Use the database as the authoritative source for missing-episode lists so
that episodes marked is_downloaded=True are never shown as missing, even
when the in-memory state is stale.
Key changes:
- EpisodeService.get_by_series() gains only_missing flag
- AnimeService uses DB-backed episodeDict and preserves downloaded episodes
during sync, skipping them when adding/removing missing episodes
- DownloadService broadcasts series_updated after marking an episode downloaded
so the frontend reflects the change immediately
- Frontend filters out series with zero missing episodes client-side and
fixes renderSeries to respect the active filter
- Unit tests updated to assert the broadcast is sent
- Add missing year field when building series list in anime_service
- Add _create_missing_nfo to generate minimal NFO for series without one
- Update perform_nfo_repair_scan to detect and create missing NFOs
- Add semaphore-protected async creation with TMDB rate limiting
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Add chardet-based _decode_html_content() to aniworld_provider. Apply
to all BeautifulSoup parsing calls to prevent decoding warnings on
pages with mismatched encoding declarations. Falls back to utf-8
with errors='replace' when confidence < 0.7.
Also fix test_enhanced_provider HLS test signature and add HLS
pattern unit tests.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- New migration script: populate year from folder (YYYY) pattern
- SerieScanner: refactor year extraction logic
- anime_service: pass year when syncing from data files
- Add create_minimal_nfo() method to NFOService for fallback when TMDB lookup fails
- Update API endpoints (single and batch) to use minimal NFO fallback on TMDBAPIError
- Document fallback behavior in NFO_GUIDE.md section 3.6
- Add unit tests for minimal NFO creation (11 tests passing)
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Iterate providers actually advertised on the episode page (ordered by
SUPPORTED_PROVIDERS preference) instead of always re-resolving VOE.
Each candidate is HEAD-probed before yt-dlp runs, so dead links are
skipped immediately; direct video URLs use a streaming fast path that
bypasses yt-dlp; total failure now logs the exhausted provider list.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Retry count and queue status were in-memory only and lost on
restart, so failed downloads could not be safely resumed and
permanently-failed episodes silently blocked re-queueing via the
episode-id unique index.
- Add status + retry_count columns to DownloadQueueItem
- Replace unique(episode_id) with unique(episode_id, status) so
permanently_failed rows do not block new pending entries
- Add PERMANENTLY_FAILED to DownloadStatus enum
- Persist retry_count on each failure; mark permanently_failed once
max_retries reached
- QueueRepository reads status/retry_count from DB instead of
defaulting to PENDING/0
- Stop double-incrementing retry_count in retry_failed_items;
increment only happens in _process_download on failure
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Change _remove_episode_from_missing_list to set is_downloaded=True
and populate file_path via EpisodeService.mark_downloaded, instead of
deleting the Episode row. Preserves download history so queries can
distinguish series with downloaded episodes from completely unwatched
series.
- Pass serie_folder to construct file_path
- Look up series_id via AnimeSeriesService.get_by_key
- Update tests to mock mark_downloaded path
- Document episode lifecycle in docs/DEVELOPMENT.md
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Fix race condition: next_run_time only available after scheduler.start()
- Handle None gracefully in logging
- Add debug logging to _perform_rescan and _run_rescan_job
- Document scheduler troubleshooting in DEVELOPMENT.md
- Add _run_startup_health_checks() function in fastapi_app.py
- Check ffmpeg availability (warning)
- Check DNS resolution for aniworld.to and api.themoviedb.org (warning)
- Check anime_directory configuration and writability (error)
- Store startup checks in app.state for health endpoint access
- Add /health/ready endpoint for container orchestrators
- Returns not_ready with 503 when critical failures present
- Includes critical_failures list for debugging
- Update /health endpoint to include startup check results
- Status reflects worst check (error > warning > ok)
- Document health check endpoints in DEVELOPMENT.md
- Add unit tests for startup health checks
- Add unit tests for /health/ready endpoint
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- New _search_with_fallback() method tries multiple strategies:
1. Primary query with year filter (de-DE locale)
2. Alternative titles with ja-JP / en-US locales
3. English search (en-US)
4. Search without year constraint
5. Punctuation-normalized query
- create_nfo() accepts new alt_titles param for Japanese/title fallback
- Better match rate for anime with non-English titles
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- In-memory dedup in add_to_queue() using _pending_by_episode dict
- Batch-local dedup via seen_in_batch set (handles duplicates within single call)
- Database unique index on episode_id via __table_args__
- 5-minute cooldown in _auto_download_missing() to prevent rapid re-triggers
- Updated _add_to_pending_queue() and _remove_from_pending_queue() to track episode keys
- Added TestQueueDeduplication with 4 test cases
- Updated DEVELOPMENT.md and TESTING.md with queue dedup docs
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Add ffmpeg to Dockerfile.app for container HLS support
- Configure yt-dlp with --downloader ffmpeg --hls-use-mpegts
- Add startup health check warns if ffmpeg missing
- Update DEVELOPMENT.md with ffmpeg prerequisites and troubleshooting
- Add tests for ffmpeg HLS options and health check
Without a /32 route in the main table, CHECK_HOST (1.1.1.1) fell through
to the VPN default route where source-address selection was defeated by
the priority-100 'from ETH0_IP' policy rule, causing pings to bypass
wg0 and be dropped by the kill switch.
Also add secondary google.com ping to distinguish IP vs DNS failures.
- Increase max_retries from 3 to 5 with exponential backoff capped at 30s
- Add per-second rate limiter (~35 req/s) to stay under TMDB's ~40/s limit
- Replace small semaphore (4) with larger one (30) + token-bucket throttle
- Abort retries immediately on DNS/name-resolution failures
- Increase rate-limit fallback wait from default to max(delay*2, 10)s
- series.py: use regex to remove all trailing (YYYY) before appending year
- nfo_service.py: _extract_year_from_name strips all trailing year suffixes
- nfo_repair_service.py: add _read_tmdb_id() helper to extract TMDB ID from NFO
Apply the same duplicate-year prevention logic to additional code paths:
- Serie.name_with_year property: skip adding year suffix if name already ends with it
- add_series API endpoint: avoid duplicating year in folder_name_with_year
- Add integration test for Serie.name_with_year idempotency
- Add API test for add_series endpoint year deduplication
Complements the folder_rename_service fix for comprehensive coverage.
Use regex to strip all trailing year suffixes before adding the canonical
one, preventing duplication like 'Show (2021) (2021) (2021)'.
- Add regex pattern (\s*\(\d{4}\))+\s*$ to remove all existing year suffixes
- Ensure idempotent behavior across multiple folder rename runs
- Add 7 unit tests covering the bug cases and edge scenarios
Fixes: 86 Eighty Six (2021) (2021)..., Alma-chan (2025) (2025)...
- Pass app logger to yt-dlp so internal [download] progress lines
are routed through the INFO-level logger instead of stdout.
- Throttle download_progress_handler debug logging to avoid
flooding logs on every fragment tick.
- Switch key provider lifecycle messages to INFO (start/complete)
while keeping verbose details at DEBUG.
- Set debug_enabled=False in development config so dev mode
does not emit extra debug noise.
- Update config docstring example from DEBUG to INFO.
Add check_vpn_connectivity() that runs once after wg0 comes up:
- Waits for handshake (up to 15s) and prints public key if missing
- Measures RX bytes before/after curl to detect server-side routing issues
- Tests DNS resolution and dumps resolv.conf on failure
- On failure prints exact server-side commands to fix (sysctl, iptables, wg)
Add PersistentKeepalive=25 to wg0.conf to keep NAT mappings alive.
DNS OUTPUT was restricted to -o wg0, but routing decision happens
after iptables OUTPUT — so DNS to VPN-internal addresses (198.18.0.x)
was blocked before the kernel selected the outgoing interface.
Allow DNS unconditionally; routing still sends it through wg0.
Add NET_RAW capability so ping works inside the container.
- Parse AllowedIPs dynamically from WireGuard config instead of hardcoding routes
- Remove auto-created default route by wg setconf to prevent breaking endpoint connection
- Fix DNS parsing: write comma-separated DNS servers as separate nameserver lines
- Add test for AllowedIPs route verification and DNS configuration
- Update test to skip container runtime tests when not running as root
- Rotate to new VPN endpoint (91.148.236.64)
- Update private/public keys and client address
- Switch DNS to 198.18.0.1/0.2
- Add local network route preservation via PostUp/PostDown
- Align nl.conf and wg0.conf configurations
