Fixes#439, likely also #302.
Reported: on iOS, the download folder silently reverts to the default
SpotiFLAC folder after starting a download (and stays reverted after
closing and reopening the app).
_processQueue ran an iCloud/writable-path shape check against the
persisted downloadDirectory string and, if it looked like iCloud Drive
or failed the structural writable-path check, called
setDownloadDirectory() to reset it to the default folder. That call
passes no iosBookmark argument, so it also wipes the user's
security-scoped bookmark in the persisted settings - permanently, not
just for the current run.
This check ran unconditionally, even when a bookmark was already
present for that folder. A folder picked from Files can legitimately
have a persisted path that looks like iCloud Drive, or that fails this
generic structural check, while the bookmark itself still grants real
write access - the bookmark is the actual source of truth, and the
queue already has a separate, correct bookmark-resolution step right
after this one (StartAccessingIosBookmark, with its own proper
failure handling that fails the queued items with a clear message
instead of destroying the setting). The earlier path-shape check just
never deferred to it.
lib/utils/file_access.dart: extract the run/skip decision into
shouldValidateIosOutputDir(isIOS, isSafMode, outputDir,
downloadDirectoryBookmark) - skips whenever a bookmark is present -
and use it in download_queue_provider.dart in place of the inline
condition. Kept as a plain function taking isIOS as a parameter since
Platform.isIOS itself can't be exercised from a host test run.
test/file_access_ios_test.dart covers: runs for a plain app-folder
path, skips when a bookmark is present (the exact #439 case), skips
off iOS, skips in SAF mode, and skips when there's no output dir yet.
Verification: flutter analyze and flutter test (224 tests, all green,
including the 5 new ones) both clean.
The connectivity listener now stays alive while network-failed items
remain and, on reconnect, shows a debounced snackbar counting them
with a Retry action wired to retryAllFailed scoped to network
failures. Also regenerates localizations for the recent feature
strings.
Queued rows get a skip-next button that reinserts the item ahead of all
other queued items, so the next free slot picks it up. During an active
native worker run the current batch keeps its order; the new order
applies from the next run.
Stability:
- transient timeouts no longer classified as ISP blocking; they fall
through to retry backoff (hard blocks - DNS/RST/cert - still abort)
- connectivity change now closes idle Go connections in every network
mode (debounced), so pooled sockets from the old interface are not
reused after a wifi/cellular handoff
- download body reads get a 60s stall watchdog that cancels and
surfaces a retryable network error instead of hanging; distinct
from user cancellation
- ResponseHeaderTimeout 45s on all transports; IdleConnTimeout 90->60s;
dial timeout 30->10s; retry backoff gains full jitter; Retry-After
honored on 5xx and Deezer 429
Data usage / speed:
- cover downloads deduplicated with singleflight plus a 24MB/15min
in-memory cache keyed by final URL (album batches fetched the same
1800px cover once per track before)
- song.link availability cached (30min positive / 5min negative)
in front of the 9-req/min rate limiter
- uTLS Cloudflare path now pools one HTTP/2 connection per host with
a shared TLS session cache instead of a full handshake per request
Deezer artist track-count N+1 kept: counts feed the discography
download UI and are already amortized by the artist TTL cache.
- _replaceSafFileVia: op now registers produced files via addCleanup
the moment they exist, so converted temp files are deleted even
when a later step throws (old code's outer-variable finally)
- native worker history: recompute format from the final file path
at history-write time and gate bitrate on lossy output again -
the extracted helper was recording pre-conversion format/bitrate
(FLAC conversions logged as m4a with nulled bit depth on SAF)
- drop dead ignore_for_file header in the finalization part
Pure move: paths, native worker, finalization, ReplayGain, and
metadata embedding move to part files as private extensions on
DownloadQueueNotifier; queue orchestration, state, and public API
stay in the main file (7598 -> 4022 lines).
- _buildDownloadRequestPayload/_historyItemFromResult shared by the
inline and native-worker paths; _buildOutputDir now derives from
_buildRelativeOutputDir
- shouldAbortWork extended (deleteFileOnAbort, null-item check) and
used at all ten guard sites; _purgeAlbumRgEntry replaces six copies
- ProgressStreamPoller shared by download queue and local library;
openVerificationAndAwaitGrant shared by queue and track provider
Go:
- callExtensionScript generic replaces the ~40-line perf/lock/run/
timeout skeleton in ten wrapper methods; PostProcess V1/V2 share
postProcessCommon (V1 keeps its postProcess-only probe)
- attemptExtensionDownload extracts the duplicated provider download
attempt from the source-extension and priority-loop paths;
overlayStr/overlayInt collapse the enrichment overlay chains;
buildDownloadFilename shared by both output-path builders
Dart (download_queue_provider):
- _replaceSafFileVia wraps the SAF copy/operate/publish/cleanup
roundtrip used at eight sites
- decrypt and external-LRC finalization unified between the inline
and native-worker pipelines (divergences parametrized: AC-4 repair,
per-stage error messages, base-name resolution, extension-state
snapshot); HIGH/container conversion keep separate lenient/strict
policies and share only the SAF mechanics
The 1s snapshot poll re-read and re-parsed the whole worker state file
— which retains every completed item's result (history row, possibly
full synced lyrics; 1.5-5 KB each) for the entire run — then
re-serialized it across the channel and re-iterated it on the Dart UI
isolate. Late in a 1000-track batch that is megabytes parsed several
times per second, on the Android main thread.
Pollers now echo back the state_serial of the last snapshot they fully
processed; when the state has not advanced, the native side serves a
cached compact header (no items, no results, no settings) merged with
the small progress delta — steady-state polls are O(1) regardless of
batch size, and the full payload is delivered exactly once per state
transition. The channel handler also reads and parses off the main
thread now, matching its neighbours.
- Verification challenge no longer abandons the batch: batch mates the
worker cancel would mark skipped are requeued and fenced from later
snapshot applications, and the adoption loop restarts the queue for
requeued items when the run ends
- Pause vs cancel race: a cancelled result with no pause flag set waits
up to 1.5s for the pause intent to land on the main looper before
being classified as a permanent skip
- Stale Go cancel flag: new ResetDownloadCancel export (wired through
Android and iOS bridges) drops a pre-registered cancel with no active
download; retryItem/retryAllFailed call it so the first retry of a
cancelled-before-start item no longer aborts instantly
- Reconcile-loop errors now cancel the native worker and clear the run
id before marking items failed, instead of leaving the service
downloading a batch the UI already wrote off
Stale snapshot (queue wedged forever): the service is START_NOT_STICKY,
so a process kill or reboot mid-batch froze the state file at
is_running:true with nothing ever rewriting it. Adoption then set
isProcessing and polled that flag indefinitely; restored items sat at
downloading/queued until the user intervened. Adoption and the
reconcile poll now check DownloadService.isServiceRunning() (three
consecutive dead polls in the loop), reconcile whatever the worker
finished, requeue in-flight items, clear the run id, and hand the batch
back to the Dart queue.
Adoption gate race (duplicate executors): adoption was gated on
_canUseAndroidNativeWorker, which reads extension state that loads
asynchronously after startup and settings toggles the user may have
flipped mid-run. A background batch surviving task removal was then
orphaned while the Dart queue re-downloaded the same items. Adoption
now only requires Android; request contexts that cannot be built yet
(extensions still loading) are retried on each reconcile poll.
Queue replacement race (superseded worker kept running): startNativeWorker
only cancelled the coroutine, which cannot interrupt the blocking
gomobile call, and it reset the shared cancel flag the old loop checks -
so a replaced worker kept downloading its old batch, and its finally
block wrote is_running:false over the new run's snapshot and stopSelf'd
the service underneath it. Replacement now cancels the old run's Go
downloads and FFmpeg work first, and workers carry a generation token:
a superseded generation exits at the next checkpoint and skips the
final snapshot write and service teardown. An invalid replacement
payload no longer stops the service while a run is active.
zarzet