept eating the same ~1 req/s per-IP budget.
Only Spotify/Last.fm/Genius were in the yield set; MusicBrainz, Deezer, iTunes,
Discogs etc. kept grinding through downloads.
Policy (new core/enrichment/yield_policy, tested):
- downloads active -> ALL enrichment workers yield (post-processing touches
every metadata source). listening-stats (local-only) and repair
(user-scheduled) intentionally keep running.
- discovery active -> the API-contention five yield (spotify/itunes/deezer/
discogs/hydrabase) — discovery never paused anything before, despite the
pause helper literally defaulting to label='discovery'.
- user overrides and user-paused bookkeeping keep their existing semantics;
the dashboard yield_reason label now says WHICH foreground work caused it.
Observability (the 4-minute silence can never come back):
- every source lookup is timed; >2s logs a warning NAMING the source and
duration (core/metadata/source.py _call_source_lookup)
- the pipeline always logs "Metadata enhancement took X.Xs" per track
7 policy tests (incl. the motivating case: MB yields to downloads, keeps
running during discovery); 277 pipeline/enrichment tests pass.
A user reports ~0.7 MiB/s RSS growth; the one theory offered so far
(connection leak) was debunked, so instead of guessing: measure. New
core/diagnostics/memory_tracker wraps tracemalloc behind three GET endpoints
the user can drive from a browser:
/api/debug/memory/start begin tracing + baseline snapshot (idempotent)
/api/debug/memory/report top allocation sites by GROWTH since the baseline
(?top=N), with traced totals + process RSS so we
can see how much of the real growth tracing
accounts for; 15-frame tracebacks name the caller
/api/debug/memory/stop end tracing, free trace bookkeeping
Opt-in by design — tracemalloc shadows every allocation while active, so it
never runs by default. RSS via psutil with a /proc fallback.
Tests: report-without-tracking returns a hint (not an error); a real
start->hog->report->stop roundtrip attributes a genuine 5MB allocation to the
test file (fun fact encoded in the test: 'x'*1000 constant-folds into ONE
shared string and traces as ~40KB — the hog must allocate at runtime); the
stat formatter is duck-typed and unit-tested.
YouTube now gates downloadable formats behind JS challenges (nsig); yt-dlp
needs a JavaScript runtime (Deno — its only default-enabled one) to solve
them, and its STABLE channel can lag months behind a breaking YouTube change.
Users hit "Requested format is not available" on every stream and music-video
download with no hint why. Neither piece is fixable via requirements.txt:
Deno isn't a Python package, and a version floor can only ever resolve stable.
- Dockerfile: install Deno in the runtime image (official installer,
auto-detects amd64/arm64, build fails early via `deno --version` if it
breaks; unzip added for the installer) and build with the yt-dlp NIGHTLY
channel (`pip install -U --pre "yt-dlp[default]"`) on top of requirements
- core/youtube_client.py: one-time startup warning when deno isn't on PATH,
naming the exact failure it causes and the install command — instead of
letting users debug a cryptic yt-dlp format error
- requirements.txt: annotate the yt-dlp line with the stable-lag caveat, the
nightly upgrade command, and the Deno requirement
- README: Deno + nightly notes in Prerequisites and the Python (No Docker)
install section; Docker bundles both automatically
"Liked Songs" isn't a real Spotify playlist — no playlist URI exists for it;
the web UI invents the virtual id 'spotify:liked-songs' and Spotify serves the
collection via the saved-tracks endpoint. The playlist DETAIL endpoint special-
cases that id, but the mirrored refresh path resolves stored ids through
get_playlist_by_id, which fed the virtual id straight into sp.playlist() ->
"http status: 400 ... Unsupported URL / URI" on every sync cycle, silently.
get_playlist_by_id now special-cases the virtual id at the client seam (every
by-id resolver benefits, not just the mirror adapter): it builds the Playlist
from the existing get_saved_tracks() pagination, with the real owner name and
track count. New LIKED_SONGS_PLAYLIST_ID constant owns the magic string.
Safety: get_saved_tracks swallows fetch errors into [] — indistinguishable
from "no likes" — and the virtual playlist is only ever offered when likes
exist. An empty result therefore resolves as a FAILED refresh (None) instead
of a valid-looking empty playlist a mirror sync might propagate by clearing
the server-side copy.
Tests: virtual id resolves from saved tracks and never touches the playlist
endpoint, real ids still do (regression), the mirrored adapter seam returns a
full PlaylistDetail, and empty saved-tracks -> None. 473 passed across the
playlist/mirror/spotify families.
On path-mapped setups (Docker mounts etc.) the scan checked a bare
os.path.isfile() on the raw DB path — false for EVERY track — while the apply
handler resolves container/host mismatches. With a cover-art mode set, the
cover_action kept the album past the "anything to do?" gate, so every album
produced a finding with an empty tracks list whose apply could only ever fail
with "No tracks to re-tag in finding".
- the scan now resolves each track path with the same resolver the apply
handler uses (resolve_library_file_path) before reachability checks and
current-tag reads; plans carry the resolved path
- a finding can never be created with zero tracks — cover-action albums with
no usable tracks are skipped, with a debug log of why (unreachable/unmatched
counts) and the counts surfaced in the finding description
- unmatched-but-reachable tracks now get an art-only plan (empty db_data) so
album cover art covers ALL the album's files, not just source-matched ones —
apply_track_plans already treats empty db_data as a pure cover embed and
counts a failed cover download as skipped, never failed (now locked by tests)
- cover-only findings are titled "(cover art, N track(s))" instead of the
puzzling "(0 track(s))"
Tests: +5 (mapped paths resolve into plans, cover-with-nothing-reachable
creates no finding, unmatched -> art-only plan, art-only plan embeds cover,
failed cover download -> skipped). 87 passed across retag/repair/tag_writer.
Since the per-listener stream sessions refactor (Phase 3b), every browser gets
its own stream session — but the 1s 'tool:stream' socket broadcast still read
the legacy GLOBAL state (the DEFAULT session no real browser uses), so it told
every client "stopped" forever. The frontend skipped HTTP polling whenever the
WebSocket was up, so it only ever saw that wrong broadcast: the backend prep
downloaded the track, moved it into the session's stream folder and sat at
"ready" while the mini player showed nothing. Proxy users whose WebSockets
don't connect fell back to HTTP polling (session-correct) and streamed fine —
which is why this hid so well.
Fix: stream status is inherently per-listener, so stop pretending a global
broadcast can carry it —
- web_server.py: remove the 'tool:stream' emit from the tool-progress loop
(the broadcast thread has no request context; it can only ever see DEFAULT)
- media-player.js: the status poller always polls /api/stream/status (resolves
the caller's own session from the cookie); drop the dead broadcast handler
- core.js: unwire the 'tool:stream' socket listener
Observability fix that made this undebuggable: core/streaming/prepare.py used
getLogger(__name__) — outside the soulsync.* namespace where handlers attach —
so every prep log line (including failures) vanished from app.log. Moved to
get_logger("streaming.prepare") + a regression test locking the namespace.
34 streaming tests pass; ruff clean; web_server compiles; JS syntax-checked.
A single enriched against the deluxe gets every source ID pointing at the deluxe,
so the organizer filed it as e.g. track 2 of a 10-track album. Root cause: the
canonical resolver only ever scored the editions already linked — the correct
single was never even a candidate, and the misfit deluxe scored so low (0.1,
below the 0.5 floor) that nothing got pinned and the priority-walk grabbed the
deluxe anyway.
Fix, in three tested layers:
- resolve_canonical_for_album gains a fetch_alternates seam: when no linked
edition clears the floor, it scores the source's OTHER editions of the same
release and re-picks by best fit (dedup, injected, pure).
- default_fetch_alternates lists the artist's editions and keeps the same-release
ones (edition-blind name match: Deluxe / - Single / [Remastered] all collapse),
returning their tracklists. Favors recall; the scorer is the precision gate.
- _resolve_source does the misfit check inline: it fit-scores the walked edition
and only on a clear misfit searches for a better edition, then persists the pin
on apply (Track Number Repair + future runs agree). Cost-neutral and behavior-
identical for well-fitting albums (no extra API calls); strict_source and the
#758 manual lock are never overridden.
Tests: +4 resolver (expand/no-expand/dedupe/back-compat), +7 alternates (name
matcher + fetcher over fake APIs + cap), +3 organizer end-to-end (misfit->single
+pin, well-fit->no-expand, strict->no-expand). 300 passed across the reorganize
+ canonical family, lint clean.
User report: "Not authenticated with Spotify" daily + workers paused. The
log was misleading — is_spotify_authenticated() returns False for five
distinct reasons (no creds / rate-limit ban / post-ban cooldown / no
cached token / probe failure), and all five API call sites logged the
same bare "Not authenticated with Spotify". So a routine rate-limit ban
read as a logout, and the real cause (logged only at DEBUG) was invisible.
New pure describe_spotify_unavailable() maps the real state to a clear
message (priority matches is_spotify_authenticated): not-configured →
rate-limited ("ban ~Nm left (not a logout)") → post-ban cooldown →
no-token ("not connected — re-authenticate") → "auth check failed (token
refresh may have failed)". A side-effect-free client method
(_auth_unavailable_reason) gathers the live state (reads the cached token,
no API probe) and the 5 sites log it.
Now a daily ban is identifiable as a ban, and a genuine logout is
identifiable as one — so reports like this are diagnosable from the log
alone. 7 tests pin the priority/messaging. Full suite clean.
#798 follow-up. The worker's 500/day budget is a REAL-API ban shield, but
when it was hit the worker paused outright — even for a Spotify-Free user
with the uncapped free source available. So "I'm on Spotify Free" still
got capped overnight. The intuition is right: if it's ever using Spotify
Free, the budget shouldn't apply.
Fix: spent budget now becomes a third "use free" trigger (alongside
no-auth and rate-limited). When the real-API budget is exhausted and the
free source is available, the worker switches to free (uncapped) for the
rest of the day instead of pausing, then reverts to real-first on the
daily reset.
- should_use_free_fallback gains a budget_exhausted arg (free activates on
no-auth OR rate-limited OR spent-budget).
- the worker sets _budget_exhausted_use_free on ITS OWN client (a separate
instance from the search client — verified, so user searches still use
real auth), and clears it when the budget resets; _free_active() honors
the flag.
- get_stats() using_free reports the budget-bridge too, and the dashboard
bubble shows "Running (Spotify Free)" instead of "Daily Limit Reached"
(budgetStuck = exhausted AND not bridging).
A no-free user still pauses on the budget (nothing to bridge to). A pure
free-only worker never increments the budget at all. New gate test pins
the budget_exhausted trigger. Full suite clean.
When a file is quarantined (AcoustID / integrity / bit-depth), the source
is moved away and _mark_task_quarantined sets _quarantine_entry_id on the
context. A second post_process_matched_download call with the same
file_path (caused by a monitor re-poll or concurrent dispatch before the
context is cleaned up) then hit the race guard — "source file gone, no
known destination" — and overwrote the in-flight retry with a failed status.
Fix: check _quarantine_entry_id before firing the race guard. If it is
set, the file is legitimately in quarantine and this is a stale duplicate;
return silently so the quarantine retry already in flight can proceed.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The existing fallback (pipeline.py:1084) only ran inside
post_process_matched_download_with_verification — i.e. when a file *was*
downloaded and AcoustID retries were fully exhausted. If the retry *search*
itself found zero valid candidates (source returned nothing, or all failed
HiFi validation), the task was marked not_found and the fallback was never
reached, even though the quarantine already held N version-mismatch entries.
Fix: add try_version_mismatch_fallback to TaskWorkerDeps; in the "no valid
candidates" path of task_worker, invoke it before marking not_found when
is_quarantine_retry. Wired in _build_task_worker_deps via a new helper
(_try_version_mismatch_fallback_for_worker) that calls
try_accept_version_mismatch_fallback directly with the track's title and
artist and a reprocess lambda over _post_process_matched_download_with_verification.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Some tracks don't exist on the sources in the wanted cut — every copy is, say,
the instrumental. The retry engine correctly rejects each (version mismatch) and
gives up, leaving the track missing. New opt-in fallback: once a track's AcoustID
retries are fully exhausted, if every quarantined candidate for it failed the
SAME version mismatch (same matched version, e.g. all instrumental) and there are
>= N of them, accept the best (first-tried = oldest = highest-confidence) one.
Safety rules (core/imports/version_mismatch_fallback.py):
- Version mismatches only. Audio/artist mismatches (different recording) and
integrity/duration failures (truncated/wrong file) never participate.
- All qualifying entries must share the same matched version; a mix
(instrumental + live) is ambiguous → no acceptance.
- Re-import bypasses ONLY the AcoustID gate; integrity/duration/bit-depth still
run, so a truncated or genuinely wrong file is never let through here.
- Reuses the existing quarantine approve_quarantine_entry + re-verify dispatch.
Wired at the AcoustID give-up point in the verification wrapper. Two new
post_processing settings surfaced in the Retry Logic tile (default off):
accept_version_mismatch_fallback + version_mismatch_min_count.
Pure decision core + orchestration covered by tests (11). Acceptance logged at
WARNING with track + matched version.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The cross-script alias bridge (#442/#586) silently returned [] for some
artists ("Sawano Hiroyuki"). Root cause: the mb-only escape — built exactly
for the case where local string similarity is ~0 (romaji↔kanji) but MB's own
score is decisive — inspected scored[0], the COMBINED-score leader. When an
unrelated same-script decoy outranks the real artist on combined score (decoy:
sim 0.82 + mb_score 83 → combined 0.82, just under the 0.85 bar; real '澤野弘之':
sim 0 + mb_score 100 → combined 0.30, sorted last), the gate saw the decoy's
mb_score 83 (< 95), failed both paths, and cached an empty alias result.
Verification then scored the kanji artist 0% against the romaji expected name
and quarantined every correct file.
Evaluate the MB-SCORE leader independently of combined ranking for the mb-only
escape, and pull aliases from whichever entity actually passed (combined leader
for the combined path, MB-score leader for the mb-only path). The unambiguity
check now compares the top two raw MB scores. Same-script and single-result
paths are unchanged (regression-guarded by the existing #442/#586 tests).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The cached-first retry (8d98b755) abandoned a source after a single query:
the first run returns as soon as ONE query starts a download, so
cached_candidates held only that query's results. On a quarantine retry the
whole source was then excluded from re-search (via searched_sources), so the
later queries (e.g. "artist + album") never hit that source again — it jumped
to the next source after one query instead of exhausting all queries per
source.
Track searched QUERIES (searched_queries) instead of whole sources. A
quarantine retry now skips only the already-run queries (their candidates are
walked via cached-first) and still searches the not-yet-run queries against the
same source. Budget-exhausted sources (exhaustive mode) stay excluded, so the
source switch still fires when a source is genuinely spent.
Removes the now-dead searched_sources state (written but no longer read).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Each AcoustID/integrity quarantine retry re-ran the FULL search (all queries,
all sources) before picking the next-best candidate — so a track that failed
verification a dozen times re-queried Soulseek a dozen times (~3 min/cycle in
the field). The next-best pick was already sitting in cached_candidates.
Now the monitor flags the re-queue as a quarantine retry; the worker walks the
already-found candidates first (skipping used + budget-exhausted sources) and
hands them straight to the download path — no search. A source is searched
exactly once: once its candidates are cached, later quarantine retries exclude
it (searched_sources) so the hybrid chain falls through to a not-yet-searched
source instead of re-querying the spent one. Fresh downloads and the monitor's
dead-connection/stuck retries clear searched_sources and search fresh, so the
only re-search is for a genuinely new source or a dead peer.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
In exhaustive retry mode, a source that spent its whole per-source budget
(query_count × retries_per_query) gave up and failed the track outright —
never trying the other configured sources. For tracks where Soulseek has a
deep pool of wrong peers (e.g. an AcoustID title mismatch every copy shares),
the budget tripped long before HiFi/Tidal/… were ever reached.
Now, when a source's budget is spent, the monitor marks it exhausted on the
task and re-queues so the worker excludes it from the next hybrid search,
falling through to the next source in the chain. Each new source spends its
own fresh budget. The task only fails once no fallback source remains (or the
absolute total ceiling trips) — single-source mode still fails immediately,
since there's nothing to fall back to.
task_worker folds the exhausted-source set into both the orchestrator search
exclusion and the hybrid-fallback source list.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds an opt-in exhaustive mode to the quarantine-retry path. Default
behaviour is unchanged: a single global cap (MAX_QUARANTINE_RETRIES=5).
When post_processing.retry_exhaustive is on, each source gets its OWN
retry budget sized as query_count x retries_per_query. Soulseek peers
collapse to one 'soulseek' bucket; streaming plugins keep their name.
The worker now records query_count on the task; the budget scales with
the track's real query count. Loop protection is threefold: per-source
cap, used_sources exhaustion (the natural terminator), and an absolute
ceiling (MAX_TOTAL_QUARANTINE_RETRIES=100).
New settings (config + WebUI): retry_next_candidate_on_mismatch (master),
retry_exhaustive, retries_per_query (default 5).
Tests: 6 new cases covering per-source budgeting, source separation,
Soulseek-peer bucketing, query_count default, and the absolute ceiling.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
When a downloaded file is quarantined because AcoustID verification or the
integrity/duration check fails, the task no longer dead-ends as failed — it
re-runs the worker on the next-best candidate, skipping the quarantined source.
Reuses the monitor's existing transfer-error retry machinery (used_sources +
cached_candidates + worker re-dispatch), just triggered from the post-process
verification wrapper's two quarantine branches instead of only on transfer
errors. Universal across sources (Soulseek, HiFi, Tidal, etc.) since all
batch/sync downloads funnel through post_process_matched_download_with_verification.
- monitor.requeue_quarantined_task_for_retry(): marks bad source used, resets
task to searching, resubmits worker. Guards: manual picks, cancelled tasks,
missing source id, and a MAX_QUARANTINE_RETRIES=5 loop cap.
- Opt-out via post_processing.retry_next_candidate_on_mismatch (default on).
- Manual quarantine approve is unaffected (_skip_quarantine_check='all' bypasses
the checks, so no quarantine flag, so no retry).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Users manually match an album to the regular edition, but enrichment/
repair keeps treating it as the deluxe (missing songs, renumbered tracks).
Root cause: an album has TWO identities — the enrichment match
(spotify_album_id, which manual-match sets and the worker already honors)
and a SEPARATE canonical version pin (canonical_album_id, added by #777).
The canonical pin is what track-number repair / reorganize / missing-track
detection actually read, and library_manual_match never wrote it — so it
was resolved independently and landed on the deluxe edition.
(So #777 did NOT solve #758: it added canonical pinning, but manual
matches didn't write the pin.)
Fix: a manual ALBUM match on a canonical-recognised source now also pins
AND locks the canonical version to the chosen release:
- new canonical_locked column (same migration pattern as the other
canonical cols).
- set_album_canonical(..., locked=False) gains an atomic WHERE-clause
guard: an auto write can't overwrite a locked pin; a manual write
(locked=True) always wins. get_album_canonical exposes `locked`.
- library_manual_match pins canonical for album matches via the pure
should_pin_manual_canonical(entity_type, source).
The auto resolve job already skips already-pinned albums, so the lock is
protected on two fronts; the new guard also covers any future
re-resolution. A new manual match still overrides.
18 tests: the pure gate (+ a sync-invariant test vs _ALBUM_ID_COLUMNS)
and the DB lock seam (auto can't clobber a manual lock; manual overrides;
auto-over-auto still works). Additive — locked defaults False, so the
auto path is unchanged unless a manual lock exists. Full suite clean.
#798 follow-up. The Spotify enrichment worker's daily budget and post-ban
cooldown both exist to protect the REAL authenticated API from bans. But
the worker bridges to the no-creds Spotify Free source during a ban (a
different, anonymous path), and those guards weren't free-aware:
- the budget guard slept the worker when the daily cap was hit, blocking
free work for no reason,
- free-served items still incremented the budget counter, draining the
real-API cap with calls that never touched the real API (so you'd
return from a ban with budget already spent), and
- the post-ban cooldown slept the worker even when serving via free.
Fix: compute free_serving = client._free_active() once per loop
(defensively wrapped → False on error → original behavior). When
free_serving:
- skip the daily-budget guard,
- skip the post-ban-cooldown guard,
- don't increment the daily budget.
So the budget is now strictly a cap on REAL Spotify API usage; free runs
unthrottled by it (the free client keeps its own inter-call pacing).
The decision input (_free_active) is already pinned by the gate-model
tests (free True exactly when rate-limited+Free / Free-primary, False when
authed+healthy). Full suite clean (only pre-existing soundcloud /app env
failures remain).
#798 follow-up. When the real Spotify API is banned but the worker keeps
matching via the no-creds Spotify Free source, every status surface still
read the literal rate_limited=True flag and showed "Rate Limited /
waiting Nm" — so the dashboard bubble looked paused/stuck even though the
worker (visible in Manage Workers) was actively matching.
- spotify_worker.get_stats() adds a `using_free` flag: rate_limited AND
is_spotify_metadata_available(). Computed ONLY when rate-limited, where
is_spotify_authenticated() returns False without an API probe, so the
2s status loop pays no quota cost.
- Dashboard bubble (enrichment.js): when using_free, the bubble is
'active', the tooltip says "Running (Spotify Free)" and "Now: X (via
Spotify Free)" instead of "Rate Limited / Waiting Nm". Clicking it
pauses (works) rather than hitting the resume-blocked toast.
- Manage Workers (enrichment-manager.js): status pill shows "Running
(Spotify Free)"; the warning banner is replaced with a calm "matching
via Spotify Free until the ban lifts" note.
The flag flows through both feeds (the /api/enrichment/spotify/status
poll and the WebSocket enrichment:* push) since both serialize
get_stats(). Genuinely-stuck (no-free) workers still show "Rate Limited".
#798 follow-up. The enrichment worker's own loop already bridges to the
no-creds Spotify Free source during a rate-limit ban (its guard checks
is_spotify_metadata_available()). But the resume button's pre-check
(_spotify_resume_pre_check) blocked resume on ANY rate-limit with no
awareness of Free — so a Free-opted-in user who got rate-limited was
locked out of restarting the worker, unable to fall through to the free
API.
Fix: the resume guard now mirrors the worker. Block only when
rate-limited AND nothing can serve (plain auth, no Free) — where resuming
would just sleep out the ban. When Free is available it serves during the
ban (is_spotify_authenticated() is False while banned, so
is_spotify_metadata_available() reports the free source), so resume is
allowed and the worker bridges via Free, then returns to real auth once
the ban lifts. Stays real-API-first; Free is only the bridge.
The rule is pinned in a pure helper should_block_rate_limited_resume()
next to the other gate functions, with 3 tests. Full suite clean (only
pre-existing soundcloud /app env failures remain).
A mis-grouped library track sits under a 'Various Artists' / '[Unknown
Album]' record while the file itself is correctly tagged. Write Tags
reads the DB and stamped that junk over the file — destroying the
correct tags. (Rematching never helped: a match only stores a source-ID
pointer, it never changes the local name/title the writer reads.)
Guard: never replace a real file value with a placeholder. Added at both
seams so preview and write agree:
- build_tag_diff marks such a field protected/no-change (the preview no
longer shows a wrong overwrite, and has_changes reflects reality),
- write_tags_to_file reads the file's current values and skips the
placeholder-over-real fields, preserving the file.
Field-agnostic and direction-safe: the guard fires ONLY when the DB
value is a placeholder AND the file holds a real one. A legitimate value
still writes — including a genuine 'Various Artists' album artist on a
real compilation, where the file has no conflicting real value, so the
guard doesn't fire. Every write_tags_to_file caller writes DB->file as a
correction, so blocking placeholder-over-real is correct for all of them
(the download post-process uses a different path, embed_source_ids).
23 new tests (placeholder detection, the guard fn, build_tag_diff on the
screenshot-#2 scenario, end-to-end FLAC write preserving real values and
still overwriting with real ones). 113 existing repair/retag/tag tests
pass.
The post-scan reconcile previously ran AFTER the worker's 'finished'
signal, which flips db_update_state status to 'finished'. Automations
wait for a scan by polling that status, so they stopped waiting before
the reconcile ran — and the dashboard/Tools card showed "Completed" then
flipped to "Reading file tags…". For incremental scans this was
invisible (sub-second); for a full refresh it was a real gap (a chained
automation would fire minutes before the IDs were filled).
Fix: the worker now routes completion through _emit_finished(), which
runs self.post_scan_hook (the reconcile) FIRST, then emits 'finished'.
The hook is injected by the web layer (it owns path resolution). So:
- status stays 'running' through the reconcile,
- the reconcile pushes its phase ("Reading file tags for N new tracks…")
and per-track progress through the SAME db_update_state callbacks the
scan already uses — so automations, the dashboard card, and the Tools
page all see it for free and wait for it,
- 'finished' is emitted exactly once, AFTER the reconcile — race-free, no
status blip a poll could catch,
- best-effort: a hook exception never blocks 'finished', so a scan can't
get stranded as perpetually 'running'.
Both scan entry points (_run_database_update_task, _run_deep_scan_task)
set the hook before run()/run_deep_scan(); the redundant post-run calls
are removed.
5 ordering tests pin the contract (hook-before-finished, finished still
fires without a hook, hook exception doesn't block finished, hook gets
the worker). Full suite clean (only pre-existing soundcloud /app env
failures remain).
Extends the manual "Import IDs from File Tags" backfill so newly-scanned
files get their embedded provider IDs pulled into the DB automatically —
no button press needed to keep up with new music.
How it works:
- insert_or_update_media_track now returns 'inserted' / 'updated' / False
(truthy-compatible; existing `if track_success` callers unaffected) so
the scan worker can tell a genuinely new row from an update.
- DatabaseUpdateWorker collects the ids it newly INSERTED this run
(self._new_track_ids) across all insert paths (Plex/Jellyfin/deep).
- After run()/run_deep_scan(), web_server calls _reconcile_after_scan(),
which gap-fills embedded IDs for just those new tracks. Runs as a
post-scan pass (the scan loop itself is untouched/fast — the media
server API never exposes these custom IDs, so the file must be read
once regardless; batching at the end keeps it out of the hot loop and
best-effort so it can never abort a scan). A progress phase ("Reading
file tags for N new tracks…") surfaces the full-refresh tail.
Shared engine:
- New reconcile_library() in core does the paging + lazy parent-map
loading (only loads albums/artists actually referenced — cheap when
scoped to a few new tracks) + per-page commits. BOTH the manual button
and the scan hook call it, so there's one tested orchestration, no
duplication. The backfill job was refactored onto it.
Same hardened safety: gap-fill only, atomically guarded against
overwrite, schema-introspected, idempotent. Scoped to new arrivals for
incremental/deep; full refresh re-inserts everything as new (recovering
the IDs a full-refresh wipe destroys).
+10 reconcile tests (reconcile_library scope/idempotency/progress/stop +
the engine). Full suite clean (only pre-existing soundcloud /app env
failures remain).
Files SoulSync (or MusicBrainz Picard) already tagged carry Spotify /
iTunes / MusicBrainz / Deezer / Tidal / AudioDB / Genius / Last.fm IDs in
their metadata. Enrichment workers gate their queues on
{provider}_match_status IS NULL, so reading those IDs back and gap-filling
the {provider}_id + match_status='matched' columns lets the workers skip
the API lookup entirely — big API savings on an already-tagged library.
New manual job in Tools -> Database & Scanning ("Import IDs from File
Tags"): scans every library file, reads embedded IDs, fills any that are
missing in the DB. Background job + progress card, mirroring the
write-tags-batch pattern.
core/library/embedded_id_reconcile.py (pure + tested):
- plan_reconcile(): gap-fill plan for a track + its album + artist. Only
empty id columns are planned; a disagreeing embedded id is a conflict,
never applied.
- apply_reconcile_plan(): one guarded UPDATE per id column —
WHERE id=? AND (col IS NULL OR col=''). The guard makes the fill atomic:
if an enrichment worker matched the same entity between our read and
this write, the UPDATE affects 0 rows instead of clobbering it. Columns
are introspected so a schema missing a provider's columns is skipped.
- reconcile_track_row(): per-track orchestration (id extraction, plan ->
apply, keeping the in-memory parent maps fresh for sibling tracks).
Job hardening: paged track scan (bounded memory), per-page commits (don't
starve concurrent workers), per-file try/finally (one bad file can't abort
the run), counters from real rowcount.
Scope: 19 column-fills across 8 providers. MB *recording* (track) id is
left out (UFID frame the reader doesn't surface; Vorbis key ambiguous) —
MB album+artist are covered. Amazon/ASIN deliberately excluded (ASIN is a
different namespace than the worker's amazon_id). All target columns
verified against the live schema.
Purely additive: new module, two new endpoints, one new Tools card —
no existing behavior changed. 20 unit tests (incl. the concurrency guard).
Full suite clean (only pre-existing soundcloud /app env failures remain).
AcoustID returns a recording's title/artist in their ORIGINAL script
(e.g. "久石譲" for Joe Hisaishi) while SoulSync's expected metadata is
romanized/English. A correct download then fails verification on two
walls: the title can never clear the 0.70 similarity bar cross-script,
and the only skip path that ignores the title required a near-perfect
0.95 fingerprint plus a resolved alias. Result: every non-English
artist trips it. Two complementary fixes, per the reporter's two ideas.
Graceful fix (automatic):
- New pure core/matching/script_compat.py detects when two strings are
in genuinely different writing systems (CJK/Hangul/Cyrillic/Greek/
Arabic/Hebrew/Thai vs Latin). Accented Latin (Beyoncé, Sigur Rós)
stays Latin — no false trigger.
- acoustid_verification.py: when the EXPECTED artist and the matched
artist span scripts AND the artist is confirmed via the existing
MusicBrainz alias bridge, SKIP instead of quarantine, without the
0.95 floor (the 0.80 trust floor already gates the fingerprint).
- Deliberately narrow: keyed on the ARTIST spanning scripts + being
confirmed. A same-script artist with only a cross-script title keeps
the stricter 0.95 floor, so the #607 wrong-file protection (Kendrick
R.O.T.C, low-fingerprint Japanese-title) is untouched.
Per-request toggle (manual escape hatch):
- New "Skip AcoustID verification" checkbox in the download-missing
modal beside "Force Download All".
- skip_acoustid threads request -> batch -> per-track track_info ->
download context (same path as _playlist_folder_mode), landing on
the existing _skip_quarantine_check='acoustid' bypass. No new
mechanism; only the AcoustID gate is bypassed (integrity/bit-depth
still run).
Tests:
- tests/matching/test_script_compat.py — script-boundary cases.
- test_acoustid_skip_logic.py — Joe Hisaishi SKIPs at 0.85; unconfirmed
cross-script artist still FAILs; same-script low-fingerprint still
FAILs.
- test_downloads_candidates.py — toggle injects the bypass; absent
toggle keeps verification.
Full suite: 5169 passed; only pre-existing soundcloud /app env failures
remain. Zero regressions.
Per the cleaner model: the free source only runs for users who explicitly picked
'Spotify Free' — not for every connected user. _free_wanted() is now just
_free_selected() (dropped the has-credentials auto-trigger). So:
- Plain 'Spotify' user, rate-limited -> waits out the ban as before (no surprise
background scraping, no ToS exposure for people who never chose free).
- 'Spotify Free' user, no auth -> free serves.
- 'Spotify Free' user who also connects an account -> official when healthy,
free bridges only during a rate-limit, then switches back.
Rewrote the metadata-source help text as a plain per-source list with a clear
note on how Spotify Free + a connected account interact. Gate tests updated to
pin the opt-in behavior (plain-Spotify ratelimit = no bridge; Spotify-Free
ratelimit = bridge).
Consistency fix: Spotify Free is now its own entry in the metadata-source
dropdown (alongside Spotify / iTunes / Deezer / MusicBrainz) instead of a
side-toggle. Stored as fallback_source='spotify' + spotify_free=true so all
downstream 'spotify' routing and the spotify_* columns are unchanged.
Refined gate model (no toggle):
- Connected user (has credentials) -> official; bridges to free AUTOMATICALLY
during a rate-limit ban (no opt-in needed).
- No-auth user -> must pick 'Spotify Free' in the dropdown; then free serves.
- Never opted into Spotify (no creds, didn't pick it) -> free never runs, so no
surprise scraping. _free_wanted() = has_credentials OR picked-spotify-free is
the guard.
- AUTHED + healthy -> official always; free never opens.
UI: dropdown gains 'Spotify Free (no credentials)' (selectable when the package
is installed — surfaced via status.free_installed, since selecting it is the
opt-in and can't depend on having selected it); load/save map the dropdown value
to the (fallback_source, spotify_free) pair; old checkbox removed.
Gate model pinned by 6 scenario tests (connected/healthy, connected/ratelimited
bridge, no-auth picked, no-auth not-opted-in, package-missing). 117 tests green.
When Spotify Free is enabled, it now also bridges an official rate-limit ban for
authenticated users instead of stalling — search already did this (the gate
opens on no-auth OR rate-limit); this extends it to the enrichment worker.
- spotify_worker: the rate-limit guard now sleeps only when free CAN'T cover
(is_spotify_metadata_available() is False). Purely additive — with Spotify
Free off, that's False during a ban and the worker sleeps exactly as before.
Verified: toggle OFF + rate-limited -> sleeps (original); toggle ON -> bridges.
- Reframed the Settings toggle so connected users know it also covers rate-limits
("Use Spotify Free when Spotify is unavailable or rate-limited").
The official auth path is untouched; free never runs while authed Spotify works
normally.
Adds an opt-in no-creds Spotify metadata path: SpotifyClient serves SpotipyFree
data (real Spotify IDs) when metadata.spotify_free is enabled AND SpotipyFree is
installed AND there's no real Spotify auth. The data lands in the SAME spotify_*
columns, so the enrichment worker, search, and #775 lookups work UNCHANGED —
they just receive free-sourced data. The worker's only change is its availability
gate.
- core/spotify_free_metadata.py: SpotifyFreeMetadataClient + normalize_artist +
pure gate fns (should_use_free_fallback / should_offer_spotify_metadata /
spotify_free_installed).
- SpotifyClient: _free_enabled (opt-in, default OFF) / _free_available /
is_spotify_metadata_available / _free_active + in-client routing for
search_artists/tracks + get_album/artist/track_details/album_tracks/artist_albums.
- 3 scoped availability gates use is_spotify_metadata_available(): enrichment
worker loop, search resolve_client, watchlist. The ~40 discovery/user-library
sites stay auth-only (no sprawl, no user-data risk).
Authed users are untouched (gate closed when auth healthy). UI surfacing comes
next. Pure gates + normalizer tested; orchestrator resolve test added.
A manually-fixed mirrored track silently reverted to 'Wing It' after re-running
discovery. Two compounding causes:
- extra_data is MERGED on save (update_mirrored_track_extra_data), and the
manual-fix DB write (web_server.py) didn't clear the prior wing_it_fallback
flag — so a track fixed after being a Wing It stub kept wing_it_fallback=True.
- the Playlist Pipeline pre-scan checked wing_it_fallback BEFORE manual_match
(if/elif), so the stale flag won: the track was re-discovered and, on a miss,
fell back to Wing It — discarding the user's pick.
Fix: extracted the pre-scan gate into core.discovery.manual_match.should_rediscover
(manual_match checked FIRST = authoritative, regardless of leftover flags), and
the manual-fix write now also clears wing_it_fallback/unmatched_by_user. Behavior
is identical for every other branch — only the manual-vs-wing-it ordering changes.
Tested at the seam incl. the exact regression (wing_it_fallback + manual_match
both set -> skip). 227 discovery tests green.
Two complementary fixes to stop distinct artists ending up with the same source
id (the near-name collisions: ODESZA/odessa, Blance/Blanke, Lady A/Lady Gaga,
plus MusicBrainz's combined-score weak matches like Grant/Amy Grant):
- core/worker_utils.accept_artist_match() / source_id_conflict(): one shared,
tested gate. Rejects artist matches below 0.85 (stricter than the 0.80 used
for album/track titles, since short artist names false-positive easily) AND
refuses to store a source id a DIFFERENTLY-named artist already holds. A
same-named holder (one act across two media servers) is still allowed.
- Routed every artist-match worker through it: deezer, qobuz, tidal, discogs,
itunes, spotify (its scorer now uses the 0.85 threshold), audiodb, and
musicbrainz (conflict guard only — its matcher is combined-score, so the
guard is the net that catches its weak-name matches).
Centralizing in worker_utils avoids the copy-paste that let the original
album/track overwrite bug live in four workers at once. 17 new gate tests.
core/maintenance/dedupe_source_ids.py + scripts/dedupe_source_ids.py: find
source-id clusters held by differently-named artists (the enrichment-corruption
signature) and clear the id + match-status on those rows so the now-name-checked
workers re-derive each correctly on the next enrichment pass. Same-name
duplicates (one artist across two media servers) are left untouched.
Dry-run by default; --apply to write. 8 seam tests cover detection (corrupt vs
legit), dry-run safety, apply behaviour, and the no-op case.
The blind 'correct the parent artist's source id from an album/track match'
logic was copy-pasted into four enrichment workers; the Deezer fix only covered
one. AudioDB, Qobuz, and Tidal had the identical bug and would corrupt their own
id columns (and re-corrupt after any cleanup).
All three now gate the correction on a name match between the result's artist
and the parent artist (audiodb reads result['strArtist']; qobuz/tidal thread the
result artist name in from their callers, as Deezer does). Regression tests
cover mismatch-skips and match-corrects for each.
Root cause of the duplicate deezer_id corruption: when enriching an album or
track, _verify_artist_id 'corrected' the parent artist's deezer_id to the
search result's primary-artist id whenever they differed — with NO name check.
For a collaboration/compilation track (e.g. one our library credits to Jorja
Smith that lives on Kendrick Lamar's curated 'Black Panther' album), the result
resolves to Kendrick's album, so Kendrick's id (525046) got written onto Jorja,
Vince Staples, SOB X RBE, etc. — many artists ending up with the same id.
Now the correction only fires when the result's primary-artist NAME matches the
parent artist (the album/track-artist path now mirrors _process_artist, which
already name-checks). Mismatches are logged and skipped as collab/compilation.
Note: this prevents new corruption; existing wrong ids in a library aren't
auto-repaired (per-artist enrichment preserves an existing deezer_id).
A pasted Deezer artist link (or any Deezer-source artist click) opened the
wrong artist's header: deezer_id 525046 is stamped on 4 library rows (Kendrick
+ 3 others — an enrichment-corruption bug), and the library-upgrade lookup did
WHERE deezer_id=? LIMIT 1, grabbing an arbitrary row (Jorja Smith) while the
discography loaded fresh from Deezer (Kendrick) — a Frankenstein page.
find_library_artist_for_source now detects when a source id maps to >1 library
artist and refuses to guess: it skips the id-based upgrade (still allowing the
name fallback), so the caller renders the source artist directly — landing on
the correct artist. Unique ids are unaffected (no regression).
The underlying enrichment bug that writes one source id onto multiple artists
is separate and still worth a follow-up.
Spotify/Apple/MusicBrainz/Deezer artist links now resolve via each source's
get-by-id (get_artist / Deezer get_artist_info), shaped to the artist card and
rendered as an artist result that opens the artist detail page through the
existing flow. Album/track link handling is unchanged; bare IDs still rejected.
Follow-up to the bare-ID footgun: a bare number like 525046 carries no
source and no entity type, so it resolved to whatever album happened to own
that id (a user pasting Kendrick's Deezer artist id got an unrelated album).
Now the resolver accepts provider URLs (and the explicit spotify: URI) only;
a bare/unrecognized string is rejected and the dropdown surfaces a hint to
paste a full link. URL parsing + album/track resolution are unchanged.
New 'Link / ID' input on the Search page: paste a Spotify / Apple Music /
MusicBrainz / Deezer URL (or a bare ID) and it's looked up directly on the
owning source — no fuzzy search, no scoring.
- core/search/by_id.py: source-agnostic parser (URL domain/path or bare-ID
format -> source,kind,id; numeric IDs fan out, first hit wins) + per-source
get-by-id dispatch + adapters projecting each provider's dict onto the
standard album/track card shape.
- /api/enhanced-search/by-id: thin additive route over resolve_identifier.
- Frontend: dedicated input that adopts the resolved source as active and
renders through the existing dropdown + download/import flow.
Purely additive — existing files are insertion-only; the resolver runs only
behind the new route. 29 seam tests cover parsing, shaping, fan-out, and
not-found.
The album-bundle path COPIES slskd's completed files into private staging (then
on to the library) but never removed slskd's originals, so they piled up in the
download folder. (copy, not move, is correct for the torrent/usenet bundle paths
— those clients keep seeding — so the shared copier can't just always delete.)
Add an opt-in remove_source to copy_audio_files_atomically that deletes each
source ONLY after it copies successfully (never on a failed stage), and set it
for the Soulseek path only. Torrent/usenet keep their originals.
Tests: keeps source by default / removes when requested / keeps on failed copy.
The scoring best-of only helps if the right candidates were returned. File/CSV
titles ('Artist - Title') made the search query carry the artist prefix; add
canonical-title search queries so the correct tracks are actually found, then
the scorer best-of matches them. Additive (extra queries only when the title
canonicalizes differently).
#768 added canonical_source_track to the live-sync matcher and the playlist
editor reconcile, but NOT to the two paths that actually run for file/CSV
mirrored playlists: the discovery worker (core/discovery/playlist.py) and the
DB-only matcher (core/discovery/sync.py). YouTube playlists are cleaned at
ingest, so they matched; file playlists fed the raw 'Arctic Monkeys - Do I
Wanna Know?' title into search+scoring and never matched the library's clean
'Do I Wanna Know?' → reported missing / shown as 'extra'.
Add a conservative canonical best-of to both: score with the raw title AND the
canonicalized one, keep the better. canonical_source_track only strips an
'<artist> - ' prefix when it equals the artist, so it can only add a candidate.
Tests: _canonical_best_score seam (file-style match / clean title scored once /
keeps original when better).
start_playlist_sync validated the resolved mode with 'if sync_mode not in
(replace, append): sync_mode = replace' — a pre-existing clamp two lines below
the config read I added, which silently downgraded a configured 'reconcile' to
'replace'. So config=reconcile resolved correctly then got clobbered, and every
sync ran replace regardless of the setting (and incognito didn't help — it's
backend, not cache).
Replace the hand-rolled clamp with a pure normalize_sync_mode(requested,
configured) helper (VALID_SYNC_MODES includes reconcile) so the resolution is
testable and can't silently drop a mode again. Regression tests cover
reconcile-from-config, request-overrides-config, and unknown->replace.
Replace mode (default) deletes + recreates the server playlist every sync,
which wipes its custom image, description, and identity. Add an opt-in
'reconcile' sync mode that edits the existing playlist in place — adds the
tracks now in the source, removes the ones gone — without destroying the
object, so the user's custom art/description survive.
- Pure planner plan_playlist_reconcile(current, desired) -> {add, remove}.
- Per-client reconcile_playlist: Plex addItems/removeItems on the same object;
Navidrome Subsonic updatePlaylist delta (songIdToAdd / descending
songIndexToRemove); Jellyfin add + remove-by-PlaylistItemId on /Playlists/{id}/Items.
- sync_service: reconcile branch with a replace FALLBACK (if a server's in-place
edit is unavailable/fails, sync still succeeds destructively — logged loudly).
- Default stays 'replace' (no behavior change). New Settings > Playlist sync mode
picker (replace/reconcile/append) backed by playlist_sync.mode; per-request
sync_mode still overrides.
- Reconcile skips the post-sync source-image push so a custom poster isn't
re-clobbered (the bug).
Tests: planner (add/remove/dedupe/order/empty) + reconcile-or-replace dispatch
(success / false-fallback / exception-fallback / no-method). Per-server in-place
API calls need dev validation against real Plex/Jellyfin/Navidrome.
NOTE: opt-in only; default behavior unchanged.
Consistency follow-up: the Filler picked art via metadata source-priority +
its own prefer_source knob, ignoring metadata_enhancement.album_art_order —
the explicit cover-art source list that post-process embed and the Library
Re-tag job now use. So 'cover art sources' meant two different things.
Prefer select_preferred_art_url (the configured order) first; fall back to the
existing prefer_source / source-priority loop when no order is configured
(the default) — non-breaking, existing behavior unchanged for those users.
Help text updated. Test: configured order wins + skips the source loop.
User feedback (Sokhi): after changing cover-art sources, re-tag should
re-download fresh covers from THEM. The job took cover_url only from the
matched metadata source's album image, ignoring the user's configured
cover-art order. Now prefer select_preferred_art_url (the same
metadata_enhancement.album_art_order the post-process embed honors), falling
back to the source image when no order is configured (non-breaking).
'replace' cover mode already force-refreshes art on every matched album, and
the embed replaces existing art (no duplicate pictures) — so 'replace' + a
configured art order = fresh covers from those sources. Help text updated.
Tests: prefers configured source URL / falls back to source image when unset.
Find & Add on the playlist-sync page only wrote sync_match_cache, which is
DELETEd wholesale after every DB scan — so the source->library pairing (and
the user's manual matches) reverted to 'extra'/red-dot on the next shallow
scan. The three match stores (sync_match_cache, manual_library_track_matches,
discovery extra_data) were disconnected and all pointed at tracks.id, which a
rescan re-keys (esp. Jellyfin/Navidrome GUIDs).
Unify the match so it's one durable fact, recorded once, honored everywhere:
- Find & Add also writes a durable manual_library_track_matches row (one-way;
the manual-match tool has no playlist to act on, so no reverse). Carries the
library file path.
- New library_file_path column (idempotent migration) + find_track_id_by_file_path:
re-resolve a stale library_track_id after a rescan re-keys the track, and
self-heal the row.
- The sync compare display's override lookup now falls back to the durable
manual match (resolve_durable_match_server_id) when sync_match_cache misses —
so the pairing persists across a scan instead of reverting to a red dot.
Purely additive: only adds matches when the cache returns nothing.
Tests: durable resolver (valid / stale-reresolve+self-heal / no-match / not-in-
playlist / missing-methods), file_path persistence + find_track_id_by_file_path.
A bare host like '192.168.1.5:8080' or 'qbittorrent.lan:8080' (no scheme)
is what users naturally type, but requests then raises 'No connection
adapters were found for ...' — it can't pick an http/https adapter, and a
bare host:port even gets misparsed as scheme=host. This surfaced as the
generic 'qbittorrent probe failed' with a 'login error: No connection
adapters were found' in the logs.
Add normalize_client_url() in torrent_clients/base: default a missing scheme
to http:// (+ trim trailing slash), and route all three adapters'
_load_config through it. Transmission normalizes the base before appending
/transmission/rpc.
Tests: normalizer unit cases + per-adapter regression (bare host -> http://).
Note: usenet adapters (sabnzbd/nzbget) share the same pattern and need the
same treatment in a follow-up.
Follow-up hardening to #789. The selection was keyed purely by folder name,
so renaming a music folder in Navidrome silently reverted the scan to all
libraries. Now persist the folder id (stable across renames) as the primary
key alongside the name (kept for display + back-compat), and restore by id
first with a name fallback. Self-heals on reconnect: pre-id installs and
drifted/renamed names get the id + fresh name written back, so the settings
dropdown keeps highlighting the right folder.
Tests: restore-by-id-after-rename (+ name heal), name-fallback self-heals id,
no-drift writes nothing.
The saved music-folder selection was silently dropped on every reconnect.
_setup_client's restore step called the public get_music_folders(), which
starts with ensure_connection() — but we're already inside ensure_connection()
at that point (_is_connecting=True, _connection_attempted not yet set), so the
re-entrant call bailed and returned []. The restore matched nothing,
music_folder_id stayed None, and the per-call musicFolderId filters all
no-op'd → scans imported every library regardless of the user's choice.
Surfaces after any restart or settings save (reload_config resets the state).
Split get_music_folders() into the public method (does the connection check)
and a non-reentrant _fetch_music_folders() seam; the restore now calls the
seam directly (connection is already established + ping succeeded by then).
Regression + seam tests added.
- depth setting (light = core tags + matched source ids; full = same
multi-source enrichment cascade a fresh download gets, run additively
via embed_source_ids). Threaded through scan/finding/auto-apply and the
repair_worker fix handler.
- source now defaults to 'auto' (= your source priority / active source)
instead of blank.
- give native <option> popups a solid dark background (were white-on-white).
- tests for full-depth full_meta payload + enrich invocation + light no-op.
The job was the odd one out — auto_fix=False, no dry_run setting, so it never
showed the 'Dry Run' badge the other jobs do (the badge keys off
settings.dry_run === true). Aligned it to the standard pattern:
- auto_fix=True + dry_run setting defaulting True. Default behavior is unchanged
(findings only, nothing written) AND it now shows the Dry Run badge.
- Turning dry_run off makes the scan auto-apply in place (result.auto_fixed),
no finding — the opt-in 'just retag it' mode.
- Extracted a shared apply_track_plans() used by both the scan auto-apply and
the repair_worker fix handler (handler now resolves Docker paths then
delegates — one code path, no duplication).
Tests: dry_run=False auto-applies + writes + no finding; existing dry-run
finding/skip/apply tests still green. 410 passing.
The old per-download Retag Tool was limited (only native-pipeline downloads,
100-group cap, manual per-group) and did the wrong thing — it moved/reorganized
files instead of just tagging. It's superseded by the new Library Re-tag job
(whole-library, in-place) + the enhanced-library 'Write Tags' button.
Removed: the post-download record_retag_download ingestion hook (stops writing
retag_groups on every download), core/library/retag.py, the web_server state +
deps + /api/retag/* endpoints + the tool:retag WebSocket emit, the dashboard
card + both modals (index.html), the core.js socket handler, and the tools-page
wiring + help entry (wishlist-tools.js). Updated the import-pipeline test.
Verified: web_server parses, app + core imports OK, 392 tests pass, no live
references to removed symbols.
Left as inert (harmless) for a careful follow-up sweep: the retag_groups/
retag_tracks tables + their DB CRUD methods (no longer written/read), and the
now-orphaned retag JS helper functions (no entry point/wiring/socket calls them;
interspersed with wishlist functions, so not blind-deleted).
write_tags_to_file wrote the core fields + cover but never the source IDs
(Spotify/iTunes/MusicBrainz) the import post-process embeds. Added a focused
source.embed_known_source_ids() that writes ALREADY-KNOWN ids (from db_data)
via the canonical, Picard-compatible frame writer the import uses
(_write_embedded_metadata) — no API re-fetch, frames correct by construction.
write_tags_to_file now calls it whenever db_data carries id keys.
Fed from both paths: the enhanced-library 'Write Tags' button now carries the
track's spotify/itunes/musicbrainz ids, and the Library Re-tag job stamps the
matched album/track source ids onto each track. So both now write the full tag
set, not a subset.
New 'Library Re-tag' repair job (default-OFF, opt-in; weekly when enabled):
- Scans every source-matched album (spotify/itunes/deezer/musicbrainz album id),
pulls fresh metadata + tracklist from that source, reads each local track's
current tags, and uses the planner to compute per-field diffs.
- Dry-run by design: scan only CREATES findings — nothing touches a file. Each
finding is highly detailed: per-track old->new for every changed field, the
source used, the mode, a cover-art action, and any unmatched tracks, plus a
summary description. Settings: mode (overwrite | fill_missing), cover_art
(replace | fill_missing | skip), source override.
- Apply handler (_fix_library_retag in repair_worker): writes each track's
planned tags in place via tag_writer.write_tags_to_file (+ batch-embeds cover,
refreshes cover.jpg). Only adds/overwrites planned fields — no moves/renames/
re-match. Resolves Docker paths; read-only/unreachable files counted, never
crash. Media-server-only / unreachable tracks are skipped.
Registered in the job list + fix dispatch. The old per-download Retag Tool is
left untouched alongside this for now.
The testable core for the new library-wide re-tag job. Given a source album's
metadata + tracklist and the library tracks' current file tags, it:
- matches source tracks to library tracks (disc+track number, then title sim),
- computes the per-field diff (old -> new) for the dry-run finding,
- builds the minimal write_tags_to_file payload — only fields that actually
change under the chosen mode (overwrite vs fill-missing), so applying never
touches unrelated/unchanged tags.
No IO/network/DB — 10 unit tests cover matching, both modes, blank-source
fields, and the album-artist/track-count payload mapping.
Verification found a non-additive edge: embed_album_art_metadata uses FLAC
add_picture(), which APPENDS — so applying to an album where some tracks already
had art would have added a duplicate embedded picture. The apply now checks each
file and skips any that already carry art (shared _audio_has_art helper), so it
only ever ADDS art to files missing it. Test covers the skip (no re-embed).
Previously the filler only flagged albums whose DB thumb_url was empty and, on
apply, only updated that DB thumb_url — so albums whose files had no embedded
art and no cover.jpg (but whose DB row had a URL) were never found, and even
'applying' art never touched the files. That's the reported 'doesn't scan all
albums' gap.
New core.metadata.art_apply (reuses the post-processing standard so the user's
album_art_order is honored):
- album_has_art_on_disk(): cheap-first check — folder cover.jpg/folder.jpg
sidecar, then embedded art in a representative track (FLAC/ID3/MP4/Vorbis).
- apply_art_to_album_files(): embeds via embed_album_art_metadata + writes
cover.jpg via download_cover_art; only ADDS art (never rewrites the user's
tags); read-only/unwritable files are skipped + counted, never crash.
Scan now examines every titled album and flags it when art is missing in the DB
OR on disk. Apply embeds into the album's audio files + writes cover.jpg in
addition to the DB thumbnail (media-server-only albums fall back to DB-only).
Tests cover sidecar/embedded detection, the cheap-first short-circuit, and the
apply orchestration (embeds each file + cover.jpg; read-only failures counted).
The title/artist fallback search took results[0]'s artwork unconditionally, so
a loose full-text match returned the wrong album's cover (the 'new sources give
incorrect art' reports). Now it pulls a few results and only accepts one whose
title matches (subset, to allow Deluxe/Remaster) AND whose artist matches
exactly — the artist being the strong guard against wrong covers. Falls back to
an exact title match when a result carries no artist.
The album's own stored source-id path is unchanged (that id is authoritative).
Tests: wrong-artist rejected, skips wrong result for a matching one, + unit
coverage of the matcher (deluxe/feat/stopwords accepted, wrong artist/title
rejected).
qBittorrent 5.2.0 changed /api/v2/auth/login to return HTTP 204 (No Content)
on success instead of HTTP 200 with body 'Ok.'. The adapter required the body
to equal 'Ok.', so every login on 5.2.0+ failed with 'HTTP 204 body=' — the
connection probe and all torrent actions were broken.
Treat login as successful on the SID auth cookie and/or a success body: 'Ok.'
(<=5.1) or an empty HTTP 204 (>=5.2.0). Still reject bad creds, which
qBittorrent reports as HTTP 200 + 'Fails.' (not a 4xx).
Tests: 204-empty -> success, SID-cookie+empty-body -> success, 'Fails.' (even
with a stale cookie) -> failure.
- Stale-cache check (playlistTrackCacheIsStale) compared raw track_count to the
filtered/cached track list, so any playlist with local or unavailable tracks
always looked 'stale' and refetched + re-mirrored on every modal open. Now it
compares the upstream snapshot_id (stored at cache time in the shared fetch
choke point), and returns not-stale when no snapshot is available — explicit
invalidation on refresh still handles real changes.
- organize_download: guard executor.submit so a refused job cleans up the batch
instead of stranding it in 'analysis' (holding a limited analysis slot).
- Removed the dead, deprecated, unused mirrorSpotifyPlaylistTracks.
resolve_mirrored_playlist tried the mirrored-playlists primary key FIRST for
any all-digit ref. Deezer upstream ids are all-numeric, so a Deezer playlist id
was mistaken for the PK and the organize-by-playlist toggle resolved a wrong row
(or nothing) — the toggle silently wouldn't save / 'Open in Mirrored' missed.
Resolve by (source, source_playlist_id) first, fall back to PK only when the
source lookup misses. Thread the batch/wishlist source through the download-path
callers so numeric upstream ids resolve correctly there too. Spotify (base62
ids) is unaffected.
Seam tests: numeric Deezer id resolves by source (not PK), spotify alphanumeric
by source, PK fallback still works, profile-scoped, empty refs -> None.
The worker's WARNING observability proved the '38 errors' were almost all
MusicMap returning 404 (artist has no map page) — a genuine not-found, not a
fetch failure. But iter_musicmap_similar_artist_events flattened every
RequestException to status_code 502, and the worker maps 400/404 -> not_found
/ everything-else -> error, so these inflated the error count.
Surface the real HTTP status from the exception's response (404 stays 404),
falling back to 502 only when there's no response (timeout/connection drop,
which is correctly still an error eligible for retry).
Regression tests: 404 -> 404 (not_found), timeout -> 502 (error), 500 stays
error, plus an end-to-end worker check that a 404 result marks 'not_found'
and stores nothing.
Verified against live data: 1312/1313 stored similars carry a metadata source id,
but 1 slipped through name-only (a match on a source with no id column, e.g.
discogs). Enforce the standard: process_artist now SKIPS any similar whose match
doesn't map to a storable id column (spotify/itunes/deezer/musicbrainz) instead
of writing a useless name-only row. Regression test covers discogs-match + no-id
cases. Now 100% of newly-stored similars are actionable.
The kettui move: 38/79 fetches errored on the first live run, but they were
logged at DEBUG only — invisible in app.log, so the cause (rate-limit vs
no-providers vs bug) is unprovable. process_artist now returns a (status, count,
detail) triple carrying the error reason (status code + message / exception),
and the worker logs the first 15 errors per session at WARNING (rest DEBUG) +
keeps _last_error. No blind pacing tweak — let it run, read the real reason, then
fix the proven cause. Seam tests updated + assert the reason is captured.
- get_stats now reports PERSISTENT counts from the DB (matched/not_found/pending
+ a progress.artists breakdown) instead of in-memory session counters, so the
dashboard orb tooltip and the Manage modal agree (was showing 0 vs 14 after a
restart) and it survives restarts — same approach as the other workers.
- Orb tooltip reads progress.artists ('Artists: 14 / 15 (93%)') like the rest.
- Worker now defaults to ON (running) instead of opt-in-paused; still honors a
saved pause across restarts. It self-paces (~3s/artist) and backs off on
MusicMap outages, so the orb spins/active like the others when there's work.
10 seam tests pass.
SoulSync standalone matches library tracks without Plex fetchItem,
reports missing counts correctly, and skips server playlist writes.
Automation re-syncs when the mirror grows; after sync finishes, starts
organize download (organize-by-playlist) or wishlist processing.
UI: Spotify URL playlist-folder controls, organize toggle layout in the
discovery modal, reload organize preference when reopening Download Missing.
Co-authored-by: Cursor <cursoragent@cursor.com>
Closes the gap where similar artists only existed for WATCHLIST artists: a new
background worker populates them for the whole LIBRARY, slotting into the
existing enrichment-worker pattern (bubble + Manage Enrichment Workers modal,
status/pause/resume, matched/not_found/pending/errors).
Per source-matched library artist → get_musicmap_similar_artists(name, 25)
(the same matcher the artist-detail page uses: fetches MusicMap names, matches
each to the user's source chain — primary + active fallbacks — returns only
matched artists) → store via add_or_update_similar_artist keyed by the artist's
metadata source id, the SAME key the watchlist scanner + artist map use, so the
two cooperate (idempotent upsert + retry_days window).
- core/similar_artists_worker.py: pure seams (pick_source_artist_id,
map_payload_to_store_kwargs, process_artist) + the threaded worker; skips
artists not yet source-matched; classifies not_found vs transient error
(retry after 30d).
- DB migration: similar_artists_match_status / _last_attempted on artists
(mirrors every other source worker's tracking columns).
- Registered in EnrichmentService + instantiated in web_server, DEFAULT-PAUSED
(opt-in) like Amazon — MusicMap is scraped/outage-prone + this is library-wide.
- SERVICE_ENTITY_SUPPORT['similar_artists']=('artist',) so the modal breakdown
('artists with / without similars') + Retry work; manual-match (inapplicable
to a relationship) is gated out via relationship:true.
- 10 seam tests; existing 80 enrichment tests still pass.
Note: keys under profile 1 (single-profile setups); multi-profile is future work.
Persist organize_by_playlist on mirrored playlists and run playlist-folder
downloads from the auto-sync pipeline instead of the global wishlist phase.
Register SoulSync library rows after playlist-folder post-processing, route
failed organize batches to the wishlist correctly, and skip sync-time
unmatched wishlist only when organize download handles retries.
Invalidate stale playlist track caches on refresh (Spotify and Deezer ARL),
re-mirror on refetch, and improve standalone playlist modals (re-analysis,
Open in Mirrored). Add filesystem missing-track detection and tests.
Co-authored-by: Cursor <cursoragent@cursor.com>
- Verified end-to-end: fetch_public_playlist_full pulled all 236 tracks of the
test playlist via SpotipyFree (the library handles the client-auth that 429'd
the raw approach). Name + tracks correct.
- requirements.txt: declare spotipyFree>=1.1.2 as a normal pip dependency (like
spotDL, also MIT — aggregation, not vendored) + websockets (a transitive dep
SpotipyFree/spotapi needs that pip doesn't pull automatically). Code still
soft-imports + falls back to embed, so it's never a hard runtime requirement.
- meta fetch uses limit=1 (name/owner only) so we don't pull the whole list
twice. 9 tests green.
The in-house anonymous-token path is blocked by Spotify (429 without the web
player's rotating client-auth). Switch the full-fetch to SpotipyFree — the
maintained no-creds spotipy drop-in spotDL uses, which tracks that machinery.
- core/spotify_public_api.fetch_public_playlist_full now uses a SpotipyFree
client (playlist + playlist_items + next), normalising the spotipy-shaped
items to the embed scraper's shape. Injectable client_factory keeps it
unit-testable without the library or network. Dropped the dead in-house
token/pagination code.
- Licensing: SpotipyFree is GPL-3.0, so it is NOT bundled/required (SoulSync is
MIT). Optional, user-installed: the import is soft, and on ImportError (or any
failure) fetch_spotify_public falls back to the embed scraper (~100). So the
shipped project stays cleanly MIT and the link path never regresses.
- requirements.txt: documents it as a commented optional extra
(pip install SpotipyFree) with the GPL/MIT rationale.
- 9 tests: normalisation, pagination past 100, library-missing -> raises (->
fallback), and the embed-fallback orchestration.
Needs a live click-through with SpotipyFree installed to confirm the exact
class/method names match (SpotipyFree.Spotify / playlist / playlist_items).
The full-fetch's logs used a bare module logger that app.log doesn't capture,
so we couldn't see whether the API path succeeded or why it fell back. Route
them to 'soulsync.spotify_public' and log: token found?, embed parsed?, the
API HTTP status on a non-200, and pagination result. Lets us see the exact
failure (e.g. 401 vs 429) on the next link-tab test.
Live debugging the 'shows 100' report:
- The full playlist page no longer embeds an accessToken, and get_access_token
/ server-time now 403/404. The EMBED page (open.spotify.com/embed/playlist/{id})
still ships a usable anonymous token. Was fetching the wrong page -> no token
-> raised -> embed fallback (100). Now reads the embed page for the token.
- Confirmed live: token extraction + embed parse work; the token is accepted by
the Web API (429 rate-limit, not 401). Could not show >100 from here because
the test IP got rate-limited from probing; needs a clean-IP click-through.
While in there, made it more robust against the rate-limiting that's clearly in
play:
- Refactored scrape_spotify_embed -> reusable parse_embed_html.
- fetch_public_playlist_full now does ONE embed fetch for token + name + first
page (no separate metadata call = fewer requests = less 429 surface), then
paginates the API. If the API is unavailable/rate-limited, it keeps the embed
page's tracks (<=100) instead of raising — so the result is always >= today's
behaviour, never worse.
- 12 tests incl. the new API-fails-but-embed-tracks-survive path.
Caveat unchanged: rides Spotify's undocumented embed-page token; degrades to the
embed fallback, never crashes.
The no-auth 'add by link' path scrapes Spotify's embed widget, which only ever
contains ~100 tracks and can't paginate — so big public playlists got
truncated. This adds an in-house anonymous fetch that pulls the FULL list:
- core/spotify_public_api.py: reads the anonymous web-player accessToken Spotify
already embeds in its own open.spotify.com page HTML (no app credentials, and
no rotating TOTP secret for us to maintain), then paginates
/v1/playlists/{id}/tracks 100 at a time until the whole playlist is pulled.
Returns the embed scraper's exact shape. Pure helpers + injected http_get so
it's unit-testable without the network.
- core/spotify_public_scraper.fetch_spotify_public(): tries the full fetch for
playlists; on ANY failure (or for albums) falls back to scrape_spotify_embed.
Worst case == today's behaviour, so the link path can't regress.
- web_server: the link-tab endpoint and the authed flow's last-resort scrape
now both go through fetch_spotify_public.
Scoped entirely to the spotify_public_* (no-auth) path — the authenticated
playlist sync is untouched. 11 tests (token extraction, normalisation,
pagination past 100, and the embed-fallback orchestration).
Caveat: rides Spotify's undocumented page-embedded token — expected to break
when they change their page; it degrades to the embed fallback, never crashes.
Needs a live click-through to confirm the token path works end to end (can't
hit Spotify from the test env).
- #1 Unconfigured-source banner: when a source has enabled=false, show a
notice that browsing works but matches/retries won't run until it's set up.
- #2 Rate-limit detail: when rate_limited, surface 'resumes in ~Xm' (from the
status payload) instead of just a pill.
- #3 Richer rows: unmatched items now show parent context — an album's artist,
a track's album — via a parent expression in the query (+ test).
- #4 Bulk select: per-row checkboxes + a bulk bar to retry several at once
(capped concurrency), reusing the /retry item endpoint.
- #5 Remember last worker: selection persists in localStorage and is restored
on open; openEnrichmentManager(workerId) supports future deep-linking
(bubbles left on their pause-on-click behaviour).
- #6 Keyboard nav: ArrowUp/Down moves focus between rows; actions are native
buttons (Enter/Space) and Escape closes — list isn't poll-refreshed so focus
is stable.
53 enrichment tests green; JS syntax clean.
Per-worker processing-order override + UI polish.
Feature — pin an entity group to enrich first:
- Each worker normally runs artist -> album -> track. A user can pin one
group (artist/album/track) to run first from the modal; the worker keeps
that group first until it's exhausted, then resumes the normal chain.
- core/worker_utils.py: read_enrichment_priority() (reads
<service>_enrichment_priority each loop, live) + priority_pending_item()
(shared, whitelisted query returning the worker's expected item shape;
Spotify/iTunes get album_individual/track_individual via a type map).
- A guarded ~6-line hook at the top of all 11 workers' _get_next_item.
CRITICAL: when nothing is pinned (default) the hook returns immediately,
so default enrichment order is byte-identical to before. Discogs (no track)
and Genius (no album) only honor their supported entities.
- core/enrichment/api.py: GET/POST /api/enrichment/<id>/priority (+ config_get
hook); POST validates the entity against what the source enriches.
- 14 new tests (helper shapes, exhaustion, route get/set/clear/validate).
UI:
- Refined hero header: identity + inline status left, single Pause right,
'now enriching' quiet sub-line; overall coverage % moved into the stats
section ('82% matched · 1,203 of 1,460'). Hero gently pulses while running.
- New processing-order strip: artist→album→track steps showing the live phase
(pulsing 'now'), pinned group ('first' + 📌), and done/remaining; click a
step to pin it, click again for auto.
py_compile clean across all 11 workers; 52 enrichment tests green.
Fixes a correctness bug and adds bulk re-queuing.
- Bug: per-row 'Retry' used clear-match, which sets an item to not_found
with last_attempted=NULL. The worker only retries not_found items where
last_attempted < (now - 30d), and 'NULL < cutoff' is false in SQLite, so
those items were never re-queued. Fixed by resetting match_status to NULL
(pending), which every worker's queue picks up on the next pass.
- New POST /api/enrichment/<id>/retry with scope 'item' | 'failed'
(failed = re-queue every not_found item of an entity type), backed by a
pure whitelisted build_reset_query + MusicDatabase.reset_enrichment().
- UI: per-row Retry now hits /retry; a 'Retry all failed' bulk button appears
when the current entity has not-found items (confirm + count toast); a hint
line explains retry/match/auto-retry behaviour.
- 11 new tests (38 enrichment tests total, all green).
Dashboard 'enrichment bubbles' could pause/hover but offered no way to
*manage* a worker. This adds a full management modal opened from a new
header button, covering all 11 enrichment sources.
Backend (testable core helper + seam tests; no live-DB dependency):
- core/enrichment/unmatched.py: pure, whitelisted SQL builders for the
unmatched browser. service/entity validated against a support map (never
interpolated raw); search + pagination bound as params; tracks join albums
for artwork; limit capped at 200.
- database/music_database.py: get_enrichment_unmatched() +
get_enrichment_breakdown() (the breakdown splits matched/not_found/pending,
which the existing get_stats().progress lumps together).
- core/enrichment/api.py: GET /api/enrichment/<id>/{unmatched,breakdown} on
the existing blueprint + a db_getter hook.
- web_server.py: wire db_getter=get_database.
- tests/enrichment/test_unmatched.py: 19 tests across builders, DB methods,
and Flask routes.
Frontend (vanilla, matches app conventions):
- webui/static/enrichment-manager.js: worker rail with live status + coverage
micro-bars, accent-themed detail panel (hero header, segmented matched/
not_found/pending stat cards, current item, pause/resume), and a searchable
paginated unmatched browser with inline manual match (reusing
search-service + manual-match) and retry (clear-match re-queues).
- Polish: entrance/exit motion, scroll-lock, Escape, refresh control,
flicker-free polling (in-place updates), skeleton loaders, relative
timestamps, per-worker accent theming, real dashboard logos reused at
runtime (with the same invert/circle treatment), responsive rail.
- index.html: header button + script include. style.css: full styling.
Reuses existing pause/resume, status, and manual search+assign endpoints.
Backend tests green (19 new + 11 existing enrichment tests).
- B023: default_fetch_tracklist built a per-item lambda closing over the loop
variable `it`. Replaced with a module-level _item_get(item, key, default)
helper (takes the item as a param — no closure). Behavior unchanged; the
dict/object normalization test still passes.
- S110: the two best-effort guards in the canonical job (skip-already-pinned
read, estimate_scope active-server read) now carry `# noqa: S110 — <reason>`,
matching the repo's existing convention for intentional swallow-and-continue.
ruff check passes on all canonical files + tests; 30 affected tests green.
The canonical source_selection setting was rendering as a free-text box — easy
to typo an invalid mode. Added a generic choice mechanism so it's a dropdown:
- RepairJob.setting_options: {key: [allowed values]} (default {} — opt-in).
- CanonicalVersionResolveJob declares source_selection's three modes.
- repair_worker.get_all_job_info() includes setting_options in the job payload.
- enrichment.js renders a <select> (options prettified, current value selected)
for any key listed in setting_options; everything else renders by value type
as before. The save path already reads <select>.value as a string, so no
change needed there.
Generic — any future job can get dropdowns the same way. Jobs that don't
declare setting_options are untouched (empty dict -> existing input rendering).
Tests: source_selection exposes the 3 options and its default is one of them.
23 repair-job/worker + canonical tests pass (other jobs unaffected).
Per request, pack each finding with everything available WITHOUT extra API
calls (kettui: reuse what's already fetched, read the album row we already
loaded, degrade per-field, keep it tested):
- Pinned release's track titles — already fetched during scoring, so free
(capped at 60 to bound details_json).
- From the album row (free): year, DB track count, total duration, genres-free
context, and the album's currently-linked source IDs.
- file_track_titles (your library's titles) for a side-by-side with the release.
- Artist + album thumbs (artist via the guarded lookup) and names.
_describe_pin now renders: "Artist — Album (year)", the fit breakdown, "Currently
linked: … → pinning X", "Beat: <alternatives>", and the release tracklist — so
the card is judge-able at a glance, and the structured fields are in details for
a richer UI.
NOT included (would cost an extra per-album API fetch, left as opt-in): the
*release's* own year/type/cover/URL from get_album_for_source, vs the library's.
Tests: _describe_pin rich-render (year/linked/tracklist), resolver release-titles,
orchestration free-context fields. 94 canonical + reorganize regression pass.
Findings now carry artist_thumb_url alongside album_thumb_url (same key the
track-repair findings use, so the findings UI already renders it).
Fetched via a guarded _lookup_artist_thumb() — checks the artists table has a
thumb_url column first and swallows any error — rather than adding ar.thumb_url
to the shared load_album_and_tracks SELECT. The shared-loader approach was
tried first and REVERTED: it crashed reorganize on schemas whose artists table
has no thumb_url column (caught by 40 orchestrator tests). The lookup only runs
for albums that actually resolve, so it adds no cost to the no-source-id
short-circuit majority.
Tests: orchestration test asserts artist_name + album_thumb_url + artist_thumb_url
flow through. 47 canonical + 104 canonical/reorganize regression tests pass.
Live-run feedback: "Best-fit release: deezer (665666731), score 1.0" is too thin
to trust/accept. Each finding now explains WHY:
- score_release_detail() exposes the per-signal breakdown (count/duration/title)
instead of just the blended score.
- resolve_canonical_for_album returns an enriched result: the breakdown,
file_track_count vs release_track_count, and a `candidates` list of every
source it scored (so a finding can show what the winner beat).
- resolve_and_store adds album/artist/thumb context from the row it already
loaded (no extra query). Storage still only reads source/album_id/score.
- The job builds a real description via _describe_pin(), e.g.:
"Pin deezer release 665666731 (confidence 100%).
Fit to your library: 11 files vs 11 tracks on this release — track count
100%, durations 100%, titles 100%.
Beat: spotify 65% (17 tk)."
and a clearer title ("Pin deezer as canonical: <artist> — <album>").
Tests: resolver enrichment (breakdown + candidate comparison fields), and
_describe_pin (judge-able text incl. the beaten alternatives, and honest "n/a"
for a missing signal). 42 canonical tests pass.
Note: the description string carries the judge-able info regardless of UI; how
the findings tab renders the extra details keys (thumb image, candidates table)
is still UI-dependent and unverified.
Feedback from the live dry-run: the job was pinning whichever source best fit
the files regardless of which source it was, which was surprising — users
expect it to respect their active metadata source. Made it a per-job setting
instead of a baked-in policy.
source_selection (default 'active_preferred'):
- active_preferred — use the active/primary metadata source's release when the
album has an ID for it AND it clears the score floor; otherwise fall back to
the best-fit among the other sources. Respects the configured source but
self-heals when that link is clearly broken (below floor / no ID).
- active_only — only ever the active source; never considers others.
- best_fit — previous behavior: whichever source matches the files best.
resolve_canonical_for_album gains mode + primary_source; the orchestration
threads the primary source through; the job reads source_selection from its
settings. Note: active_preferred respects the active source as long as it clears
the floor, so it will NOT override a deluxe-vs-standard mismatch on the primary
(#767-Bug2) — that's what best_fit is for; the choice is now the user's.
Tests: per-mode coverage in test_canonical_resolver.py (active_preferred uses
primary when it fits, falls back when primary is below floor, keeps primary even
when another fits better; active_only pins primary / never falls back; best_fit
unchanged), orchestration default-mode test, and the setting default. 39
canonical tests pass.
The populate trigger that turns the (until now dormant) feature on. Until a user
enables and runs this job, no album has a canonical -> both read sides (Stages
3-4) fall back -> zero behavior change. So the whole feature ships safely off.
- core/repair_jobs/canonical_version_resolve.py — "Resolve Canonical Album
Versions". Iterates the active server's albums, skips ones already pinned, and
calls the tested resolve_and_store_canonical_for_album per album. Opt-in
(default_enabled=False) and dry-run-by-default: resolving compares an album's
candidate releases across sources (metadata-source API calls, once per album),
so it's deliberately user-triggered. Dry run reports a finding per album it
would pin; live mode stores. Registered in _JOB_MODULES.
- core/metadata/canonical_resolver.py — resolve_and_store gains store=True; the
job's dry run passes store=False to resolve-without-writing.
Tests: tests/test_canonical_version_job.py (6) — registered, opt-in + dry-run
defaults, live resolves+stores (auto_fixed), dry run creates findings without
persisting, already-pinned albums skipped. Registry loads all 19 jobs cleanly.
145 tests across the full feature + reorganize/track-repair/DB regression pass.
_resolve_album_tracklist gains a Fallback -1: if the album has a pinned
canonical (source, album_id), use it before the existing 6-level cascade — so
Track Number Repair resolves the SAME release the Reorganizer does (Stage 3) and
the two stop contradicting each other (#765, the Spotify-4 vs MusicBrainz-3
conflict).
Gated + additive: the entire existing cascade is untouched for albums without a
canonical, so this job's all-01-album rescue (which relies on the MusicBrainz/
AudioDB fallbacks for albums with no DB source ID) is fully preserved — that's
the regression we explicitly refused to take in a reactive fix.
New helper _lookup_canonical_from_db() mirrors _lookup_album_ids_from_db
(file-path -> track -> album), returns None when no DB / no match / columns
absent / unresolved.
Tests: tests/test_track_repair_canonical.py (4) — returns canonical when pinned,
None when unresolved / file untracked / no DB. Existing track_number_repair
tests still pass (no regression).
_resolve_source now prefers the album's pinned canonical (source, album_id) when
set, before the source-priority walk. So once an album's canonical is resolved,
reorganize agrees with Track Number Repair (Stage 4) and stops mislabelling a
standard album as deluxe (#767-Bug2).
Gated + side-effect-free: only changes behavior for albums that already carry a
canonical (none do until the populate step runs), an explicit user source pick
(strict_source) still wins over the canonical, and a failed canonical fetch
falls through to today's priority walk. So this stage is behavior-neutral until
canonical is populated.
Tests: tests/test_reorganize_canonical_source.py (4) — canonical preferred over
priority, fetch-failure falls back, strict_source ignores canonical, no-canonical
unchanged. 113 reorganize-orchestrator/tag-source/unknown-artist tests still pass
(no regression).
Completes Stage 2's populate path. Still dormant — no consumer calls it yet.
- resolve_and_store_canonical_for_album(db, album_id, ...): loads the album's
source IDs + its tracks' (duration_ms, title) from the DB via the SAME
loader the Reorganizer uses (load_album_and_tracks + _extract_source_ids), so
the canonical is chosen over exactly the source IDs the reorganizer sees;
scores off the DB track rows (the library's view of the files — no per-file
disk reads), resolves the best fit, and persists it. Returns the stored result
or None when unresolved.
- default_fetch_tracklist(): production fetcher wrapping
get_album_tracks_for_source, normalising to {title, track_number, duration_ms}
(duration best-effort; sec->ms; absent -> scorer leans on count+title).
Design note: chose LAZY resolution (Stages 3-4 consumers call this when they hit
an album with no canonical) over a standalone backfill repair job — no new
scheduling/UI surface, resolves only when a tool actually needs it, and stays
gated (NULL canonical = today's behavior).
Tests: tests/test_canonical_orchestration.py (5) — end-to-end on a real temp DB
(11 files pick the 11-track release over a 17-track deluxe and persist it),
no-source-ids -> None, missing-album -> None, and default_fetch_tracklist
normalization (dict items, seconds->ms) + failure -> None. All canonical +
DB-migration tests green.
Turns the Stage-1 scorer into an end-to-end resolver + persists the result.
Still DORMANT — no consumer reads it yet, so zero behavior change.
- core/metadata/canonical_resolver.py — resolve_canonical_for_album(): builds
candidate releases from the album's per-source IDs (in source-priority order),
fetches each tracklist via an INJECTED fetch_tracklist (so it's unit-testable
without live APIs), scores them with pick_canonical_release, and returns the
best-fit {source, album_id, score}. Skips sources with no id / failed fetch;
returns None when there are no files, no candidates, or nothing clears the
confidence floor.
- database/music_database.py — set_album_canonical() / get_album_canonical()
write/read the Stage-1 columns. get returns None when unresolved, which every
consumer will treat as "fall back to today's behavior".
Tests: tests/test_canonical_resolver.py (7) — best-fit beats priority, priority
breaks true ties, skips missing-id/failed-fetch sources, None on
no-candidates/no-files/below-floor, score rounding. tests/test_canonical_db.py
(4) — set/get round-trip incl. timestamp, unresolved -> None, overwrite,
missing-album -> False. 34 canonical + DB-migration tests pass.
Remaining for Stage 2 (the trigger): read on-disk file durations/titles for an
album, gather its source IDs, call the resolver, store — wired via a backfill
repair job + an enrichment hook. Then Stages 3-4 wire the Reorganizer and Track
Number Repair to READ the pinned canonical.
First stage of the canonical-album-version fix (#765 + #767-Bug2). Pins ONE
canonical (source, album_id) per album, chosen by best-fit to the user's actual
files, so the Reorganizer, Track Number Repair, and tagging stop re-resolving
independently and contradicting each other.
Ships DORMANT — nothing reads or writes the new data yet, so zero behavior
change. Later stages populate (Stage 2) and consume (Stages 3-4) it.
- core/metadata/canonical_version.py — pure scorer (the testable heart):
score_release_against_files() rates a candidate release by track-count fit +
duration alignment (greedy nearest within ±3s) + title overlap, dropping and
renormalizing missing signals so it never crashes on sparse metadata.
pick_canonical_release() takes candidates in source-priority order, picks the
best fit, breaks ties toward the earlier (higher-priority) candidate so the
choice is DETERMINISTIC — that determinism is what makes every tool agree
(#765), while count/duration fit picks the right EDITION (#767-Bug2). A
confidence floor (default 0.5) means a low-confidence guess is never pinned.
- database/music_database.py — additive, nullable columns on albums
(canonical_source / canonical_album_id / canonical_score /
canonical_resolved_at), guarded by the existing PRAGMA-table_info pattern.
NULL = unresolved = every consumer falls back to today's behavior.
Tests: tests/test_canonical_version.py (11) — edition discrimination (11 files
-> standard, 17 -> deluxe), deterministic priority tiebreak, duration
disambiguation on count ties, graceful degradation (no durations / counts only /
fuzzy titles), confidence floor, empty-input safety. tests/test_canonical_
columns_migration.py (4) — fresh DB has the columns, they're nullable w/ NULL
default, migration is idempotent, and it ALTERs them onto an old albums table.
60 DB/schema regression tests still pass.
A source row with no art of its own (e.g. a YouTube source, which provides
none at mirror time) now borrows the cover from its MATCHED server track, so
both sides of the sync editor show an image.
The endpoint already had a borrow fallback (_server_art_map), but it matched by
an exact normalized "{artist}|{title}" key — so a YouTube-shaped row like
"Arctic Monkeys - Do I Wanna Know?" never matched the library's "Do I Wanna
Know?" and stayed blank even though the server had the cover. This borrow is
keyed off the ACTUAL source<->server pairing the reconcile already computed, so
it works for those rows once #768's canonical matching pairs them.
Done in the pure reconcile_playlist (final pass), so no frontend change is
needed — the editor already renders source_track.image_url. Guarded so it only
fills an EMPTY source image (Spotify/CDN art is never overwritten) and only when
the matched server track actually has a thumb.
Composes with the rest: #766 made the server cover URL work, #768 made the
YouTube row match, this makes the matched source row borrow that cover — so an
artless YouTube row matched to a Navidrome track with art shows on both sides.
Tests: tests/test_playlist_reconcile.py (+4) — artless source borrows the
matched cover; source with its own art keeps it; unmatched source has nothing to
borrow; borrow skipped when the server track has no thumb. 15 reconcile + 59
sync/navidrome tests pass.
The sync editor renders server covers as <img src="/api/navidrome/cover/{id}">,
but no Flask route ever served that path — so every Navidrome cover 404'd, on
every album, art or not. The source (left) side then went blank too: a source
row with no native art (e.g. YouTube, which provides none at mirror time) falls
back to borrowing the matched server track's cover — i.e. that same dead route.
So both sides collapsed to nothing.
Fix:
- New NavidromeClient.build_cover_art_url(cover_id) — builds the absolute,
Subsonic-authenticated getCoverArt URL (base_url + token/salt), keeping
credentials server-side. Uses a FIXED cover-art salt so the URL is
deterministic for a given (server, password, cover_id): a rotating salt (as
in _generate_auth_params) would make every request a unique URL → image-cache
miss every time + a dead, never-reused cache row per fetch. Token auth doesn't
require a unique salt, and the password is never exposed (only its salted md5).
- New route /api/navidrome/cover/<cover_id> — resolves that URL and streams the
image through the shared image cache (same pattern as /api/image-proxy), with
a private max-age so the browser caches by the stable route URL.
Effect: server side works for any album that has art in Navidrome; matched
source rows with no native art now borrow the (now-working) server cover.
Unmatched YouTube rows stay blank — no image exists anywhere to show.
Tests: tests/test_navidrome_cover_url.py (8) — URL structure + salted-token auth
(never the raw password), determinism (same id -> same URL so the cache hits;
different id/password -> different URL), optional size, and the not-connected /
no-id / no-credentials guards.
Caveats: not executed against a live Navidrome (no server in CI) — the URL
builder is unit-tested; the route's cache→HTTP→bytes round-trip is read-verified
only. Scope is the sync editor's Navidrome route; Plex/Jellyfin server-cover
branches and any other modals using a different mechanism are untouched.
The reorganize preview (dry run) was physically creating destination album
folders, littering the library with empty dirs and making "changes" before the
user ever hit Apply.
Cause: preview_album_reorganize calls build_final_path_for_track purely to
COMPUTE the destination path string — but that shared helper has 9 os.makedirs
side effects (it's also the live download/import path builder, where creating
the dir is correct). So computing the preview path created "Lenka (Expanded
Edition)/" on disk.
Fix: build_final_path_for_track gains create_dirs=True; all 9 makedirs now route
through a gated helper. The reorganize PREVIEW passes create_dirs=False, so a
dry run computes the exact destination path with zero filesystem side effects.
Everything else keeps the default True:
- the download/import post-process flow (still writes files into the dir),
- retag,
- the reorganize APPLY path — verified it goes through post_process_fn (the real
pipeline → build_final_path_for_track with create_dirs=True), so live moves
still create their destination dirs. The gate only silences the dry run.
Tests: tests/imports/test_import_paths.py — create_dirs=False computes the
correct path (matching the reported "01 - The Show.flac") but writes NOTHING to
disk (not even the Transfer root); create_dirs=True still creates folders; both
yield an identical path. Updated two reorganize-orchestrator test doubles to
accept the new kwarg. 148 reorganize/paths/retag/pipeline tests pass.
Does NOT fix the second half of #767 (Expanded Edition picked over the standard
album). That is NOT a reorganizer bug: the library album row was linked to the
deluxe release at enrichment time (its stored spotify_album_id/itunes_album_id/
deezer_id points at "Lenka (Expanded Edition)"), and the reorganizer faithfully
reorganizes to whatever the album is linked to. The real fix is in album
enrichment's edition preference — tracked separately.
Three compounding bugs hit tracks whose source metadata is YouTube/streaming-
shaped — title "Artist - Song", artist "Official Artist"/"Artist - Topic"/
"ArtistVEVO" (reported: "Arctic Monkeys - Do I Wanna Know?" by "Official Arctic
Monkeys"). Server-agnostic — affects Plex/Jellyfin/Navidrome, not just the
reporter's Navidrome.
Bug A — the match fails. The confidence scorer and the editor's reconcile both
compared the raw "Artist - Song" title against the library's clean "Song"; the
length-ratio penalty + floor drove it to ~0.18 (NO-MATCH), so the track showed
unmatched while its server copy showed as an orphan "extra". New pure
core/text/source_title.py (clean_source_artist / strip_artist_prefix /
canonical_source_track) strips the channel/video decoration, applied at BOTH
matching seams: services/sync_service._find_track_in_media_server (tries raw
then canonical, keeps the best) and the editor reconcile. Conservative: a title
prefix is stripped only when it equals the artist, so "Self-Titled", "Jay-Z",
and "Marvin Gaye" (by another artist) are untouched, and the canonical form is
an additional best-of candidate so it can only help.
Bug B — manual matches never persisted. get_server_playlist_tracks built the
per-source entry WITHOUT source_track_id, so "Find & add" posted an empty id
and _persist_find_and_add_match returned early. The match reverted to "extra"
on reload and re-adding looped. The editor's 3-pass matcher is now lifted to a
pure, tested core.sync.playlist_reconcile.reconcile_playlist that includes
source_track_id (the frontend at pages-extra.js:1836 already reads + sends it).
Bug C — manual match duplicated + delete wiped all copies. "Find & add" always
inserted, so linking a source to an already-present server track appended a
duplicate (pos 72, 73...); remove filtered out EVERY entry with the target id.
New pure core.sync.playlist_edit (plan_playlist_add: link-don't-duplicate when
the target is already present; remove_one_occurrence: drop a single copy) wired
into the Plex/Jellyfin/Navidrome add + remove branches.
Tests (extreme): tests/test_source_title.py (35), tests/test_playlist_reconcile.py
(11 — incl. the reported case, parity for override/exact/fuzzy/extra, and
duplicate-server handling), tests/test_playlist_edit.py (12). 286 matching/sync
tests still pass.
Caveats: the sync_service change and the add/remove/editor endpoints are
read-verified, not executed against a live media server (none in CI). The pure
cores they call are exhaustively unit-tested; output-shape parity of the
reconcile lift is covered. Delete removes the first matching copy (duplicates
are identical, so harmless).
Tracks NOT in the library were matched to a DIFFERENT song by the SAME artist
and reported with high confidence instead of as missing — e.g. "Dani
California" -> "Californication" (Red Hot Chili Peppers), "Under The Bridge"
-> "Around the World".
Root cause: _calculate_track_confidence scores 0.5*title + 0.5*artist. A
same-artist comparison always yields artist = 1.0, so the title score is the
only thing that can tell two of an artist's songs apart — but that score is a
SequenceMatcher CHARACTER ratio, which over-credits unrelated titles that
share a long substring ("californi…" = 0.67) or just a stopword ("the" =
0.62). With the flat 0.5 artist term, anything clearing the weak 0.6 char
floor lands at ~0.81-0.83, well over the 0.7 sync threshold. Reproduced on
dev: both reported pairs score 0.81/0.83.
Fix: new core/text/title_match.py:titles_plausibly_same, called in
_calculate_track_confidence right before the floor. It accepts a pair only
when it's near-identical char-wise (>=0.85, so typos / punctuation / casing
like "Beleive"->"Believe", "HUMBLE."->"Humble" still match) OR the titles
share at least one significant (non-stopword) word. Two different songs by the
same artist share no content word, so they're rejected and the real track is
correctly reported missing. ("the" is a stopword — that's what leaked "Under
The Bridge"/"Around the World".)
Scoped deliberately: the word-overlap test fires ONLY when at least one side
has 2+ content words. For single-word titles there is no other word to share,
so it defers to the existing char floor — otherwise legitimate stylized
spellings ("Grey"/"Gray", "Tonite"/"Tonight", "4ever"/"Forever") would become
new false-negatives. Verified those still match. The few single-word variants
that do score low (Ok/Okay, Thru/Through) were already rejected by the
pre-existing length-ratio penalty, not by this gate.
Both reported false positives now score 0.33/0.31 -> missing. Does NOT address
the harder case of two different same-artist songs that DO share a content
word (e.g. "Believe"/"Believer") — pre-existing and unworsened. Any residual
error fails safe: a false-missing is re-downloaded/wishlisted, vs the old
behavior which silently substituted the wrong song.
Tests: tests/test_title_match_guard.py (14) — pure-guard unit tests + a
13-pair battery driving the REAL _calculate_track_confidence (genuine matches
stay >=0.7, same-artist different songs drop below), plus an explicit
no-regression test for stylized single-word spellings. 292 matching/sync tests
pass.
The manual-import routes (album + singles) call post_process_matched_download
directly. When the pipeline quarantines a file — integrity / AcoustID / FLAC
bit-depth — or hits the race guard, it sets a context flag and RETURNS
NORMALLY (it only marks the task failed + notifies when there's a task_id,
which manual imports don't have). So the inner pipeline raised no exception,
and routes.py counted `processed += 1` for a file that had just been moved to
ss_quarantine, not the library. Result: the UI shows a green "Done" while the
track silently vanished — exactly the #764 report (Coldplay - Yellow.flac ->
ss_quarantine, but "Done").
The download path already handles this in
post_process_matched_download_with_verification (it reads the same flags and
marks the task failed); only the manual-import routes were missing the check.
Fix: new pure helper import_rejection_reason(context) returns a human-readable
reason for any terminal rejection (_integrity_failure_msg / _acoustid_quarantined
/ _bitdepth_rejected / _race_guard_failed) or None for a clean import. Both
manual-import routes now consult it: album_process reports the track in
`errors` instead of counting it processed; process_single_import_file returns
("error", reason) instead of ("ok", ...). Verified every move_to_quarantine
call site (4, all in pipeline.py) sets one of those flags, so no quarantine
path slips through. This also delivers the "direct display of the error" the
reporter asked for — the reason now surfaces in the response `errors` list.
Does NOT address the reverse symptom ("failed even though it moved correctly")
— not yet root-caused — nor the separate bit-depth hole on the download-path
wrapper.
Tests: tests/imports/test_import_rejection_reason.py (10) — each trigger
detected, falsy flags ignored, deterministic ordering, plus two route-level
tests driving the REAL process_single_import_file (quarantine -> "error";
clean -> "ok").
enhance_file_metadata rebuilds tags from scratch: for FLAC it calls
clear_pictures(), for MP3/MP4 it clears the whole tag block — and it does
this UP FRONT, then saves the file, long before it tries to fetch and embed
the replacement art. So every way the re-embed could come up empty left the
file saved with the original art destroyed and nothing put back:
- extract_source_metadata returns nothing -> early save, no embed
- no album-art URL / art download fails / rejected by the min-size guard
-> embed_album_art_metadata returns early without adding a picture
- art embedding disabled in config -> embed skipped entirely
- embed raises mid-enrichment -> file left cleared on disk
This is the "cover art gets corrupted/destroyed during import" half of #764
(continuation of #755); distinct from #750's truncated-cache DISPLAY bug.
Fix: new core/metadata/art_preservation.py snapshots the existing art
(the live Picture / APIC / MP4Cover objects, so they re-apply verbatim)
BEFORE the clear, and restores it before each save IFF the file currently
has none. Wired into all three exit paths in enhance_file_metadata
(no-metadata early return, the final save, and the except handler). The
restore is a strict no-op when art is already present, so the happy path —
new art embedded — is byte-for-byte unchanged: it never clobbers or
duplicates a freshly-embedded cover. embed_album_art_metadata now returns a
bool so the intent (embedded / didn't) is explicit.
Tests:
- tests/test_art_preservation.py (5) — snapshot/restore round-trips through
real mutagen FLAC + ID3 objects; restore no-ops when new art is present.
- tests/test_enrichment_art_preservation.py (4) — runs the REAL
enhance_file_metadata over a real FLAC with embedded art and asserts the
art survives on disk for missing-metadata / failed-embed / embed-raises,
and is correctly REPLACED (exactly one picture, new bytes) on success.
1019 tests pass across the metadata/enrichment/imports/acoustid suites.
The DB-update + deep-scan automation monitor used a hard 2-hour TOTAL cap
(while elapsed < 7200). It tracked progress but only used it to print a stall
warning — the only thing that actually timed out was wall-clock. So a large
library that scans for >2h while progressing fine (reported: 4781 artists) trips
the cap and the automation card flips to 'error: timed out after 2 hours' even
though the scan thread is healthy and still running (the timeout never cancels
it, which is why it keeps progressing in the logs after the 'error').
Time out on STALL, not total runtime:
- 30 min with NO progress -> error ('stalled'); catches a genuinely hung scan.
- 10 min idle -> warning (repeats); unchanged heads-up.
- 24h absolute backstop, purely a runaway-loop guard.
- An actively-progressing scan keeps resetting the idle clock, so it never
times out no matter how many hours the whole library takes.
- Progress is judged on (processed, progress, current_item) so a slow stretch
where the rounded % holds steady (but the artist keeps changing) isn't a
false stall.
The decision is extracted into a pure, testable scan_wait_action(); both the
deep-scan and full-refresh handlers share the monitor loop, so both are fixed.
Tests: tests/test_scan_wait_action.py (9) — headline regression (5h/12h total
but progressing -> 'continue', not timeout), finished/stall-warn/stall-timeout/
abs-cap thresholds, and ordering. 280 automation tests still pass.
Defense-in-depth follow-up to #760. Even with the entrypoint chown fix, if the
album-bundle staging dir ever can't be created/written (permissions, read-only
mount, disk full), the dispatch caught the plugin exception and marked the whole
batch failed — even though the album had already downloaded (the #715 symptom:
'release finishes downloading but the batch fails').
Now an OSError from the plugin is flagged fallback-eligible, so the dispatch
returns to the per-track flow instead of hard-failing. OSError covers the
staging/filesystem failure that motivated this (#760's PermissionError) and, by
Python's IOError==OSError aliasing, any propagated transient I/O error —
falling back is never worse than hard-failing, and per-track is the universal
graceful path. Programming errors (TypeError, KeyError, RuntimeError, …) are
NOT OSError and stay terminal, so genuine bugs still fail loudly — the existing
'plugin exception => failure' contract and its test are preserved.
Test: new test_dispatch_staging_oserror_falls_back_to_per_track (PermissionError
on the staging dir -> result False, phase 'analysis', not failed). Existing
RuntimeError-is-terminal test still passes. 131 album-bundle/plugin tests green.
After an update, installs became unusable: the Amazon enrichment worker runs by
default, the default public T2Tunes proxy (t2tunes.site) was returning
503 'Amazon Music API is not initialized', and the worker treated every album
as an individual error -- logging an ERROR per item, churning network + DB
continuously across the whole library, and marking every row 'error' (a state
the retry tiers never re-attempt, so even after the proxy recovered nothing
re-enriched). The reporter couldn't reach the UI to turn it off.
Two-part fix:
1. Source-outage circuit breaker (core/amazon_outage.py, pure + tested):
- is_source_outage(exc) distinguishes a whole-source outage (HTTP 5xx,
'not initialized', connection failure, non-JSON error page) from a real
per-item miss (404, transient 400, etc.).
- On an outage the worker now leaves the item UNTOUCHED (so it's retried once
the proxy recovers instead of being permanently burned to 'error'), logs
ONCE per streak, and backs off with next_poll_delay_seconds() -- escalating
30s -> 60s -> ... capped at 30 min -- instead of grinding every 2s. It
auto-resumes the normal cadence the moment the source answers (success OR a
non-outage error both clear the streak).
- AmazonClientError now carries status_code so detection doesn't rely on
message parsing.
2. Opt-in by default (web_server.py): amazon_enrichment_paused now defaults to
True. Because enrichment depends on an external public proxy that can be
down, it stays paused unless the user explicitly enables it -- a proxy outage
can no longer take down installs that never opted in. (Behaviour change:
anyone on the old auto-on default is now paused; re-enable in Settings.)
Together: on update the worker is paused -> no flood -> UI accessible; opted-in
users are protected from future outages by the breaker.
Tests: tests/test_amazon_outage.py (12) pin the classifier across every error
surface (incl. the exact 503 'not initialized' case) and the back-off schedule
(monotonic, capped). 157 Amazon tests pass; lint clean.
Note: could not reproduce the exact 'UI fully unreachable' mechanism remotely
(WAL + 8 gthreads shouldn't hard-lock); the fix removes the flood/churn that is
the practical cause and defaults the feature off.
The Duplicate Detector's 'Keep Best' auto-selection ranked copies by highest
bitrate -> duration -> track number, with no notion of format. A FLAC whose
bitrate the library scan never populated (a common gap) therefore lost to a
282 kbps MP3: 282 > 0, so the MP3 was kept and the FLAC deleted (reported on
Havok 'Prepare For Attack', and again on Kendrick GNX).
Fix: rank by format/lossless tier FIRST, then bitrate, duration, track number.
A lossless file now always beats a lossy one regardless of the recorded
bitrate; bitrate/duration/track# only break ties within the same format.
- core/library/duplicate_keep.py (new): pure, importable pick_duplicate_to_keep
+ duplicate_keep_sort_key + format_rank_for_path (extension rank mirroring
auto_import_worker._quality_rank: flac=10 ... mp3=5 ... unknown=1).
- core/repair_worker.py: _fix_duplicates auto-pick now calls
pick_duplicate_to_keep instead of the bitrate-first max().
- webui/static/enrichment.js: the KEEP/REMOVE recommendation mirrors the same
format-first ranking so the badge matches what the backend will delete.
Parity: Python uses '.ext' keys (os.path.splitext), JS uses 'ext'
(split('.').pop()) -> identical results; both keep the first copy on a full
tie. Verified the only other dedup path (the standalone Duplicate Cleaner
automation, core/library/duplicate_cleaner.py) was already format-priority-first
and correct -- no change needed there.
Tests: tests/test_duplicate_keep.py (11 -- incl. the exact FLAC-with-missing-
bitrate vs 282 kbps MP3 case, format ranking, within-format tie-breakers, and
edge cases). 147 repair/duplicate tests still pass.
Note: why FLAC bitrate is NULL in the DB is a separate library-scan gap;
format-first ranking makes the keep decision correct regardless.
Follow-up to the preferred-art feature. Real test runs showed a source could
win on priority while handing back a small cover: Cover Art Archive is
volunteer-uploaded with no size floor, so CAA-first gave a 599x531 (Taylor
Swift) and a 600x600 (Kendrick GNX) -- front-1200 only caps the max, so a
~600px upload stays ~600px -- and Deezer/iTunes lower in the order never got a
turn.
Fix:
- Minimum-resolution guard: artwork._min_size_art_validator builds the
resolver's validate hook -- it fetches each candidate, caches the bytes (so
the winner isn't fetched twice), and accepts art only when its shortest side
>= metadata_enhancement.min_art_size (default 1000px; 0 disables). Art that's
too small is a miss, so the resolver falls through to the next source instead
of winning on priority. Unmeasurable images are accepted (don't over-reject;
fallback is still today's art). Wired into both embed_album_art_metadata and
download_cover_art.
- iTunes art upgraded to /3000x3000bb/ (was the 600px default) so it
contributes high-res when it wins.
- select_preferred_art_url gains a validate passthrough to the resolver.
- config default metadata_enhancement.min_art_size: 1000.
Effect: with an order like caa > deezer > spotify > itunes, a ~600px CAA upload
is now skipped and Deezer's ~1900px wins -- consistent big art. (Spotify art
often maxes ~640px, so it's skipped at the 1000 floor in favor of bigger
sources; lower min_art_size to ~640 to allow it.)
Tests: tests/metadata/test_art_min_size.py (6 -- incl. the real 599x531 and
600x600 cases, shortest-side logic, unmeasurable-accept, no-bytes-reject,
0-disables) + iTunes max-res upgrade test. Full metadata suite green (617).
Lets users pick which providers' cover art to use and in what priority,
generalizing the single prefer_caa_art toggle into an ordered, mix-and-match
list (Sokhi's request). Fully opt-in: default album_art_order is [], so every
existing install is byte-for-byte unchanged until the user enables sources.
How it works:
- Per album, walk the user's ordered sources top-to-bottom; the first source
that actually has THIS album's cover wins. A miss falls through to the next;
if all miss, the download's own art is kept (today's default). The worst case
is always exactly the cover you'd get today -- never wrong art, never an
error into the download.
- Connection-gated: a source is only tried when the user is connected to it
(free sources CAA/Deezer/iTunes/AudioDB always; Spotify only when
authenticated). Tidal/Qobuz/HiFi deferred (cover-URL construction + no clean
core accessor -- not shipping unverified extraction).
- Album-match validated: a source's art is used only when the album it returns
matches the requested artist+album (significant-token subset, tolerant of
Deluxe/Remastered/articles/feat./multi-artist). A loose top search hit for a
different record is treated as a miss -> guarantees no wrong-album art.
- The list supersedes the legacy prefer_caa_art toggle: when album_art_order is
non-empty it is the sole authority (add 'caa' to the list to use Cover Art
Archive), and prefer_caa_art is neutralized for both the embedded-tag art and
cover.jpg paths. With an empty list, prefer_caa_art behaves exactly as before.
Implementation:
- core/metadata/art_sources.py: pure resolver -- effective_art_order (config +
legacy back-compat) and resolve_cover_art (ordered walk + fallback,
exception-safe per source). No network/config/DB; fully unit-testable.
- core/metadata/art_lookup.py: availability gating, per-source lookups against
existing clients (Deezer/iTunes/AudioDB/Spotify search + CAA via MBID),
album-match validation, per-album caching, and select_preferred_art_url --
the single gate the pipeline calls (no-op unless an explicit list is set).
- core/metadata/artwork.py: wired into embed_album_art_metadata and
download_cover_art, gated so no configured list == current behavior.
- web_server.py: GET /api/metadata/art-sources (connected sources only).
- config/settings.py: default album_art_order: [].
- webui (index.html + settings.js): reorderable list in Core Features reusing
the hybrid-source-list pattern + real service logos (with emoji fallback);
load/save wired through the existing metadata_enhancement settings flow.
loadArtSourceOrder populates the saved order synchronously (filtered to known
sources, not availability) so a save before the availability fetch resolves,
or a temporarily-disconnected source, can never wipe the saved order.
Tests: 40 unit/seam tests (resolver ordering/fallback/back-compat, availability,
per-source extraction, album-match validation incl. wrong-album/wrong-artist
rejection, caching, exception-safety, the off-by-default gate). Full metadata
suite still green (610 passed) -- the gated integration changes nothing when no
list is configured.
Note: the settings UI (DOM-heavy, not unit-testable in the JS harness) and the
live per-source art-fetch quality are validated by manual testing.
A torrent-first (or usenet-first) hybrid download would freeze at
"Torrent searching for release 0%" and never move to the next source when
Prowlarr returned no results for the album. Reported by Cezar:
[Album Bundle] torrent flow failed for '...': No torrent results found
Cause: the album-bundle dispatch only returns to the per-track flow (which,
in hybrid mode, tries the next configured source) when the plugin's failure
outcome carries fallback=True; otherwise it marks the batch failed and stops.
Both plugins set fallback=True on their 'results found but none matched the
album' branch, but the adjacent 'no results at all' branch set only an error
and no fallback flag -- so zero results hard-failed while wrong results fell
back. Backwards, and soulseek's plugin already defaults fallback=True for
exactly this reason.
Fix: set fallback=True on the no-results branch in torrent.py and usenet.py.
The dispatch's fallback handling (return False -> per-track flow) was already
correct and is unchanged.
The only consumer of download_album_to_staging is the dispatch, which reads
the result via .get('fallback'), so the change is additive and locally
contained.
Tests: new test_torrent_album_to_staging_no_results_flags_fallback and
test_usenet_album_to_staging_no_results_flags_fallback assert the plugins now
emit fallback=True on an empty search; the existing torrent no-results test is
extended with the same assertion. Existing dispatch tests already pin
fallback=True -> per-track flow. Full downloads/plugins/adapters sweep: 690
passed.
When no media server is connected, discovery/sync patches sync_service's
matcher with a database-only implementation. sync_service calls it as
_find_track_in_media_server(track, candidate_pool=...) (a per-artist candidate
cache), but the database-only override took only (spotify_track) — so every
sync raised 'database_only_find_track() got an unexpected keyword argument
candidate_pool' and aborted.
Lift the override from a nested closure to a module-level
_database_only_find_track(spotify_track, candidate_pool=None) so it (a) accepts
the kwarg for interface parity with the real matcher and (b) is importable and
unit-testable. The DB-only path queries the library directly via
check_track_exists, so it accepts but doesn't need the candidate cache. Also
dropped the dead original_find_track local.
Tests: signature includes candidate_pool; called with candidate_pool={} returns
(None, 0.0) on no match; returns the match when the DB has it.
New GET /api/downloads/task/<id>/detail merges the live download task with its
library_history row (the data the Download History cards show) into one payload
the upcoming track-detail modal renders: status kind, title/artist/album,
source, quality, final location, AcoustID verdict, and expected-vs-downloaded.
Assembly + status classification live in core/downloads/track_detail.py as a
pure, importable build_track_detail()/classify_status_kind() (9 unit tests);
the endpoint is thin glue that looks up the matching history row by track.
An invalid API key (and rate limits / missing chromaprint / fingerprint
failures) all collapsed into the same None as a genuine no-match, so:
- every download showed a benign 'AcoustID: Skipped', and
- the 'Test API key' button reported a dead key as VALID
because test_api_key trusted 'no exception = valid' but fingerprint_and_lookup
swallows the error and returns None. A broken AcoustID setup looked completely
normal — which cost a real debugging session to untangle.
- New AcoustIDClient.lookup_with_status() returns a structured result that
distinguishes ok / no_match / error / no_backend / fingerprint_error /
unavailable. fingerprint_and_lookup() stays dict-or-None (library scanner /
auto-import / their tests unchanged) as a thin wrapper over it.
- verify_audio_file() uses it: a real error -> new VerificationResult.ERROR
(-> _acoustid_result='error' -> the existing red 'Error' history badge),
a genuine no-match -> SKIP 'No match in AcoustID database'. ERROR never
quarantines (an outage/bad key must not punish good files).
- test_api_key() now validates via the authoritative direct API call (error
code 4 = invalid key) instead of the swallowed-exception path.
Tests: structured-status distinction, legacy-wrapper contract, verify ERROR vs
SKIP, and test_api_key invalid-vs-accepted. Existing verify tests updated to
stub lookup_with_status (a stub returning just recordings is inferred as ok).
Clicking a quarantined track's status used to open the generic search modal,
identical to a plain failure — no way to review or recover the file. It now
opens a chooser:
- Listen: streams the file in-app via a new /api/quarantine/<id>/stream
endpoint (range-supported; the real audio Content-Type is recovered from the
sidecar since the on-disk file ends in .quarantined).
- Accept & Import: existing /approve (restore + re-import, gates bypassed).
- Search for a different result: the existing candidates modal (old behavior).
Non-quarantine failures (not_found / failed / cancelled) are unchanged — a
single click listener routes by dataset set at render time, so a task that
fails then later quarantines can't end up double-bound.
Also fixes the Accept failure on Windows: the Listen stream holds an open file
handle, so the subsequent restore move hit WinError 32 ('file in use') and the
endpoint mislabeled it 'thin sidecar'. Accept now releases the audio handle
before approving, and approve/recover moves retry briefly on transient OS locks
(_move_with_retry). Accept also auto-falls-back to Recover-to-Staging for
genuinely thin/orphaned sidecars.
Tests: stream-info resolution (sidecar + filename-fallback + missing), and
_move_with_retry success/give-up.
post_process_matched_download_with_verification pops task_id/batch_id out of
the context before running the inner pipeline (so the inner doesn't fire its
own task notifications). But _mark_task_quarantined runs inside that inner call
and reads context['task_id'] — which is now None — so it silently no-op'd.
Result: every download through this wrapper (album-bundle / staging path)
quarantined WITHOUT recording quarantine_entry_id on the task, so the UI had no
handle to manage the file (the status click just fell back to the search modal).
_mark_task_quarantined now also stashes the entry id on the context (survives
the pop), and the wrapper applies it to the real task in both quarantine
branches (integrity + AcoustID). Direct (non-wrapper) callers are unchanged.
Tests: unit coverage for the stash-with/without-task_id behavior, plus a
wrapper-level test proving the entry id reaches the task on integrity quarantine.
HiFi assembles FLAC from HLS segments and demuxes with `ffmpeg -c copy`,
which preserves total_samples=0 in the STREAMINFO of Tidal's fragmented/
streamed FLAC. Every audio frame is present and the file plays fine, but
mutagen computes length = total_samples / sample_rate = 0, so the integrity
check rejected it as 'zero-length audio' and quarantined nearly every HiFi
download. Users confirmed the quarantined files play normally once restored.
Length 0 is not proof of corruption at that point: the file already passed
the size gate, was identified as a real audio format, and has a valid info
block — a genuinely empty/truncated/stub file fails one of those earlier
checks instead. Treat length 0 as 'length unknown': accept the file and skip
the duration cross-check we can't perform without a length. mutagen never
decoded/validated frame data anyway, so this doesn't weaken real corruption
detection — size, parse, format, info-block, and duration-drift guards all
remain.
Tests: a large valid-parse length-0 file (streamed-FLAC signature) is now
accepted; a tiny length-0 stub still fails (size gate fires first).
A Soulseek album bundle stages whichever single folder scored best. If that
folder doesn't contain every track the album needs, the missing tracks were
marked not_found with no fallback — even in hybrid mode where later sources
(Deezer, YouTube, etc.) could fill them. The staging-miss short-circuit fired
for Soulseek because 'soulseek' was lumped into the torrent/usenet source set
when album bundles were added, and album_bundle_partial only reflects whether
the files found IN the folder downloaded, not whether the folder had every
needed track.
Drop 'soulseek' from the short-circuit (keep torrent/usenet). A track not
claimed from the staged Soulseek folder now falls through to the normal
per-track Soulseek search and, in hybrid mode, onward down the configured
chain. Unlike torrent/usenet — where per-track search re-adds the same
release — Soulseek per-track search is a genuine per-file network search, so
this is correct and cheap. Realizes the original author's stated intent
('keep partial bundles from blocking per-track fallback') robustly, since the
partial flag couldn't detect a folder that was simply missing tracks.
Only affects tracks NOT claimed from staging — fully-staged albums claim every
track via try_staging_match and never reach this gate, so working albums are
unchanged. Likely also mitigates #755 (all-album-import failures now fall
through to per-track instead of dying).
Tests: rewrote the two Soulseek staged-miss tests to assert fall-through
(single + hybrid-first); kept the torrent guard; added a usenet guard test.
The user-facing Search-for-Match / Fix popup runs non-strict MB searches.
That path built a bare "track artist" query with no field scoping, so the
artist was just a free fuzzy term — covers and karaoke whose TITLES contained
the artist name outranked the canonical recording. Reproduced live: searching
"Say You Will" / Foreigner returned cover artists with Foreigner absent, and
"Sweet Child O Mine" / Guns N Roses returned only covers (Presnyakov, PMJ…),
never the Guns N' Roses original.
Keep the track/album side loose (no phrase quotes → diacritic + bracketed-
suffix recall, the reason non-strict exists) but field-scope the artist as
artist:(...) so it constrains. The artist value is Lucene-escaped via
_escape_lucene() — without it, names like "Sunn O)))" or "Anthony Green
(Saosin)" would close the artist:( group early (returning unrelated artists)
or break the query (zero results). Same fix applied to search_release.
Verified against the live MB API: both reporter queries now return the real
artist top-to-bottom; diacritic recall is preserved (artist:(Bjork) folds to
Björk); and paren/?/!-laden artist names produce valid, balanced queries.
Tests pin the constructed query string (no network): non-strict scopes and
escapes the artist while keeping the track loose/unquoted; strict path
unchanged; plus _escape_lucene unit coverage.
The save endpoint coerced library_track_id with int(), which rejected
every non-numeric id with "Invalid library track id". Library ids are
str(ratingKey) — numeric for Plex but GUIDs/hashes for Navidrome,
Jellyfin, and other Subsonic servers — and are stored in the TEXT
tracks.id column, so the coercion broke manual matching on every
non-Plex server.
Replace the int() coercion with a normalize_library_track_id() helper
that trims and rejects only empty input, passing the opaque string id
straight through. Plex numeric ids are unaffected (SQLite INTEGER
affinity still stores a clean numeric string as an int, so existing
matches are byte-identical) and no schema migration is needed (the
INTEGER column already stores non-numeric ids as text).
Tests: pure-helper cases (numeric/GUID/whitespace/empty) plus a real-DB
round-trip proving a GUID id saves, reads back unchanged, and enriches.
Investigation (not assumption): the cache's TTL eviction + junk cleanup ARE
correct and DO run automatically every 6h (CacheEvictorJob, auto_fix=True).
The real gap is there's NO SIZE CEILING — TTL-only eviction means 'how big can
it get' = 'however much you fetch within the 30-day window', so heavy
discovery/enrichment legitimately grew metadata_cache_entities to ~1.8M rows /
7.6 GB, bloating the main DB (a factor in the corruption incident).
Fix — add a bounded LRU cap:
- entities_to_evict_for_capacity(total, max_rows): pure decision fn (cap<=0
disables), unit-testable like core.db_integrity.prune_backups.
- MetadataCache.evict_over_capacity(): deletes the least-recently-ACCESSED rows
(uses the already-stored last_accessed_at; NULL = never-touched = evicted
first) down to the ceiling. Default 250k rows, tunable.
- Wired as Phase 5 of CacheEvictorJob — runs LAST, after TTL/junk/orphan/null
cleanup, so it only trims a still-oversized HEALTHY cache.
Verified safe to bound/wipe: audited every cache reader (get_entity/
get_entities_batch/get_search_results/get_entity_detail/browse) — all degrade
to None/[]/empty on miss, treated as 'go fetch'. Nothing depends on a row
existing, so eviction can't break callers.
Tests: tests/metadata/test_cache_capacity_eviction.py (8) — pure-fn coverage +
real temp-DB proof that it drops the LRU rows specifically (not arbitrary) and
NULL-access rows go first. 18 adjacent cache tests still green; ruff clean.
Follow-ups (separate phases, scoped): (2) move the cache to its own bounded
metadata_cache.sqlite3 (no JOINs to library tables — confirmed clean to split;
invalidate-and-rebuild rather than migrate the 7.6GB), (3) kill the
raw_json + 22-extracted-column double storage.
Post-incident hardening. A WAL-mode DB corrupted (most likely an interrupted
write during a hard restart), and the backup routine made it unrecoverable:
it (a) never checked integrity, so src.backup() faithfully copied the corrupt
pages into every rolling backup, and (b) pruned oldest-by-mtime, so each new
corrupt backup evicted the last good one. Result: all snapshots poisoned.
New core/db_integrity.py (pure, unit-tested):
- quick_check()/is_healthy(): fast read-only PRAGMA quick_check probe.
- safe_backup(): verifies the SOURCE is healthy BEFORE the Online-Backup copy
and the RESULT after; refuses + discards rather than save a corrupt copy.
- prune_backups(): rotation that NEVER deletes the most-recent verified-healthy
backup, even to honor max_keep — so a run of bad backups can't drop your last
good snapshot.
Wired into BOTH backup paths (the /api/database/backup endpoint and the
auto_backup_database automation handler) — they now refuse on integrity failure
(409 / error status, existing backups untouched) and prune safely.
Tests: tests/test_db_integrity.py (8) using REAL temp DBs incl. a physically
corrupted one — proves refuse-corrupt-source, discard-corrupt-result, and the
exact incident scenario (newest backups corrupt -> the older healthy one is
protected from pruning). Existing maintenance-handler backup test still green
(29 passed). compile + ruff clean.
NOTE: this prevents silent backup poisoning; it does NOT stop the underlying
corruption. Follow-ups still worth doing: WAL-checkpoint on clean shutdown +
a periodic live-DB integrity alert (so corruption is caught on day 1).
Compared my #730 fix against contributor PR #731 (same independent design).
Grafted their good idea — a confidence bonus when the album's full core phrase
appears intact in the release title (rescues long multi-word names whose token
coverage gets diluted) — and kept my accent-folding, which #731 lacks (their
normalize drops accented chars: Bjork -> 'bj rk').
IMPORTANT: implemented the phrase bonus WORD-BOUNDARY anchored, not as a raw
substring. My first cut used 'phrase in norm_title' (matching #731) and it
immediately reintroduced the substring bug #730 exists to fix — 'heroes'
matched 'superheroes' and the wrong album scored 0.9/passed. PR #731 has this
latent flaw. The regex anchors the phrase to word boundaries so the bonus
fires for real matches only.
Verified: substring trap (Superheroes/Scary Monsters) rejected; edition
suffixes + intact-phrase albums kept. +1 phrase-bonus test (incl. the
word-boundary guard). 126 plugin tests pass; ruff clean.
Co-authored-by: Tyler Richardson-LaPlume <170156756+IamGroot60@users.noreply.github.com>
Review caught that artist_name was added to pick_best_album_release's signature
and threaded through both call sites but never actually used — dead, misleading
code. Removed it from the helper + both callers. Artist-aware gating would be a
deliberate future feature (titles carry the artist inconsistently, so a hard
artist gate would risk the same false-negative class I just fixed); the album
relevance gate already resolves the reported wrong-release bug. No behavior
change. 127 plugin tests pass; compile + ruff clean.
Self-review found a false-negative in the title-relevance gate I just added:
it scored 'fraction of the ALBUM-NAME's words present in the title', so a
stored album name with an edition/remaster suffix the torrent lacks
('Currents (Deluxe)', 'Heroes (2017 Remaster)') scored BELOW the 0.6 floor and
the correct release was wrongly refused -> fell back to per-track. The very
first issue example ('Heroes 2017 Remaster') would have regressed.
Fix: strip edition/format/year NOISE words (deluxe, remaster, edition, flac,
years, bitrates, ...) before scoring, via _significant_words(), with a fallback
to the raw words so an album literally named '1989' or 'Deluxe' isn't emptied
to match-everything. Verified both directions: edition suffixes now KEPT, while
the wrong-album rejection (Scary Monsters for a Heroes request, Superheroes)
still scores 0.
Tests: +2 regression tests (edition-suffix kept; noise/number-only album name).
125 album-bundle/dispatch/plugin tests pass; compile + ruff clean.
Reporter (IamGroot60): requesting an album via a Prowlarr-backed source
(Usenet/Torrent) could download a DIFFERENT album — e.g. asking for Bowie's
'Heroes' downloaded 'Scary Monsters' because the picker ranked purely by
seeders/grabs -> quality -> size with NO title check, and the wrong album had
~16x the grabs. (Confirmed the old picker chose the wrong release on exactly
this scenario.)
Fix (the reporter's proposal):
- album_title_relevance(candidate_title, album_name): word-coverage match,
accent-folded (Bjork != bj rk) and WORD-BOUNDARY (Heroes != Superheroes), so
a wrong album that shares no title words scores 0.
- pick_best_album_release gains album_name/artist_name params and a relevance
gate (floor 0.6) applied BEFORE the seeders/quality/size ranking. When
album_name is given and NOTHING clears the floor, returns None.
- torrent.py + usenet.py call sites pass album_name/artist_name and set
result['fallback'] = True on None, so the dispatcher (source-agnostic
fallback routing) hands off to the per-track flow instead of grabbing a
wrong album. Matches what Soulseek already did via its preflight scorer.
No album_name -> no gating (old behavior preserved for callers without a
title). Tests: 9 new in test_album_bundle.py (relevance math incl. the
substring trap + accent fold, the exact Bowie refuse-and-fallback scenario,
None-when-no-match, and no-gate-without-name). 125 album-bundle/dispatch/plugin
tests pass; compile + ruff clean.
Reporter: album covers render as a top strip then solid grey ('break' on
import) — and it happens regardless of the album-art toggles. That ruled out
the import embed/cover.jpg paths (all toggle-gated) and pointed at the DISPLAY
cache, which every cover view goes through.
Root cause: ImageCache._fetch_and_store streamed the body to a tmp file and
committed it as status='ok' with only a 'total <= 0' (empty) guard. A
dropped/short connection makes requests' iter_content END EARLY WITHOUT
raising, so a PARTIAL image was cached permanently and served forever as a
half-decoded cover. The high-res art change in 2.6.4 (bigger images) makes a
mid-stream cutoff more likely, especially on the reporter's LXC.
Fix: capture the declared Content-Length and, after streaming, reject when
fewer bytes arrived (unlink the tmp file, raise ImageCacheError) so nothing
broken is cached and the next request retries fresh. When the server omits
Content-Length (chunked), we can't detect truncation, so we don't reject —
behavior unchanged there.
Tests (tests/test_image_cache.py): truncated download raises + caches nothing +
a later good fetch still works (differential-verified it's silently cached
without the guard); positive control (declared==actual) caches normally;
no-Content-Length still caches. 6 image-cache tests pass.
Strong-candidate fix: it's a real defect that produces exactly this symptom,
but I can't reproduce the reporter's LXC network to prove it's THE cause.
listening_history was populated ONLY from the media server; the web player
recorded nothing. Now a play heard ~10s logs to listening_history AND bumps
tracks.play_count/last_played — so the existing 'recently played' query reflects
actual SoulSync listening, and the Phase-2 smart-radio recency signal gets real
data.
- core/playback/play_log.build_play_event(): pure, DB-agnostic normalizer from
player payload -> listening_history event shape. Caller supplies the
timestamp (stays pure). Composite/streamed ids never become the int
db_track_id; bool ids rejected; missing title -> skip. 9 unit tests.
- MusicDatabase.record_web_player_play(): inserts the history row + increments
play_count/last_played for the library track in one call.
- /api/library/log-play: thin endpoint, server-side timestamp, best-effort
(logging failure never 500s / never affects playback).
- Frontend: npMaybeLogPlay on timeupdate fires once per track at the 10s
threshold (flag reset in setTrackInfo, set-before-fetch so it can't
double-fire), fully fire-and-forget.
Pure builder is unit-tested; the DB write can't run in-sandbox (real DB throws)
so it's a thin straightforward insert+update. JS + web_server parse clean.
Foundation for multi-listener playback. Today web_server.py keeps ONE global
stream_state dict + one lock (web_server.py:747), so the whole server shares a
single 'currently playing' — every tab/device is a remote for the same
playback and two listeners collide. That global is woven through ~22 sites and
isn't unit-testable where it lives.
Lifted into core/streaming/state.py WITHOUT changing behavior:
- StreamSession: one playback's state, dict-compatible (s['k'], s.get,
s.update, 'k' in s) so existing call sites work unchanged, each with its
OWN RLock so distinct sessions never block/clobber each other.
- StreamStateStore: registry of named sessions; lazy + race-safe create;
DEFAULT session reproduces today's exact single-global behavior. Also
drop()/active_ids()/session_ids() for the eventual per-listener wiring.
web_server.py now binds (DEFAULT) and
. Drop-in: every .update()/[k]/.get()/ site behaves identically. _set_stream_state routes a reassign
through session.replace() so the store's session stays the live object (it's
effectively dead — prepare.py only mutates in place — but safe now).
Honest scope: this is the PROVABLE half of Phase 3. The remaining half (3b:
derive a per-browser session id, per-session Stream/ staging, executor
concurrency, disconnect cleanup) is browser-coupled and can't be verified
without driving 2+ live clients — deferred to a live session. The store API is
already shaped for it.
Tests (tests/streaming/, 33 total):
- test_stream_state_store.py (19): session dict-compat, isolation, lazy
create, drop rules, active_ids, concurrent-create race safety.
- test_stream_state_callsite_compat.py (7): every real web_server access
pattern (library/play, stream/start, status, audio guard, stop, prepare
in-place mutation, set->replace) against the exact object web_server binds.
- test_prepare.py +1: real prepare worker drives an actual StreamSession.
76 streaming+radio tests green; ruff clean; web_server.py parses.
Replaces radio's pure ORDER BY RANDOM() with weighted ranking. Each tier now
fetches a generous random POOL (4x the needed count, floored) and
core/radio/selection ranks it before the collector keeps the best:
score_candidate = play_count(log-damped, w=1.0)
+ lastfm_playcount(log-damped, w=0.5)
- recently_played penalty(w=2.0)
+ stable per-id jitter(w=1.0, hash-derived so runs vary but
tests stay reproducible)
Modest weights so popularity guides without burying lesser-played tracks, and
jitter keeps radio from being identical every run. All intelligence is in pure
functions (rank_candidates / score_candidate) so it's tunable + unit-testable
without SQL.
Defensive: the DB method probes PRAGMA table_info(tracks) and omits
play_count/lastfm_playcount from the SELECT when absent (older DBs predating
the listening-history migration) — the scorer treats missing signals as 0, so
radio degrades to jitter-only instead of crashing on 'no such column'.
Tests (tests/radio/, 43 total):
- score_candidate / rank_candidates: deterministic unit coverage (popularity
ordering, lastfm contribution, recency penalty, garbage→0, stable jitter).
These CANNOT pass against pre-Phase-2 code.
- DB end-to-end: ranking surfaces the heavily-played track first out of a
decoy pool (wiring proof — probabilistic vs old random, documented honestly);
plus a no-rank-columns DB proving the defensive degrade path.
- All Phase-0a behavioral/refactor-equivalence tests still green.
60 radio + adjacent-DB tests pass; ruff clean.
First step of the stream/player/radio revamp (see revamp_plan.md). The radio
algorithm lived inline inside database.music_database.get_radio_tracks as raw
SQL tangled with selection logic — untestable without a live DB (which also
throws in the dev sandbox). Lifted the pure DECISIONS into core/radio/selection.py:
- parse_tags / merge_tags — JSON-or-CSV tag fields → ordered deduped list
- same_artist_cap — tier-1 30%-floored-at-5 cap
- build_like_conditions — OR-of-LIKEs SQL fragment + params per tier
- RadioCollector — dedup + cap + exclude-set + NOT-IN placeholder/value tracking
The DB method keeps the cursor work and now delegates every decision to these
helpers. Faithful extraction, not a rewrite — behavior unchanged.
This is the kettui foundation move: radio is now unit-testable, so Phase 2
(smart ranking — play-count / recency / feature seeding) becomes 'evolve a
tested function' instead of 'rewrite SQL and pray'.
Tests (tests/radio/):
- test_selection.py (22): unit coverage of every extracted helper
- test_get_radio_tracks_db.py (7): drive the REAL get_radio_tracks against
in-memory sqlite — tier fallback, dedup, exclude, file_path filter.
Behavior-pinned: these 7 pass against BOTH old inline and new extracted
code (refactor-equivalence proof). 52 adjacent DB+radio tests green.
The $year template variable was a blind release_date[:4] slice. When
something upstream poisoned release_date with a non-date value — the album
NAME — that slice emitted garbage: 'Mantras (Deluxe)'[:4] -> 'Mant', so
every download landed in 'Mantras (Deluxe) (Mant) [Album]/' instead of
'(2026)' (Tacobell444's screenshot).
Add _extract_year_from_release_date(): returns the leading 4 chars only
when they're a plausible year (isdigit, 1900 < y <= 2100), else ''. Matches
the guard the codebase already uses in soulid_worker._extract_year. A
non-year resolves to '' and the template's existing empty-() cleanup drops
it, so a poisoned release_date can never write rubbish into the path again.
This is the shared post-process path builder
(core/imports/paths.build_final_path_for_track) that DOWNLOADS, reorganize,
and imports all route through, so the guard covers every surface at once.
Defensive fix only — it stops the SYMPTOM regardless of which upstream
writes the album name into release_date. Pinning that upstream needs the
reporter's metadata source + the release_date value from app.log (the
Soulseek + AcoustID + future-dated-album combo is the discriminator);
tracked separately.
Tests (tests/imports/test_import_paths.py): unit coverage for the helper
(real dates kept, names/sentinels/short values rejected) + an integration
test reproducing #745 — a poisoned release_date yields 'Mantras (Deluxe)
[Album]' not '(Mant)' — differential-verified it produces the exact
'(Mant)' folder without the fix. Positive control keeps real (2026). 395
import + reorganize tests green.
The Duplicate Cleaner moves de-duplicated files into <transfer>/deleted/.
If a user's media server scans the transfer folder (e.g. a /music root
holding both the library and the transfer dir), those quarantined files
get real track rows in SoulSync's DB. Reorganize is purely DB-driven —
it acts on each track's stored file_path — so it would dutifully move a
quarantined file back OUT of /deleted to the template location, exactly
what Tacobell444 reported.
We can't stop the rows from existing (they come from the media server,
which the app doesn't control), so the fix is bounded to Reorganize, as
the reporter asked: skip any track whose resolved path is under
<transfer>/deleted. Surfaced as a non-matched 'In deleted/quarantine
folder — skipped' in the preview; apply mirrors it (post-process never
runs, file left in place, counted as skipped).
Detection is anchored to the <transfer>/deleted PREFIX (not a bare
substring) so a real album like 'Deleted Scenes' is kept; falls back to
an exact 'deleted' path-segment match when transfer_dir is unavailable
(mirrors the cleaner's own 'if deleted in dirs' skip). The one
unavoidable ambiguity — an artist folder named exactly 'deleted' at the
transfer root — is pinned in a test as intentional.
Guard added once where both consumers see it: preview_album_reorganize
and the apply worker (_RunContext gains transfer_dir).
Tests: tests/test_reorganize_deleted_quarantine.py (8 unit) +
test_library_reorganize_orchestrator.py (preview + apply integration,
differential-verified they fail without the fix). 128 adjacent
reorganize tests still green.
When the preflight-selected Soulseek folder produces zero usable files —
every transfer failed/aborted/stalled (the Slipknot dead-peer case: all
tracks 'Completed, Aborted' at 0 bytes) — _poll_album_bundle_downloads
returns []. download_album_to_staging used to return that with
fallback=False, so try_dispatch marked the whole batch failed and nothing
was retried elsewhere until the next wishlist run.
Flip that branch to fallback=True so the existing, proven per-track flow
takes over and re-searches every missing track across ALL sources/peers.
This reuses the per-track multi-source robustness instead of reimplementing
candidate-folder retry inside the bundle.
Tests: tests/test_soulseek_album_fallback.py drives the preflight-reuse path
with a stubbed poll — empty poll -> fallback=True (differential-verified it
fails without the fix), healthy poll -> fallback stays False. Downstream
routing (fallback=True -> per-track) already covered by
test_album_bundle_dispatch.py.
Live testing surfaced that slskd reports a peer-side abort as 'Completed,
Aborted' at 0 bytes (peer accepts then drops every transfer). That string
contains 'Completed', so the poll's completed-branch ran first and misread it as
'completed but file missing' — routing it into the #715 unresolved/download_path
grace (gives up after 45s with a misleading 'download_path mismatch' log)
instead of recognizing it as a failure.
Add 'Aborted' and 'Cancelled' to the failure-token check (which runs before the
completed branch), so these resolve immediately and correctly as failed. Test
added for the all-aborted folder.
Two related bugs from the Slipknot album never finishing.
1) _poll_album_bundle_downloads hung when the peer stalled. The finish check
needs every transfer terminal (completed/failed); the #715 grace only covers
'slskd says Completed but file not on disk'. A transfer stuck InProgress /
Queued, or dropped by slskd, is none of those — so it blocked both the finish
AND the grace exit, and the poll spun to the full ~6h timeout.
Add a bundle-level stall guard: track a progress marker (#terminal transfers,
total bytes across pending). If NOTHING advances for _stall_grace (180s) —
no terminal transition AND no pending byte movement — the peer has stalled;
mark the stuck transfers failed so the existing finish/all-failed checks
resolve the bundle with whatever completed (missing tracks then fall back to
the per-track matcher). Conservative: only trips when EVERYTHING is frozen,
so a slow-but-progressing or still-queued transfer is unaffected.
2) Failed batches lingered in the UI forever ('No tracks loaded'). The
auto-cleanup gate removed only complete/error/cancelled phases — 'failed'
(e.g. an album-bundle hard failure) was missing, so it never aged out. Add
'failed' to the terminal set so it's removed after 5 minutes like the others.
Tests (tests/test_soulseek_album_poll_stall.py): stalled peer → gives up with the
completed subset (not the deadline); progressing bundle not falsely stalled;
all-stalled → empty; dropped transfers stall out; clean finish unaffected.
124 download/soulseek tests pass; ruff clean.
make_wishlist_batch_row / _run_wishlist_cycle annotate params with Optional,
but the typing import only had Any/Callable/Dict. Slipped past py_compile +
tests because 'from __future__ import annotations' makes annotations strings
(never evaluated at runtime), but ruff flags it statically (F821).
Stage 2: the manual 'Download Wishlist' flow now calls the same
_run_wishlist_cycle engine the auto timer uses, so a manual scan runs the exact
same code path as an auto scan. The old bespoke manual orchestration (build
payloads + SERIAL inline dispatch) is deleted — its grouping/dispatch was a
near-duplicate of auto's that had already drifted.
Behavior changes (all intended, discussed):
- Manual now dispatches album bundles in PARALLEL (album pool) like auto, instead
of serially on one thread. A single cycle='albums' engine call covers the whole
selection (albums bundled, singles/ungroupable -> per-track residual), so no
'both cycles' pass is needed.
- The manual placeholder batch_id is reused as the engine's first sub-batch
(first_batch_id), so the modal's existing poll target stays valid.
- WishlistManualDownloadRuntime gains album_bundle_executor (wired in web_server,
falls back to the shared pool when unset).
- 'Don't start manual while auto is running' is unchanged — the existing route
guard (is_wishlist_actually_processing -> 409) already covers it; no queue added.
NOT touched: process_wishlist_automatically's behavior (proven by test_automation
staying green in Stage 1) and the per-track download mechanics.
test_manual_download.py rewritten to characterize the new behavior (engine
dispatch via the executor, parallel, placeholder reuse, album-context). Full
wishlist suite green (131); wishlist + automation = 392 passed.
Stage 1 of unifying the auto + manual wishlist flows. Extract the
group -> per-album+residual batches -> register -> dispatch logic that lived
inline in process_wishlist_automatically into a standalone _run_wishlist_cycle()
engine (built on make_wishlist_batch_row). The auto path now just calls it.
Per-flow differences are arguments (auto_initiated stamps the auto-only fields +
selects auto vs manual naming/logging; first_batch_id lets a caller reuse a
pre-created placeholder). Album batches dispatch to the dedicated album pool,
residual to the shared pool (unchanged from #740).
Auto behavior is PROVABLY unchanged: its full characterization suite
(test_automation.py) stays green (10/10), and the whole wishlist suite passes
(131). This commit does NOT touch the manual flow yet (Stage 2) and does not
change what auto does — it only moves auto's logic behind a shared entrypoint
the manual flow will call next.
The auto and manual wishlist flows each built the same ~20-field
download_batches row in separate places (auto album, auto residual, manual
placeholder, manual sub-batches) — four near-identical literals that could (and
did) drift apart, producing subtly different batch shapes between the flows.
Extract make_wishlist_batch_row() as the single source of truth: it emits the
consistent core field set, with the genuinely per-flow differences as explicit
arguments — initial phase ('queued' for auto / 'analysis' for manual), the
auto-only auto_initiated/auto_processing_timestamp/current_cycle via
extra_fields, and album-vs-residual contexts. All four sites now go through it,
so every wishlist batch has an IDENTICAL shape (this also removes the field
drift that confused the modal-hydration code).
Deliberately NOT unified — and left explicit in each caller, per the
'don't cargo-cult genuinely-different code' principle: the grouping decision
(auto groups only on the albums cycle), batch-id allocation (manual reuses the
caller's placeholder id for the first sub-batch), and dispatch (auto
parallel-submits album batches to the dedicated pool + residual to the shared
pool; manual runs them serially on one thread). Those are real behavioral
differences, not duplication.
Behavior-preserving: verified safe to normalize the row shape (grep confirmed
every reader uses .get() with defaults, no key-presence checks). The existing
auto (test_automation.py) and manual (test_manual_download.py) characterization
suites stay green = differential proof of identical behavior. Adds
test_batch_factory.py (core fields, album/residual, extra_fields, no shared
mutable state, consistent key shape). 131 wishlist tests pass.
A 2.6.3 change (c3b88e69) split the wishlist albums cycle into one batch per
album. Each album batch runs an INLINE-BLOCKING soulseek/torrent/usenet
album-bundle download (album_bundle_dispatch.try_dispatch ->
download_album_to_staging) that holds its worker thread for the whole
search+download. All of these were submitted to the shared 3-worker
missing_download_executor -- the same pool used for per-track downloads AND the
manual 'Download Wishlist' analysis.
So a large Album-Completeness 'Fix all' (-> ~819 wishlist tracks -> ~82 per-album
batches) saturated all 3 workers with blocking album downloads; the manual
wishlist analysis could never get a thread ('Library Analysis' stuck on
Pending), the other ~79 batches sat in phase='queued' forever, and auto-cleanup
(which only evicts terminal-phase batches) never cleared them -> jam until
restart. Fixing batch STATUS would not help: the threads are blocked inside the
download, not waiting on a phase flip.
Fix: add a dedicated bounded album_bundle_executor (max_workers=3) and route the
AUTO per-album bundle batches to it, keeping the shared pool free for analysis /
per-track / the manual wishlist (which always starts now). Hung/slow album
downloads can only delay other album downloads, never the user-facing path.
Additive and decoupled; the submit site falls back to the shared pool when the
album pool isn't wired (older callers / tests) so behavior is unchanged there.
The manual path is untouched (it already runs album bundles serially on a single
thread, by design).
Tests (tests/wishlist/test_automation.py): album sub-batches route to the
dedicated pool while the residual per-track batch stays on the shared pool;
fallback-to-shared-pool when no album pool is wired. Existing auto-processing
tests still green (fallback preserves prior behavior). 707 passed across
wishlist + downloads suites.
The same provider ID is stored under inconsistent column names across tables
(deezer_id vs deezer_artist_id vs album_deezer_id vs similar_artist_deezer_id;
spotify/itunes keep an entity qualifier, others don't; musicbrainz uses three
nouns), so code checks 2-5 name variants everywhere.
Add core/source_ids.py as the single source of truth for (provider, entity) ->
column, with accessors that read an ID from a dict/sqlite3.Row robustly
(canonical column first, then known aliases). NO database columns are renamed —
these are the real names today; the registry just centralizes the knowledge.
Targeted adoption (behavior-identical, verified):
- core/artist_source_lookup.SOURCE_ID_FIELD now derives from the registry
instead of duplicating the mapping (values unchanged).
- web_server.py artist-detail builds artist_source_ids via source_id_map(...)
instead of a hand-rolled per-source .get() dict.
Broader call-site adoption deferred as clearly-scoped follow-up.
Tests: tests/test_source_ids_registry.py (canonical columns, alias fallback,
canonical-preferred, sqlite3.Row, source_id_map, SOURCE_ID_FIELD unchanged).
Existing artist_source_lookup + artist_full_detail suites still green.
Reported by CubeComming: importing media keeps the track artist correct
(e.g. Billie Eilish) but changes the album-artist tag ("Albuminterpret") to
"Unknown Artist", breaking grouping in the media server.
Cause: in extract_source_metadata (core/metadata/source.py), album_artist is
seeded from the resolved track artist, then overridden by the album CONTEXT's
first artist. When the album lookup comes back unresolved, that first artist is
the literal "Unknown Artist" placeholder — which is truthy, so it clobbered the
real artist.
Fix: treat "Unknown Artist" (and empty) as a non-value — only let the album
context override the album_artist when it names a real artist. A genuine album
artist (e.g. "Various Artists") still overrides as before.
Tests: tests/metadata/test_album_artist_unknown.py — placeholder doesn't
clobber, real album artist still used, no-album-context falls back to track
artist, empty doesn't clobber. (Pre-existing test_album_mbid_cache.py failures
are an unrelated sandbox DB disk-I/O issue.)
Follow-up to the poll fix, covering the two things that blocked a
successful end-to-end album import once the poll itself stopped
freezing:
1. Staging dir permissions
The album-bundle private staging path defaults to
'storage/album_bundle_staging' -> /app/storage, but /app/storage was
never created or chowned by the image (unlike /app/Staging,
/app/Transfer, etc.), and /app is root-owned. The copy failed with
"[Errno 13] Permission denied: 'storage'" under the non-root soulsync
UID. Added /app/storage to the Dockerfile build-time mkdir+chown and
the entrypoint PUID/PGID chown, exactly like the sibling runtime dirs.
2. Client->local path resolution
Usenet/torrent clients report save paths from inside THEIR OWN
container (e.g. SAB '/data/downloads/music/<album>'); SoulSync often
mounts the same files at a different point ('/app/downloads/<album>').
Feeding the client path straight to the audio walker yields
"No audio files found" though the files are physically present.
New resolve_reported_save_path():
a. use the reported path as-is if readable (mirrored mounts),
b. apply explicit download_source.usenet_path_mappings
({from,to}, Sonarr/Radarr-style) for non-shared layouts,
c. basename fallback under SoulSync's own download roots —
zero-config for the standard shared-volume arr setup.
Wired into both call sites in usenet.py AND torrent.py
(download_album_to_staging + _finalize_download), logging any
translation and including the resolved path in the no-audio error.
Tests: resolver verbatim / explicit-mapping / basename-fallback /
priority / not-found / empty / mapping-miss-then-basename. ruff +
compileall + pytest green (645 in the download suites).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The earlier #721 fix tolerated a ~10s "completed but no save_path"
window, but the real production stall sits upstream of that: SABnzbd
removes a finished download from the queue and runs par2 verify /
repair / unpack *in History*, exposing the live stage in the slot
`status` ('Verifying' / 'Repairing' / 'Extracting' / 'Moving' / ...)
with `storage` empty until the final move. `_parse_history_slot` mapped
EVERY non-'Failed' status to 'completed', so a still-extracting 1.7 GB
FLAC album looked "completed with no save_path" the instant download hit
100%. The poll burned its completed-no-path budget mid-PP and bailed,
freezing the UI on the last download emit (the stuck-at-99%/100%
signature). SAB then finished fine — which is why the job shows
Completed in History but SoulSync never staged it.
Root fix
- `_parse_history_slot` routes `status` through `_map_state`, so PP
stages stay NON-terminal: the poll keeps waiting (as 'downloading')
for as long as post-processing takes and only a real 'Completed'
flips to terminal success. `save_path` is trusted only on true
completion (mid-PP path fields may point at the incomplete dir).
Supporting / defensive
- `UsenetStatus.incomplete_path`: surfaced separately from save_path
(SAB `incomplete_path`) and used by the poll loops as a LAST RESORT
after the completed-no-path window, to recover the case where
`storage` never lands but the files are physically on disk.
- `poll_album_download`: dedicated, configurable completed-no-path
window (~120s via `download_source.album_bundle_completed_no_path_seconds`)
decoupled from the ~10s transient-miss window; incomplete_path
fallback; a 30s heartbeat log so the previously-silent poll loop is
diagnosable.
- `usenet.py` `_download_thread`: per-track parity — it was erroring
immediately on the first completed-no-path read.
- `album_bundle_dispatch.py` / `status.py` / `monitor.py`: use the
project `get_logger` so download-flow logs land in app.log under the
`soulsync.*` namespace (they were console-only before, which hid the
`[Album Bundle] flow failed` line during triage).
Tests
- PP-history state mapping; end-to-end Hunky Dory PP regression
(download -> Verifying/Extracting in History past both budgets ->
Completed+storage -> success); completed-no-path window +
incomplete_path fallback; per-track thread parity. ruff + compileall +
pytest all green (the only local failures are environmental: missing
tzdata + local tools/ffmpeg.exe, neither present on CI).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Final cluster: the four structurally-identical snapshot endpoints
(discover_downloads, artist_bubbles, search_bubbles, beatport_bubbles) ->
core.discovery.endpoints.save_bubble_snapshot(...), wired via
_save_source_bubble_snapshot. All four validate a payload key, persist via
db.save_bubble_snapshot(kind, items, profile_id=...), and return a count +
timestamp; they differ only by:
- payload_key ('downloads' for discover, 'bubbles' for the rest) + its
no_data_error message.
- snapshot_kind, success_noun, and the info/except log subject + noun
("downloads"/"artists"/"albums/tracks"/"charts").
get_database / get_current_profile_id injected; get_json (request.json) invoked
inside the try, preserving the original 400/500 behavior incl. traceback dump.
Tests: +5 (missing key 400, None body 400, happy path with kind/profile/count/
timestamp, discover_downloads variant, exception -> 500). Full discovery suite:
210 passed.
web_server.py: -98 lines.
Ninth cluster: update_<source>_playlist_phase for the five sources sharing the
identical validation + full-message response (Tidal, Deezer, Qobuz,
Spotify-Public, YouTube) -> core.discovery.endpoints.update_playlist_phase(...),
wired via _update_source_playlist_phase + the _PHASE_LIST/_PHASE_LIST_YT
constants.
Per-source params:
- valid_phases — YouTube additionally allows 'parsed'.
- apply_extra_fields — Deezer/Qobuz/Spotify-Public also persist
download_process_id / converted_spotify_playlist_id from the body; Tidal and
YouTube do NOT, so they pass False (kept strictly 1:1 — the generic won't
apply those keys for them even if a caller sent them).
- not_found_message / error_label; get_json invoked inside the try.
NOT folded in: iTunes-Link — uses data.get('phase') (no "Phase not provided"
400) and returns a no-message payload.
Tests: +7 (404, missing-phase 400, invalid 400, happy path with extra-fields
suppressed, extra-fields applied when enabled, YouTube 'parsed' allowed,
exception -> 500). Full discovery suite: 205 passed.
web_server.py: -123 lines.
Eighth cluster, the heavyweights (~110 lines each). The fix-modal
update_<source>_discovery_match for the four sources with the identical
structure (Tidal, Deezer, Qobuz, Spotify-Public) ->
core.discovery.endpoints.update_discovery_match(...), wired via
_update_source_discovery_match. Applies the user-selected Spotify track to the
discovery result (status/artist/album/duration/spotify_data/match-count) and
writes the manual fix to the discovery cache.
Per-source pieces are params:
- source_log_label / error_label.
- original_track_key ('tidal_track' / 'deezer_track' / ...).
- original_artist_getter: Tidal handles string-or-object artists
(first_artist_str_or_obj); the rest assume strings (first_artist_plain).
- web_server helpers (join/extract artist, build_fix_modal_spotify_data,
cache-key, get_database, active-discovery-source) injected.
- get_json passed as a callable and invoked INSIDE the try, preserving the
original's "request.get_json() inside try" behavior (malformed body -> 500).
NOT folded in (genuinely divergent): iTunes-Link (saves spotify_data directly
via a different cache signature), YouTube (multi-key original_track fallback),
ListenBrainz (entirely different unmatch-capable structure, no cache write),
Beatport.
Tests: +9 (extractors; 400/404/400 guards; full happy path with result
mutation + duration formatting + match-count + cache-save args; no-increment
when already found; cache error swallowed; get_json raise -> 500). Full
discovery suite: 198 passed.
web_server.py: -400 lines.
Seventh cluster: start_<source>_sync for the five sources with the identical
flow (Tidal, Deezer, Qobuz, Spotify-Public, YouTube) ->
core.discovery.endpoints.start_sync(...), wired via _start_source_sync.
Validates phase, converts discovery results, seeds sync state, posts a
"... Sync Started" activity item, and submits to the sync executor. Per-source
pieces are params:
- sync_id_prefix (f"{prefix}_{key}"), not_found/not_ready messages, convert_fn.
- name/image accessors: Tidal reads an object (playlist_name_obj/
playlist_image_obj), the rest a dict (playlist_name_strict/playlist_image_dict).
- activity_label vs error_label DIFFER for Spotify-Public ("Spotify Link
Sync Started" activity, "Spotify Public" logs).
- submit_sync_task glue (_submit_sync_task) closes over sync_executor /
_run_sync_task / get_current_profile_id so the helper stays global-free.
NOT folded in: iTunes-Link (no final info log), ListenBrainz (submits the
task WITHOUT a playlist_image_url arg), Beatport (extra debug logging, chart).
Tests: +6 (404, not-ready 400, no-matches 400, full happy path with
state/sync-infra/submit/activity assertions, resync phases allowed,
exception -> 500). Full discovery suite: 189 passed.
web_server.py: -172 lines.
Sixth cluster: the bulk-hydration get_<source>_playlist_states endpoints for
the five sources that build the identical per-entry dict + {"states": [...]}
shape (Tidal, Deezer, Qobuz, Spotify-Public, iTunes-Link) ->
core.discovery.endpoints.get_playlist_states(states, *, error_label,
info_log_label=None), wired via _get_source_playlist_states.
iTunes-Link is the only one of the five without the "Returning N stored ..."
info log, so info_log_label is optional (iTunes passes None to suppress it).
NOT folded in: the YouTube/ListenBrainz get_all_*_playlists endpoints. They
return {"playlists": [...]} (different key) with a different field set
(url / created_at / playlist, no discovery_results) and filter out
mirrored_/profile-scoped entries — genuinely divergent, kept as-is.
Tests: +4 (list build + last_accessed bump + exact shape, empty, optional ids
default None, missing-required-field -> 500). Full discovery suite: 183 passed.
web_server.py: -116 lines.
Fifth cluster: reset_<source>_playlist for the four sources with byte-
identical bodies (Tidal, Deezer, Qobuz, Spotify-Public) ->
core.discovery.endpoints.reset_playlist(states, key, *, label,
not_found_message), wired via _reset_source_playlist. Resets phase/status to
'fresh', clears discovery/sync fields, cancels any discovery_future, and
preserves the original playlist payload.
Left with their own bodies (genuinely divergent):
- YouTube: status -> 'parsed' (not 'fresh'), no download_process_id, logs the
playlist name, "reset to fresh state".
- ListenBrainz: status -> 'cached', logs playlist title, returns
{"success": True, "phase": "fresh"} (different payload), _lb_state_key.
- iTunes-Link: state.update(...), no info log, "iTunes Link reset to fresh
phase".
Tests: +4 (404, full clear + playlist preserved + future cancelled, no-future
path, exception -> 500). Full discovery suite: 179 passed.
web_server.py: -100 lines.
Fourth cluster: get_<source>_discovery_status (all eight sources, Beatport
included) -> core.discovery.endpoints.get_discovery_status(states, key, *,
not_found_message, error_label), wired via _get_source_discovery_status.
Unlike sync-status, the discovery-status response shape is byte-identical
across every source (phase/status/progress/spotify_matches/spotify_total/
results/complete), so Beatport folds in here too. Only the 404 string
("... discovery not found" vs "... playlist not found" vs "Beatport chart
not found") and the except-log label vary. ListenBrainz key via _lb_state_key.
NOT touched this cluster: get_*_playlist_state (the sibling endpoints).
Those genuinely diverge per source — different id-key name (playlist_id /
url_hash / playlist_mbid), presence of url / created_at / download_process_id,
Tidal's playlist.__dict__ serialization, and YouTube's strict (non-.get)
field access. Folding them would need a flag pile that wouldn't be a clean
1:1, so they keep their own bodies.
Tests: +4 (404, full response + last_accessed bump, complete=False when not
'discovered', missing-field -> 500). Full discovery suite: 175 passed.
web_server.py: -155 lines.
Third cluster: the get_<source>_sync_status routes (Tidal, Deezer, Qobuz,
Spotify-Public, iTunes-Link, YouTube, ListenBrainz) -> core.discovery.
endpoints.get_sync_status(...), wired via _get_source_sync_status glue.
This cluster carries the real per-source quirks, all captured 1:1 as params:
- not_found_message (iTunes-Link uses "iTunes Link not found").
- error_label vs activity_subject — these DIFFER for Spotify-Public: the
activity feed says "Spotify Link playlist ..." while the except log says
"Error getting Spotify Public sync status".
- playlist-name accessor, three styles lifted verbatim as named helpers:
playlist_name_attr_or_unknown (Tidal: object .name), playlist_name_strict
(Deezer/Qobuz/Spotify-Public/iTunes: state['playlist']['name'], can raise),
playlist_name_safe (YouTube/ListenBrainz: .get default). The strict getter
preserves the original's behavior of raising -> 500 AFTER phase/sync_progress
were already mutated.
- ListenBrainz key via _lb_state_key (caller-resolved).
Beatport stays separate (different payload: status not sync_status, sync_id,
no lock, chart key).
Tests: +9 (3 name accessors incl. raise/fallback semantics; status 404s,
running-no-mutation, finished+activity, error+revert+activity, and strict-
getter-missing -> 500 after partial mutation). Full discovery suite: 171 passed.
web_server.py: -244 lines.
First cluster of the per-source playlist-discovery deduplication. The
convert_<source>_results_to_spotify_tracks functions (Tidal, Deezer, Qobuz,
Spotify-Public, YouTube, ListenBrainz) plus the already-generic
_convert_link_results_to_spotify_tracks were byte-identical apart from the
source label used in their log line.
Lift the shared body into core/discovery/endpoints.py as
convert_results_to_spotify_tracks(results, source_label); the 7 web_server
functions become 1-line delegations (names/signatures unchanged, so all
callers and behavior are identical — 1:1).
Beatport is intentionally NOT folded in: its converter coerces artist
objects to strings and emits a different track shape (source field, album
dict), so it keeps its own implementation.
Tests: tests/discovery/test_discovery_endpoints.py (12) pin both input
shapes (manual spotify_data / auto spotify_track+found), optional
track/disc numbers, falsy-0 omission, field defaults, skip-on-neither,
order preservation, if/elif precedence, empty input.
web_server.py: -209 lines. Full discovery suite: 151 passed.
Follow-up to the album-art resolution fix. That change upgraded MusicBrainz
Cover Art Archive thumbnails (/front-250) to the bare /front original — but
/front redirects to archive.org, which is unreliable: probing release-group
covers showed intermittent HTTP 500s (same URL 500s one second, serves the
next) and multi-MB originals (2.9 MB seen). The result was the user-reported
flakiness: cover art that "sometimes works, sometimes shows nothing", and a
huge image embedded into every track when it did work.
The sized thumbnails (/front-250, -500, -1200) are served by CAA's own CDN,
not the archive.org redirect — which is why /front-250 (240p) was always
reliable. Upgrade to /front-1200 instead: 1200x1200 is a massive jump from
240p, reliably CDN-served, and a sane ~40 KB instead of multi-MB.
Applied in all three CAA spots for consistency: the _upgrade_art_url helper
(embed + cover.jpg paths) and both prefer_caa ("CCA") blocks, which fetched
the bare /front directly with no fallback — so CCA-on users hit the same
flakiness. _fetch_art_bytes still falls back to the original /front-250 if
/front-1200 is ever refused.
Tests updated to assert the 1200px target, idempotency, and that the bare
/front original is intentionally left untouched.
User report: embedded album art came out ~600x600 while the cover.jpg in
the folder was high-res. The cover.jpg path upgraded the source CDN URL
to its highest resolution, but the tag-embed path fetched the raw URL —
so iTunes art embedded at its 600x600 default, Spotify at 640, Deezer at
1000. The "Write Tags to File" retag path had the same gap (Deezer-only
upgrade), and MusicBrainz art was worse still: every Cover Art Archive
URL is built as the /front-250 thumbnail, so MB-sourced downloads
embedded 250x250.
Factor the resolution upgrade + fetch into two shared helpers in
core/metadata/artwork.py and route every art path through them:
_upgrade_art_url(url) — bump to the source's highest resolution:
- Spotify (i.scdn.co) -> original master (~2000px+)
- iTunes (mzstatic.com) -> 3000x3000
- Deezer (dzcdn) -> 1900x1900
- Cover Art Archive -> /front original (was /front-250)
_fetch_art_bytes(url) — upgrade, fetch, and fall back once to the
original size if the CDN refuses the larger one (non-regressive).
Now consistent across: embed-into-tags (post-process), folder cover.jpg
(post-process), and the enhanced-library "Write Tags to File" retag flow.
The YouTube path already upgraded via Album.from_spotify_album, unchanged.
De-duplicates the per-source upgrade code that was copied across sites
and drops the now-unused urllib import from tag_writer.
Not covered (follow-up): Last.fm / Amazon / Tidal / Qobuz have no
explicit upgrade yet — some already serve full-res, others may hand over
a capped size that passes through unchanged.
Tests: new tests/metadata/test_artwork_resolution.py pins every upgrade
(Spotify 300/640->master, iTunes 100/600->3000, Deezer->1900, CAA
thumbnail->original, unrecognized/empty unchanged) and the fetch
fallback. Updated the two tag_writer fallback tests to patch the network
at its new home in artwork.
Artist-detail discography from MusicBrainz fetched releases via the
artist lookup (`/artist/<mbid>?inc=release-groups`), which MusicBrainz
hard-caps at 25 embedded release-groups and which ignores the `limit`
param entirely. Prolific artists had ~85% of their catalogue silently
dropped — Kendrick Lamar has 167 release-groups on the site but only the
first 25 ever reached SoulSync. Reported by Sokhi: "a lot of albums are
missing when searching vs what's showing on the site."
Switch `get_artist_albums` to walk the paginated browse endpoint
(`/release-group?artist=<mbid>`, offset loop) — the same pattern the
basic-search path already uses — fetching the full catalogue up to the
caller's limit. No type filter and no studio-only filter here: the
artist-detail page wants every primary/secondary type so its tabs mirror
musicbrainz.org. Verified live: now returns all 167 for Kendrick.
Adds 7 tests covering pagination past the cap, offset advance,
short-page stop, limit cap, cross-page dedup, type->bucket mapping, and
a regression pin asserting the capped inc=release-groups lookup is no
longer the discography source.
Two things in this commit. Functional download / matched-download
behaviour is untouched — same JS handlers, same routes for the
download actions, same album-expand interaction.
VISUAL REDESIGN
- Glass search-bar card with accent radial wash + focus ring + pill
primary search button
- Source chip row above the search bar (see below)
- Always-visible compact filter pill row (Type / Format / Sort) —
pills carry both ``bs-filter-pill`` (new visual) and ``filter-btn``
(legacy class for ``resetFilters`` + ``applyFiltersAndSort`` in
wishlist-tools.js to keep working)
- Accent-tinted status pill matching the dashboard / auto-sync look
- Album result cards: glass card with accent left-edge stripe,
52px brand-tinted cover icon, chevron expand indicator, pill
action buttons (Download / Matched Album), accent glow on hover
- Track result cards: glass row with accent stripe, 44px icon,
pill action buttons (Stream / Download / Matched Download)
- Multi-disc separators inside expanded album track lists styled
with the accent treatment
- Responsive: action button columns stack vertically below 900px
New CSS lives in a self-contained ``webui/static/basic-search-v2.css``
sheet linked from index.html. Selectors are scoped to
``#basic-search-section`` for any class that already exists in
style.css (``.album-result-card``, ``.album-icon``, ``.track-*``,
etc.); the new ``bs-*`` prefixed classes for the search bar /
filters / source row / status are unscoped because they only exist
in the new markup. ``!important`` is used on the card-level rules
to defeat the original unscoped ``.album-result-card`` etc. rules
in style.css that would otherwise leak heavyweight padding /
box-shadow / 56px icon styles into the new design.
Also removed ``overflow: hidden`` from the original
``.album-result-card`` and ``.track-result-card`` rules in style.css
— those two classes only render in ``downloads.js`` basic search
results (verified via grep, two render sites only), so the
removal can't impact any other UI.
SOURCE PICKER (hybrid mode)
- New ``GET /api/search/sources`` endpoint returns the list of
active sources from the orchestrator's chain (or the single
active source in single-source mode).
- Frontend renders a chip row above the search bar. Click a chip
to target that source for the next search; the chip's brand
accent fills.
- In single-source mode the lone chip is rendered as a dashed-
border label so the user always knows what they're searching
but can't accidentally try to switch to sources that aren't
configured.
- ``/api/search`` accepts an optional ``source`` body param. When
set, ``core/search/basic.py:run_basic_search`` resolves the
client directly via ``orchestrator.client(source)`` and calls
its ``.search()`` instead of going through the hybrid chain.
- Backwards compatible: omitting ``source`` falls through to the
original ``orchestrator.search()`` call exactly as before.
Unknown source names also fall back to the default — typo
protection.
TESTS (5 new + 6 pre-existing = 11 total in test_search_basic.py)
- source param routes to specific client, NOT orchestrator chain
- no source param preserves original orchestrator-default behaviour
- unknown source name falls back to orchestrator default
- ``run_basic_soulseek_search`` backwards-compat alias preserved
- source-targeted path serialises albums + tracks correctly
101 search-suite tests pass.
Reporter @Sokhii: downloading the Mushoku Tensei Original
Soundtrack II via Apple Music metadata + Tidal download
produced duplicate library entries — same audio file landed
under multiple track positions in the album view.
Root cause (verified by direct probe + isolated repro):
``MusicMatchingEngine.normalize_string`` correctly skipped
unidecode for CJK text (kanji→pinyin would have produced
gibberish — see the inline comment at line 74-76), but then
ran ``re.sub(r'[^a-z0-9\s$]', '', text)`` which stripped EVERY
CJK character. Every Japanese title normalised to ``''``.
``similarity_score`` has an early-out guard
if not str1 or not str2: return 0.0
so EVERY CJK-vs-CJK title comparison returned 0.000.
Downstream effect: the matcher fell back to duration+artist
alone. For an OST album with 24 tracks all by the same artist
with similar durations, multiple iTunes track queries landed
on the SAME Tidal candidate. SoulSync wrote each download to
a different output filename (per the iTunes track position),
so on disk there were N copies of the same audio under
different track numbers. The user's library showed 34 entries
for an album with 24 actual tracks.
Probed iTunes album 1753240110 directly — 24 distinct tracks,
zero (disc, track_number) collisions, both US + JP storefronts.
So the duplicate origin was definitely downstream of metadata
fetch.
Fix: when CJK is detected upstream, the alphanumeric-strip step
also preserves CJK Unified Ideographs + radicals
(⺀-鿿), Hiragana + Katakana (-ヿ), Halfwidth
/ Fullwidth forms (-), and Hangul syllables
(가-). CJK titles now produce a comparable normalised
form instead of an empty string. ``similarity_score`` works as
intended:
'命の灯火' vs '命の灯火' → 1.000 (was 0.000)
'命の灯火' vs '無職転生' → 0.000 (was 0.000, but now from
actual char comparison
not from the empty-string
guard)
Latin-only normalisation is completely unchanged. ``has_cjk``
is False for Latin input, so both the CJK-lowercase branch AND
the new CJK-preserve strip branch are skipped — Latin titles
go through the original unidecode + lowercase + strip path
verbatim. Tested via 4 regression tests that pin the Latin
baseline (simple, unidecode target, $-preservation, identical
+ different similarity scores).
16 new unit tests in ``tests/test_matching_engine_cjk.py``:
- Kanji / Hiragana / Katakana / Hangul / Chinese all survive
- CJK-only strip still removes Latin punctuation in the
CJK branch
- Mixed Latin + CJK lowercases the Latin half
- Identical CJK titles → 1.0
- Disjoint CJK titles → near 0
- Partially overlapping CJK titles → midrange
- CJK doesn't falsely match unrelated Latin
- 4 Latin-baseline regression pins
- Real-world Mushoku Tensei OST scenario
371 text + imports + new CJK tests pass after the fix.
Follow-up to the 2.6.3 queue→history handoff fix (#706). User
@IamGroot60 reported in #721 that on 2.6.3 the bundle still gets
stuck mid-flight: SoulSync UI sits on "Usenet downloading release
61%" forever, SAB History shows the job as Completed 2+ minutes
ago, files are physically present in the slskd downloads folder
but never copied into ``storage/album_bundle_staging/<batch>/``.
Root cause: a second-stage gap in the SAB pipeline. SAB flips a
job's ``status`` to ``Completed`` in History as soon as par2 +
unrar finish, but its post-processing pipeline writes the final
``storage`` field a few seconds LATER (the move-to-final step).
``poll_album_download`` saw the first ``Completed`` read with
``save_path=None`` and bailed:
if status.state in complete_states:
return last_save_path # ← None at this point
``download_album_to_staging`` got ``save_path=None``, set
``result['error']`` and returned. The bundle was marked failed but
the LAST progress emit before the failure was ``downloading
progress=0.61``, so the UI froze on "61%" — the terminal ``failed``
emit never registered on the user's screen because the renderer
holds the last-known progress.
Fix
- ``poll_album_download`` now tracks a separate transient counter
for "complete state seen, save_path not yet set." Up to
``transient_miss_threshold`` (default 5) consecutive reads in
that state are tolerated before the poll bails. SAB writes the
``storage`` field within 2-10 seconds of the History flip in
practice — the default 5 × 2s = 10s window covers it.
- When save_path eventually lands, return it normally.
- When the threshold is exhausted with save_path still empty,
emit terminal ``failed`` with an explicit message pointing at
the missing save_path field — no more 6-hour silent spin.
- Earlier ``downloading`` reads with a non-empty ``save_path``
(qBit / Transmission set this from the start of the download)
remain "sticky" — if the eventual ``completed`` read has empty
save_path, the cached one applies. So torrent flows aren't
affected by the retry path.
SAB adapter (``_parse_history_slot``)
- Widened the save_path field fallback chain:
storage → path → download_path → dirname → incomplete_path
Covers SAB version differences (older builds populated ``path``)
and forks that expose ``download_path`` or ``dirname``.
``incomplete_path`` is the last resort — SAB's in-progress dir
before the final move — so the bundle plugin at least has a
path to scan when nothing else lands.
- Whitespace-only values are skipped.
- Loud debug log when none of the known fields land — users on
SAB versions / forks with novel field names need to see this in
logs so we can grow ``_HISTORY_SAVE_PATH_KEYS``.
Tests
- ``test_album_bundle.py`` (3 new):
- tolerates_completed_with_late_save_path_arrival — the #721
scenario; first Completed read has no save_path, third has
it; poll returns the path normally
- gives_up_when_completed_with_no_save_path_persists — past
the threshold the poll fails loudly instead of silent-spinning
- uses_save_path_from_earlier_downloading_emit_if_completed_lacks_one
— sticky save_path keeps torrent flows working
- ``test_usenet_client_adapters.py`` (6 new):
- falls back to ``path`` when ``storage`` empty
- falls back to ``download_path``
- prefers ``storage`` when multiple fields present
- returns ``None`` when all fields empty (the #721 gap window)
- ignores whitespace-only values
- uses ``incomplete_path`` as last resort
132 album-bundle + usenet tests pass.
Branch is on dev parented at 2.6.3 — user @IamGroot60 offered
to test on dev, so this is a candidate cherry-pick for either
a 2.6.4 hotfix or merge straight into dev for the next release.
Two related leaks in ``storage/album_bundle_staging/<batch_id>/``:
1. **Soulseek bundle cleanup was excluded.** The per-batch cleanup
at the end of a bundle download gated on:
(album_bundle_source or '').lower() in ('torrent', 'usenet')
The comment justified it as "slskd keeps its own completed
folders" — but the Soulseek bundle path ALSO copies completed
files into the private staging dir (``soulseek_client.py:1599``,
``copy_audio_files_atomically(completed, Path(staging_dir))``)
for the per-track workers to claim. Those copies persisted
forever; long-running installs accumulated stale GB. Extended
the cleanup gate's allow-list to include ``soulseek`` so the
per-batch dir is removed on bundle completion — same code path
that already worked for torrent / usenet.
2. **No sweep for orphan dirs.** Any leftover ``<batch_id>``
subdir from a previous-session crash, an errored batch, or a
pre-fix Soulseek bundle stayed on disk forever. Added
``sweep_orphan_album_bundle_staging(staging_root, active_batch_ids)``
that runs ONCE at server startup, before any batch can register
a staging dir. Removes every ``<batch_id>``-shaped subdir
whose id isn't in the active set. Safe by construction:
- Only touches subdirs of the configured staging root.
- Name-shape check (``entry.name == _safe_batch_dirname(entry.name)``)
rejects hand-placed dirs like ``.git`` or stray docs.
- ``shutil.rmtree`` errors log + continue — sweep must not
crash app startup over a permission glitch.
- active_batch_ids normalised through ``_safe_batch_dirname``
so colon-bearing batch_ids match their on-disk form.
Wired into the web_server startup right after the stuck-flags
diagnostic so it fires before anything else touches batches.
Tests
- ``test_downloads_lifecycle.py`` gained one regression test
pinning that Soulseek bundles now have their staging dir
cleaned (sibling to the existing torrent test).
- ``test_album_bundle_staging_sweep.py`` (NEW, 11 tests)
covers: orphan removal with no actives, active dirs preserved,
special-char batch_id normalisation, no-op on missing /empty
/empty-string staging root, non-dir entries skipped, unsafe-
name dirs preserved (.git etc.), partial rmtree failure doesn't
abort the rest, listdir failure returns 0 cleanly, default
None active set, defensive against empty / None entries in
the active set.
488 downloads tests pass.
For users with an existing "clean up old files" automation pointed
at this dir: stop pointing it there if you want — the auto-cleanup
+ startup sweep cover it now. Or leave it as belt-and-suspenders
with a relaxed (1h+) mtime threshold so it can't race a mid-batch
download.
finished the release (#715)
Symptom (user @pavelcreates / @IamGroot60 on 2.6.2):
- Click Download on an album in the search modal
- slskd starts + completes every track of the release
- 22+ minutes after the last completed download, batch flips
to "failed" with no clear log line explaining why
- Per-track Soulseek downloads on the same machine were fine
Root cause: ``core/soulseek_client._resolve_downloaded_album_file``
probed three hard-coded candidate paths to locate each downloaded
file in the slskd download dir:
candidates = [
download_path / remote_filename,
download_path / basename,
download_path / *normalized_path_parts,
]
On the common slskd config ``directories.downloads.username = true``
slskd writes files at ``<download_dir>/<username>/<filename>`` —
none of the three candidates carry a username segment, so the
resolver returned None for every file even though the file was
physically present in a subdir one level deeper. ``_poll_album
_bundle_downloads`` saw 0 completed_paths, kept spinning, and
hit the master deadline (~30 min) before bailing the batch.
Why per-track worked: ``web_server._find_completed_file_robust``
already does a recursive walk-by-basename + path-confirm against
the remote directory components, so any layout slskd writes ends
up resolved. The bundle path didn't go through it.
Fix
- Lifted the robust finder into ``core/downloads/file_finder.py``
as a pure function ``find_completed_audio_file(download_dir,
api_filename, transfer_dir=None) -> (path, location)``. Zero
globals; recursive walk; handles slskd dedup suffix
``_<10+digit-timestamp>``, YouTube / Tidal ``id||title`` encoded
filenames, the AcoustID-quarantine subdir skip, basename
collisions disambiguated by remote-path components, and a
fuzzy-basename fallback above 0.85.
- ``_resolve_downloaded_album_file`` keeps the three-candidate
fast path (cheap probe for the slskd-flat default) but now
delegates to the new helper when none hit, instead of giving up.
- ``_poll_album_bundle_downloads`` tracks "slskd reports
Completed but local resolver returns None" per key. When every
remaining key has been in that state past a 45-second grace
window, the poll exits early with an explicit error pointing at
the likely ``soulseek.download_path`` mismatch instead of
silently spinning until the master deadline.
- ``web_server._find_completed_file_robust`` becomes a thin
delegate so both callers share one finder. Legacy inline impl
kept as ``_find_completed_file_robust_legacy`` for reference;
to be removed next release.
- Fixed misleading ``"(0 tracks, quality=)"`` log on the preflight-
reuse path — was reading attrs off a None ``picked`` object.
Tests (17 new in tests/downloads/test_file_finder.py)
- Flat slskd layout
- Username-prefixed (the #715 case)
- Full remote tree preserved
- Deeply nested username + tree
- File genuinely missing returns None
- Basename collision disambiguated by remote dirs
- Single basename match wins regardless of dirs
- slskd dedup suffix match
- Short ``_<digits>`` (year) not treated as dedup
- AcoustID quarantine subdir skipped
- YouTube / Tidal ``id||title`` encoded filenames
- transfer_dir fallback
- Both dirs miss → (None, None)
- Non-audio files ignored
- Empty api_filename
- Fuzzy match on punctuation variant
- Fuzzy rejects below threshold
475 downloads tests pass after the lift.
Pipeline-driven Auto-Sync runs against any ListenBrainz playlist
(Weekly Jams, Weekly Exploration, Top Discoveries, etc.) would sit
on ``Refreshing: "<name>"`` with no UI updates for 5-7 minutes
before the pipeline progressed. Two real bugs stacked:
1. **Double discovery.** The refresh handler called
``_maybe_discover`` (matching engine, per-track Spotify/iTunes/
Deezer matches) inline for any source returning
``needs_discovery=True`` tracks. Phase 2 of the pipeline then
ran the SAME matching engine via ``run_playlist_discovery_worker``
on the same tracks. The refresh-side run blocked the loop with
zero progress emission; Phase 2's already has the timed
progress-poll pattern. So LB tracks discovered twice, the first
time silently.
Pipeline now sets ``skip_discovery=True`` on its refresh config.
The handler honors the flag and lets Phase 2 handle discovery
end-to-end. Standalone callers (Sync-page tab, registration
action) leave the flag unset so they still get matched_data
on refresh.
2. **No targeted LB refresh.** The LB adapter's ``refresh_playlist``
called ``manager.update_all_playlists()`` — the only refresh
entry-point the manager exposed — which re-pulls every cached
LB playlist's details from the API (~12+ round-trips) even
when only one playlist needed refreshing. Wasteful;
tax-on-everyone for one-playlist work.
Added ``LBManager.refresh_playlist(mbid)`` — reads the cached
playlist_type, fetches just that playlist's details, runs the
normal ``_update_playlist`` upsert path. Defaults type to
``user`` for un-cached mbids so new-playlist discovery still
works. Skips ``_cleanup_old_playlists`` and
``_ensure_rolling_mirrors_from_cache`` (wasted work for a
single-playlist refresh).
Also: killed a silent ``except Exception: pass`` in the LB
adapter's old refresh wrapper that was masking every LB API
failure as a stale-cache hit. Refresh errors now log with full
traceback at warning level and propagate ``None`` so the outer
handler at ``refresh_mirrored.py:104`` counts the error and
surfaces it to the run-history error tally.
Pinned with 12 new unit tests across:
- ``tests/test_listenbrainz_manager.py`` (8): targeted refresh
happy path, unauthenticated guard, empty-mbid guard, upstream
``None`` return, default playlist_type for unknown mbid,
exception propagation, cost guard skipping cleanup, skipped-
when-unchanged signal
- ``tests/test_playlist_sources_adapters.py`` (3): adapter uses
targeted call (not legacy), adapter returns ``None`` on manager
error (not silent swallow), adapter resolves synthetic series
ids before calling the manager
- ``tests/automation/test_handlers_playlist.py`` (1):
skip_discovery flag bypasses ``_maybe_discover`` end-to-end
Residual per-track wishlist downloads (single tracks from different
albums, below the album-bundle threshold) were producing folders
without a year subfolder whenever the wishlist row carried a stale
``track_number=1`` from an older payload default.
Why: ``core/downloads/candidates.py`` had a single API-fetch branch
that served two concerns — resolving the track position AND
hydrating the lean ``spotify_album_context`` (release_date /
total_tracks / cover image) — gated entirely on track_number being
unresolved. When the wishlist row's ``track_number`` happened to
be 1 (a poisoned default rather than a real value), the gate
short-circuited and the album hydration the same call would have
done was skipped. Deezer-sourced discovery matches don't ship
release_date in their search-result album shape, so without the
backfill the folder lost its year.
The two concerns split:
- track_number resolution keeps its track_info → track object →
API precedence chain. track_info defaults still win.
- album hydration runs whenever release_date or total_tracks are
missing, independent of where (or whether) track_number was
resolved.
The single API round-trip still serves both — the cost contract
is preserved. The side-effect coupling is gone.
Lifted into ``core/downloads/track_metadata_backfill.py``
(``hydrate_download_metadata``) so the precedence chain is pinned
in isolation. 24 unit tests cover the precedence chain, the
poisoned-tn=1 regression case, defensive non-dict/None inputs,
the cost guard (API called at most once per invocation), and
disc_number resolution.
Also lands the upstream piece: ``core/wishlist/routes.py:_build_track_data``
no longer defaults ``track_number=1`` / ``disc_number=1`` /
``total_tracks=1`` / ``release_date=''`` when the library-modal add
payload omits them. Missing values now flow through as ``None`` so
the downstream pipeline can detect-and-recover instead of locking
to a fake position.
Real-world regression triggered by the album-bundle work earlier in
2.6.3. Tracks with full Spotify metadata were importing as
``01 - <title>`` under ``Artist - Album/`` (no year), even when the
source filename carried the correct track number and Spotify's
release_date was available.
Investigation via DB inspection of stored wishlist rows:
```
"Never Gonna Give You Up" → track_number=None, release_date=""
"idfc" → track_number=1, release_date=""
"No Sleep Till Brooklyn" → track_number=1, release_date=""
```
Source-of-truth Spotify metadata had release_date AND real track
positions, but the wishlist row was poisoned. Three regressions
compounded the loss:
**Fix A — ``track_object_to_dict`` (``core/wishlist/payloads.py:295``)
preserved only album.name during Track→dict conversion.**
Pre-fix:
```python
album_name = "Unknown Album"
if hasattr(track_object, "album") and track_object.album:
if hasattr(track_object.album, "name"):
album_name = track_object.album.name
else:
album_name = str(track_object.album)
result = {
...
"album": {"name": album_name}, # ← release_date / images / etc. all dropped
...
}
```
When a wishlist payload arrived as a Track dataclass instead of a
raw spotify_data dict, the Track→dict conversion stripped
release_date, images, album_type, total_tracks, id, and album-level
artists. Every wishlist row added through this path landed in the
DB with ``album={'name': X}`` only.
Post-fix: three branches handle the three album shapes
- ``album_attr`` is a dict → ``dict(album_attr)`` preserves every key
- ``album_attr`` is a sub-object → pull all common Album-dataclass
attrs (id, release_date, album_type, total_tracks, images, ...)
- ``album_attr`` is a bare string → build a dict from the track
object's adjacent attrs (release_date, album_id, album_type, ...)
and surface ``image_url`` as ``album.images``
**Fix B — ``core/discovery/playlist.py:309`` only added
``track_number`` / ``disc_number`` keys when truthy.**
Pre-fix:
```python
matched_data = { 'id': ..., 'name': ..., ... } # no track_number / disc_number
if track_number:
matched_data['track_number'] = track_number
if disc_number:
matched_data['disc_number'] = disc_number
```
Deezer-sourced matches always hit this branch with ``track_number=None``
because the cache enrichment at line 304 reads ``_raw.get('track_number')``
literally, but Deezer's raw shape uses ``track_position``. So the key
was omitted from ``matched_data``, downstream consumers couldn't
distinguish "missing key" from "value is 1", and the chain silently
filled 1.
Post-fix: keys are ALWAYS present (None when unknown). Also adds a
``best_match.track_number`` fallback so the Track-dataclass-mapped
value (which DOES include ``track_position``→``track_number``
mapping) gets used when the cache lookup misses.
**Fix C — Pipeline only consulted ``album_info.track_number`` before
falling to the filename (``core/imports/pipeline.py:645``).**
VA-collection source files like ``417 Fountains of Wayne - Stacys
Mom.flac`` have a leading playlist-position number that isn't the
album track number. The previous chain (album_info → filename →
floor-1) couldn't recover the real position because the filename
extractor either returned 417 (wrong) or None (caught by the floor).
But the wishlist payload's ``track_info.spotify_data.track_number``
HAD the right answer all along — Spotify says Stacy's Mom is track
3 on Welcome Interstate Managers.
Post-fix: resolution chain extracted into ``core/imports/track_number.py:resolve_track_number``
as a pure function:
1. ``album_info.track_number`` (album-bundle dispatch authoritative)
2. ``track_info.track_number`` (per-track flow payload)
3. ``track_info.spotify_data.track_number`` (nested fallback)
4. ``extract_explicit_track_number(file_path)`` (filename, returns
0 when no numeric prefix — vs the default helper that returns 1)
5. Caller (pipeline) applies the final >=1 floor
Each step coerces to a positive int or falls through to the next.
Pure function = unit-testable in isolation = single place to fix
the rule.
**Test coverage (37 new tests):**
- ``tests/wishlist/test_payloads.py`` (+4) — Track→dict conversion
preserves full album dict (dict / object / string album shapes) +
None-track-number stays None.
- ``tests/discovery/test_discovery_playlist.py`` (+2) — matched_data
always includes track_number/disc_number keys (None when unknown)
+ falls back to best_match attrs when cache misses.
- ``tests/imports/test_track_number_resolver.py`` (+16) — every
resolution-chain branch pinned: album_info-wins, track_info
fallback, spotify_data nested, JSON-string parsing, garbage-string
fall-through, zero / negative / non-numeric / string-numeric
coercion, filename fallback, explicit extractor vs default
extractor semantics, defensive None inputs, VA-collection
filename behaviour, all-sources-missing → None.
1571 wider-suite tests pass (wishlist + imports + discovery +
downloads + metadata). Ruff clean.
**Migration note:** existing wishlist rows that were saved under
the OLD ``track_object_to_dict`` (with stripped album metadata) still
have ``release_date=''`` in the DB blob. Those won't self-heal — the
next attempt loads from the poisoned blob. Users can remove + re-add
those tracks to refresh, or wait for the next sync run that
re-discovers them with full metadata. No automatic migration shipped
in this PR (scope creep — the forward path is fixed, backfill is a
separate concern).
PR 4 of 4 in the wishlist-album-bundle issue series. UI fix only —
zero behavior change.
User's 26-track wishlist run rendered all 26 sub-batches as
"Analyzing..." simultaneously. Pre-fix the rows were created with
``phase='analysis'`` BEFORE being submitted to ``missing_download_executor``
(max_workers=3 by default), so 23 batches sat in the executor queue
visually identical to the 3 actually running. Misled users into
thinking SoulSync was processing 26 in parallel; really only 3 ever
ran at once with the rest waiting their turn.
Fix:
- Wishlist auto-flow submission sites now create batch rows with
``phase='queued'``.
- The master worker (``core/downloads/master.py:328``) already flipped
phase to ``'analysis'`` as its first action on entry — that
transition becomes the real signal that the executor picked the
batch up.
- ``core/downloads/status.py`` surfaces ``analysis_progress`` for
the ``queued`` phase too so the UI has the track count to render
"Queued — N tracks" instead of an empty card.
- Frontend (``webui/static/pages-extra.js``, ``downloads.js``) renders
"Queued ⏳" for ``phase='queued'`` distinct from the spinner-laden
"Analyzing..." for ``phase='analysis'``.
Scope choices:
- Only the auto-wishlist submission sites flipped this PR
(``core/wishlist/processing.py:860`` album sub-batches +
``core/wishlist/processing.py:907`` residual). The manual-wishlist
sites at ``:451`` and ``:627`` use the same executor + worker, but
those create a caller-allocated batch_id that the frontend polls
immediately — wanted to verify the manual-poll path handles
``queued`` cleanly before flipping those. Trivial follow-up.
- Other submission sites in album_bundle_dispatch / web_server.py /
task_worker.py left untouched — they don't go through the
executor-queue pattern that causes this UI confusion.
Tests:
- Updated ``test_process_wishlist_automatically_creates_batch_for_matching_tracks``
to assert ``phase='queued'`` on creation (was ``'analysis'``); explanatory
comment names the executor-pool reason.
- New ``test_queued_phase_surfaces_analysis_progress_for_ui_count`` in
``tests/downloads/test_downloads_status.py`` pinning the new
``queued ⊂ analysis_progress`` rendering contract.
- 884 tests pass across wishlist + downloads + imports suites.
- Ruff clean on changed Python files; JS syntax OK on changed
webui files.
PR 3 (sibling-completion gate) was investigated and dropped — the
"1/26 finalized" symptom turns out to be downstream of the
staging-match bug (PR 2's instrumentation will catch it on the
user's next reproduction run), not an independent sibling-gate bug.
The gate logic itself is correct.
Two confirmed-from-code-reading bugs in the wishlist retry chain.
Both cause downstream post-process to render every retried file as
``01 - <title>`` without year in the folder path, even when the
source slskd file had the correct track number embedded and Spotify
had the album release date.
**Bug A — track_number defaults to 1 at every link in the chain.**
Pre-fix: ``.get('track_number', 1)`` defaulted at four sites:
- ``core/wishlist/payloads.py:121`` ``ensure_wishlist_track_format``
- ``core/wishlist/payloads.py:282`` Track-object conversion
- ``core/imports/context.py:421`` legacy album-info builder
- ``core/imports/pipeline.py:645`` final processing read
Each step "filled in" 1 when the upstream had dropped the key. The
downstream filename-extract fallback at ``pipeline.py:652`` ONLY
runs when the value is None — pre-filled 1 never matched, so the
fallback never fired, so the source filename's track number (e.g.
``08. No Sleep Till Brooklyn.flac``) was discarded in favour of the
default-1.
Fix: change every default from ``1`` to ``None`` along the chain.
The pipeline already has the right detect-and-recover logic — it
just needs the chain to stop poisoning it. Final ``< 1`` floor at
``pipeline.py:660`` still defaults to 1 as last resort, so callers
that genuinely have nothing still produce a valid number.
**Bug B — release_date dropped from cancelled-task wishlist payload.**
Pre-fix: ``build_cancelled_task_wishlist_payload`` only ``setdefault``ed
``name`` / ``album_type`` / ``images`` on the album dict. The
release_date field copy was load-bearing (when input was a dict, the
``dict(album_raw)`` copy preserved it), but when input was a bare
string the constructed dict had only name + album_type — no
release_date / total_tracks / etc.
Fix:
- Explicit comment on the dict-shape branch that release_date survives
via the unconditional ``dict(album_raw)`` copy + setdefault
semantics — so a future refactor that switches to a stricter copy
doesn't silently strip the field.
- String-shape branch now pulls release_date from
``track_info.album_release_date`` or ``track_info.release_date``
when present so the round-trip preserves the year for the path
template.
- track_data shape itself now carries ``track_number`` / ``disc_number``
at the top level (Bug A intersect — was dropping it entirely).
**Tests:** 4 new in tests/wishlist/test_payloads.py:
- ``test_ensure_wishlist_track_format_preserves_real_track_number``
- ``test_ensure_wishlist_track_format_keeps_missing_track_number_as_none``
- ``test_build_cancelled_task_wishlist_payload_preserves_track_number``
- ``test_build_cancelled_task_wishlist_payload_string_album_pulls_release_date_from_track_info``
14 payload tests pass; 879 across wishlist + imports + downloads
suites still green; 1410 wider suite all pass. Ruff clean.
Commits 2 + 3 of 3 in PR 2/4 of the wishlist-album-bundle issue fix
series. Commit 1 (94ba1d73) instrumented staging-match so the next
wishlist run produces the evidence we need to diagnose bug C
(staging-match silently drops album-bundle wishlist tracks); that
fix lands in a follow-up PR after the user's next reproduction run.
Pre-fix: ``try_staging_match`` silently returned False on three exit
points (empty cache, no track title, low best-score). Could not
diagnose the "track gets staged via album-bundle but never claimed
→ re-added to wishlist → infinite loop" bug from app.log because the
match-attempt + rejection was invisible.
Now every False exit logs at INFO with enough context to debug from
a single grep:
- ``[Staging] No match attempted for <track> — staging cache empty for batch <id>``
- ``[Staging] No match attempted for task <id> — track has empty title``
- ``[Staging] No match for <track> in batch <id> — best candidate <file> (title_sim=X, artist_sim=Y, combined=Z) below 0.75 threshold``
- ``[Staging] No match for <track> in batch <id> — N staging files but none had usable title variants``
Per-candidate skips (no title variants / title_sim < 0.80) log at
DEBUG so the noise stays out of INFO unless explicitly enabled.
Logs the near-miss candidate score on rejection so a 0.74 (one point
below threshold) surfaces as a different kind of bug than a 0.10
(completely wrong file in staging). Same shape SAB's adapter logs
now use for transient-vs-terminal status calls (PR #717).
Zero behavior change — pure logging. Enables the follow-up commit
that actually fixes the staging-match drop, by giving us real evidence
of WHERE the wishlist tracks are being rejected during the user's
next album-bundle run.
24 staging tests still pass; behavior unchanged.
Commit 1 of 3 in PR 2/4 of the wishlist-album-bundle issue fix
series. See ``memory/feedback_always_build_kettui_grade.md`` for
the instrument-before-blind-fix rule that drove this ordering.
Real-world wishlist case the original c3b88e69 design missed: user with
26 missing tracks from 26 different albums. Each item used to promote
to its own album-bundle sub-batch (``min_tracks_per_album=1``), which
downloaded the ENTIRE album (5-42 files) to claim one track. Confirmed
in app.log:
- "Licensed To Ill" downloaded 3 times across cycles (3-4 files each)
- "The Understanding" 17 files for 1 wishlist track
- "Alright, Still" 42 files for 1 wishlist track
- ~85% wasted bandwidth, slskd hammered with 26 concurrent searches
PR 1 of a 4-PR fix series — see commit body footer for the other PRs.
Default ``min_tracks_per_album`` 1 → 2. Single-track wishlist items
fall to ``residual_tracks`` → classic per-track batch (already works,
already efficient). Album-bundle kept for the case it was designed
for: user has 2+ tracks missing from the same album.
Override via the new ``wishlist.album_bundle_min_tracks`` config key:
- 1 = previous behaviour (bundle every item)
- 2 = new default
- 3+ = stricter, for users who want bundle only on bigger gaps
Helper ``_resolve_album_bundle_threshold`` lives in
``core/wishlist/processing.py``. Defensive shape mirrors the existing
config-driven knobs (``get_poll_interval`` / ``get_transient_miss_threshold``):
non-numeric, non-positive, or config-manager-raise all fall back to
the safe default. Three test cases pin the fallback chain.
Both wishlist entry points wired through the same helper:
- ``process_wishlist_automatically`` (auto cycle, line 812)
- ``start_manual_wishlist_download_batch`` (manual run, line 539)
Tests:
- ``tests/wishlist/test_album_grouping.py`` — old ``test_default_threshold_promotes_solo_albums`` flipped to ``test_default_threshold_demotes_solo_albums`` with explanatory docstring naming the real-world cause. New ``test_default_threshold_promotes_multi_track_albums`` pins the 2+ promotion. New ``test_explicit_threshold_one_restores_solo_promotion`` pins that the kwarg still works for opt-back-in.
- ``tests/wishlist/test_processing.py`` — 3 new tests for ``_resolve_album_bundle_threshold``: default-when-config-missing, honors-config-override, falls-back-on-garbage.
- ``tests/wishlist/test_automation.py`` — ``test_wishlist_albums_cycle_splits_into_per_album_batches`` updated to use 2+ tracks per album (5 tracks across 2 albums instead of 3 across 2 with 1 solo). ``test_wishlist_albums_cycle_residual_for_orphan_tracks`` updated to include 2 tracks from Album One so it still promotes.
- ``tests/wishlist/test_manual_download.py`` — same shape update for the manual path test.
- ``tests/wishlist/test_album_grouping.py:test_multiple_albums_emit_separate_groups`` updated to reflect new default (alb1 with 2 tracks promotes, alb2 with 1 track goes residual).
- ``tests/wishlist/test_album_grouping.py:test_nested_track_data_payloads_normalized`` pinned with explicit ``min_tracks_per_album=1`` so the test stays focused on payload-shape parsing, not the threshold rule.
114 wishlist tests pass; 866 across wishlist + automation + downloads +
album_bundle + album_bundle_dispatch suites still green. Ruff clean.
Sibling PRs queued in TaskCreate:
- PR 2 — investigate post-process staging-match miss (the second-order
bug that causes the same album to redownload every cycle when the
staging step doesn't claim the requested track).
- PR 3 — fix sibling-completion gate that fires on first sibling
instead of last (log evidence: run a4945c88 finalized 1/26 batches).
- PR 4 — UI distinguish Queued from Analyzing for batches waiting
on the executor (23/26 batches sit at "Analyzing..." while really
queued at max_workers=3).
PR 1 (commit 6ad85e27) shipped the ``next_run_at`` pure function as
foundation plumbing. PR 2 wires the engine through it and adds
``monthly_time`` as a real registered trigger type. After this PR
``core/automation_engine.py`` no longer has its own datetime
arithmetic for daily / weekly schedules — every next-run computation
flows through one function with one set of defensive fallbacks.
Net user-visible change: zero (no UI surface for monthly_time yet —
that's PR 3). New ``monthly_time`` trigger is reachable only via
direct API for now.
**Engine refactor:**
- ``_finish_run`` — collapsed three inline branches (daily_time
arithmetic, weekly_time arithmetic, fallback schedule arithmetic)
into a single ``next_run_at(...)`` call with ``_dt_to_db_str``
normalising the aware-UTC result to the engine's naive-UTC string
convention. Retry-delay short-circuit preserved. Exception
swallowing preserved (logged at debug, writes None next_run).
- ``_setup_daily_time_trigger`` + ``_setup_weekly_time_trigger`` +
new ``_setup_monthly_time_trigger`` — three near-identical methods
collapsed into one ``_setup_timed_trigger`` skeleton. Each public
method is now a one-line dispatch passing trigger_type to the
shared helper with a human-readable label for the debug log.
- Existing ``_next_weekly_occurrence`` deleted — its logic now lives
in ``core/automation/schedule.py:_next_weekly`` (lifted in PR 1).
- New ``_dt_to_db_str(dt)`` module-level helper normalises aware-UTC
→ naive-UTC string. Centralised so a tz mistake here surfaces in
one place. Aware non-UTC datetimes converted to UTC first
(defensive against a future bug that passes the wrong tz).
- New ``_resolve_system_default_tz()`` reads the server's local IANA
tz via ``tzlocal``. Cached at module import (the host's tz doesn't
change while the process runs). Falls back to UTC when ``tzlocal``
is missing — defensive for minimal Docker images.
- New ``self._default_tz`` engine attribute reads from
``automation.default_timezone`` config first, falls back to the
system-detected IANA name. Override path lets users on weird
setups pin a specific tz without touching env vars.
**Convergence fix (intentional behaviour change):**
Old ``_setup_daily_time_trigger`` / ``_setup_weekly_time_trigger``
didn't check the DB for an existing future ``next_run`` — they'd
recompute from scratch on every engine startup, overwriting manual
edits or pending retries. The interval path (``_setup_schedule_trigger``)
already had this check. The new shared ``_setup_timed_trigger``
brings daily / weekly in line: existing-future next_run wins over
freshly-computed delay. Treat this as a correctness fix, not a
breaking change — the old behaviour was an inconsistency, not a
deliberate choice.
**Backward-compat:**
- Existing ``schedule`` / ``daily_time`` / ``weekly_time`` rows
continue to work unchanged. The ``_trigger_handlers`` registry
keeps every historic key.
- Existing rows without an explicit ``tz`` field use
``self._default_tz`` (server-local IANA via ``tzlocal``) —
preserves "every Monday 09:00 server-local" behaviour on
non-UTC servers. Pre-fix the engine used naive
``datetime.now()`` which is also server-local; net effect is
identical wall-clock time, just routed through a tz-aware
pipeline that handles DST correctly (the May 2026 "next in 8h"
bug fix class).
- Engine boots even when ``tzlocal`` is missing — the resolver
falls back to UTC silently. Existing tests would catch a hard
dependency on tzlocal here.
**``tzlocal>=5.0`` added to requirements.txt** alongside
``tzdata>=2024.1`` from PR 1. Both libraries are small and stable;
``tzlocal`` returns a clean IANA name across Windows / Linux /
Docker, sidestepping the platform-specific tz detection mess.
**Tests:** 20 new in ``tests/automation/test_engine_schedule_integration.py``:
- ``_dt_to_db_str`` x3 (aware UTC, aware non-UTC converted to UTC,
naive assumed UTC)
- ``_resolve_system_default_tz`` x2 (returns IANA string, falls back
to UTC without tzlocal)
- ``_finish_run`` dispatch through next_run_at for each trigger type
(schedule, daily_time, weekly_time, monthly_time)
- Retry-delay short-circuits next_run_at
- next_run_at returns None → DB next_run cleared
- next_run_at raises → engine swallows + writes None
- Event triggers skipped (no scheduled next-run)
- ``self._default_tz`` passed through to next_run_at
- monthly_time registered in _trigger_handlers
- All historic trigger types kept registered
- ``_setup_monthly_time_trigger`` arms timer + writes DB
- ``_setup_timed_trigger`` honours existing future DB next_run
- Skip-with-log when next_run_at returns None
- End-to-end no-mock smoke for monthly_time
260 automation suite tests pass; the 240 from PR 1's branch plus 20
new integration tests. Ruff clean.
No WHATS_NEW entry — UI doesn't expose monthly_time yet (PR 3),
and the backward-compat path preserves existing daily/weekly
schedule timing.
Self-review pass on ec4a55c1 — applying the standing kettui-grade
rule (see memory/feedback_always_build_kettui_grade.md). Three issues
that would have surfaced on review:
1. Silent tz fallback to UTC
``_resolve_tz`` returned UTC when the IANA name was unknown — no
log, no warning. User on a host without ``tzdata`` who configures
``America/Los_Angeles`` got schedules running silently at UTC
offset with no way to debug. Now logs WARNING once per unknown
name (deduped via ``_UNKNOWN_TZ_WARNED`` set so a misconfigured
row doesn't spam every poll cycle) and the log line names BOTH
real causes — typo or missing tzdata — so the user can fix from
a single grep.
2. ``weeks`` unit drift from engine
I added ``'weeks': 86400*7`` to ``_INTERVAL_MULTIPLIERS`` but the
engine's existing ``_calc_delay_seconds`` only recognises
minutes/hours/days. Until PR 2 collapses both paths through this
function, any row whose config snuck through with ``unit='weeks'``
would get scheduled by the engine as 1-hour and by this function
as 7-day — drift between two live implementations. Dropped
``weeks`` from the map to match the engine. Added a comment
pinning the map to the engine's contract and a regression test
that asserts ``unit='weeks'`` falls back to the same hours
default the engine produces.
3. DST edge cases unverified
The module docstring claims DST-aware via ``zoneinfo`` but no test
pinned the spring-forward gap (02:30 LA on DST-Sunday doesn't
exist) or fall-back ambiguity (01:30 LA on fall-Sunday happens
twice). Three new tests:
- ``test_dst_spring_forward_lands_after_the_gap`` — pins that the
function doesn't crash + lands on a real instant past ``now``.
- ``test_dst_fall_back_handles_ambiguous_local_time`` — pins
zoneinfo's default-earlier-instant resolution for ambiguous
local times (01:30 PDT vs 01:30 PST → picks PDT).
- ``test_weekly_across_dst_boundary_keeps_local_wall_clock`` —
pins that a "every Sunday at 09:00 LA" schedule keeps the
local wall clock across the boundary even though the UTC
equivalent shifts by an hour. This is the exact bug class
that caused the May 2026 "next in 8h" tz mismatch.
Also loosened ``tzdata==2026.2`` to ``tzdata>=2024.1``. IANA tz data
changes a few times a year for real-world DST policy updates; pinning
to one snapshot would freeze the app's tz knowledge to the build date
and miss future government-mandated rule changes.
41 schedule tests pass (5 new); 240 across the full automation suite.
Ruff clean.
Backend plumbing for upcoming weekly + monthly Auto-Sync schedules.
PR 1 of 4 in the schedule-types feature — see
``memory/project_auto_sync_schedule_types.md`` for the full plan.
Net behaviour change in this PR: zero. The automation engine still
computes next_run via its existing inline ``_calc_delay_seconds`` /
``_next_weekly_occurrence`` helpers; this module is unused until PR 2
wires the engine through. Lands separately so the foundation can sit
on dev for a beat before the engine change.
``core/automation/schedule.py:next_run_at(trigger_type, trigger_config,
now_utc, default_tz)``:
- Pure function. ``now_utc`` injected (tests freeze time without
monkeypatching ``datetime.now``); ``default_tz`` injected (so daily /
weekly / monthly schedules compute against the USER's timezone, not
the server's — the same class of bug that produced the May 2026
"Auto-Sync next in 8h" timezone fix).
- Returns aware-UTC ``datetime`` ready to serialise to the DB
``next_run`` column, or ``None`` for unrecognised / event-based
triggers (callers should not write a next_run for those).
- Naive ``now_utc`` inputs are assumed UTC for defensive symmetry
with the engine's DB-string parser convention.
Trigger types covered:
- ``schedule``: ``{interval: N, unit: 'minutes'|'hours'|'days'|'weeks'}``
— matches engine's existing ``_calc_delay_seconds``. Unknown unit
defaults to hours; zero/negative interval clamps to 1 (preserves
the engine's guard against scheduling for the past); non-numeric
interval falls back to 1.
- ``daily_time``: ``{time: 'HH:MM', tz: '<IANA>'}`` — DST-aware via
``zoneinfo``; ``tz`` falls back to ``default_tz``; unknown IANA
string falls back to UTC; garbage ``time`` falls back to 00:00.
- ``weekly_time``: ``{time, days: ['mon',...], tz}`` — empty / all-
invalid ``days`` list means "every day" (matches engine fallback);
abbreviations case-insensitive; 8-day scan finds the next match.
- ``monthly_time``: ``{time, day_of_month: 1-31, tz}`` — NEW shape.
Day clamped to [1, 31]. Months too short for the target day clamp
to the LAST valid day rather than skipping a month (standard cron
convention; running a day early in February is less surprising
than missing the whole month). 12-iteration loop cap so a
pathological config can't infinite-loop.
Tests (36 cases, all passing):
- Interval: every unit, unknown-unit fallback, zero/negative/garbage
interval clamp, tz field ignored on interval (wall-clock-independent).
- Daily: today-at-future-time runs today, today-at-past-time rolls to
tomorrow, exact-match rolls to tomorrow (no schedule-now-then-schedule-
again-immediately), user-tz vs server-tz, default_tz fallback,
garbage time / unknown tz defensive returns.
- Weekly: same-day-still-future qualifies, same-day-past rolls to next
allowed day, wraps across week boundary, empty days = every day,
garbage abbreviations dropped, case-insensitive, tz across day
boundary (LA Wednesday evening is Thursday UTC).
- Monthly: target day this month, rolls to next month when passed,
Feb 31 → Feb 28 / Feb 29 leap year, day_of_month above 31 / below
1 clamp, Dec → Jan year roll, user-tz pre-midnight edge case.
- Result-shape contract: every returned datetime is aware UTC at
offset zero (engine relies on this when serialising to the
``next_run`` string column).
Added ``tzdata==2026.2`` to requirements.txt. Windows ``zoneinfo`` and
minimal Docker base images ship without the system tz database;
without ``tzdata`` ``ZoneInfo('America/Los_Angeles')`` raises
``ZoneInfoNotFoundError`` and the helper silently falls back to UTC.
No WHATS_NEW entry — no user-visible behaviour change in this PR.
PR 2 (engine wire-through) will land the user-facing changelog entry
when ``monthly_time`` becomes a real schedulable trigger.
Follow-up to f13d3395. Five gaps called out on self-review:
1. Per-track inline transient tolerance was duplicated between
usenet.py and torrent.py (~12 lines each, identical) and wasn't
directly tested. Extracted into ``TransientMissCounter`` in
``album_bundle.py`` — small class with ``record_miss()`` returning
True at threshold and ``reset()`` for successful reads. Both
per-track flows AND the lifted ``poll_album_download`` now use
the same counter, so the rule is in one place.
2. Threshold is now config-driven via
``download_source.album_bundle_transient_miss_threshold``
(default 5). Same defensive pattern as ``get_poll_interval`` /
``get_poll_timeout`` — non-positive / non-numeric falls back to
the default. Users with very slow servers (huge multi-disc box
sets, slow disks) can extend the tolerance window without
touching code.
3. SAB state map verified against the canonical Status enum in
``sabnzbd/constants.py`` (sabnzbd/constants.py:~95-118). Dropped
six entries I'd guessed at and couldn't verify in source
(``trying``, ``prop_paused``, ``prop_failed``, ``unpacking``,
``pp``, ``postprocessing``). Kept the verified ``deleted`` (lower-
cased from SAB's ``Deleted``) and added the one real state I'd
missed: ``Propagating`` (SAB's pre-download delay state — maps to
``queued`` since we're waiting on the NZB to be available, not
actively downloading).
4. SAB integration test exercising the queue→history gap end-to-end
through the real adapter HTTP layer. Mocks SAB's queue + history
endpoints with the exact response shapes SAB emits, runs three
gap polls (both endpoints empty), then a recovery poll where the
slot appears in history as Completed. Confirms the TransientMissCounter
absorbs the gap and ``poll_album_download`` returns the save_path
without emitting terminal failure. This was the path I had only
tested at the helper layer before — now pinned end-to-end through
the adapter.
5. SAB state mapping has new tests: every Status value from SAB's
canonical enum must map to a known adapter state (not the 'error'
default fallback), Propagating routes to queued, Deleted routes
to failed. Future SAB state additions that we miss will surface
as 'error' default → transient-miss tolerance → terminal failure
with a clear log line, but the explicit assertion list here means
we'll catch the omission in CI before users do.
Test count after: 537 download-suite tests pass; 21 new
(``TransientMissCounter`` ×4, ``get_transient_miss_threshold`` ×3,
SAB state-coverage ×3, SAB direct ``nzo_ids`` lookup ×5, SAB
queue→history integration ×1, plus the existing helper-layer
coverage from the parent commit). Ruff clean.
User reported usenet album downloads getting stuck on "downloading
release" while SABnzbd reported the job as complete. Container restart
did not help; reproducible on every usenet album download.
Three independent issues all causing the same symptom — the download
modal freezes mid-flow with no error surfaced to the user:
1. SAB queue → history transition window
SAB removes a slot from its queue BEFORE adding it to the history,
and on a busy server (par2 verify, unrar, multi-file move) that
window can span several poll iterations. The poll treated a single
None status as terminal failure ("disappeared from client") and
gave up. Now the poll tolerates up to ~10s of consecutive misses
(5 polls at the default 2s interval) before declaring the job gone.
2. SAB queue states like `Pp` were unmapped
`_SAB_QUEUE_STATE_MAP` didn't cover SAB's `Pp` (post-processing
summary), `Unpacking`, `Trying`, `Deleted`, or the `Prop_paused`
/ `Prop_failed` variants. Unmapped states fell through to the
default-'error' fallback, and the poll loop only treated explicit
'failed' / 'completed' as terminal — 'error' was neither, so the
loop spun until the 6-hour timeout. Map now covers every Status
value from SAB's `sabnzbd/api.py`, and the poll treats the default-
'error' fallback as a transient miss (warn-logged, retry within
the same tolerance window) so a brand-new unmapped state can't
infinite-loop the way `Pp` did here.
3. No terminal failure emit
The poll only logged on failure / timeout / disappeared — never
called the progress callback with 'failed', so the download modal
stayed at the last 'downloading' emit forever. Plumb a 'failed'
emit through every failure exit path so the UI flips out of the
downloading state when the poll gives up.
Plus:
4. SAB direct nzo_ids lookup instead of paging all-history
`_get_status_sync` was fetching the latest 50 history entries on
every poll and iterating to find the target nzo_id. On busy
servers (many recent downloads), the target job could roll past
the 50-entry window and look like a "disappeared" job. Replaced
with a targeted `mode=queue&nzo_ids=<id>` → `mode=history&nzo_ids=<id>`
chain. Falls back to the bulk path for SAB versions that pre-date
the nzo_ids filter — the transient-miss tolerance covers any
short-lived gap there too.
Implementation:
Lifted the album-bundle poll loop out of `usenet.py` and `torrent.py`
into `core/download_plugins/album_bundle.py:poll_album_download` —
near-duplicate implementations are now a single function with deps
injected so it's testable in isolation (kettui's extract-don't-AST-parse
standard; can't unit-test a `time.sleep` loop inside a plugin method).
The lifted helper takes:
- `get_status` callable bound to job_id, so the same loop works for
usenet UsenetStatus and torrent TorrentStatus shapes
- `complete_states` set so torrent's `{'seeding', 'completed'}` and
usenet's `{'completed'}` both Just Work
- `failed_states` set so torrent's `{'error'}` is terminal while
usenet's default-'error' fallback is transient
- `transient_miss_threshold` (default 5 ≈ 10s at 2s poll)
- `sleep` / `monotonic` injectables for deterministic tests
Per-track flows in both plugins gained the same transient-miss
tolerance inline — they don't use the emit pattern (update an
`active_downloads[id]` row dict via lock instead), so reusing the
helper would have required threading a no-op emit through. Inline
fix is small enough.
Tests:
- 11 new tests in `tests/test_album_bundle.py:poll_album_download`
cover the happy path, transient-miss tolerance with recovery,
hard-failure threshold, explicit-failed surface, timeout-emit,
default-'error' transient treatment, shutdown clean exit,
torrent's `seeding`-counts-as-complete, save_path captured across
iterations, and adapter-exception treated as transient miss.
- 521 download-suite tests pass (33 in test_album_bundle, others
pin existing torrent + usenet contracts).
- Ruff clean.
Closes#706.
Discogs uses two disambiguation conventions for duplicate artist names:
- legacy `(N)` numeric suffix: "Bullet (2)", "Madonna (3)"
- newer `*` asterisk suffix: "John Smith*", "Foo*"
Both were leaking through to the UI on artist search and album search,
and worse — through the import path into folder names on disk
(reported: importing yielded folders literally named `Foo*`).
The pre-existing cleanup only handled `(N)` and only at ONE site —
`get_user_collection` (line 469) and one path inside
`extract_track_from_release` (line 448 — `re.sub(r'\s*\(\d+\)$', '',
artist_name)`). Every other surface (artist search, album search,
album-track lookups, get_artist_albums feature matching) returned the
raw Discogs string.
Centralized into `_clean_discogs_artist_name(name)` at module top,
with regex covering both suffixes including repeated forms (`Baz**`,
`Foo (3)*`). Applied at six sites:
- `Artist.from_discogs_artist` (artist search)
- `Album.from_discogs_release` (album search — three fallbacks: array,
string, title-split)
- `Track.from_discogs_track` (track lookup — track-level + release-level
fallback)
- `extract_track_from_release` (replaces the inline `(N)`-only re.sub)
- `get_user_collection` (existing site, now also strips `*`)
- `get_artist_albums` (artist_name used for primary-vs-feature matching;
cleaning prevents `Beyoncé*` from failing equality vs `Beyoncé`)
- `get_album` (artists_list + per-track artists in the tracklist projection)
Tests:
- New `test_clean_discogs_artist_name` parametrized over 14 cases
covering `(N)`, `*`, repeated `**`, combined `(N) *`, whitespace
handling, empty/None defensive returns.
- New `test_get_user_collection_strips_discogs_asterisk_disambiguation`
pinning the asterisk path end-to-end through the collection import
flow (sibling to the existing `(N)` test).
- Existing 37 discogs tests still pass.
Out of scope (separate issue): the same #634 report flagged track-count
and year fields rendering as 0 / empty in Discogs album search. Both
are inherent to Discogs `/database/search` response shape — search
results don't carry `tracklist` (only release detail does) and `year`
is often `0` in search payloads. Fixing requires lazy-fetching release
detail per row, which hits the 25 req/min unauth limit hard. Not
bundled here.
Five pre-existing lint errors on dev baseline (all introduced May 25-26
before this branch was cut) were blocking CI on this PR. Cleared as
courtesy fixes so the merge isn't gated on unrelated tech debt:
- web_server.py:22613 — F811 duplicate `urlparse` import inside
`_parse_itunes_link_url` (already imported at module top, line 20).
Removed from the inline `from urllib.parse import parse_qs, urlparse`;
kept `parse_qs` since that one is only used here.
- core/listenbrainz_manager.py:746 — S110 silenced with
`# noqa: S110 — best-effort lookup, delete proceeds either way`.
Matches the existing project convention used in web_server.py:1693,
core/watchlist/auto_scan.py:463, core/library_reorganize.py:548.
- core/playlists/sources/listenbrainz.py:236 — B905 `zip()` without
explicit `strict=`. Added `strict=False` — preserves existing
behaviour where `matched` can legitimately be shorter than
`match_indices` on partial discover failure.
- core/playlists/sources/listenbrainz.py:273 — S110 silenced with
`# noqa: S110 — caller falls back to last cached playlist on
refresh failure`.
- core/playlists/sources/soulsync_discovery.py:105 — S110 silenced
with `# noqa: S110 — manager persists last_generation_error on
failure; surface existing snapshot`. The existing multi-line
comment that already explained the swallow was rolled into the
noqa justification so the rule + reason live on one line.
Ruff `python -m ruff check .` now passes; 664 discovery + metadata
tests still pass.
Live smoke against `/api/musicbrainz/search_tracks?track=Coffee+Break&artist=Zeds+Dead`
exposed the edge case the tiebreaker implementation couldn't reach:
The canonical Zeds Dead "Coffee Break" recording (mbid 6e2d4a70, length
184000ms) lives on the Coffee Break Single release — album_type='single',
which carries a 0.85 album_type_weight in `score_track`. A sibling
length-less recording (mbid 3b89bf3c) lives on an Album release —
album_type='album', weight 1.0. After multiplying by EXACT_ARTIST_BOOST
the canonical sat at 1.275 while the length-less sibling sat at 1.5.
The previous tiebreaker only kicked in on equal scores, so the
length-less album edition wins and the user sees 0:00 first instead of
the actionable 3:04 row. Bug reproduced: ordering came out
length-less / canonical / Omar-LinX-collab.
Switched `prefer_known_duration` to a 1.25x score boost on recordings
with non-zero duration_ms. The multiplier is sized above the
album-vs-single weight spread (0.176) so length-known recordings can
overcome an album-type penalty when scores would otherwise tie on
title + artist match, but stays small enough that cover/karaoke
penalty (0.05) and variant-tag penalty (0.85) still dominate — a
length-known tribute still loses to a length-less canonical.
Post-fix live response: 6e2d4a70 (canonical, 184000ms) sits first,
8ec2ce3f (Zeds Dead + Omar LinX collab, 153000ms) second, 3b89bf3c
(length-less album edition) third.
Verified Björk diacritic fallback path unaffected — `Bjork` + `Army of
Me` still cascades strict-empty → bare and returns all 10 Björk
recordings.
122 metadata tests pass — the three `prefer_known_duration` cases were
designed to pin behaviour, not the specific multiplier value, so they
all still pass under the boost implementation: ties promote
length-known, relevance still beats length-pref, default-off behaviour
unchanged.
Self-review pass on the prior three commits — kettui-style cleanup
that should have landed first time.
**Length-preference sort ordering (real bug):**
The `search_tracks_with_artist` stable sort that promoted length-known
recordings ran in `core/musicbrainz_search.py`, but the MB endpoint in
`web_server.py:search_musicbrainz_tracks` runs `rerank_tracks` after
it — which re-sorts by relevance score and dropped the length-pref
ordering down to tiebreaker-only. For canonical-same-song MB duplicates
that all score identically the tiebreaker survived, but the
order-of-operations was wrong.
Moved into `rerank_tracks` itself via a new `prefer_known_duration`
flag. Sort key sits between relevance score and the stable-order
tiebreaker so relevance still wins (length only decides ties, never
overrides a higher-relevance match). The MB endpoint opts in via
`prefer_known_duration=True`; Spotify / iTunes / Deezer callers stay
on the default-off path since their search results always include
length. Pinned with three new `TestRerankTracks` cases:
ties-promote-length, relevance-still-wins, default-off-unchanged.
**Route logic lifted to `core/discovery/manual_match.py`:**
Two pieces lived as inline route logic in `web_server.py` — the
`derive_manual_match_provider` fallback chain (payload.source →
active source → 'spotify') used by `update_youtube_discovery_match`,
and the `is_drifted_for_redo` predicate (cached provider differs from
active AND not manual_match) used by `prepare_mirrored_discovery`.
Per kettui's "extract logic from web_server.py, don't AST-parse it"
standard, both helpers now live in `core/discovery/manual_match.py`
with 12 dedicated unit tests covering fallback resolution order,
non-dict payload defenses, manual_match exemption from drift,
absent-provider legacy default, and edge cases.
Side benefits from the lift:
- `match_source` now derived once before the cache-save try block
instead of being duplicated in try + except (the except block existed
only because the original used `match_source` later — pre-computing
killed the duplication).
- `prepare_mirrored_discovery`'s `has_cached` check now reuses
`is_drifted_for_redo` with inverted polarity instead of restating
the field whitelist inline, so a future schema change only has to
land in one place.
- The mirrored-DB persist block now gates on `matched_data is not None`
to avoid a pre-existing latent NameError if the cache-save block
raised before matched_data construction.
**Enhanced toggle localStorage key now profile-scoped:**
`soulsync-library-view-mode` was global — two admin profiles would
share one preference. Wrapped in `_libraryViewModeKey()` which appends
`:${currentProfile.id}` when a profile is loaded, falls back to the
unsuffixed key otherwise (preserves pre-multi-profile saved values).
Tests:
- 12 new in `tests/discovery/test_manual_match.py` pinning both helpers.
- 3 new in `tests/metadata/test_relevance.py` pinning the
`prefer_known_duration` semantics.
- `test_search_tracks_with_artist_prefers_results_with_known_length`
renamed to `_does_not_resort_by_length` since the sort moved out of
this method. 664 tests pass across discovery + metadata suites.
User reported that manually mapping a mirrored-playlist track via the
Fix popup (either by search or by pasting an MBID) worked end-to-end
once — match saved, library track downloaded — but the next Playlist
Pipeline run flipped the track back to "Provider Changed" and forced
them to re-do the manual map every cycle.
Three independent issues were combining to cause this:
1. Hardcoded `provider: 'spotify'` on manual-fix save
`update_youtube_discovery_match` (the endpoint the Fix popup posts
to, also used by mirrored playlists since the frontend routes
`platform === 'mirrored'` through the YouTube endpoint) always
stamped the cached match as Spotify-provided. The Fix-popup cascade
actually queries the user's primary metadata source first and falls
back to Spotify / Deezer / iTunes / MusicBrainz — so a user on
MusicBrainz primary picking an MB result still had it saved as
`provider: 'spotify'`. The next prepare-discovery call (which
compares cached_provider to the active source) then immediately
classified the match as drifted and pending re-discovery. Fixed by
deriving `match_source` from `spotify_track.get('source')` (every
*_search_tracks endpoint stamps `source` on results) with a fallback
to `_get_active_discovery_source()` for the MBID-paste path (which
uses the lean flat shape that doesn't carry source). `matched_data['source']`
and the mirrored `extra_data['provider']` both now use the derived
value. `match_source` is also recomputed in the cache-save except
handler so the downstream mirrored-DB save still has it.
2. Discovery worker re-queueing manual matches as "incomplete"
`run_playlist_discovery_worker` in `core/discovery/playlist.py`
re-adds any track to `undiscovered_tracks` when its `matched_data`
lacks `track_number` or `album.id` / `album.release_date`. The
check was designed as a legacy-fix backfill for old discoveries
that lost those fields to a Track-dataclass stripping bug. But
manual fixes from the popup are *intentionally* lean — search-
result rows don't include `track_number` (none of the search
endpoints return it), and the MBID-lookup flat shape doesn't
carry `album.id` / `release_date` (the recording lookup returns
only `album.name`). So every manual match looked "incomplete" and
got re-discovered every pipeline run, overwriting the user's pick
with whatever the auto-search ranked first. Manual matches now
short-circuit ahead of the incomplete-data branch.
3. `prepare_mirrored_discovery` ignored the `manual_match` flag
Independent of the provider-stamping fix above, the prepare-
discovery endpoint that powers the mirrored-playlist UI did its
own `cached_provider != current_provider` check and didn't honour
manual_match either. Defence in depth — even if a future code
path stamps the wrong provider on a manual match, the flag now
anchors it as cached. `has_cached` also extended so manual
matches with off-provider stamps still count toward the cached
tally for phase classification.
Tests:
- new `test_manual_match_skipped_even_when_matched_data_incomplete`
in `tests/discovery/test_discovery_playlist.py` pins the worker
short-circuit using a realistic MB-shape matched_data (album dict
without id / release_date, no top-level track_number). 16 existing
tests still green; 848 across discovery / metadata / automation
suites pass.
`/api/musicbrainz/search_tracks` powers the Fix popup's auto-search
cascade for users on MusicBrainz as primary. When both track + artist
fields were filled, `search_tracks_with_artist` always took the bare
keyword path (`<track> <artist>` joined as one query string). MB's
recording-search scorer weights title matches far above artist matches,
so for "Coffee Break" + "Zeds Dead" the top results were Emapea / The
Vidalias / West One Orchestra's "Coffee Break" — three unrelated cover-
title collisions ahead of the canonical Zeds Dead recording. The
endpoint's `rerank_tracks` pass can't fix this when the right answer
is below the API's 50-result cutoff.
Both-fields mode now uses a strict field-scoped Lucene query first
(`recording:"<t>" AND artist:"<a>"`) which anchors the artist and
prunes title-collision covers at the source. `min_score=0` because the
field-scoped query is itself precise; rerank still does final ordering.
Bare query stays as the fallback when strict returns nothing — covers
the diacritic / alias cases the original `strict=False` path was added
for ("Bjork" query vs canonical "Björk" artist where Lucene phrase
match never hits the recording).
Single-field mode (track-only or artist-only) is unchanged: still bare-
query directly, since there's no artist value to anchor.
Also stable-sort results to prefer entries with non-zero `duration_ms`.
MB has multiple recordings per song (single release, album release,
remasters, compilations) and not every recording carries length data.
Without the preference sort, the user sees a 0:00 row first while a
sibling recording with the real 3:04 sits two rows below — matches the
report where MBID-paste lookup of the canonical recording (length 3:04)
contradicted the search-result's 0:00 row for the same song.
Tests:
- new `test_search_tracks_with_artist_strict_first_when_both_fields`
pins the strict=True call when both fields present
- new `test_search_tracks_with_artist_falls_back_to_bare_when_strict_empty`
pins the Björk-style fall-through path
- new `test_search_tracks_with_artist_prefers_results_with_known_length`
pins the length-preference sort
- existing `..._keeps_low_score_for_rerank` updated to side_effect so
the bare-fallback path is exercised; behaviour pinned identically
- existing `..._uses_bare_query_mode` renamed + repurposed for strict-
first; old name's behaviour no longer accurate
The sibling-merge aggregator from 7f751202 used "least-complete
phase wins", which made the modal appear frozen during parallel
album bundle downloads. The task table is phase-gated to
downloading/complete/error in downloads.js — so whenever any
sibling was still in album_downloading, the merged phase stayed
there and tasks for the sibling that had advanced past its bundle
never rendered. User reported: both albums downloading on slskd,
modal blank until one completes fully.
Flip the rule: surface the most-advanced live phase so the modal
renders task progress as soon as any sibling reaches it. The
all-siblings-in-album_downloading case still surfaces
album_downloading (bundle progress UI is correct there); error
stays sticky.
Updated WHATS_NEW under 2.6.3 to describe the corrected behavior.
Two new tests pin the regression:
- downloading + album_downloading → downloading
- album_downloading + album_downloading → album_downloading
Phase 1c.2.1 splits each wishlist run across multiple
``download_batches`` rows (per-album bundle dispatch). The
download-missing modal opens against the original batch_id
allocated by ``start_manual_wishlist_download_batch`` /
``process_wishlist_automatically``. Pre-fix that batch_id was
just one sibling among N, so the modal went stale as soon as the
primary sub-batch finished — subsequent albums downloaded fine
but no live status reached the UI.
Fix: backend merges every sibling sub-batch's tasks +
analysis_results into the response keyed under the originally-
requested batch_id. Modal sees one unified view of the whole run
without knowing about the split. Frontend untouched.
Architecture (Kettui standards):
- ``core/downloads/wishlist_aggregator.py`` — pure
``merge_wishlist_run_status(primary, siblings)`` helper.
No IO, no runtime state, no globals. Lifted out of
``status.py`` so the merge contract can be pinned via unit
tests without standing up the live ``download_batches`` /
``download_tasks`` state.
- ``core/downloads/status.py``'s ``build_batched_status`` now
pre-indexes ``download_batches`` by ``wishlist_run_id`` inside
the existing ``tasks_lock`` snapshot, then runs the merge
helper whenever a requested batch has a sibling.
Merge rules pinned by 12 tests:
- ``track_index`` re-indexed globally 0..N-1 across the merged
``analysis_results`` so the modal's ``data-track-index`` DOM
keys don't collide between siblings. Tasks' ``track_index``
follows the same remap so the analysis-results ↔ tasks
cross-reference stays intact.
- ``task_id`` is uuid per task — no collision concern.
- Phase: error is sticky; otherwise the LEAST-complete
pre-terminal phase wins (analysis < album_downloading <
downloading). All-complete returns ``complete``; mixed
complete + active returns ``downloading`` so the modal stays
alive until every sibling lands.
- ``album_bundle``: picks whichever sibling currently has an
active bundle download (state in
``{searching, downloading, downloading_release, staging}``).
Falls back to the first non-empty bundle so a completed run
still shows a progress bar.
- ``analysis_progress`` summed across siblings.
- ``active_count`` summed; ``max_concurrent`` keeps primary's
value as the representative.
- ``playlist_id`` + ``playlist_name`` preserved from the primary
(the row the modal originally opened against).
Legacy single-batch wishlist runs (no ``wishlist_run_id`` on the
batch) skip the merge entirely — passthrough. Back-compat by
absence.
1108 tests across downloads + wishlist + automation + imports +
playlist-sources + lb-series suites green. 12 new aggregator
tests pin the merge contract.
Closes the open UX gap from the Phase 1c.2.1 ship — modal now
tracks every sibling sub-batch's progress for the full duration
of the wishlist run.
Phase 1c.2.1 splits each wishlist invocation into per-album sub-
batches so the album-bundle dispatch can engage once per album.
Side effect: the completion handler ``finalize_auto_wishlist_completion``
ran end-of-run logic (cycle toggle + state reset + automation
event emit) once per BATCH, so a 2-album run fired the cycle
toggle twice + emitted two ``wishlist_processing_completed``
events. The cycle landed at the right value either way but the
state machine had become per-batch instead of per-run.
Fix: reify "wishlist run" as a first-class concept via a shared
``wishlist_run_id`` UUID. Generated once per wishlist invocation
in both the auto- and manual-wishlist paths, stamped on every
sub-batch row in ``download_batches``.
``finalize_auto_wishlist_completion`` now reads the completing
batch's ``wishlist_run_id`` and, when present, scans
``download_batches`` for siblings still in pre-terminal phases.
If any sibling is still active, the per-batch summary records
but the cycle toggle + state reset + automation emit are
deferred. Only the last completing sibling fires the run-level
finalization. Legacy single-batch runs (no run_id field) keep
their toggle-immediately behavior — back-compat by absence.
The run_id also lays groundwork for frontend grouping (one
logical row in the Downloads view per wishlist run instead of N
sibling rows), but that UX work is deferred.
3 new tests in ``test_processing.py`` pin: defer-when-siblings-
active, toggle-when-last-sibling-done, back-compat-without-run_id.
1 new assertion in ``test_automation.py`` confirms all sub-batches
of one auto-wishlist invocation share the same run_id. 309 tests
across wishlist + automation suites green.
Notes: dispatch concurrency unchanged — sub-batches still run via
the shared download worker pool. Slskd serializes per-uploader at
its own layer (same uploader = automatic queue, different
uploaders = legit parallel), so SoulSync-side serial enforcement
would duplicate work the right layer already handles.
The Phase-1 fix (commit c3b88e69) only extended the per-album
bundle dispatch to ``process_wishlist_automatically``. The manual
"Run Wishlist Now" path goes through
``_prepare_and_run_manual_wishlist_batch`` instead, so the
behavior didn't change for users who triggered downloads from the
Wishlist tab UI — they still saw N per-track Soulseek searches
when N missing tracks all came from one album.
Caught in a real-app test: user added Katy Perry's PRISM (Deluxe)
to the wishlist + clicked "Download Wishlist" → app log shows
``_prepare_and_run_manual_wishlist_batch:421`` running a single
batch with 16 tracks + per-track searches firing one by one
("katy perry prism deluxe legendary lovers", "katy perry prism
deluxe roar", etc.), no album-bundle dispatch.
Fix:
- ``_prepare_and_run_manual_wishlist_batch`` now runs the same
``group_wishlist_tracks_by_album`` helper after filtering. For
each detected album, it builds a sub-batch with
``is_album_download=True`` + populated album/artist context.
Residual tracks (no resolvable album metadata) land in a single
per-track residual batch.
- The first sub-batch re-uses the caller-allocated ``batch_id``
so the frontend's existing poll against it keeps working;
additional sub-batches get fresh ids materialized into
``download_batches`` so they show up in the Downloads view.
- Sub-batches dispatch serially — each ``run_full_missing_tracks_process``
call blocks until the album-bundle staging + per-track tasks
complete before the next album's bundle search fires.
New test ``test_manual_wishlist_splits_into_per_album_sub_batches``
pins the contract — multi-album wishlist content with
nested-spotify_data shape produces N master-worker calls (one per
album), each batch carries the album_context, first sub-batch
re-uses the original batch_id. 106 wishlist tests + 1099 across
the broader suite green.
Adding 16 Katy Perry PRISM tracks to wishlist + clicking download
should now fire ONE slskd album-bundle search for the release
instead of 16 individual searches.
Auto-wishlist's "albums" cycle used to dump every missing album
track into one batch and run per-track Soulseek / Prowlarr searches
for each (~50 searches for a typical scan). The album-bundle
dispatch (introduced in 2.5.9 for explicit album downloads) was
gated on ``is_album_download=True`` + populated
``album_context``/``artist_context``, none of which the wishlist
batch ever set — so wishlist runs always took the per-track flow
even when 12 missing tracks all belonged to the same album.
Fix: split wishlist albums-cycle tracks into per-album sub-batches
at submission time. Each sub-batch carries its own album context,
trips the existing dispatch gate, and engages one slskd / torrent
/ usenet album-bundle search per album. Tracks the helper can't
group (no album metadata, no artist) fall through to a residual
per-track batch.
- New ``core/wishlist/album_grouping.py``:
``group_wishlist_tracks_by_album(tracks)`` returns
``WishlistGroupingResult(album_groups, residual_tracks)``.
Pure function — extracts album_id (or name-normalized fallback)
+ primary artist + album context from each track's nested
spotify_data, buckets, and threshold-promotes. Independent of
runtime state so it can be unit-tested without the wishlist
executor.
- ``core/wishlist/processing.py``: when ``current_cycle ==
'albums'``, run the grouping helper, submit one batch per album
with ``is_album_download=True`` + the group's album/artist
context, then a single residual batch for orphans. Singles
cycle path unchanged.
- 9 new tests in ``test_album_grouping.py`` pin the bucketing
contract (empty / single album / multi album / orphan / threshold
/ nested payloads / no-id fallback / no artist).
- 2 new tests in ``test_automation.py`` exercise the per-album
split end-to-end through ``process_wishlist_automatically``:
multi-album batch → two sub-batches each with album context;
mixed orphan + real album → one bundle batch + one residual.
1099 tests across wishlist + imports + downloads + automation +
playlist-sources + staging-provenance + track-number-repair
suites green. WHATS_NEW entry added under 2.6.3.
Now when an auto-wishlist scan finds 12 missing tracks from
Ryoto's "Cha-La Head-Cha-La", it runs ONE slskd / Prowlarr
album-bundle search for the release instead of 12 per-track
searches.
Soulseek album-bundle (and any other release-staging path) was
importing every file with ``track_number=1`` because the staging
metadata reader used the auto-import-flavor filename extractor:
``extract_track_number_from_filename`` returns 1 when the basename
has no ``NN -`` prefix. That's the right default for the loose
auto-import flow (single file in, no upstream metadata to lean
on), but completely wrong for staging-cache reads:
- For an album-bundle download the user has authoritative track
numbers in the Spotify track list flowing through to
``track_info`` for each task.
- ``try_staging_match`` in ``core/downloads/staging.py`` was
meant to use those numbers when the staged file's own metadata
doesn't have them.
- But the staging cache populated ``track_number=1`` for every
untagged bare-title file (e.g. ``Cha-La Head-Cha-La.flac``), the
album-bundle resolution branch reads file-side first, sees 1,
and short-circuits the rest of the chain.
Fix:
- New ``extract_explicit_track_number`` in
``core/imports/filename.py`` — strict variant that returns
``0`` when no numeric prefix is visible. Docstring explicitly
contrasts with the legacy 1-defaulting helper so future
callers pick the right one.
- ``read_staging_file_metadata`` in ``core/imports/staging.py``
now uses the strict extractor, so the staging file dict
carries ``track_number=0`` ("unknown") instead of ``1`` for
untagged bare-title files.
- The legacy ``extract_track_number_from_filename`` keeps its
1-default behavior so auto-import callers + the post-process
template fallbacks are unchanged; it's now implemented in
terms of the strict variant.
- Tag-side parsing also tightened to require ``> 0`` before
overriding the filename-derived value.
3 new tests pin the contracts:
- ``test_extract_explicit_track_number_returns_zero_when_no_prefix``
- ``test_read_staging_file_metadata_returns_zero_track_when_unknown``
- existing ``test_extract_track_number_from_filename_handles_common_patterns``
now explicitly comments why bare filenames keep returning 1.
758 tests across imports + downloads + repair + staging-provenance
suites green. WHATS_NEW entry added under 2.6.3.
Reported against an album-bundle download of Ryoto's
"Cha-La Head-Cha-La" where slskd staged 15 untagged FLAC files
named after the song titles only.
The bulk rolling-mirror ensure path was instrumented with INFO
lines + a WARNING on SELECT failure (commit 5378b726) while we
chased down why only one rolling mirror was being created — turned
out the issue was simply needing two refresh cycles after the
rolling code shipped. Diagnostic served its purpose, removing
the noise from every LB refresh now.
- Dropped per-walk + per-match + summary INFO lines from
``_ensure_rolling_mirrors_from_cache`` — the loop is silent.
- Reverted the outer SELECT failure catch from ``logger.warning``
back to ``logger.debug``.
- Kept the per-placeholder ``Pre-created rolling mirror
placeholder`` INFO line in ``_ensure_rolling_series_mirror``
since it's a genuine one-shot event (only fires when a new
placeholder is actually inserted, not on every refresh).
Temporary instrumentation — bulk ensure path silently created
only one rolling mirror despite multiple known series members
existing in the LB cache. Promotes the bulk-ensure summary +
per-title match notes to INFO level so the next refresh
surfaces in the server log:
- ``[LB Rolling] Bulk ensure walking N cached titles for profile X``
- ``[LB Rolling] Title matched series: <title> -> <series_id>``
- ``[LB Rolling] Bulk ensure done — M/N titles matched a series``
Plus the outer ``except`` is bumped from debug to warning so a
genuine SELECT failure stops being invisible.
Once the root cause is identified the noise can drop back to
debug.
Two paths were leaving rolling mirror placeholders uncreated:
1. ``_update_playlist`` short-circuits with status "skipped" when
the cached track count matches the API result (the smart-
comparison fast path). The Phase 1c.2.1 ``_ensure_rolling_series_mirror``
call sat after the short-circuit, so any user whose LB cache was
already up-to-date got zero rolling placeholders inserted —
their Auto-Sync sidebar showed no ListenBrainz group after
refresh.
2. First-time install of the rolling-mirror code on top of an
existing LB cache: every per-playlist call goes "skipped"
because nothing has changed, so even with fix#1 the user
needs a per-playlist trigger to populate. No good.
Fix:
- ``_update_playlist`` now runs ``_ensure_rolling_series_mirror``
on the skip path too (with an explicit ``conn.commit()`` since
the insert needs to land before the connection closes).
- ``_cleanup_old_playlists`` gains ``_ensure_rolling_mirrors_from_cache``
— a one-shot bulk pass that walks every cached LB title and
ensures the matching rolling mirror exists. Cheap (single
SELECT + idempotent INSERT OR IGNORE per row) and catches the
first-run + skipped-everything cases.
Make the rolling Weekly Jams / Weekly Exploration / Top Discoveries
/ Top Missed Recordings mirror entries appear in Auto-Sync's
sidebar the moment ListenBrainz first publishes any member of the
series — without requiring the user to manually discover a per-
period card first.
Previously the rolling mirror was only created on discovery
completion, so users with cached LB playlists but no discovery
history saw an empty ListenBrainz group in the Auto-Sync manager
and couldn't schedule the rolling entries.
- ``_ensure_rolling_series_mirror(cursor, title)`` new helper on
``ListenBrainzManager``: detect_series + ``INSERT OR IGNORE``
the matching ``mirrored_playlists`` row with the synthetic
source_playlist_id, the canonical name, and zero tracks.
Idempotent — no-op when the rolling mirror already exists or
when the title doesn't belong to a series.
- ``_update_playlist`` now calls the helper after the cache row
is inserted/updated, so every LB refresh that lands a per-
period series member guarantees a rolling mirror exists.
First Auto-Sync schedule fired against an empty rolling mirror
populates tracks through the existing LB adapter +
``_maybe_discover`` hook — synthetic id resolves to the latest
cache row, tracks come back with needs_discovery=True, matching
engine runs, mirror gets tracks. No extra wiring needed.
236 tests still green.
User-visible behavior: at most 10 mirrored Last.fm Radio rows
exist at any time. When the cache prunes the 11th-newest +
older lastfm_radio rows, the existing cascade-delete hook
(``_cascade_delete_mirrored_for_mbids``) removes their matching
``source='lastfm'`` mirror rows in the same transaction.
5 was too aggressive — users seeding multiple radios in a row
were losing earlier downloads' provenance before they had time
to act on the tracks. 10 gives a few weeks of breathing room
without letting the Mirrored tab balloon.
ListenBrainz publishes "Weekly Jams for X" / "Weekly Exploration
for X" with a fresh MBID every week, and "Top Discoveries of YYYY
for X" / "Top Missed Recordings of YYYY for X" with a fresh MBID
every year. Auto-mirroring those per-period yielded one mirrored-
playlist row per week/year — useless for Auto-Sync schedules
because the underlying LB playlist never updates, only a brand new
playlist replaces it. The user accumulates 100+ dead Weekly Jams
rows per year if they discover regularly.
This commit collapses each family into a single ROLLING mirror
keyed by a synthetic ``source_playlist_id`` (e.g.
``lb_weekly_jams_Nezreka``). Each new period UPSERTs into the same
row, so the user gets one stable Auto-Sync schedule per series
that automatically picks up the latest period's tracks on every
refresh. Non-series LB playlists (user-created, collaborative,
Last.fm radios for a specific seed) continue to mirror under
their per-playlist MBID as before. Per-period LB playlists are
still visible + usable on the LB Sync tab — only the mirror layer
collapses.
- ``core/playlists/lb_series.py`` (new) — series-detect helper
with regex patterns + canonical-name + LIKE-pattern template
for each known LB family. Exposes
``detect_series(title)``, ``is_series_synthetic_id(id)``, and
``list_series_synthetic_ids()`` so both the JS auto-mirror hook
and the LB adapter can speak the same language.
- ``GET /api/listenbrainz/series-detect?title=...`` — thin HTTP
shim around ``detect_series`` so the auto-mirror JS doesn't
duplicate the regex.
- ``ListenBrainzPlaylistSource.get_playlist`` now recognizes
synthetic series ids — it queries the LB cache for the newest
cache row whose title matches the series' LIKE pattern and
resolves to that row's MBID before fetching tracks. The mirror's
meta keeps the synthetic id so refreshes always re-resolve to
the latest period.
- ``_mirrorListenBrainzAfterDiscovery`` (sync-services.js) calls
the new detect endpoint when discovery completes — if a match
comes back it swaps the per-period MBID for the synthetic id +
the canonical name. Existing Last.fm radio routing logic stays
intact (Last.fm radios aren't a series).
- ``ListenBrainzManager._cleanup_per_period_series_mirrors`` —
one-shot consolidation sweeper runs in ``_cleanup_old_playlists``
+ deletes any legacy per-period mirror rows so the consolidated
rolling mirror is the only one left. Idempotent — only matches
per-period titles ("Weekly Jams for ..., week of ...") and never
the canonical rolling-mirror titles ("ListenBrainz Weekly
Jams").
- 11 new tests pin the detector + synthetic-id helpers; 236 total
across adapter + automation + lb-series suites green.
Two-part fix for Last.fm Radio playlists showing up in the
ListenBrainz group of the Auto-Sync manager + Mirrored tab
instead of their own Last.fm group:
1. **Mirror-creation hook** (sync-services.js): the
``_mirrorListenBrainzAfterDiscovery`` helper hardcoded
``source='listenbrainz'`` on every auto-mirror call, even for
Last.fm Radio playlists (which share the same MB-track shape +
discovery worker but should land under ``source='lastfm'``).
``save_lastfm_radio_playlist`` always prefixes the playlist name
with "Last.fm Radio: <seed>", so the helper now keys on that
prefix to pick the right mirror source + owner fallback. Going
forward, new Last.fm radios mirror correctly the moment
discovery completes.
2. **Backfill** (listenbrainz_manager.py): legacy mirror rows
created before the fix above are stuck under
``source='listenbrainz'``. Added
``_retag_misrouted_lastfm_radio_mirrors`` to ``_cleanup_old_playlists``
so the next LB refresh re-tags any row whose name starts with
"Last.fm Radio:" but is still on ``source='listenbrainz'``.
Idempotent — UPDATE only matches misrouted rows.
ListenBrainz auto-rotates the user's "For You" playlists weekly:
"Weekly Jams for X, week of 2026-05-25 Mon" gets a fresh MBID
every Monday, and the prior week's playlist gets dropped from
ListenBrainz's API after ~25 weeks. The LB manager already mirrors
that retention policy in ``_cleanup_old_playlists`` (keeps the 25
most-recent per category). The Sync-tab auto-mirror flow, though,
created a ``mirrored_playlists`` row for each unique MBID — so the
user's Mirrored tab would accumulate 100+ dead Weekly Jams /
Weekly Exploration rows per year, each pointing at an LB playlist
the cache had already pruned.
Fix: when LB manager removes a cached LB playlist (either via the
periodic ``_cleanup_old_playlists`` rotation or an explicit
``delete_cached_playlist`` call), also delete the matching
``mirrored_playlists`` row + its tracks. Downloaded tracks stay
in the library — only the mirror row + track refs go.
- New ``_cascade_delete_mirrored_for_mbids(cursor, mbids, source)``
helper runs in the same transaction as the LB cache delete so
the two stay consistent.
- ``_cleanup_old_playlists`` now selects ``playlist_mbid`` alongside
``id`` from the stale rows + passes the mbids through the cascade
helper before committing.
- ``delete_cached_playlist`` looks up the playlist's type first
(so it knows whether to target ``source='listenbrainz'`` or
``source='lastfm'`` mirrored rows), then cascades.
Cleanup is best-effort: any cascade error logs a warning but
doesn't roll back the LB cache delete itself. Losing the
cache→mirror link in a rare edge case is preferable to crashing
the LB update loop.
Adds ``discover_tracks(tracks) -> List[NormalizedTrack]`` to the
PlaylistSource interface. Sources whose tracks already carry
provider IDs (Spotify, Tidal, Qobuz, YouTube, Deezer, Spotify
public, iTunes link, SoulSync Discovery) inherit a no-op default;
ListenBrainz + Last.fm override to run the matching engine.
This closes the last gap before LB / Last.fm / SoulSync Discovery
can land as Sync-page mirror sources: the refresh handler now
calls ``source.discover_tracks(...)`` whenever a source returns
tracks with ``needs_discovery=True``, so mirrored LB rows arrive
already discovered + ready for the sync pipeline. Previously, LB
playlists ran through a separate state-machine worker tied to the
Discover-page UI, with results stored in ``discovery_cache``
instead of ``mirrored_playlist_tracks.extra_data``.
Changes:
- ``core/playlists/sources/base.py`` — PlaylistSource switches from
Protocol to ABC so a concrete default for ``discover_tracks``
can live on the base class. The four real-work methods stay
``@abstractmethod``; instantiating an adapter that forgets one
fails loudly at construction.
- ``core/discovery/matching.py`` (new) — pure ``match_mb_tracks``
helper that runs Strategy-1-only matching-engine queries against
Spotify (primary) or iTunes (fallback). No state machine, no
discovery-cache writes, no wing-it stub — that richer flow stays
in ``core/discovery/listenbrainz.py`` for the Discover-page UI.
- ``ListenBrainzPlaylistSource`` + ``LastFMPlaylistSource`` take
an optional ``discover_callable`` constructor arg. Last.fm reuses
the LB implementation since the track shape is identical.
- ``bootstrap.build_playlist_source_registry`` accepts a
``discover_callable`` kwarg and wires it into LB + Last.fm
adapters.
- ``web_server.py`` boot constructs the discovery callable from the
existing matching engine + ``_discovery_score_candidates`` +
Spotify / iTunes clients, passes through to the registry.
- ``refresh_mirrored.py`` adds a small ``_maybe_discover`` helper
that calls ``source.discover_tracks(...)`` between fetch and
``to_mirror_track_dict`` projection — only fires when at least
one track has ``needs_discovery=True``, so the normal Spotify /
Tidal / etc. refresh path stays a zero-cost pass-through.
Tests:
- 5 new adapter tests: default no-op pass-through, LB discovery
with mixed matches/misses, LB no-callable fallback, Last.fm
shares the LB implementation, mirror-dict spotify_hint emit.
- 1 new automation test: end-to-end LB refresh with a stub
discover_callable proves the matched_data lands in
``mirror_playlist_tracks.extra_data`` after the registry
refresh + discover hop.
225 tests across adapter + automation suites green.
Phase 1a of the Discover-to-Sync unification. The mirrored-playlist
refresh handler used to branch per-source through a ~190-line
if/elif chain (Spotify, Spotify public, Deezer, Tidal, YouTube).
Each branch hand-built its own ``extra_data`` JSON for the matched-
data block. With every new source we considered for Sync-page mirror
support (ListenBrainz, Last.fm radio, SoulSync Discovery, iTunes
link), that chain would have grown a new elif.
This commit lifts the per-source logic into the existing adapter
layer and collapses the dispatch to a registry lookup:
- ``core/playlists/sources/deezer.py`` — new adapter so the registry
covers every source the refresh handler previously branched on.
- ``core/playlists/sources/bootstrap.py`` — single helper that builds
a populated registry from injected getter callables. Both
``web_server.py`` boot and the automation test fixtures call it,
so the two construction paths can't drift.
- ``core/playlists/sources/base.py`` — ``to_mirror_track_dict``
projection helper centralises the NormalizedTrack → DB-row
conversion (including the discovered/matched_data and
spotify_hint extra_data shapes the downstream sync + wishlist
consumers already expect).
- Spotify adapter now populates ``extra['discovered']`` + an
``extra['matched_data']`` block when fetching via the authed API,
so Spotify mirrors keep landing pre-discovered (matches the
pre-refactor contract pinned by
``test_spotify_refresh_writes_to_db``).
- Spotify-public adapter populates ``extra['spotify_hint']`` so the
discovery worker can skip its search step and jump straight to
enrichment for the known track ID.
- All artist-name fields now project to first-artist-only across
every adapter — matches the pre-refactor mirror_playlist DB shape
(``t.artists[0]``).
``refresh_mirrored.py`` shrinks ~190 → ~80 lines and keeps:
- the file/beatport unrefreshable-source filter,
- URL extraction from ``description`` via ``require_refresh_url``
for spotify_public + youtube,
- the Spotify-public → authed-Spotify fallback when the user is
signed in (handler-level branch, not in any adapter),
- the Tidal-not-authenticated soft-skip log (skip, not error),
- existing-extra_data preservation across refreshes,
- the ``playlist_changed`` automation event emit on track-set delta.
Test scaffolding:
- ``_build_deps`` in ``tests/automation/test_handlers_playlist.py``
now builds a default registry from the passed clients via
``build_playlist_source_registry``, so existing refresh tests
exercise the same path without per-test changes. New tests cover
Tidal-not-authed soft-skip, Deezer refresh writes plain tracks,
YouTube refresh reads URL from description, and Spotify-public
uses authed Spotify when signed in.
- 4 new adapter tests for Deezer projection +
``to_mirror_track_dict`` (minimal track, Spotify matched_data,
Spotify-public spotify_hint).
- ``playlist_source_registry`` field on ``AutomationDeps`` defaults
to ``None`` so the other 5 automation test files (which don't
exercise refresh_mirrored) keep working unchanged.
220 tests across automation + adapter suites green.
Groundwork for unifying Discover-page playlists (ListenBrainz, Last.fm
radio, SoulSync Discovery) with Sync-page playlists (Spotify, Tidal,
Qobuz, YouTube, Spotify public, iTunes link). All nine sources now
expose the same `PlaylistSource` Protocol so callers stop having to
branch per-source.
This commit only adds the abstraction — no dispatch sites collapse to
the registry yet, no DB or UI changes. Adapters wrap existing clients
via injected getter callables to avoid eager imports of web_server.py
globals.
- core/playlists/sources/base.py — PlaylistMeta, NormalizedTrack,
PlaylistDetail dataclasses + PlaylistSource Protocol with
supports_listing / supports_refresh / requires_auth capability
flags. needs_discovery flag on NormalizedTrack marks tracks that
carry raw MB metadata (LB, Last.fm) vs tracks already matched to a
provider ID (everything else).
- core/playlists/sources/registry.py — thread-safe lazy-factory
registry with instance caching + re-register invalidation.
- nine adapters in core/playlists/sources/ wrapping SpotifyClient,
TidalClient, QobuzClient, spotify_public_scraper, the YouTube +
iTunes-link parsers (via injected callables), ListenBrainzManager,
Last.fm radio rows in the ListenBrainz cache, and
PersonalizedPlaylistManager.
- tests/test_playlist_sources_adapters.py — 18 tests covering each
adapter's field projection with fake backing clients, plus
registry lazy-construct + cache + re-register invalidation.
Phase 1 will collapse refresh_mirrored.py's per-source if/elif chain
to a registry lookup and surface ListenBrainz as a Sync-page tab.
