Switch nightly discovery refresh to scrape Audible's curated HTML storefronts (popular, new releases, category pages) while keeping real-time user paths on the JSON catalog API. Add robust HTML resilience knobs (increased retries, capped jittered backoff, AdaptivePacer changes and per-batch cooldowns) to avoid failing nightly jobs during 503 storms. Implement multi-narrator capture via a new extractAllNarrators helper and update parsers to preserve all narrator anchors. Introduce two-pass dedup: in-memory deduplicateAndCollectGroups + collapseByExistingWorks that consults the works table, export metadataScore for consistent representative selection, and persist dedup groups (fire-and-forget). Wire collapseByExistingWorks into search/author/series routes and make defensive dedup in the refresh processor. Add HTML parsing helpers, runtime/lang-aware parsing, jitteredBackoff cap, and tests for the new behaviors.
The duration column (Int/int4, max ~2.15B) overflows when storing
millisecond values for items with large durations from Audiobookshelf
or Plex backends. Change to BigInt (int8) and wrap duration calculations
in BigInt() at the Prisma write boundary.
Changes:
- prisma/schema.prisma: PlexLibrary.duration Int? → BigInt?
- plex-recently-added.processor.ts: BigInt(Math.round(...)) wrapping
- scan-plex.processor.ts: same BigInt wrapping
- documentation/backend/database.md: updated duration type notation
Fixes#193
Co-Authored-By: Oz <oz-agent@warp.dev>
Adds request_grabbed event that fires when a torrent/NZB is successfully
handed off to the configured download client, filling the gap between
request_approved (pre-search) and request_available (fully imported).
- Add request_grabbed to NOTIFICATION_EVENTS with titleByRequestType
(Audiobook Grabbed / Ebook Grabbed), info severity, Details messageLabel
- Add NotificationEventConfig interface and update getEventMeta() return
type to expose messageLabel to all providers without TypeScript errors
- Add messageLabel: 'Reason' to issue_reported event
- Fix all 4 providers (Discord, ntfy, Pushover, Apprise) to derive message
field label from meta.messageLabel ?? 'Error' instead of hardcoded
isIssue ternary — prevents grab details showing as Error
- Trigger request_grabbed in download-torrent.processor.ts after
client.addDownload() succeeds; message carries torrent title, indexer,
and download client name; requestType sourced from request.type
- Update notifications.md documentation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Restructure and expand tests for AudibleService: replace a single hoisted axios client mock with separate htmlClientMock and apiClientMock, update axios.create to return clients in initialization order, and remove the fs mock. Add reusable fixture helpers (makeProduct, makeProductsResponse, apiResponse) and many new/spec-complete test cases organized into describe blocks (initialization, search, mapping, series rules, author search, popular/new releases, categories, and audiobook details). Improve assertions for pagination, deduplication, field mapping, error handling, and region/config behavior; reset and clear mocks in beforeEach to ensure isolation.
Move Audible catalog operations from HTML scraping to Audible's unauthenticated JSON catalog API (/1.0/catalog/*) while keeping Audnexus as the primary per‑ASIN detail source. audible.service.ts: remove cheerio parsing, add apiClient/htmlClient split, CATALOG_RESPONSE_GROUPS constant, catalog response types, stripHtml and mapCatalogProduct mappers, and paging (API is 0-indexed) + author-ASIN client-side filtering. Update search, popular, new-releases and author endpoints to call the catalog API, use apiClient for retries/backoff, and preserve htmlClient only for series-page scraping and link generation. Improve retry logic to accept an Axios client, move to jittered/exponential backoff for API/external calls, and adjust delays/AdaptivePacer usage. Documentation updated to reflect architecture, data sources, region handling, and gotchas.
Add a new client-side Path Mapping Helper page at src/app/path-helper/page.tsx. Implements a multi-step wizard to help users configure Docker volume mappings for download clients and ReadMeABook (RMAB): select clients, enter container save paths, enter host/container volume mappings (with optional remote path mapping), and generate recommended RMAB docker-compose volume snippet. Includes utility functions to compute common roots and relative paths, UI components (step indicator, info/warning boxes, code block), and logic to derive RMAB download directory, per-client custom paths, and verification instructions. No API calls — purely client-side helper with sensible defaults for supported clients.
Adds an info icon button (top-right of each card) in the Requests
Awaiting Approval section. Clicking it opens AudiobookDetailsModal
with full book details (cover, description, narrator, series, genres,
etc.) and embeds the Approve / Search / Deny action buttons so admins
can review and act without navigating away from the admin panel.
Implementation:
- AudiobookDetailsModal: adds optional `adminActions` prop rendered as
a second row inside the existing sticky action bar
- admin/page.tsx: adds detailsAsin/detailsRequestId state, info button
per card (conditional on audibleAsin presence), and AudiobookDetailsModal
wired with admin action buttons matching the card button behaviour
- Documentation updated: request-approval.md, components.md, TABLEOFCONTENTS.md
Closes#157
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add support for selecting individual audio files during manual and bulk imports and pass that selection through the scan, API, job queue, processor and organizer.
Key changes:
- API: scan now returns audioFiles for each discovered book and emits a new 'grouping' progress phase; execute and manual-import routes accept file lists (audioFiles / selectedFiles) and validate them.
- Scanner: group loose audio files by metadata (title/author/narrator), deduplicate multi-part sets (CD1/CD2) across folders, and return audioFiles + groupingKey; add concurrency limit for ffprobe reads and merge groups post-scan.
- Job queue & processor: OrganizeFiles payload now includes selectedFiles; processors forward selectedFiles to the FileOrganizer and to cleanup logic.
- File organizer & cleanup: filter to only selectedFiles when organizing; cleanup now deletes only the selected files (if provided) instead of removing the whole directory.
- UI: Manual import browser and bulk import wizard updated to show per-file selection, track checkedFiles, toggle all, and send selected files to the API; ConfirmPhase updated to allow checking/unchecking files and prevents starting import with no files selected.
- Filesystem browse: removed expensive per-subfolder stats to keep browsing responsive (now lists subdirectories without nested stat calls).
Overall this change enables finer-grained imports, reduces accidental deletion of unselected files, and improves scan grouping for multi-folder audiobooks.
Add sessions_invalidated_at to users (migration + Prisma schema) to support immediate session revocation. Set sessionsInvalidatedAt when an admin revokes a user's login token and enforce revocation checks in auth middleware and the refresh endpoint (compare token iat against sessionsInvalidatedAt). Add optional iat fields to JWT payload types. Scrub token from browser history after token-login. Consolidate rate-limiting logic into src/lib/utils/rateLimit.ts (rename/merge previous auth/apiToken rate limiter implementations), remove the old apiTokenRateLimit.ts, and update imports and tests to use the new module.
Introduce a Bulk Import feature for admins to scan server folders, match discovered audiobook folders against Audible, review matches, and queue batch imports.
What changed:
- Added documentation: documentation/features/bulk-import.md and TABLEOFCONTENTS update.
- Backend: SSE scan endpoint (POST /api/admin/bulk-import/scan) streams discovery and matching events; execute endpoint (POST /api/admin/bulk-import/execute) validates paths, creates/resolves audiobook & request records, and queues organize_files jobs. Both endpoints enforce admin-only access and validate allowed root directories (download_dir, media_dir, /bookdrop).
- Frontend: Modal wizard and steps for folder selection, scan progress, and match review (BulkImportWizard + ScanFolderStep, ScanProgressStep, MatchReviewStep + shared types).
- Utilities: bulk-import-scanner for folder discovery and ffprobe metadata extraction; shared types for scanned books/events.
- UI: Added Bulk Import quick action to admin dashboard (src/app/admin/page.tsx).
Key details:
- Audible searches are rate-limited (≈1.5s) and matching results include library/request status checks.
- Reuses existing organize_files job queue and manual-import pipeline; no new database tables introduced (state is ephemeral during the wizard).
- Includes error handling, path normalization, and security checks for allowed directories.
This commit wires frontend, backend, and docs together to provide an admin-only multi-step bulk import workflow.
Introduce a per-user "ignored audiobooks" feature to suppress auto-requests. Changes include:
- Database: add Prisma model IgnoredAudiobook and SQL migration to create ignored_audiobooks table with indexes and FK to users.
- Backend: new API routes to list, add, delete, and check ignored audiobooks (/api/user/ignored-audiobooks, /check/:asin, /:id). Add annotateWithIgnoreStatus utility and integrate it into multiple audiobook list endpoints (popular, new-releases, category, search, authors, series).
- Request creator: add ignore-list check (with sibling-ASIN expansion) and a bypassIgnore option for manual requests; return an 'ignored' reason when blocked.
- Frontend: hooks (useIsIgnored, useToggleIgnore, useIgnoredList) and UI updates — AudiobookCard shows an "Ignored" indicator and AudiobookDetailsModal adds an ignore toggle and propagates local state changes.
- Misc: adjust deduplication duration tolerance (to 5% / min 10 minutes), tweak SWR refresh intervals for shelves/syncing, and small logging/info updates.
- Tests: add unit tests for request-creator ignore logic and update existing tests/mocks to account for ignore annotation; extend prisma test helper with ignoredAudiobook mock.
This commit implements the ignore-list end-to-end (DB, server, client, and tests) so users can ignore specific ASINs and have auto-request flows respect that preference.
Include the requesting user's ID as an additional argument when enqueueing immediate shelf sync jobs so the job has user context. Updated the route implementation and adjusted affected tests (goodreads-shelves-id, hardcover-shelves-id, and hardcover-shelves routes tests) to expect the extra 'user-1' parameter.
Introduce an autoRequest boolean on Goodreads and Hardcover shelves (default true) so users can pause/resume automatic request creation. Schema, API handlers, hooks and types were updated to accept and persist autoRequest when creating or updating shelves; add endpoints only trigger an immediate resync when the feed/token changes. The shelf sync core and service code now respect autoRequest (skipping request creation and annotating logs when disabled). UI updates include an AddShelf toggle, manage/update payload changes, shelf list props, and visual indicators + toggle actions in the shelf cards.
Introduce file and directory permission settings (fileChmod, dirChmod) end-to-end. UI: new controls in Paths settings with octal validation and defaults (664/775). API: GET exposes defaults; PUT validates octal strings and upserts configuration keys (file_chmod, dir_chmod) and clears related cache keys. Runtime: read config values in file utilities and services (FileOrganizer, direct-download, chapter-merger, epub-fixer) to apply mkdir modes and chmod files/dirs; FileOrganizer now accepts fileMode/dirMode and getFileOrganizer reads/parses DB settings. Docker: add UMASK option to docker-compose and propagate/apply UMASK in entrypoint/app-start scripts. Tests: update mocks to account for config service usage.
Add support for passing sourceHeaders when fetching NZB/torrent files: extend AddDownloadOptions and SABnzbd AddNZBOptions, forward headers in sabnzbd and nzbget clients, and populate sourceHeaders in download-torrent.processor (injecting Prowlarr API key as X-Api-Key for proxy URLs). Make OIDC request scope conditional: only include the 'groups' scope when group-based access control or admin-claim is enabled (update provider logic, add tests, and update setup UI text). Also remove explicit take:100 in Plex processors and add CLAUDE guidance about requesting approval before implementing code changes.
kikootwo