chore(release): prepare v0.20.0

2026-09-23 05:16:23 -07:00 · 2026-09-23 01:15:20 -07:00
parent e93b696e71
commit 7fbc36dc41
48 changed files with 419 additions and 190 deletions
@@ -1,4 +0,0 @@
-type: docs
-area: launcher
-
- Documented the bundled-runtime launcher install, legacy launcher migration, package-managed updates, and the bundled Bun runtime's MIT and LGPL notices (installed under `/usr/share/licenses/subminer-bin` by the AUR package and included in `subminer-assets.tar.gz`).
@@ -1,4 +0,0 @@
-type: changed
-area: launcher
-
- Every installed and downloadable launcher now uses the Bun runtime bundled with SubMiner instead of a system Bun. Recognized legacy launchers migrate automatically, Windows gets a `subminer.cmd` download, and first-run setup is reduced to a single optional launcher control with runtime repair guidance shown only when needed.
@@ -1,4 +0,0 @@
-type: internal
-area: ci
-
- Removed duplicate source and launcher smoke executions from the reusable quality gate while preserving every distinct test lane and failure artifact.
@@ -1,4 +0,0 @@
-type: internal
-area: verification
-
- Replaced mislabeled dist source reruns with a small Electron-runtime smoke check for compiled stats startup, HTTP service, native SQLite, port conflicts, and cleanup.
@@ -1,4 +0,0 @@
-type: changed
-area: sync
-
- Sync now uses compressed, incremental rsync transfers between compatible macOS and Linux machines, caching the last received snapshot per peer to cut traffic on later syncs. Machines without compatible rsync (including Windows) fall back to compressed scp, and older peers still work without the upload cache. Transfers abort after 30 minutes.
@@ -1,4 +0,0 @@
-type: fixed
-area: anki
-
- Added `ankiConnect.fields.wordAudio` so word audio is read separately from the sentence-audio destination, fixing animated images that started moving immediately when `fields.audio` pointed to `SentenceAudio`.
@@ -1,4 +0,0 @@
-type: fixed
-area: config
-
- Validate direct AnkiConnect, Kiku, and Senren settings before admitting them to runtime config, with warnings and defaults for invalid values.
@@ -1,4 +0,0 @@
-type: fixed
-area: startup
-
- Fixed first launch exiting on macOS when the SubMiner config directory did not yet exist by creating it before acquiring the startup lock.
@@ -1,4 +0,0 @@
-type: fixed
-area: jellyfin
-
- Jellyfin streams no longer leak URL-derived titles or credential-bearing stream URLs into metadata lookups, Anki source fields, Discord presence, stats identities, or AniList retry keys. Previously cached credential-bearing parser metadata is cleaned up without touching watch history or library assignments.
@@ -1,7 +0,0 @@
-type: fixed
-area: jellyfin
-
- Jellyfin playback, subtitle, artwork, and remote-control URLs now authenticate with the `ApiKey` query parameter instead of the legacy `X-Emby-*` headers, so the integration works on Jellyfin 12 where legacy authorization is disabled by default.
- "Play on SubMiner" keeps working on Jellyfin 12: the cast-target websocket answers keep-alive requests and reconnects when the server stops replying instead of silently dying after about a minute.
- The Jellyfin "now playing" bar clears when you close or finish a cast video instead of running on to the end of the episode.
- Anki cards mined from Jellyfin playback get the episode title in the misc info field again instead of "Unknown media".
@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
- Clicking a subtitle sidebar cue releases row focus, and Space no longer seeks back to a focused cue. Enter still seeks the focused cue, and Space keeps its configured playback action.
@@ -1,4 +0,0 @@
-type: fixed
-area: stats
-
- Reject malformed resource IDs and partly invalid ID lists before stats library mutations or cover backfills run.
@@ -1,4 +0,0 @@
-type: fixed
-area: stats
-
- Stats server port conflicts are reported through status notifications instead of crashing SubMiner. Startup and shutdown are also more robust: concurrent startup requests are shared, background stop no longer disconnects foreground dashboards, and shutdown bounds how long it waits for active requests.
@@ -1,4 +0,0 @@
-type: fixed
-area: anki
-
- Closing the overlay while media timing review is still loading now cancels setup and modal retries, restores playback if the review paused it, and cleans up the hidden preview player.
@@ -1,4 +0,0 @@
-type: fixed
-area: anki
-
- Treat `ankiConnect.media.maxMediaDuration: 0` as unlimited for stats dashboard mining, matching overlay mining and configuration.
@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
- Keep Hyprland recovery dialogs above SubMiner windows so overlay placement updates do not cover their Wait and Close buttons.
@@ -1,4 +0,0 @@
-type: docs
-area: subtitles
-
- Documented subtitle generation: model choice, VAD behavior, reference-track guided splitting, fallback behavior, and known limits.
@@ -1,7 +0,0 @@
-type: added
-area: subtitles
-
- Generate Japanese SRT subtitles locally with whisper.cpp from a modal (Ctrl+Shift+G), the empty subtitle sidebar's generation button, or `subminer generate-subs`, with progress, cancellation, and automatic loading into mpv.
- Pick an official multilingual model (including quantized variants) with size and accuracy guidance and download it in-app, or point Settings at an existing model. `large-v3-turbo` is recommended when CUDA support is detected, `small` otherwise. whisper-cli, ffmpeg, and ffprobe are found on PATH unless overridden, and missing tools are named before any download starts.
- Optional "Focus on spoken dialogue" mode uses a separately downloadable Silero VAD model, keeping uncertain audible sections so dialogue under music is not dropped (songs may be transcribed too).
- Long passages are split near detected speech starts or quiet pauses, guided by an eligible embedded or external subtitle track already loaded in mpv when one is available, to reduce early subtitle timing. Each passage runs in a fresh Whisper process to avoid repeated-character output.
@@ -1,4 +0,0 @@
-type: docs
-area: jellyfin
-
- Clarify that Windows mpv playback and Jellyfin casting can use a configured executable path instead of PATH.
@@ -1,4 +0,0 @@
-type: fixed
-area: jellyfin
-
- Honor the configured mpv executable when Jellyfin starts playback, allowing casting when mpv is installed outside PATH, and detect portable plugins beside the selected executable.
@@ -1,4 +0,0 @@
-type: docs
-area: jellyfin
-
- Documented how Jellyfin media titles and stats identities keep stream credentials out of metadata.
@@ -1,4 +0,0 @@
-type: added
-area: jimaku
-
- Jimaku modal gains Anime / Live action tabs so searches can pull from Jimaku's live action catalogue instead of only anime entries (Arrow Left / Right switch tabs).
@@ -1,4 +0,0 @@
-type: docs
-area: mining
-
- Document independent screenshot selection in media timing review.
@@ -1,4 +0,0 @@
-type: added
-area: mining
-
- Choose a still screenshot independently of the audio in media timing review, with a live preview, a separate time slider, and decoded-frame stepping. Local video and seekable remote streams such as Jellyfin are supported.
@@ -1,4 +0,0 @@
-type: added
-area: overlay
-
- The overlay picks up non-conflicting keyboard bindings from mpv defaults, `input.conf`, and loaded scripts. SubMiner controls and explicitly disabled bindings take precedence; discovered bindings are session-only and not listed in the help menu.
@@ -1,4 +0,0 @@
-type: changed
-area: release
-
- Reduced installer and unpacked app size by dropping demo media, source maps, TypeScript sources, test fixtures, and unused Koffi binaries from the package, and sharing one Japanese UI font across windows. Release builds now publish package size reports with comparisons against the previous release.
@@ -1,4 +0,0 @@
-type: fixed
-area: overlay
-
- Cancel pending Linux overlay window replacements during teardown so a delayed close callback cannot reopen the overlay.
@@ -1,4 +0,0 @@
-type: fixed
-area: anki
-
- Keep word-card sentence furigana in sync with full stats-search context and expanded timing-review selections. Clear stale readings if regeneration fails.
@@ -1,4 +0,0 @@
-type: docs
-area: config
-
- Clarified live-setting save feedback, mixed restart warnings, and subtitle-generation reload behavior.
@@ -1,5 +0,0 @@
-type: fixed
-area: config
-
- Settings marked LIVE now use the same reload policy as save results, fixing false restart warnings for notifications and subtitle generation.
- Mixed saves apply live changes and list only sections with changed fields that require a restart.
@@ -1,4 +0,0 @@
-type: added
-area: overlay
-
- Select dialogue across subtitle sidebar rows and copy it without timestamps using Ctrl/Cmd+C or the Copy button. Selecting does not seek or require mining a card.
@@ -1,4 +0,0 @@
-type: changed
-area: stats
-
- The stats server now rejects requests from non-loopback hosts and browser origins and requires `application/json` for mutation bodies. The in-app stats overlay loads from the local server so it shares the same protection. Reverse-proxied or Tailscale Serve dashboards are unsupported; scripts that POST must set a JSON content type.
@@ -1,4 +0,0 @@
-type: docs
-area: config
-
- Documented the subtitle selector setting, shortcut override, and primary/secondary track controls.
@@ -1,5 +0,0 @@
-type: added
-area: overlay
-
- Added an optional Catppuccin subtitle selection modal for primary and secondary mpv tracks. Enable it in Settings under Behavior, then press g followed by s. Disabling it restores mpv's subtitle selection binding.
- Single-key actions take priority over configured sequence prefixes. Conflicting sequences are disabled with a warning, and the existing y commands stay reserved.
@@ -1,4 +0,0 @@
-type: fixed
-area: subtitles
-
- Keep the subtitle sidebar near playback during gaps when the subtitle file has a cue starting at zero.
@@ -1,4 +0,0 @@
-type: docs
-area: sync
-
- Documented compressed transfers, incremental sync cache storage, and compatibility with older peers.
@@ -1,4 +0,0 @@
-type: docs
-area: stats
-
- Documented TMDB linking, provider reassignment, merge compatibility, and credential command caching.
@@ -1,6 +0,0 @@
-type: added
-area: stats
-
- Live-action dramas and movies in the stats Library get posters, synopses, and titles from TMDB. Release builds include a project key; `tmdb.apiKey` (or `tmdb.apiKeyCommand`) overrides it and is required when running from source.
- Titles AniList cannot match are looked up on TMDB automatically when the parsed filename matches a Japanese live-action title exactly; otherwise use the new **Link to TMDB** action. Entries linked to the same TMDB title merge into one card, and the Library kind selector gained a Live Action option.
- Provider reassignment keeps the previous link and artwork if the replacement download fails. Merges and sync keep AniList and TMDB identities separate, and the merge dialog explains mixed selections instead of failing.
@@ -1,4 +0,0 @@
-type: docs
-area: anki
-
- Documented the separate word-audio field mapping; existing animated images need regeneration to pick up the fix.
@@ -1,4 +0,0 @@
-type: changed
-area: yomitan
-
- Updated bundled Yomitan with upstream 26.9.8 changes, including historical Japanese kana transformations, Ukrainian language support, and improvements to Anki duplicate searches and audio retrieval.
@@ -1,4 +0,0 @@
-type: docs
-area: stats
-
- Document YouTube channel filtering and video statistics in the Library.
@@ -1,5 +0,0 @@
-type: added
-area: stats
-
- YouTube channels are now a separate Library media kind. Existing channel entries migrate automatically without changing viewing history or manual video assignments.
- Added All Titles, Anime, and YouTube Library filters. Channels stay out of AniList matching, season repair, and duplicate recommendations, and merges or video moves cannot mix an anime entry with a YouTube channel.