## v0.9.22 ### Fixed: Image Cache settings hint showed the current path instead of the actual default **The "Default:" label under Settings → Image Cache echoed `EffectiveImageCachePath`**, which resolves to whatever path is currently set — so once a user picked a custom folder, the label just repeated that custom folder back at them labeled "Default," which is meaningless. Added `AppSettings.DefaultImageCachePath` (the real static `%APPDATA%\AbigailsMediaRenamer\art_cache` fallback) and pointed the hint at that instead, with clearer wording ("Default (used when this is left empty): ..."). --- ## v0.9.21 ### Renamed "Automatically repackage MP4 files as MKV" setting to "Convert Incoming Video Files to MKV" **The label and description in Settings → Transcode still said "MP4" even though the underlying `ConvertToMkvService.RemuxableExtensions` list was widened to `.mp4/.avi/.mov/.wmv/.m4v` a while back** — the feature already converted anything not already MKV, the wording just hadn't caught up. Updated the checkbox text and helper text in `SettingsWindow.xaml` to match, and renamed the backing property `AppSettings.AutoRemuxMp4ToMkv` → `AutoConvertToMkv` (and its two call sites in `AutoRenameService.RenameAndMoveAsync`/`RenameService.RenameAsync`) so the name doesn't lie either. No functional change to the conversion logic itself. Note: renaming the property changes its JSON key in `settings.json`, so the saved on/off state resets to the default (off) once — a deliberate tradeoff since there's no back-compat need here. --- ## v0.9.20 ### Added automated test coverage for the pure-logic parsing/scoring/path-building services **New `AbigailsMediaRenamer.Tests` xUnit project** covering `ParserUtilities` (JaroWinkler, non-video title cleaning/scoring, audiobook duration/candidate scoring, tag completeness), `MediaParser` (movie/TV title cleaning, casing, storage-name formatting), `TvParser`/`MovieParser`'s `ScoreConfidence` formulas, `NamingConventionService` (path building, live preview, template validation), and `DirectoryCacheService` (fuzzy folder-name matching). 191 tests, deliberately adversarial sample data (mixed dot/underscore/space separators, boundary confidence values, colon-vs-comma title conventions, tiny/substring/duration-mismatched audiobook candidates, malformed templates) chosen specifically to catch the shape of bug that's shipped before rather than just happy-path cases. Added `.slnx` entry, `AbigailsMediaRenamer/AssemblyInfo.cs` `[InternalsVisibleTo("AbigailsMediaRenamer.Tests")]`, and bumped a handful of `private` scoring/normalization methods to `internal` specifically so the suite can exercise them directly (`TvParser`/`MovieParser.ScoreConfidence`, `DirectoryCacheService.ComputeMatchScore`/`NormalizeForComparison`) — no behavior changes, visibility only. `MediaParser.FormatTitleForStorage` had its logic extracted into a new `internal static` overload taking `movePrefixesToEnd` as a parameter, since the instance method otherwise required constructing a full `MediaParser` (and transitively `MetadataProviderRegistry`, which reads/writes the real `%APPDATA%` settings file — not something a test run should touch). ### Fixed: comma-article title convention ("Walking Dead, The - Dead City") didn't fuzzy-match its colon-subtitle equivalent **Found by the new test suite, not a user report.** The v0.9.15 fix added leading- and trailing-article stripping to `DirectoryCacheService.NormalizeForComparison` specifically so `"Walking Dead, The - Dead City"` (on-disk storage convention) and `"The Walking Dead: Dead City"` (freshly-parsed title) would normalize to the same string and fuzzy-match as the same folder. They didn't, for exactly the scenario the fix targeted: once the generic punctuation-to-space pass converts the comma and dash to spaces, `"The"` ends up in the *middle* of the string (`"walking dead the dead city"`), past where either the leading or trailing article regex can reach it — only a `the`/`a`/`an` at a true string boundary gets stripped. Fixed by adding an explicit `,\s*(?:the|a|an)\b` strip that removes the comma-article pair as a unit *before* the generic punctuation pass runs, so it no longer depends on where the article happens to land afterward. --- ## v0.9.19 ### Fixed: movie title cleaning left fragments like "DDP5 1 H 264" behind instead of stripping them **Scene tags and dotted quality tokens (`DDP5.1`, `H.264`, the built-in `5.1`/`7.1` entries, or any custom tag a user adds with a literal `.` in it) were being matched against the title AFTER punctuation had already been normalized to spaces.** `MediaParser.CleanMovieTitle` converted `.`/`_` separators to spaces before running the scene-tag stripping loop, so a tag pattern like `5\.1` (from `Regex.Escape("5.1")`) no longer had a literal dot to match against — by that point the text read "5 1", not "5.1" — leaving mangled fragments in the cleaned title instead of removing them. Moved the dot/underscore-to-space normalization to run after scene-tag stripping (and the release-group suffix strip) instead of before, so tags are matched against the original text while their internal punctuation is still intact. TV title cleaning (`CleanTvTitle`) doesn't do scene-tag stripping at all, so it wasn't affected. --- ## v0.9.18 ### Audiobook matching, round 2 — adopted AudiobookShelf's actual techniques (still no AI) **Investigated AudiobookShelf's real matching source** (`BookFinder.js`, `Audible.js`, `Audnexus.js`, `scandir.js` from github.com/advplyr/audiobookshelf, not just a summary) at the user's request, since ABS finds audiobooks far more reliably than this app did even after last session's fixes. **Search now hits Audible's own catalog directly, not just Kindle-store scraping.** Found that ABS's "Audible integration" is nothing more than Audible's own public, unauthenticated catalog endpoint (`api.audible{tld}/1.0/catalog/products`) — no partner API, no secret key, the same endpoint Audible's own site uses. Our primary search source was `KindleSearchClient` scraping Amazon's *Kindle e-book* store results (not audiobook-specific, fragile HTML parsing). Added `AudibleCatalogClient` hitting that same endpoint as the new primary search tier in `AudiobookProvider`, with each returned ASIN immediately enriched via the existing Audnexus client (gives every candidate full data, including duration, right from search). Kindle-scrape and Audnexus-search are now fallback tiers 2 and 3, unchanged otherwise. **Duration is now part of match confidence, weighted 70%.** ABS's `calculateMatchConfidence` blends `0.7×duration + 0.2×title + 0.1×author` — duration dominates because recording length is close to a fingerprint for a specific edition, where title/author text can't distinguish abridged vs. unabridged, wrong-but-similar-titled books, or omnibus editions. Added `ParserUtilities.ScoreDurationMatch` (same piecewise-linear curve ABS uses) and blended it into `ScoreAudiobookCandidate` with the same weights when both the local file's duration (read via TagLib) and a candidate's runtime are known — falls back to last session's title/author-only blend otherwise. Guarded against a real gotcha: a single file's duration is only meaningful for a consolidated single-file audiobook (typically `.m4b`) — for a folder split into many chapter/part files, one file's duration has nothing to do with the book's total runtime, so that case is detected and duration scoring is skipped for it. **Folder-derived author guesses are now validated instead of trusted blindly.** Added `AudnexusClient.ValidateAuthorAsync` (mirrors ABS's `Audnexus.findAuthorByName`) — checks a guessed author name against Audnexus's real author-search endpoint before using it as a search filter; a bad guess (e.g. a series name mistaken for an author) now gets dropped instead of polluting the search. Tag-sourced authors are already trustworthy and skip this. **Structured folder-name tokens**, mirroring ABS's `scandir.js` grammar: trailing `{Narrator}`, a leading `(YYYY) - ` / `YYYY - ` published year, and a leading `Book N -` / `Vol. N` / bare-number series sequence — each now extracted from the folder name as fallback signals (last session only used the whole folder name as a title fallback). Caught and fixed a regex edge case during implementation where a plain year-like title (e.g. "1984") would have been misparsed as a bogus sequence number — the sequence pattern now requires an explicit `.`/` - ` separator after the number, matching ABS's own documented exclusion rule. **Small scoring guards**: a substring short-circuit (if the query is literally contained in the candidate title/author or vice versa, treat as a strong match rather than trusting edit-distance alone) and a tiny-query guard (titles under 5 characters now require an exact match rather than a fuzzy one) — both taken directly from `BookFinder.js`'s `filterSearchResults`. Deliberately not adopted: ABS's full multi-provider fan-out (Google/iTunes/OpenLibrary/FantLab/AudiobookCovers — FantLab is Russian-only, AudiobookCovers is cover-art-only and out of scope this round) and its combinatorial fuzzy title×author retry cascade (too large/risky a change for uncertain marginal benefit now that the primary search-source mismatch is fixed). --- ## v0.9.17 ### Improved audiobook identification/matching quality (no AI involved) **Investigated `philipvox/audiobook-tagger-web`, an AI-based AudiobookShelf metadata cleaner, at the user's request** ("our audiobook lookups are REALLY not great"). It turned out to be a post-identification cleanup tool (assumes AudiobookShelf already matched the book, then uses GPT/Claude to reconcile fields) rather than a better matcher — not adopted, and no LLM/AI dependency was added, per explicit instruction. Comparing its prompt-engineered rules against our own `AudiobookParser`/`ParserUtilities` did surface several concrete, non-AI gaps that were fixed: **Search results were never re-ranked — we always took the first hit.** `MetadataService.SearchAudiobookAsync` returns a full candidate list, but `AudiobookParser` blindly used `hits[0]` regardless of whether its author matched the file's tags, so a same-ish-titled book by a different author could win over the actually-correct match. Added `ParserUtilities.ScoreAudiobookCandidate` (title similarity blended with author similarity when both sides have one) and score every returned candidate instead of just the first. **Junk-suffix stripping was thin.** `CleanNonVideoTitle` (shared by Audiobook/Book/Music/Comic parsers) only stripped leading track numbers — now also strips "(Unabridged)", "(Abridged)", bitrate tags (128/256/320kbps), "(MP3)", "(M4B)", "(Full Cast)", "(Complete)", "(HQ)" in bracketed or bare form. **Narrator was never read from the filename.** Added "Read by X" / "Narrated by X" / "Performed by X" extraction from the raw filename as a last-resort fallback (ASIN and API-sourced narrator data still take priority). **Folder structure was never consulted.** Audiobook releases are commonly organized `Author/Book Title/file.m4b` or `Author/Series/Book Title/file.m4b`. When the embedded tag title is missing or generic ("Track 1", "Audiobook", "CD 2", a bare number), the parser now falls back to the folder name; when no author tag exists, the grandparent folder is used as a soft signal for search/scoring only — never overwrites real tag data. **Series names could drift across books in the same series.** Added `AudiobookParser.NormalizeSeriesName`, which checks the library for this author's other known series spellings (reusing the `MovieParser`/`TvParser` precedent of a parser opening its own `MediaDbContext`) and normalizes a newly-fetched series name to the already-established one when they're a near-exact Jaro-Winkler match but not identical — so book 3 doesn't end up with a subtly different series string than books 1-2. --- ## v0.9.16 ### Comprehensive Kodi/Jellyfin metadata export pass for Movies and TV **Poster art now writes `poster.jpg` for Kodi setups too, not just Jellyfin.** Art always downloaded to `cover.jpg`, and was only ever copied to the Kodi-standard `poster.jpg` filename when a Jellyfin-specific setting was on — Kodi-only setups never got recognized poster art at all. `NfoWriterService` now writes `poster.jpg` whenever Kodi export is enabled, independent of the Jellyfin setting. **Auto-rename NFOs now carry full metadata instead of a bare stub.** The automatic watcher pipeline previously wrote `movie.nfo`/`tvshow.nfo` straight from the parsed filename — no rating, director, IMDB id, or cast/crew — even when the database already had all of that from a prior manual fetch. It now looks up the just-saved database row and reuses the same full export path the manual "Export NFO" action uses. **Added an "auto-download art during auto-rename" option (off by default).** Previously art only ever downloaded via manual Library Manager actions — the watcher never touched poster/fanart/logo/banner/clearart on its own. New Settings → Metadata Export toggle wires the watcher into the same art pipeline. **Fixed TV shows and movies identified via TVDB never getting an IMDB id or cast.** Only the TMDB identification path populated either — the TVDB path silently left both empty. Added TVDB's `remoteIds`/`characters` data to close the gap. **Added per-episode `.nfo` export — the biggest gap.** No `.nfo` was ever written for individual episodes anywhere in the app, so episode synopsis/air-date/rating already sitting in the database, and guest-star/episode-crew data TMDB/TVDB can provide, never reached Kodi/Jellyfin at all. Two new settings (both off by default): one writes a Kodi `` `.nfo` next to each episode during auto-rename and library backfill; the other opts into an extra per-episode TMDB/TVDB request for guest stars, episode director/writer, and the episode thumbnail image — both providers now supported, not just TMDB. **Wired up a dead setting.** `ExportIncludeCast` existed in Settings but was never actually read anywhere — cast now genuinely gets included or excluded from NFOs based on it. **Rounded out remaining NFO fields.** Original title, tagline, MPAA/content certification, keywords, trailer, and movie collection/set data — all things TMDB already returns but the app previously never requested or stored — now flow through to `movie.nfo`/`tvshow.nfo` when available. Deliberately did *not* add non-standard tags Kodi doesn't recognize (producer/composer/cinematographer/editor credits, sorttitle, outline, country, userrating, votes, top250) — that data stays in the database/UI but isn't invented as NFO XML nothing will read. --- ## v0.9.15 ### Fixed: colon-subtitled TV/movie titles (e.g. "The Walking Dead: Dead City") didn't match their existing library folder, creating a mangled duplicate instead **Reported: an existing folder `Walking Dead, The - Dead City (2023)` was ignored for new episodes of "The Walking Dead: Dead City"** — the app instead created `Walking Dead Dead City, The (2023)`, with "Walking Dead" and "Dead City" mashed together and "The" stranded at the very end. Two independent bugs compounded. **`MediaParser.FormatTitleForStorage` moved a leading article ("The"/"A"/"An") to the end of the entire title string without first splitting off a colon-separated subtitle.** For "The Walking Dead: Dead City" this produced `"Walking Dead: Dead City, The"` — the colon (not a valid filename character) then got silently stripped by `NamingConventionService.SanitizeComponent`, collapsing the double space left behind and mashing the two title segments together with no separator at all. Fixed by splitting off the `": Subtitle"` portion first, moving the article only on the base title, then rejoining with `" - "` — reproducing the existing on-disk convention (`"Walking Dead, The - Dead City"`) instead of destroying it. **`DirectoryCacheService.FindBestMatchingDirectory` (the "does an existing series folder already exist?" check used before building a new one) normalized names via a completely separate, simpler function that only stripped a leading article and no punctuation at all.** So `"walking dead, the - dead city"` tokenized into `dead,`/`-`/`the` fragments that didn't line up with the (already-mangled) query, dropping the match score below threshold even before the bug above is accounted for. `NormalizeForComparison` now also strips `,`/`:`/`-` before word-splitting and strips a trailing `"... the"`/`"... a"`/`"... an"` in addition to a leading one, so both the `"The X: Y"` and `"X, The - Y"` conventions normalize to the same word set. Both fixes are in shared, single-implementation functions used for movies and TV alike, so movies with colon subtitles (e.g. "Kill Bill: Vol. 1") get the same fix. --- ## v0.9.14 ### Added: "Fetch Metadata…" button for Movies and TV, automatic TV episode-detail backfill, and wider Convert-to-MKV support **"Fetch Metadata…" button.** Both `MoviesLibraryControl` and `TvLibraryControl` previously only exposed the "Fetch & Compare" flow (side-by-side review before applying) as a visible button — the direct-apply fetch (search, pick a match, apply immediately) only existed buried in the card context menu as "Fetch Metadata…". Added a matching "🔍 Fetch Metadata…" button next to "🔍 Fetch & Compare" in both controls' detail panels, reusing the existing `MovieLibraryOperations.FetchMetadataAsync` and the equivalent episode-level fetch+`SaveEpisodeAsync` path — no new fetch/apply logic, just a shortcut to the flow that already existed. **Automatic TV episode-detail backfill.** Added `TvLibraryOperations.StartAutoBackfillTimer()`, a settings-gated (`AppSettings.AutoBackfillEpisodeDetails`, default off; `AutoBackfillEpisodeIntervalHours`, default 12) periodic timer that reuses the existing `RescanEpisodesAsync` sweep (previously only reachable via the manual "Re-fetch" button) to keep every identified show's episode titles/overviews/air dates/ratings current automatically, for as long as the TV library view is open. New toggle + interval field added to Settings → Downloads/TV episode tracking section. **Convert to MKV now handles more than .mp4.** `Mp4RemuxService.RemuxAsync` itself never restricted source format — the `.mp4`-only gating was hardcoded at five separate call sites (`AutoRenameService`, `RenameService`, `MovieMenuBuilder`, `TvMenuBuilders`, `MovieLibraryOperations`), so `.avi`/`.mov`/`.wmv`/`.m4v` source files never got the option even though `mkvmerge` remuxes them just as well. Added `Mp4RemuxService.RemuxableExtensions`/`IsRemuxable()` and switched all five gates to use it. ### Fixed: Movies "Fetch Metadata…" opened the edit form and required a manual Save click **The new "Fetch Metadata…" button (above) initially called `BuildEditor(entry, null)` after applying, which opens the movie edit form and leaves it waiting for the user to press Save — even though `FetchMetadataAsync` had already persisted the fetched values to the DB.** Changed `AutoFetchMovie_Click` to call `PopulateDetail(movie)` (the plain, non-editing detail view) instead, matching what the fetch already did: one click, no confirmation step. The TV episode equivalent already saved in one click and didn't need this fix. ### Fixed: art/match changes left the currently open detail panel stale until navigating away and back **Comics "Download Art" wrote the new cover to disk but never updated anything in memory** (`ComicsLibraryManagerWindow.xaml.cs` `DownloadArt_Click`) — the series card and the open issue panel kept showing the old cover (or placeholder) until a rescan or re-navigation forced a reload. Now loads the downloaded file into a `BitmapImage` immediately and pushes it into both `series.CoverImage` (grid card, via its existing `PropertyChanged` subscription) and `FilmstripSeriesCover.Source` when that series' issue panel is the one currently open. **TV "Edit Metadata" (re-identify/rematch a series) only refreshed the top-level Series grid.** `OpenFullEditor` called `ReloadSeriesAsync()`, which rebuilds the Series grid with fresh card instances but never touches the currently drilled-into Seasons/Episodes panel — if a series was re-matched while its own Seasons view was open, the breadcrumb, reorganize button, and hero banner kept showing the pre-match title/poster. Now also refreshes the breadcrumb, reorganize button label, and re-runs `PopulateSeasonCards()` (which rebuilds the hero banner) when the just-edited series is the one currently drilled into. Movies/TV/Music/Audiobooks/Books "Assign Art"/"Download Art" and the inline metadata-save/fetch flows were already found to self-refresh correctly (poster-path null-reset + explicit reload, or `PopulateDetail`/`PopulateEpisodeDetail` after save) — only the two cases above were actually stale. ### Renamed: `Mp4RemuxService` → `ConvertToMkvService` **The class name still said "MP4" after it was widened to remux `.avi`/`.mov`/`.wmv`/`.m4v` too, which misdescribed what the type actually does.** Renamed the file and class to `ConvertToMkvService` (matches the existing "Convert to MKV" UI/menu naming) and updated all eight call sites (`MainWindow.xaml.cs`, `TvLibraryControl.xaml.cs`, `AutoRenameService.cs`, `RenameService.cs`, `MovieLibraryOperations.cs`, `MovieMenuBuilder.cs`, `TvMenuBuilders.cs`). Log tags switched from `[Remux]` to `[ConvertToMKV]`; internal parameter/comment wording generalized from "mp4" to "source file". No behavior change. --- ## v0.9.13 ### Fixed: proposed name for any content type collapsed to just one folder segment when an optional token (e.g. Year) had no value **Any item missing an optional field used in a naming convention template — most commonly ebooks/PDFs with no Year metadata — got a proposed name with the required title dropped entirely, e.g. `D:\ebooks\GinaKleinworth.pdf` instead of `D:\ebooks\Gina Kleinworth\Air Fryer Cookbook for Two.pdf`.** `NamingConventionService.BuildSubfolderStructure` treated any `{Key?}` conditional token with no value as a signal to drop the *entire path segment* it appeared in, including any other required tokens sharing that segment. The built-in default templates all mix a required token with a trailing optional one in the same segment (`{Title} ({Year?})` for Books, `{SeriesTitle} ({Year?})` for TV, `{Title} ({Year?})` for Movies, `{Series} ({Year?})` for Comics) specifically so the `(Year)` part is dropped gracefully when unknown — instead the whole segment vanished, leaving only the segment before it (e.g. just `{Author}`), which then got treated by `PathBuilderService.ResolveConventionPath` as a single-segment (flat file) convention. This went unnoticed in the Settings naming-convention live preview because its sample data always includes a Year value. Fixed by having a missing `{Key?}` token substitute to empty text instead of nuking the segment, then cleaning up the now-empty wrapper punctuation it leaves behind (`"Title ()"` → `"Title"`, `"Harry Potter #"` → `"Harry Potter"`) — applied to both `BuildSubfolderStructure` (actual renames) and `GetPreview` (Settings UI) so they now agree. ### Fixed: Ebook proposed names ignored the {Series}/{SeriesNumber}/{ISBN}/{Genre} naming tokens **The Books hold-queue's proposed name always rendered `{Series}`, `{SeriesNumber}`, `{SeriesNumberPadded}`, `{ISBN}`, and `{Genre}` as blank, even when the parser had already found those values.** `PathBuilderService.BuildGenericMetadata`'s `"Books"` branch hardcoded those tokens to `""` instead of reading them off the parsed item — `ParsedBookMedia.Series`/`SeriesPosition`/`Isbn` (populated by `BookParser` from tag/API lookups) and `parsed.Genres` were simply never wired in, so any naming convention template using those tokens silently dropped them from the proposed filename regardless of what metadata the queue displayed. Added a `FormatSeriesPosition` helper (mirrors whole vs. fractional series positions, e.g. `3` vs `3.5`) and wired `Series`/`SeriesNumber`/`SeriesNumberPadded`/`ISBN`/`Genre` through to the actual parsed values. ### Fixed: Books grid context-menu actions could silently operate on a different queue's item **Right-clicking in the Books grid (Fetch Metadata, Rename, Edit Metadata, Download Art, Assign Art) could act on whatever item was last right-clicked in the Music, Audiobook, or Comics grid instead of the book actually clicked.** All four non-video grids share a single `_contextMenuTarget` field (`MainWindow.xaml.cs`), refreshed only by `NonVideoGrid_PreviewMouseRightButtonDown` — but that handler was wired up on `MusicGrid`, `AudiobookGrid`, and `ComicsGrid` and missing from `BooksGrid` (`MainWindow.xaml`). Since `GetNonVideoItemFromContextMenu` just returns the stale field with no per-grid check, right-clicking a book row never updated the target, so context-menu actions kept using the last item right-clicked elsewhere — e.g. fetching metadata for a book could pull in a completely different book/audiobook's data (ISBN, cover, description), and the overwrite-collision check would then correctly report *that* stale item's destination already existing. Fixed by wiring `PreviewMouseRightButtonDown="NonVideoGrid_PreviewMouseRightButtonDown"` on `BooksGrid` to match the other three grids. --- ## v0.9.12 ### Automatic ASIN detection + Hold Queue ASIN lookup for audiobooks **`AudiobookParser` (the auto-watcher's initial parse) now detects an ASIN from tags or the filename and, when found, does a direct Audnexus lookup before falling back to the fuzzy title/author search.** Some Audible rips (e.g. from Libation) embed the ASIN in the filename (`Title [B017V4IM1G].m4b`) or as a freeform tag; previously nothing read either. `AudiobookParser.TryExtractAsinFromTags` reads ID3v2 `TXXX:ASIN` (mp3) and — the gap that mattered most, since Audible's native format is m4b — the MP4 `----:com.apple.iTunes:ASIN` freeform atom via `TagLib.Mpeg4.AppleTag.GetDashBox`, which nothing in the codebase read before (`AudiobookLocalMetadataReader`, used by the library "Fetch and Compare" flow, had the same mp3-only gap — fixed there too). `TryExtractAsinFromFilename` matches a bracketed/braced 10-char token or a bare `B0` + 8 chars pattern. A successful ASIN lookup sets confidence to 100 and populates title/author/narrator/series/year/genres/overview/cover in one shot — more reliable than fuzzy matching, since the ASIN is close to a ground-truth identifier. **Extended ASIN-aware lookup to the Hold Queue's "Fix Match".** Last session's fix only updated the main manual queue's "Fetch Metadata…"; the Hold Queue (`HoldFixMatch_Click` in `MainWindow.xaml.cs`) had its own separate audiobook branch still going through the generic fuzzy-search dialog. Split it into its own `ContentType.Audiobook` branch using `UnifiedMetadataHandler.FetchAudiobookMetadata` (same `AudiblePickerDialog` as the manual queue), applying the full `AudiobookDetails` (including `Asin`/`Series`/`SeriesPosition`) instead of just title/creator. --- ## v0.9.11 ### Audiobook fixes: list-view Reorganize, stale "Open in Explorer" path, ASIN lookup + rename token; manual metadata editing for every queue type **Audiobook list-view Reorganize was missing.** The per-card context menus (`AudiobookAuthorMenuBuilder`/`AudiobookBookMenuBuilder`) already had "📁 Reorganize Files", but the separate list-level context menus (`BuildAuthorContextMenu`/`BuildBookContextMenu` in `AudiobooksLibraryControl.xaml.cs`, used by `AuthorsListView`/`BooksListView` right-click) were hand-built and never got the item — added it to both, wired to the same `ReorganizeAuthorWithPreview`/`ReorganizeBookWithPreview` methods. **"Open in Explorer" opened a stale path after Reorganize.** `AudiobookLibraryEntry.FolderPath`/`FolderName` (and `AudiobookAuthorEntry`'s) were `init`-only, so after `AudiobookLibraryOperations.ReorganizeBook`/`ReorganizeAuthor` moved the folder on disk and updated the DB row, the in-memory entry the UI was still holding kept pointing at the old, now-nonexistent path — and the post-reorganize UI refresh (`_vm.RefreshBooksAsync()`) ran fire-and-forget, so clicking Explorer before it completed hit the stale path. Both properties are now settable and are updated in place immediately after the `Directory.Move`, so the same `AudiobookLibraryEntry`/`AudiobookAuthorEntry` instance the UI already holds a reference to is correct right away regardless of refresh timing. **Manual-queue audiobook metadata lookup couldn't use an ASIN.** The queue's "Fetch Metadata…" routed through the generic `NonVideoFixMatchDialog` (title/author fuzzy search only, shared with Music/Books). `AudiblePickerDialog` (used elsewhere in the library view) already had ASIN-aware lookup (`AudiblePickerDialog.AsinPattern` + direct Audnexus lookup via `MetadataService.GetAudiobookDetailsAsync("audiobook", asin)`) but was never reachable from the manual queue. Added `UnifiedMetadataHandler.FetchAudiobookMetadata` and `ManualRenameOperations.FetchAudiobookMetadata`, and pointed the Audiobooks grid's "Fetch Metadata…" item at it instead — paste a 10-character ASIN and it does a direct lookup instead of a fuzzy title search. **Added `{Asin}` as an Audiobooks naming-convention token**, since there was previously no way to include it in a folder/file pattern even though the field already existed on `AudiobookLibraryEntry`/DB. Wired through `NamingConventionService` (token list + live-preview sample data), `PathBuilderService.BuildGenericMetadata`, and all four `AudiobookLibraryOperations` metadata dicts (`PreviewReorganizeAuthor`/`Book`, `ReorganizeAuthor`/`Book`). Also added an `Asin` property to `ParsedMedia`/`ParsedAudiobookMedia` so it flows from the new ASIN-lookup dialog through to a manual-queue rename. **New: manual "Edit Metadata…" for every manual-queue grid** (Movie/TV/Music/Audiobook/Book/Comic) — previously the only way to correct a bad or failed automatic match was Fetch Metadata's fuzzy search; there was no way to just type the right values in by hand. Added `Views/ManualMetadataEditDialog` (a small generic label+textbox dialog driven by a per-content-type field list built by the caller — no per-type XAML needed) plus `ManualRenameOperations.EditMetadataManually` for Music/Audiobook/Book/Comic and dedicated `EditMetadataMovie_Click`/`EditMetadataTv_Click` handlers in `MainWindow.xaml.cs` for Movie/TV (mirroring the existing Fetch Metadata code paths), wired into every manual-queue grid's context menu. Saving recomputes `NewPath` via the same `PathBuilderService` call the "Rename All" loops use, so edits actually affect where the file lands. --- ## v0.9.10 ### Per-copy progress bars, individual Stop buttons, and a "pause queue" panic button **Previously there was a single shared `GlobalProgressBar`** fed by two independent copy pipelines (the auto-watcher queue in `AutoRenameService` and the manual "Rename All" loops in `MainWindow`/`ManualRenameOperations`, both funneling through `RenameService`/`FileCopyService`). When more than one copy happened to be in flight at once — e.g. a manual batch rename running while the watcher auto-processed another file — the one bar visibly jumped between whichever copy last reported progress, and there was no way to cancel a specific stuck copy or pause the queue without closing the app. Added `CopyOperationsTracker` (`Services/CopyOperationsTracker.cs`) as the single choke point both `RenameService.RenameAsync` and `AutoRenameService`'s internal `RenameAsync` now go through for every copy — since literally every copy path in the app (manual single/batch renames, hold-queue retries, the delayed auto-queue) calls one of those two methods, this required no changes to the ~20 individual "Rename All" call sites in `MainWindow.xaml.cs`. - Each in-flight copy gets its own `CopyOperation` (`Models/CopyOperation.cs`) — a labelled, colour-coded progress row in a new panel above the log, with its own **⛔ Stop** button. - **Stop** cancels that copy's `CancellationToken`, deletes the (now-partial) destination file, leaves the source file completely untouched, and — for auto-queue items — sends the item to the Hold Queue with reason "Cancelled by user" so it's easy to find and retry. - New **⏸ Pause Queue** button pauses the start of any *new* copy for 5 minutes (via `CopyOperationsTracker.WaitWhilePausedAsync`, awaited at the top of both `RenameAsync` methods before the file stream opens); copies already past that point keep running and must be stopped individually. Click again (now "▶ Resume Queue") to resume early; the button shows a live countdown. - Removed the old `GlobalProgressBar`/`_progressSmoothTimer`/`AutoRenameService.CopyStarted`/`CopyProgressChanged`/`CopyFinished` machinery, now fully superseded by the tracker. --- ## v0.9.9 ### Restored Edition field to the movie metadata editor **The inline "Edit Metadata" screen for movies (`MoviesLibraryControl`'s dynamically-built editor) never had an Edition field**, even though `Edition` was fully wired everywhere else (DB column, `MediaParser.ExtractEdition`, `PathBuilderService`'s Plex `{edition-…}` tag, and the older standalone `MetadataDetailWindow`). Users could no longer set/see a movie's Edition (e.g. "Director's Cut", "Extended Edition") from the library's main edit form, so `PlexEditionsEnabled` naming had no way to be populated for existing library items. Added an Edition row to `BuildEditor`/`AddEditorRow` in `MoviesLibraryControl.xaml.cs`, and added an `edition` parameter to `MovieLibraryOperations.SaveMovieAsync` (previously discarded) so it's persisted via the existing `entry.Edition`/`row.Edition` path in `PersistMovieToDbAsync`. --- ## v0.9.8 ### Fixed duplicate auto-downloads and repeated quality-upgrade re-triggers **Automatic downloads and quality upgrades were sometimes being queued more than once for the same item.** `app.log` showed repeated `InvalidOperationException`s ("a second operation was started on this context instance" and "collection was modified") thrown out of `WantListService.CheckLibraryAndRemoveAcquiredAsync` — caused by `WantListService` and `QualityUpgradeService` sharing a single `MediaDbContext` (`App.Database`) with the rest of the app (UI thread) and with `EpisodeSyncService`'s own independent periodic timer. EF Core `DbContext` instances aren't thread-safe, so concurrent access crashed the "remove items already in the library" cleanup step mid-run; the item then stayed in the want list, still marked not-acquired, and got re-downloaded on a later sweep. Separately, `WantListService`'s and `QualityUpgradeService`'s periodic timers had no reentrancy guard, so a slow provider search could overlap with the next tick and queue the same item twice. Fixes: - `WantListService` and `QualityUpgradeService` now each own a **dedicated `MediaDbContext`** instead of sharing `App.Database`, isolating their background-timer DB access from the rest of the app. - Both `CheckAutoDownloadsAsync` and `ScanAndQueueAsync` now have a **reentrancy guard** (`SemaphoreSlim`) that skips a tick instead of racing an already-running one. - Every raw `_db` call inside `WantListService` is now guarded by a fine-grained `SemaphoreSlim` gate, since `EpisodeSyncService.PruneWantListAcquiredAsync()` calls into `WantListService` from its own independent timer and could otherwise still race `WantListService`'s own periodic check. - `CheckLibraryAndRemoveAcquiredAsync` now snapshots the want list before iterating, avoiding a "collection was modified" crash if an item is added/removed mid-scan. **New: explicit 24-hour "queued download" block.** Once a torrent/NZB is successfully queued for a want-list item (including quality-upgrade candidates), `DownloadQueuedAt` is stamped and the item is excluded from future auto-download sweeps for `DownloaderSettings.QueuedDownloadBlockHours` (default 24h) — independent of, and longer than, the existing `RetryIntervalHours` "no results" cooldown (default 6h), which was short enough that a download still in progress could get re-queued again before it had time to land in the library. If the download stalls or fails and nothing lands within the block window, the item becomes eligible for a fresh search again. When the file does land, the existing `NotifyTvEpisodeAcquired`/`NotifyMovieAcquired` removal (now made reliable by the fixes above) clears the item from the want list immediately, so it can't be re-downloaded. New setting exposed in Settings → Downloads as "Queued download block (hours)". --- ## v0.9.7 ### Restored Edit Tracks and Transcode to x265 for TV, with proper hierarchy **These context-menu actions existed for Movies but were never carried over to the TV library control** during the Comics-style migration, so they silently disappeared for TV series/season/episode. `TvLibraryOperations` gains `TranscodeSeries`/`TranscodeSeason` (bulk-queue every video file in the folder for x265 encoding, no per-file dialog — mirrors the existing `NormalizeAllAudio`/`NormalizeSeasonAudio` pattern) plus `TranscodeEpisodeAsync` and `EditEpisodeTracksAsync` (single-file, with the `TrackEditWindow` picker — mirrors the Movie behavior). "Transcode to x265" now appears at Series, Season, and Episode level; "Edit Tracks" is Episode-only and only offered for a single selected episode (remuxing multiple files at once via one track selection doesn't make sense). --- ## v0.9.6 ### Auto-downloader no longer re-downloads the same episode every 30 minutes in manual mode **In manual mode (`AutoRenameEnabled = false`), the want-list auto-downloader kept re-downloading episodes it had already fetched.** `CheckAutoDownloadsAsync` ran on its own 30-minute timer completely independent of manual/auto mode, and downloaded files in manual mode sit in the watch folder / manual-review grid until the user clicks Rename — they never land in `TvRootFolder`/`MovieRootFolder`. Since `IsItemInLibrary` only recognizes files already inside the library folders, the episode never registered as acquired, so each periodic sweep treated it as still missing and queued another download. `CheckAutoDownloadsAsync` now also checks `AppSettings.AutoRenameEnabled` and skips the run entirely when the app is in manual mode, alongside the existing `DownloaderSettings.AutoDownloadEnabled` check. --- ## v0.9.5 ### Comics: series year now correct in proposed names, incomplete metadata held from queue **`{SeriesFirstYear}` was always blank in the manual queue.** The watcher-path parser was setting `result.Year` from the CV volume's start year but never writing to the separate `SeriesFirstYear` field — so the proposed folder was `X-Force ()` instead of `X-Force (1991)`. The fix populates `SeriesFirstYear` directly from the API volume search result, the issue details, and pinned-series DB lookups. The `{Year}` and `{SeriesFirstYear}` tokens remain semantically distinct: `{SeriesFirstYear}` is the year the series first published (for the folder name), `{Year}` is the issue cover year (for display). The library reorganize path had the same bug — `BuildComicMeta` was mapping the issue's `CoverYear` to the `{SeriesFirstYear}` token; this now correctly uses the series's `SeriesStartYear` field. **Comics no longer appear in the queue until they have a fully qualified proposed name.** Files are held if `Publisher`, `SeriesFirstYear`, or `IssueNumber` are missing after a full metadata scrape. Held files are automatically retried every 5 minutes so a rate-limiter pause or a slow API response doesn't lose the file. Any file held for more than 15 minutes is force-queued regardless of completeness so nothing disappears permanently. --- ## v0.9.4 ### Missing albums now actually appear — DiscographyEnabled wired up end to end **Missing albums were never showing because DiscographyEnabled was always false.** `DbArtist.DiscographyEnabled` defaults to `false` (DB column `DEFAULT 0`), and `RefreshAllArtistsAsync` filters on it — so no discography fetch ever ran, `MusicReleases` stayed empty, and the missing albums panel was always blank. The fix has two parts: (1) Fetching metadata in the Music Library for an artist now automatically sets `DiscographyEnabled = true` the moment a MusicBrainz ID or Discogs ID is found, so existing artists pick up tracking the next time you click "Fetch Metadata". (2) The Artist Editor window now shows a "Track discography (fetch missing albums)" checkbox that loads the current value and saves it back — so you can also toggle it manually or turn it off for artists you don't want tracked. --- ## v0.9.3 ### Reorganize now catches episodes loose in the series root **TV Reorganize now handles episodes not yet in season subfolders.** Previously, `PreviewReorganize` and `ReorganizeEntry` only scanned season subdirectories — any video files sitting directly in the series folder (e.g. `House of the Dragon (2022)\S01E01 - Title.mkv`) were invisible and the operation would report "nothing to do." Both methods now also sweep the series root for video files matching the `S##E##` pattern, create the season subfolder if it doesn't exist, and move the files in. The preview dialog now shows relative paths (e.g. `Season 01\S01E01 - The Heirs of the Dragon.mkv`) so it's obvious when a file is being moved into a subfolder rather than just renamed in place. Movies are unaffected — each movie already has its own folder and the reorganize correctly processes files within it. --- ## v0.9.2 ### AudioDB removed; Upcoming Albums section now visible on the General tab **AudioDB removed.** TheAudioDB has been dropped as a metadata provider — MusicBrainz and Discogs cover everything it did, better. `AudioDbClient.cs` is deleted. The AudioDB ID fields are gone from the Artist Editor window, the `DbArtist` and `DbAlbum` entities, and artist NFO exports. Fetch Metadata now runs Last.fm, MusicBrainz, and Discogs in parallel; artist photo auto-pick now uses Discogs only (via the existing art aggregator). The `AudioDbArtistId` and `AudioDbAlbumId` columns remain in existing databases to avoid destructive migrations — they are simply ignored. **Upcoming Albums section was missing from the General tab.** The ViewModel logic and data binding were complete but the XAML section was never added to `MainWindow.xaml`. The "UPCOMING ALBUMS" cards now appear below the TV Shows section, gated by Music tab visibility and the "Show upcoming music" setting. Each card shows square album art, artist name, album title, a type badge (Album/EP/Single), and the relative release date, with "Add to Want List" and "Ignore" buttons. --- ## v0.9.1 ### Reorganize button now scopes to the current level — no more accidental whole-library renames **Reorganize targets what you're looking at, not the whole library.** In the TV library, clicking Reorganize while inside a series (Seasons or Episodes view) now reorganizes only that series — showing the same preview confirmation dialog the right-click context menu uses — instead of silently reorganizing every series in the library. In the Movies library, clicking Reorganize while a movie's detail pane is open now reorganizes only that movie. At the top-level grid view (no series or movie selected) the button still batch-reorganizes the whole library as before. **Button label reflects scope.** When you navigate into a TV series, the toolbar button changes from "📁 Reorganize All" to "📁 Reorganize 'Series Name'" so it's always clear what clicking it will affect. Navigating back to the series grid restores "📁 Reorganize All". Movies work the same way — the button reads "📁 Reorganize 'Movie Title'" while you're in the detail pane. --- ## v0.9.0 ### Music discography tracking, upcoming album releases, and missing album detection **Artist discography tracking from MusicBrainz and Discogs.** When a music artist in your library has a MusicBrainz or Discogs artist ID (assigned via Library Manager → Music → Edit Artist), you can now opt-in to discography tracking per-artist. The new `MusicDiscographyService` fetches the full release group list from MusicBrainz (paginated, ~1 req/sec) or Discogs releases (fallback when no MBID), stores them in a new `MusicReleases` table, and then fuzzy-matches each release against your local `Albums` table to determine what is and isn't in your library. Cover art thumbnails are fetched from the Cover Art Archive during the scan and stored alongside each release. Refresh runs in the background on startup and on a configurable interval (default: every 7 days). **Upcoming album releases on the General tab.** The General tab now includes an "Upcoming Albums" section showing releases from tracked artists that fall within a configurable lookahead window (default: 180 days). Each card shows cover art, artist, album title, type badge, and relative date. "Add to Want List" queues the album with a music-specific search query (`"Artist Album Year FLAC"`). "Ignore" permanently hides the release (stored in settings). Album release events are also merged into the weekly episode calendar view, appearing as secondary rows on matching days. **Missing albums panel in the Music Library.** Navigating into an artist in the Music Library now reveals a collapsible "Missing Albums" expander at the bottom of the artist drill-down view when discography tracking is enabled for that artist. Each row shows the release year, album title, type badge, and buttons to add to the Want List or open Download Search pre-filled with the artist and album name. **Music Tracking settings.** A new "Music Tracking" expander in Settings → Downloaders controls all discography settings: enable/disable tracking, show upcoming music, show missing albums, refresh interval, upcoming lookahead, and which album types to include (studio albums only by default — EPs, Singles, Live, Compilations off). A "Refresh Discography Now" button triggers an immediate re-scan of all enabled artists. **Want List music support.** `WantListItem` now has an `ArtistName` field. `DisplayTitle` renders music items as "Artist — Album" and `SearchQuery` produces "Artist Album Year FLAC". `WantListService.AddMissingAlbum` de-duplicates by artist+title and `IsItemInLibrary` checks the `Albums` DB table for music items. --- ## v0.8.5 ### TV folder routing by ID + MP4→MKV filename fix in logs and toasts **TV folder matching now uses TMDB/TVDB ID instead of fuzzy name matching.** When a TV series folder has been identified in Library Manager (assigned a TMDB or TVDB ID), new episodes for that series always route to the correct existing folder regardless of what the folder is named on disk. Previously, fuzzy name matching would fail when the TMDB title differed from your folder name (e.g. "Last Week Tonight with John Oliver" not matching your "Last Week Tonight (2014)" folder). The preferred metadata provider (`Settings → Metadata → Preferred TV Provider`) controls which ID is tried first; the other is used as a fallback. For shows with no DB entry yet (first-time arrivals not yet scanned into Library Manager), a fresh folder is created using the TMDB/TVDB title — no fuzzy guessing. **MP4→MKV log and toast now show the correct filename.** When `Auto Remux MP4 to MKV` is enabled, the "Moved" toast and rename history entry previously used the original `.mp4` filename instead of the actual `.mkv` destination. `RenameService.RenameAsync` now returns the actual destination path used (which may differ from the input `newPath` when remux fires), and all callers use that path for toasts, logs, and rename history. --- ## v0.8.4 ### Release-readiness: log level, plugin trust, art cache, first-run wizard **Default log level changed to Minimal.** New installs start with `LoggingLevel.Minimal` instead of `Extra`, so the log only records errors, warnings, and rename outcomes by default. Existing users who have already saved a log level preference are unaffected. **Newznab plugin auto-authorized on first load.** Any DLL whose filename contains "newznab" (case-insensitive) is now automatically added to `TrustedPluginHashes` without a user prompt. All other third-party plugins still require explicit user consent. The auto-authorization is logged as `[PluginManager] … — auto-authorized (first-party bundled plugin)`. **Art cache default directory changed to `art_cache`.** `EffectiveImageCachePath` now resolves to `%AppData%\AbigailsMediaRenamer\art_cache\` (previously `ImageCache`). The directory is created automatically on first access. Users with a custom `ImageCachePath` in settings are unaffected. Note: any art already cached in the old `ImageCache` folder will not be moved; if you need the old art, copy it manually. **First-run setup wizard.** On a fresh install (no settings file, or settings file with no watch/API key configured), a three-step setup wizard appears before the main window. Step 1 (Folders): choose watch folder, movies root, and TV root — all three are required and must exist. Step 2 (API Keys): enter TMDB and TVDB API keys with links to the sign-up pages. Step 3 (Naming): adjust Movie and TV naming templates using the same token/preview UI as Settings → Naming. On completing the wizard ("Go"), settings are saved, the file watcher starts, and the main window opens. Existing users are detected via the presence of a configured watch folder or API key and never see the wizard. --- ## v0.8.3 ### Movies — Artwork Tab inline + Cast & Crew repositioned **Artwork manager is now an inline tab** on the movie detail panel instead of opening a separate window. The movie detail view has two tabs: **Info** (poster + metadata, same as before) and **Artwork** (poster, fanart, banner, clearlogo, clearart slots — right-click each to download, assign from file, or clear; "Download All Art" button downloads everything at once). The artwork slots lazy-load the first time you switch to the tab. **"🎨 Artwork Manager" context menu item** on movie cards now navigates to the detail view and switches directly to the Artwork tab rather than opening a separate window. **Cast & Crew moved** from below the two-column layout (where it was a full-width block) to inside the right info column, directly underneath the plot summary. Director, writers, and cast now appear inline with the rest of the metadata. **"⬇ Download Art" (Info tab poster button)** refreshes the Artwork tab's poster slot too if it's already loaded. --- ## v0.8.2 ### Movies — Artwork Manager + Poster Picker **"Fetch Art" now shows the picker dialog instead of auto-downloading.** Clicking "⬇ Download Art" (inline panel or context menu) previously silently downloaded the first TMDB poster. It now opens the `ArtPickerDialog` with all TMDB + Fanart.tv poster candidates, matching the TV and movie `MetadataDetailWindow` behaviour. **New "🎨 Artwork Manager" button and context menu item** on the Movies detail panel. Opens `MetadataDetailWindow` for the selected movie — giving access to the full Artwork tab (poster, fanart, logo, banner, clearart, season posters) just like TV series already had. **After closing the Artwork Manager**, the inline panel's cover image refreshes automatically if `cover.jpg` changed on disk. **`MovieMenuBuilder.Actions`** gains an `OpenArtworkManager` delegate alongside `DownloadArt` and `AssignArt`. Both the inline button and the context menu use shared helper methods (`ShowPosterPickerAsync`, `OpenArtworkManagerForCard`) so the logic isn't duplicated. --- ## v0.8.1 ### Discogs Artist Disambiguation + Bio Source Switch **Discogs ID now used directly for art fetching.** Previously, "Fetch Artist Photo" always did a Discogs name search and took the first result — so if you had manually corrected the Discogs ID (e.g. to "Blazy (2)"), the picker ignored it and fetched the wrong artist's photos. Both the context-menu picker and the Artist Editor "🖼 Fetch Photo" button now pass the stored `DiscogsArtistId` to the aggregator, which bypasses the search entirely and hits `/artists/{id}` directly. **New `ICanProvideArtistArtById` interface.** `DiscogsClient` implements this alongside `ICanProvideArtistArt`. `ArtAggregatorService.GetArtistArtCandidatesAsync` now accepts an optional `discogsArtistId` parameter — when set, providers implementing `ICanProvideArtistArtById` use the ID path; other providers (AudioDB) still use the name. **Discogs name search now returns candidates from all matching artists.** When no ID is stored and a name search is needed, up to 5 matching Discogs artists are queried in parallel. Each artist's photos are labelled with their Discogs name (e.g. `"Discogs (Blazy)"` vs `"Discogs (Blazy (2))"`) so the picker makes disambiguation visible at a glance. **Discogs is now the primary bio source.** Last.fm bios are unreliable for disambiguation — when multiple artists share a name, Last.fm blends results from all of them. `FetchArtistMetadataAsync` and the Artist Editor's "🔍 Fetch Metadata" button now use the Discogs `profile` field as the bio (fetched via `GetArtistInfoAsync`, using the stored ID directly if known), falling back to AudioDB biography only if Discogs returned nothing. Last.fm still provides tags, listener count, and Last.fm URL — just not the bio. **Manually set Discogs IDs are now preserved.** Both "Fetch Metadata" flows previously overwrote `DiscogsArtistId` with whatever the name search returned, even if you had corrected it manually. Now the ID is only filled in if the field is currently empty. **New `DiscogsClient.GetArtistInfoAsync(name, knownId?)`** — convenience method returning `(string? Id, string? Profile)`. Uses the known ID directly if supplied; otherwise searches by name. Replaces separate `SearchArtistAsync` + `GetArtistDetailAsync` calls at the fetch sites. **`DiscogsArtistDetail` DTO gains `Profile` field** — maps to the `profile` JSON key returned by the Discogs artist detail endpoint. --- ## v0.8.0 ### Art Capability System — All Content Types Now Have Art Pickers This is a large infrastructure change that brings multi-source art discovery and a picker dialog to every content type in the library. **New: `ArtAggregatorService`** — a central service that queries all art-capable providers in parallel and merges the results. Replaces ad-hoc single-source downloads everywhere. **New art capability interfaces** (`IArtCapabilities.cs`): `ICanProvideArtistArt`, `ICanProvideAlbumArt`, `ICanProvideMovieArt`, `ICanProvideTvArt`, `ICanProvideBookArt`, `ICanProvideAudiobookArt`, `ICanProvideComicArt`. Providers opt-in by implementing the interfaces they support. **Providers now contributing art:** **Music artist photos:** TheAudioDB (existing) + Discogs (new — searches by name, fetches `/artists/{id}` for primary/secondary images) **Album covers:** Deezer (existing) + MusicBrainz/CoverArtArchive (existing) — now both surface in the same picker **Movie art (poster/fanart/logo/banner/clearart):** TMDB + Fanart.tv — both surface through the aggregator **TV art:** TMDB + Fanart.tv — both surface through the aggregator **Book covers:** Google Books + Open Library + GoodReads **Audiobook covers:** Audible/Audnexus (via Amazon CDN cover URL from ASIN) **Comic covers:** ComicVine + Metron **Picker dialog wired to:** **Music artist "🖼 Fetch Artist Photo"** (new context-menu action) — picks from TheAudioDB + Discogs **Artist Editor "🖼 Fetch Photo" button** — now shows picker instead of silently downloading first result **Music "🔍 Fetch Metadata" auto-path** — now uses aggregator (covers AudioDB + Discogs) for auto-pick on first fetch **Books "Download Art" action** — now shows picker from Google Books + Open Library + GoodReads **Audiobooks "Download Art" action** — now shows picker from Audible **Comics "Download Art" button** — was a stub; now shows picker from ComicVine + Metron **New client: `AudioDbClient`** — split out from `LastFmClient`. AudioDB is now its own service. `LastFmClient` is Last.fm only. **`DiscogsClient`** gains `GetArtistDetailAsync(id)` hitting `/artists/{id}` and implements `ICanProvideArtistArt`. --- ## v0.7.9 **Discogs Artist ID now fetched during "Fetch Metadata".** The artist metadata fetch (context menu, hero banner re-fetch, and the new editor button) now also queries Discogs `database/search?type=artist` and stores the matched artist's Discogs ID. Requires a Discogs token in Settings (same token used for album metadata). **"Fetch Metadata" button added to Artist Editor window.** The "Edit Metadata" dialog now has a "🔍 Fetch Metadata" button that fires Last.fm, TheAudioDB, MusicBrainz, and Discogs in parallel and populates all editable fields (MBID, AudioDB ID, Discogs ID, bio, tags, listeners, Last.fm URL, country, formed, disbanded, style). Fields are overwritten with fresh API data; hit Cancel to discard if the results look wrong. **Fixed MusicBrainz crash when `life-span.ended` is JSON null.** Calling `GetBoolean()` on a null JSON element threw "target element has type 'Null'"; now checks `ValueKind == JsonValueKind.True` instead. ## v0.7.8 **Artist metadata fetch now pulls from three sources simultaneously.** "Fetch from Last.fm" has been renamed to "🔍 Fetch Metadata" everywhere (context menu, inline hero button, re-fetch button). Clicking it now fires three providers in parallel: Last.fm provides bio summary, genre tags, listener count, and Last.fm URL (unchanged); TheAudioDB provides AudioDB Artist ID, photo thumbnail, biography (fallback when Last.fm bio is empty), formed/disbanded dates, country, and style/genre; MusicBrainz provides the Artist MBID, ISO country code (takes precedence over AudioDB text country), and formed/disbanded dates. **All new fields are saved to the DB and are immediately visible if you re-open Edit Metadata for that artist.** ## v0.7.7 **Music library: MusicBrainz, AudioDB, and Discogs IDs stored in DB.** `DbArtist` gains `MusicBrainzArtistId`, `AudioDbArtistId`, `DiscogsArtistId`, `Formed`, `Disbanded`, `Country`, and `Style`. `DbAlbum` gains `MusicBrainzAlbumId`, `MusicBrainzAlbumArtistId`, `MusicBrainzReleaseGroupId`, `AudioDbAlbumId`, `DiscogsAlbumId`, `DiscogsUrl`, `Description`, `Style`, `Mood`, `ReleaseDate`, and `Compilation`. All new columns are added via the standard `TryAddColumn` migration so existing databases upgrade silently on first launch. **Artist Editor now exposes all new fields.** The "Edit Metadata" dialog for artists has been expanded to include editable fields for MusicBrainz Artist ID, AudioDB Artist ID, Discogs Artist ID, Formed, Disbanded, Country, and Style. The Bio text box has been retained and the window height adjusted to fit all fields. **NFO export for music: `artist.nfo` and `album.nfo`.** Two new methods (`WriteArtistNfoAsync`, `WriteAlbumNfoAsync`) added to `NfoWriterService`. Output complies with both Kodi and Jellyfin music NFO specs. `artist.nfo` includes: ``, ``, ``, ``, ``, ``, ``, `