Spaces:
Sleeping
Sleeping
Commit History
26 days ago
-
fix: eliminate duplicate movie entries, prevent moving dubbed releases across language folders, and enable dynamic backdrop rotation in spotlight
Hermes Agentaade177 -
Add permanent deletion handling: on_deleted_messages, /delete command, and Jellyfin Web deletion persistence
Hermes Agent5955a8b -
Fix Kattalan resolution, MMAX tagging, and purge Ms.Rose from plugin databases
Hermes Agent2cd8606 -
feat(ai): add Omniroute AI candidate disambiguation with automated hardcoded fallback
Hermes Agentec8cd47 -
fix(metadata): purge unwanted Ms.Rose & Mr.Cane and resolve Obsession to official 2026 Curry Barker release
Hermes Agent5ff93ff
27 days ago
-
feat: inject 1080p HD thumbnails and randomize homepage spotlight movies on each restart
Hermes Agent415ebcc -
fix(tmdb): strictly prevent old English movies from matching modern Indian releases like Patriot 2026
Hermes Agent6a4197a -
fix(tmdb): fix mismatched movie thumbnails and ensure HD backdrops and posters
Hermes Agent341e7f3 -
fix(metadata): improve TMDB matching accuracy, upgrade HD backdrops & logos, purge TV English folder and blacklist demo episodes
Hermes Agentd4c144c -
feat(perf): implement direct-play bypass, HD artwork pipeline, multi-bot pooling, WAL checkpoints and read-ahead buffering
Hermes Agent2bc80a8 -
fix(streamer): resolve fatal player error by correcting inverted msg_id/chat_id in ingestion, cache lookup, and stream routes
Hermes Agentc52a48e -
feat(streamer): add multi-client fallback on expired references and batch ID range scanning for old messages
Hermes Agent81c25f6 -
feat(ui): integrate Abyss theme and Spotlight hero banner
Hermes Agent8ca1a05 -
fix(gateway): route Jellyfin plugin endpoints and referer requests to Jellyfin instead of fallback 404
Hermes Agent7c02102 -
feat(ui): reset Jellyfin branding and web assets to stock default
Hermes Agent5cabfcf -
feat(ui): restore original clean Jellyfin layout with pure OLED pitch black theme
Hermes Agentd080af5 -
feat(apex): robust any-title TMDB resolution
Hermes Agent- Tier-0: embedded IMDb ids (Radarr-style {imdb-tt...}, [tt...]) resolve deterministically via TMDB /find, bypassing fuzzy text search entirely. - best_tmdb_match: merge original_title/original_name tokens so non-Latin primary titles still match transliterated queries; year is now a soft proximity signal (exact +0.3, +/-1 +0.15, far -0.15); original_language affinity bonus; roman-numeral normalization (III -> 3). - Language-biased search: detected release language (ISO 639-1) is passed to TMDB search, with a dedicated retry in the native language when English search misses. - Diagnostics: no-match warning now logs candidates + lang tried. - Repair sweep: restored cache entries that never resolved are re-enriched once with the improved matcher (bounded, skips tmdb_id/tmdb_title entries). - 12 new unit tests (58 total in module).541097d -
fix(meta): self-heal stale NFOs with ASCII-mangled original titles
Hermes AgentNon-ASCII original titles (Malayalam/Tamil/Hindi scripts) were written as a run of '?' by an earlier writer and never repaired: the enrichment idempotency gate treated any existing NFO as cached and the movie/episode NFO writes were guarded by os.path.exists. Now: - nfo_needs_repair() flags any NFO whose originaltitle/original_name is a run of '?'); such NFOs are excluded from the idempotency gate and are rewritten on the next enrich pass - movie/episode NFO writes are unconditional when a fresh detail was fetched (self-healing), tvshow.nfo rewrites when malformed - streamer launched with PYTHONUTF8=1 PYTHONIOENCODING=utf-8701239e -
fix(meta): channel-marker junk + AI web-title resolution
Hermes Agent- tmdb_meta: drop '6CH'/'5.1ch' channel markers from search queries so 'Obsession.2025...6CH.x265...' searches 'obsession' (was 'obsession 6ch' which returns 0 TMDB results) instead of No-Match. - tg_streamer: AI-title prompt now resolves the REAL official title from model knowledge (Indian + international OTT incl. Malayalam), verified live: 'Mollywood.Times...JHS...' -> 'Mollywood Times' (2026). - tg_streamer: AI tier-2 retries TMDB single-type then /search/multi, dropping year if filtered search is empty; new AI-tier-3 publishes a minimal canonical-title NFO when TMDB has no record, so Jellyfin never keeps a release-junk folder name.4a485e9
28 days ago
-
feat: AI title-extraction fallback for TMDB enrichment
Hermes Agent- Cascade TMDB search across rule-parser candidates before giving up - omniroute AI tier-2 fallback (auto/best-fast) only when TMDB misses - Prefer AI-extracted year over filename year for matching - Adds 11 regression tests for OTT tails, series parts, short release tags7157236 -
fix: TMDB enrichment for new files - dotted group tails, numeric titles, short release tags
Hermes Agent9c18a42 -
fix: organize media by detected language
Hermes AgentUse TMDB original language as a fallback when filenames do not contain language tags, and relocate published STRM files into the canonical language folder.\n\nCo-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>817b0d1 -
Integrate Jellyfin alongside FileToLink
Hermes AgentRun the Apex Jellyfin gateway, Telegram streamer, and Jellyfin server beside the existing FileToLink service with isolated ports, storage, startup supervision, and separate playback bot variables.\n\nCo-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>0e94626