xpsony/UmlautAdaptarrEX 1.1.1 on GitHub

1.1.1 — 2026-05-25

Lidarr/Readarr sync no longer crashes on duplicate album/book titles: Library rows are now keyed by normalize("{artist} {album}") for Lidarr and normalize("{book} {author}") for Readarr instead of just the album/book title. Previously libraries with multiple albums or books sharing a title across artists/authors ("Greatest Hits", "Live", "Best Of", "Self-Titled", …) collided on the SearchItem unique constraint and aborted the sync with Prisma P2002. As a side effect the legacy /api?t=search&cat=3000… getByExternalId lookup now actually resolves, because Prowlarr's ?q=Artist Album query normalizes to the same key the row was stored under.
Sync dedup defense-in-depth: persistAndReindex now deduplicates items by externalId before writing, so a stray duplicate in the upstream payload no longer aborts a 50-item chunk transaction. Duplicates are counted and warned in the log; SyncRun.itemsCount reflects the deduped count.
Scheduler provider gate: POST /api/admin/sync and the scheduled tick no longer return no_provider for Lidarr/Readarr-only setups. The gate now only blocks when at least one Sonarr/Radarr instance is enabled, because only those two consult a title provider during sync.

No manual resync required. On the next sync (automatic or via "Sync now" in the dashboard) the stale rows with old externalIds are removed