vavallee/bindery v1.9.0

on GitHub

Added

• Book metadata can be remapped from the Book Detail page (#590) — Books with ABS or stale metadata now show an Improve metadata action that searches upstream providers or accepts a direct provider ID. New POST /api/v1/book/{id}/map applies the upstream title, cover, language, ratings, genres, and provider ID while preserving local status, files, media type, ASIN, narrator, selected edition, and exclusion state.

• Calibre-Web-Automated (CWA) ingest (#417) — A new
  Settings → General → Calibre-Web-Automated (CWA) field configures a
  shared ingest folder. When set, every successful ebook import is also
  copied into that folder so a sibling
  CWA container
  can auto-ingest it. Bindery keeps its own copy and never moves the file,
  so a misconfigured CWA can't take bindery's library with it. No Calibre
  runtime dependency is added to the bindery container — only the file
  drop is in scope. Audiobook imports are unaffected since CWA is built
  around ebook libraries.

• Prowlarr search timeout is now configurable (#576) — The Prowlarr indexer
  search timeout has been raised from 15 s to 60 s and can be adjusted in
  Settings → Indexers → Prowlarr → Search timeout. Slow usenet indexers
  no longer time out on the first query.

Fixed

Importer / download clients

• qBittorrent SavePath fallback caused incorrect imports (#574) — When
  qBittorrent's content_path field was absent or empty, the importer
  fell back to SavePath (the shared download root) and could match
  unrelated files or walk directories it should not touch. The importer
  now uses content_path exclusively and aborts cleanly when it is missing.
• Default import mode changed from move to hardlink/copy (#577) —
  The out-of-box default was move, which silently broke torrent/usenet
  seeding immediately after import. Bindery now defaults to hardlink when
  source and destination are on the same filesystem (free, preserves seeding)
  or copy when they are cross-device. Upgrade note: migration 038
  clears the implicit move default written at install time; users who
  explicitly set an import mode in Settings are not affected.
• Downloads stuck in importFailed are now retried automatically (#578)
  — Previously, a download that failed during import was permanently orphaned.
  Bindery now retries up to three times before leaving it for manual
  intervention. Retry count is persisted via migration 037.
• CheckDownloads now polls all enabled download clients (#572) — Only
  the highest-priority client was polled for status updates. Secondary
  clients (e.g. a second qBittorrent instance or a fallover) were silently
  ignored. All enabled clients are now iterated in priority order.
• Bulk-grab torrent dedup race condition fixed (#573) — Grabbing multiple
  releases simultaneously could assign the same torrent_id to two
  downloads, breaking per-download tracking. AddTorrent is now serialised.

Auth

• API key authentication now grants admin role (#582) — Requests
  authenticated via API key successfully verified the key but did not set
  the admin role in the request context, causing RequireAdmin-protected
  endpoints to return 403. The role is now correctly propagated.
• Auth endpoints no longer require X-Requested-With header (#575) —
  The login endpoint enforced X-Requested-With: bindery-ui, blocking
  non-browser clients (curl, mobile apps, integrations). Auth endpoints are
  now exempt; programmatic clients should use API key auth instead of cookie
  sessions.

AudioBookShelf (ABS)

• ABS library is rescanned after audiobook import (#581) — Bindery now
  triggers POST /api/v2/libraries/:id/scan after a successful audiobook
  import so the file appears in ABS immediately rather than on its next
  scheduled scan.
• Move-mode audiobook imports no longer appear MISSING in ABS (#583) —
  The ABS rescan after import updates ABS's path knowledge, resolving the
  MISSING status that appeared when the import moved the file.
• History events include format for dual-format books (#584) — bookImported
  events for books with media_type='both' now record which format (ebook
  or audiobook) was imported, making the History page unambiguous.

Metadata

• ISBN lookups now canonicalise provider-native matches (#590) — ISBN
  searches normalise ISBN input, consult configured metadata enrichers, and
  conservatively relink provider-native results back to canonical OpenLibrary
  works when the author/title evidence is unambiguous. This improves
  translated and edition-specific matches while avoiding plausible
  wrong-title fallbacks.
• Audiobook ASIN enrichment can relink to upstream metadata (#590) —
  Enriching an audiobook now uses Audnex ASIN metadata to find a safe
  canonical upstream match, so ABS/imported audiobook rows can gain better
  titles, covers, language, search metadata, and OpenLibrary IDs while
  keeping audiobook-specific fields intact.
• ABS imports no longer trust stale secondary-author aliases or provenance
  (#590) — Existing ABS author provenance and aliases are reused only when
  they still match the local author, preventing secondary-author names from
  corrupting future imports.
• Direct book adds preserve series links (#590) — Adding a book directly
  no longer drops existing series associations during metadata
  canonicalization.
• Google Books provider settings are respected at startup (#590) —
  Bindery now prefers the UI-managed Google Books API key, keeps legacy
  setting fallback for existing installs, and treats a deliberately cleared
  UI setting as disabled.