legacy-arrflix/playbooks/subtitles/README.md

Subtitle acquisition process — v1

Last updated: 2026-05-10 Status: v3.5 — four fetch paths (plugin / OS REST / Addic7ed / YouTube auto-CC). American Dad 49/58 + Sassy 5/5. v4 WhisperX planned (ROADMAP H5).

This recipe is written for Claude Code to execute. Each step lists the exact command, what to verify, and what to do on failure. Background reference for how Jellyfin and the OpenSubtitles plugin work together lives in docs/03-subtitles.md.

Current state: COVERAGE.md is the live audit (per-show + per-movie). Regenerate at any time:

JELLYFIN_TOKEN=<admin-token> playbooks/subtitles/lib/audit-coverage.py

Run after every fetch batch so the committed file stays accurate.

Read STYLE.md first. Every fetch must hit the bar set there: one English .srt per episode, plain (no SDH / no MT / no AI / no Forced), best-quality release. The picker logic in v1/v2/v3 mirrors that bar; if a step would violate it, stop and ask before downloading.

Stop-gap exception: when the only available source is the v3.5 YouTube auto-CC path (lowercase, censored, mangled names), ship the sub but add the show to STOPGAP-SUBS.md so v4 WhisperX picks it up later.

---

Prereqs (verify before running)

Check	How
OpenSubtitles plugin v20 installed + Active	`docker exec jellyfin ls /config/plugins
Plugin creds saved (Caveman5)	docker exec jellyfin grep -E 'Username|CredentialsInvalid' /config/plugins/configurations/Jellyfin.Plugin.OpenSubtitles.xml — expect Caveman5 and false
TV library has SaveSubtitlesWithMedia=true, SubtitleDownloadLanguages=["eng"], RequirePerfectSubtitleMatch=false	curl -s -H "X-Emby-Token: $TOK" http://localhost:8096/Library/VirtualFolders
Free-tier quota remaining today (≥ episode count, else plan multi-day)	docker logs --tail 200 jellyfin 2>&1 | grep "Remaining downloads" | tail -1 (free = 20/day, resets 00:00 UTC)
Source files have audio language tag	ffprobe sample episode

If any prereq fails, stop. Fix it before running the recipe.

---

Step 1 — Probe the source

Pick one episode of the target show. Run ffprobe on it:

ssh user@192.168.0.100 'docker exec jellyfin /usr/lib/jellyfin-ffmpeg/ffprobe -hide_banner "<path-to-mkv>" 2>&1 | grep -E "Stream|Duration"'

Record in the run log:

• video codec + resolution + frame rate
• audio language tag(s)
• whether any subtitle streams are embedded
• container

Decide based on probe:

Probe result	Branch
English audio, no embedded subs	"simple" path (this recipe)
Foreign-dub audio, no embedded subs	"foreign-dub" path (deferred to v?)
Embedded English subs already present	skip — Jellyfin will use them
Embedded PGS/VobSub bitmap subs	"OCR" path (deferred to v?)

---

Step 2 — Resolve series + episode IDs

TOK=<jellyfin-admin-token>
SERIES_NAME='American Dad'
ssh user@192.168.0.100 "docker exec jellyfin curl -s -H 'X-Emby-Token: $TOK' \
  'http://localhost:8096/Items?searchTerm=${SERIES_NAME// /+}&IncludeItemTypes=Series&Recursive=true&Limit=3'" \
  | python3 -c "import json,sys; [print(x['Id'],x['Name']) for x in json.load(sys.stdin).get('Items',[])]"

Record series Id. Then list episodes:

SERIES=<series-id>
ssh user@192.168.0.100 "docker exec jellyfin curl -s -H 'X-Emby-Token: $TOK' \
  'http://localhost:8096/Items?ParentId=$SERIES&IncludeItemTypes=Episode&Recursive=true&Fields=Path,ParentIndexNumber,IndexNumber'" \
  | python3 -c "import json,sys; [print(e['Id'],'S%02dE%02d'%(e['ParentIndexNumber'],e['IndexNumber']),e['Name']) for e in json.load(sys.stdin)['Items']]"

---

Step 3 — Pick fetch path

Four paths, ordered cheapest-quota-cost-first:

Path	Cost / day cap	Coverage	Tool
v3 Addic7ed	free, no daily cap (anon)	English-only; near-complete on broadcast US shows; spotty on animated specials / niche titles	lib/sub-a7d-fetch.py
v2 OS REST	20 / day on free OS account	best overall coverage; survives any S/E numbering quirk via per-ep imdb_id	lib/sub-rest-fetch.py
v1 plugin	counts against same OS 20/day	only works when library numbering matches OS catalogue (e.g. fails on American Dad past S01E07)	lib/sub-fetch.sh
v3.5 YouTube auto-CC	free, ratelimited only	for shows distributed YouTube-first (no community subs anywhere); produces lowercase, no-punctuation, name-mangled subs — stop-gap, violates STYLE.md	lib/sub-yt-fetch.sh + lib/yt-clean.py
v4 WhisperX (planned)	local CPU/GPU time	full-quality auto-transcription, restores STYLE.md bar for niche shows	TBD lib/sub-whisperx-fetch.py (ROADMAP H5)

Default: try v3 first to spare quota; fall back to v2 for episodes v3 misses or for non-English needs. v1 stays for shows where simple plugin auto-fetch is enough. v3.5 is the stop-gap when nothing exists on community providers; v4 replaces v3.5 once the GPU node is set up.

Quick check whether v1 plugin will suffice (skip the rest if yes):

1. Pick the first episode of season 2 in the library.
2. Run curl -s -H 'X-Emby-Token: $TOK' 'http://localhost:8096/Items/$EP/RemoteSearch/Subtitles/eng' (read-only).
3. If results > 0 — v1 works.
4. If results == 0 but the show exists on opensubtitles.com — numbering mismatch (e.g. American Dad: library uses Hulu S1=7 eps; OS uses different). Use v3 then v2 for misses.

---

Step 4 — Fetch subs per episode

v3 — Addic7ed (default, free)

JELLYFIN_TOKEN=<admin-token> \
OPENSUBTITLES_API_KEY=$HOME/.config/arrflix-opensubtitles-api.txt \
playbooks/subtitles/lib/sub-a7d-fetch.py <series-id> --season N [--start E] [--end E]

Pre-flight with DRY_RUN=1. The OS REST key is used only for search (quota-free) to translate library S/E to the show's catalogue numbering.

v2 — OpenSubtitles REST (fallback for v3 misses)

JELLYFIN_TOKEN=<admin-token> \
OPENSUBTITLES_API_KEY=$HOME/.config/arrflix-opensubtitles-api.txt \
OPENSUBTITLES_USER=Caveman5 \
OPENSUBTITLES_PASS=<password> \
playbooks/subtitles/lib/sub-rest-fetch.py <series-id> --season N [--start E] [--end E]

20 / day cap, resets at 00:00 UTC.

v1 — Jellyfin plugin (when library numbering matches OS)

lib/sub-fetch.sh — see header for env. Counts against the same 20/day cap.

Verify after each batch

ssh user@192.168.0.100 'ls "<media-dir>/" | grep -c eng.srt'

---

Step 5 — Library scan + de-dup (v1 only)

If you used the v1 plugin path, the metadata-cache copy and the media-folder sidecar both register as subtitle streams in Jellyfin (counted twice). Delete the cache copies:

ssh user@192.168.0.100 'docker exec jellyfin bash -c "find /config/metadata/library -path \"*<show-name>*S0[1-9]E*.eng.srt\" -delete -print"'

v2 writes directly to the media folder so there is no cache copy to clean.

Trigger a validation-only refresh so Jellyfin sees the new sidecars:

ssh user@192.168.0.100 "docker exec jellyfin curl -s -X POST -H 'X-Emby-Token: $TOK' \
  'http://localhost:8096/Items/$SERIES/Refresh?MetadataRefreshMode=ValidationOnly&Recursive=true'"

Confirm one episode has exactly 1 external eng sub stream:

ssh user@192.168.0.100 "docker exec jellyfin curl -s -H 'X-Emby-Token: $TOK' \
  'http://localhost:8096/Items/<sample-ep-id>?Fields=MediaStreams'" \
  | python3 -c "import json,sys; subs=[s for s in json.load(sys.stdin).get('MediaStreams',[]) if s['Type']=='Subtitle']; print(len(subs),'sub streams')"

---

Step 6 — Quality gate

For the run to pass:

• Coverage: every episode has a matching <base>.eng.srt sidecar
• Sync sample: at least one episode of each season is opened in Jellyfin web and subs visually align with audio (±1 s) on a known dialogue line
• Flag check: no .sdh.srt, .forced.srt, or .hi.srt files (machine pick should have filtered)
• Stream count: Jellyfin shows exactly 1 external eng sub per episode

If any check fails, log it in runs/<show>.md under "breakage" and propose the recipe amendment in CHANGELOG.md.

---

Quota hygiene

Free OpenSubtitles.com account = 20 downloads / day, resets 00:00 UTC. Plan large series across multiple days, or switch to VIP (~$3/mo, unlimited).

Quota check:

ssh user@192.168.0.100 'docker logs --tail 200 jellyfin 2>&1 | grep "Remaining downloads" | tail -1'

When quota hits 0 the API returns 0 results, indistinguishable from a real miss. Always check quota before declaring a "no subs" failure.