Conversation
Stop probing the filesystem when computing an audiobook's wanted flag and treat DB file records as the source of truth. Make ComputeWantedFlag static and return wanted based on presence of file records (no File.Exists checks or per-file logging). Simplify the GetAll endpoint to AsNoTracking, include Files, and order results by Title (removed QualityProfile include + retry fallback). Update frontend comment to trust server semantics for wanted status and add a unit test to prevent regression where a DB file record would be treated as desired only if the file exists on disk.
Introduce request deduplication for library fetches by adding an inFlightFetch promise to the library Pinia store so concurrent fetchLibrary() calls share a single API request. Update App.vue to use the library store (fetchLibrary and audiobooks) instead of calling the API directly to avoid duplicate calls and keep wanted badge/search logic aligned with store state. Update tests to initialize Pinia when mounting App.vue and add a new library-fetch.spec.ts that verifies concurrent fetch deduplication. Bump project versions to 0.2.56.
Frontend: add computeAudiobookStatus util, AudiobookStatus type and format helper; integrate into AudiobooksView and CollectionView; derive wantedCount from hydrated library store (remove polling) and sync via SignalR onConnected; set wanted/status when files change in library store; normalize download client host input in DownloadClientFormModal and update tests; add unit tests for audiobook status and activity badge. Backend: add LibraryAudiobookListItemDto and change LibraryController.GetAll to return a slim payload with Wanted and Status computed (uses new AudiobookStatusEvaluator); add DownloadClientUriBuilder to centralize URI/authority construction and update Nzbget/Qbittorrent/Sabnzbd/Transmission adapters to use it; various logging and request improvements. Tests updated/added for adapters and controller behaviors.
Replace a broad `as any` assertion with `ReturnType<typeof useDownloadsStore>['downloads']` in fe/src/__tests__/WantedView.spec.ts to improve type safety for the mocked downloads array. This makes the test's typings accurate and helps catch type-related issues earlier.
Adds GetQueueAsync_NormalizesHostWithSchemeAndPath unit test to NzbgetAdapterTests.cs. The test uses a DelegatingHandlerMock and TestHttpClientFactory to capture the outgoing request URI when the DownloadClientConfiguration.Host contains a scheme and path (e.g. "http://192.168.50.111/nzbget"). It asserts the adapter normalizes the URI to the provided scheme/host, uses the configured port, and targets the "/xmlrpc" path, and that the returned queue is empty.
![github-code-quality[bot]](https://avatars.githubusercontent.com/u/9919?s=60&v=4){kind=small}
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 4a6506172f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
Pull request overview
This PR optimizes the library page performance and download client connection handling. For large libraries, the /library endpoint was slow due to per-audiobook filesystem existence checks and excessive data returned. The PR introduces a slim list DTO, moves status computation server-side, and introduces a shared DownloadClientUriBuilder to normalize host/scheme/port handling across all download clients.
Changes:
- Introduced a slim
GET /librarylist response (LibraryAudiobookListItemDto) with server-computed status viaAudiobookStatusEvaluator, replacing the old hybrid list/detail payload that included full file metadata and performed filesystem probes. - Added
DownloadClientUriBuilderto normalize download client URI construction (handling embedded schemes, paths, and explicit port overrides) across NZBGet, qBittorrent, SABnzbd, and Transmission in both adapters and monitor service. - Moved audiobook status computation to shared helpers (
AudiobookStatusEvaluator.csbackend,audiobookStatus.tsfrontend) and switched thewantedCountbadge from periodic polling to a store-driven computed property.
Reviewed changes
Copilot reviewed 31 out of 33 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
listenarr.api/Services/Adapters/DownloadClientUriBuilder.cs |
New shared URI builder normalizing scheme/host/port/path for all download client adapters |
listenarr.api/Services/AudiobookStatusEvaluator.cs |
New shared backend service computing audiobook list status from DB data |
listenarr.api/Models/LibraryAudiobookListItemDto.cs |
New slim DTO for GET /library list response |
listenarr.api/Controllers/LibraryController.cs |
Refactored GetAll() to use slim DTO + server status; removed filesystem checks from ComputeWantedFlag |
listenarr.api/Services/DownloadMonitorService.cs |
Replaced inline URL construction with DownloadClientUriBuilder calls |
listenarr.api/Services/DownloadService.cs |
Replaced inline qBittorrent URL construction with DownloadClientUriBuilder |
listenarr.api/Services/Adapters/NzbgetAdapter.cs |
Replaced inline URL construction with DownloadClientUriBuilder; removed credential-in-URL manual encoding |
listenarr.api/Services/Adapters/QbittorrentAdapter.cs |
Replaced all inline URL constructions with DownloadClientUriBuilder.BuildAuthority |
listenarr.api/Services/Adapters/SabnzbdAdapter.cs |
Replaced all inline URL constructions with DownloadClientUriBuilder.BuildUri |
listenarr.api/Services/Adapters/TransmissionAdapter.cs |
Replaced inline URL construction in BuildBaseUrl with DownloadClientUriBuilder.BuildUri |
fe/src/utils/audiobookStatus.ts |
New shared frontend utility for status computation and formatting |
fe/src/types/index.ts |
Added AudiobookStatus type and status field to Audiobook interface |
fe/src/stores/library.ts |
Added in-flight fetch deduplication; updated applyFilesRemoved to maintain status/wanted |
fe/src/App.vue |
Switched wantedCount from polling to computed from library store; load library on mount |
fe/src/views/library/AudiobooksView.vue |
Replaced local status computation with shared computeAudiobookStatus utility |
fe/src/views/library/CollectionView.vue |
Replaced local status computation with shared computeAudiobookStatus utility |
fe/src/components/domain/download/DownloadClientFormModal.vue |
Added normalizeHost to strip scheme/path from host field on save/test |
tests/Listenarr.Api.Tests/NzbgetAdapterTests.cs |
New tests for NZBGet host normalization and queue path |
tests/Listenarr.Api.Tests/SabnzbdAdapterTests.cs |
New test for SABnzbd host normalization |
tests/Listenarr.Api.Tests/QbittorrentAdapterTests.cs |
New test for qBittorrent host normalization |
tests/Listenarr.Api.Tests/TransmissionAdapterTests.cs |
New test for Transmission host normalization and custom RPC path |
tests/Listenarr.Api.Tests/LibraryController_WantedFlagRegressionTests.cs |
New regression test for DB-backed wanted-flag logic |
tests/Listenarr.Api.Tests/LibraryController_LibraryListSlimPayloadTests.cs |
New regression test for slim /library payload and status |
fe/src/__tests__/audiobookStatus.spec.ts |
New frontend tests for computeAudiobookStatus |
fe/src/__tests__/library-fetch.spec.ts |
New frontend test for concurrent fetch deduplication |
fe/src/__tests__/AppActivityBadge.spec.ts |
Updated test stubs for new SignalR event and Pinia setup |
fe/src/__tests__/DownloadClientFormModal.spec.ts |
Updated test to exercise normalizeHost |
fe/src/__tests__/WantedView.spec.ts |
Fixed type cast to use proper store type |
CHANGELOG.md |
Added release notes for version 0.2.57 (contains version mismatch) |
package.json / package-lock.json / fe/package.json / fe/package-lock.json |
Version bump to 0.2.56 |
Files not reviewed (1)
- fe/package-lock.json: Language not supported
You can also share your feedback on Copilot code review. Take the survey.
Add support for fileCount across frontend and backend, improve torrent/magnet handling, and expand tests. Frontend: add Audiobook.fileCount, update library store to compute fileCount and wanted/status logic, use fileCount in filter evaluators and custom filters, simplify wantedCount compute, and retry library sync on SignalR reconnect even if initial hydrate failed (new tests added/updated). Backend: include FilePath, FileSize, and FileCount in Library DTO and controller slim list payloads. Download client: introduce DownloadClientUriBuilder helpers (ResolveTorrentAddTarget, NormalizeMagnetLink, TryParseHttpOrHttpsAbsoluteUri) and update qBittorrent/Transmission adapters to uniformly handle magnet vs .torrent URLs and pre-download logic. Tests: add unit tests for audiobook status evaluator and DownloadClientUriBuilder and update existing adapter tests to reuse response objects.
Clear isMagnetTarget assignment when an indexer redirects to a magnet link in QbittorrentAdapter and TransmissionAdapter. The code now normalizes and logs the magnet URL without toggling the isMagnetTarget flag to avoid carrying redundant or potentially incorrect state into downstream logic.
Normalize and prefer a validated magnet link or an HTTP(S) torrent URL when adding torrents. Parse and normalize the magnet link, attempt to parse an absolute HTTP/HTTPS torrent URL, and prefer the HTTP URL for pre-downloading torrent bytes (so authenticated/private-tracker content can be added via file data). Use the downloaded result's magnet redirect when necessary and set torrentUrl appropriately. Move inline magnet-hash extraction into a new TryExtractMagnetHash helper and throw if neither a magnet nor a torrent URL is available. Adjusted download call to use the validated httpTorrentUrl and simplified related logic.
Introduce NormalizeTorrentUrl helper in qBittorrent and Transmission adapters to validate/normalize HTTP(S) torrent URLs and centralize parsing. Refactor pre-download flow: prefer pre-downloading a valid HTTP(S) .torrent when present (even if a magnet exists), consolidate torrentUrl selection, and add TryPredownloadTorrentFileAsync with safe exception handling and logging. Update tests to cover pre-download behavior and invalid-scheme validation, plus add test helpers for HTTP listener and response writing. This prevents accidental network calls for invalid schemes and reduces duplicated logic.
tsolo4ever
left a comment
tsolo4ever
left a comment
There was a problem hiding this comment.
Claude AI Code Review — Library Optimizations #401
Overall this is a solid improvement — the slim DTO approach meaningfully reduces payload size and moving file-existence logic out of the list endpoint is the right call. A few things worth discussing before merge:
Potential regression: ComputeWantedFlag drops legacy FilePath check
LibraryController.cs:103-114
The simplified version only checks audiobook.Files (the AudiobookFile collection). Any book tracked via the legacy FilePath field with no corresponding AudiobookFile row will now show as Wanted even though the file exists on disk. Is this intentional? If there's no migration step to backfill AudiobookFile records from existing FilePath values, users upgrading from older versions could see previously-found books flip to missing.
wanted logic is duplicated
LibraryController.cs:682
GetAll() computes var wanted = a.Monitored && !hasFiles; inline, bypassing the ComputeWantedFlag static method defined earlier in the same file. They're equivalent now, but if one changes the other won't follow. Worth consolidating to a single call.
Redundant NoFile branch in AudiobookStatusEvaluator
AudiobookStatusEvaluator.cs:38-46
Both wanted == true and !hasAnyFile return NoFile. Since wanted is derived from files.Count == 0, the second branch can never be reached when the first fires. Either one is dead code, or the intent was to distinguish "monitored with no file" from "unmonitored with no file" — in which case !hasAnyFile should return a different status.
FilePath still in the slim DTO
LibraryAudiobookListItemDto.cs:22
The slim DTO kept the legacy FilePath string field. If the intent of this PR is to make AudiobookFile records the source of truth for file state, including FilePath in the list response perpetuates client-side reliance on the legacy field. Worth being explicit about whether it stays for backward compat or gets dropped. For reference, the current frontend reads filePath in at least three places (audiobookStatus.ts, library.ts, WantedView.vue) for file-existence checks — those callers should eventually migrate to use fileCount/status instead. We'll handle the adjustments needed on our end when we rebase our PR on top of this.
nit: activeDownloadStatuses array allocated on every request
LibraryController.cs:660-667
This array is recreated on every GET /library call. A private static readonly field would be cleaner.
nit: DeriveQualityLabel lossless detection is incomplete
AudiobookStatusEvaluator.cs
The lossless check covers flac, alac, wav but misses aiff, ape, dsd, wv (WavPack). Rare in audiobooks but worth noting for completeness.
Thanks for the work on this — the query split approach (audiobooks → files → quality profiles → downloads) is the right call for keeping the list endpoint fast. Consider a brief inline comment explaining why it's structured as 4 queries rather than joins, just to help future contributors understand the intent.
…nings - audiobookStatus.ts: remove dead filePath/fileSize check inside hasFiles branch — was always quality-match since || hasFiles was always true - WantedView.vue: fallback wanted logic now prefers fileCount (slim DTO) over files[] length, drops hasPrimaryFile (legacy filePath) check - LibraryController.cs: add missing deleteFiles XML param tag (CS1573) - ManualImportController.cs: null-forgiving on metadata (already guarded above, surfaced by stricter nullable settings from PR Listenarrs#401 merge) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…on 0.2.57) Keep fork version 0.2.111 over upstream 0.2.57. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Small optimization release. I noticed that for users with larger libraries, the library page would take a long time to load. This was due to some redundant and unnecessary checks. Also the the api endpoint was returning a lot of data that was not needed for the page. Additionally I have added url resolution normalizations to download client host/ip:port connection strings so the connection tests are more robust.
Added
/libraryGET /library/librarypayload behaviorChanged
GET /libraryfrom a hybrid list/detail payload into a lighter list responseGET /library/{id}as the rich full-detail audiobook endpoint/libraryrequests into a single in-flight fetchFixed
/libraryresponses caused by per-audiobook filesystem existence checks during list loading/libraryrequests during startup and view initializationName or service not known (http:80)Removed
/librarylist pathfiles[]metadata in library list views for status calculation