fix(jellyfin): show overlay, inject plugin, and fix stats title on playback (#77)

* fix(jellyfin): show overlay, inject plugin, and fix stats title on playb

- Show visible overlay automatically during Jellyfin playback so subtitleStyle applies
- Inject bundled mpv plugin on auto-launch so keybindings work without overlay focus
- Group Jellyfin playback stats under item metadata (jellyfin://host/item/id) instead of stream URLs so episodes merge with matching local titles
- Mark ffsubsync unavailable in subsync modal for remote media paths
- Drain queued second-instance commands even when onReady throws

* fix(overlay): stabilize macOS focus handoff and sidebar Yomitan pause

- Keep overlay visible during macOS foreground probe after overlay blur
- Hold sidebar hover-pause while a Yomitan lookup popup remains open

* fix(jellyfin): fix discovery loop, device identity, tray state, and Disc

- Derive device identity from OS hostname; remove legacy configurable client/device fields
- Prevent discovery playback from reloading active item, misreporting pause state, and duplicate overlay restores
- Restart stale tray discovery sessions without re-login when server drops SubMiner cast target
- Sync tray discovery checkbox state on Linux after CLI/startup/remote-session changes
- Stop Discord presence falling back to stream URLs; prime title before tokenized stream loads
- Fix picker library discovery when log level is above info
- Fix config.example.jsonc trailing commas and array formatting

* docs(release): trim and consolidate prerelease notes for 0.15.0

- Remove breaking changes section and several redundant bullet points
- Consolidate per-platform updater notes into a single entry
- Normalize em-dash separators to hyphens in section headers

* fix(config): remove trailing commas from config.example.jsonc

- Strip trailing commas throughout both config.example.jsonc copies
- Reformat inline arrays to multi-line for JSON strictness
- Update Jellyfin subtitle preload and playback launch tests and impl

* fix(tokenizer): preserve known-word highlight when POS filters suppress

- Known-word cache matches now set isKnown=true even for tokens excluded by POS filters
- POS exclusion gate suppresses N+1, frequency, and JLPT only; known status is computed before the gate
- Jellyfin subtitle preload continues after cleanup failures instead of aborting
- Update config docs and option description to document the known-word bypass behavior

* fix(jellyfin): send explicit hide/show overlay instead of toggle

- Track overlay visibility in plugin state; y-t uses explicit hide/show commands when state is known
- Prevent paused Jellyfin playback from resuming on overlay hide
- Fix subtitle cache cleanup to only remove dirs after successful cleanup

* fix(jellyfin): fix remote progress sync, seek reporting, and startup sto

- arm active playback before loadfile with loadedMediaPath: null to suppress premature stop events
- force immediate progress report on seek-like position jumps at the mpv time-pos level
- send positionTicks and failed=false in reportStopped payload
- remove EventName from HTTP timeline payloads (websocket-only field)
- add startup grace window to drop stop events before media finishes loading

* fix(jellyfin): fix overlay toggle sync, redirect reload, and AppImage bi

- Sync visible-overlay state back to plugin via script messages to avoid toggle/hide drift
- Collapse duplicate toggle events within 250ms to prevent hide-then-show on single keypress
- Preserve manual hide across Jellyfin path-changing redirects even when media-title drops
- Rearm managed subtitle defaults on path-changing redirects
- Route toggleVisibleOverlay session binding through plugin toggle instead of app-side IPC
- Show Linux/Hyprland overlay passively (showInactive) to avoid stealing mpv keyboard focus
- Fix AppImage binary resolution to prefer $APPIMAGE env over mounted inner binary
- Add stats window layer management so delete/update dialogs appear above stats window
- Fix Jellyfin remote progress sync during Linux websocket reconnect windows

* Fix CodeRabbit review feedback

* fix(jellyfin): subtitle timing, resume progress, and overlay sync

- Add per-stream subtitle delay persistence and auto timeline-offset correction
- Strip server-selected subtitle stream from mpv load URL; suppress plugin subtitle rearm and auto-start during app-managed preload
- Fix resume position lost when mpv resets on stop; use last known position for final progress/stopped reports
- Keep Play vs Resume distinct to avoid early seek race on normal play
- Fix discovery resume when remote play sends StartPositionTicks=0 despite saved progress
- Deduplicate show/hide overlay commands using recorded visibility state
- Rewrite docs-site Jellyfin page around cast-to-device UX

* test: update lifecycle cleanup assertion

* fix: clear aborted playback state, fix overlay passthrough, and guard du

- Reset app_managed_playback_pending on lifecycle cleanup to prevent state leak into next item
- Record visible overlay action only after command succeeds, not before
- Non-native passive overlay now always click-through on re-show (fix isNonNativePassiveOverlay ordering)
- Defer activeParsedSubtitleMediaPath assignment until after prefetch completes
- Move autoplay gate release into the hide branch of toggleVisibleOverlay
- Clear active Jellyfin playback when stopping media that never loaded
- Reset managed subtitle delay and delay key when no external tracks are available
- Await async removeDir in subtitle cache cleanup
- Guard duplicate delete clicks in MediaDetailView and SessionsTab with refs
- Escape key in DeleteConfirmDialog now calls stopPropagation and stopImmediatePropagation

src/core/services/tokenizer/annotation-stage.test.ts

@@ -425,6 +425,21 @@ test('shouldExcludeTokenFromSubtitleAnnotations keeps lexical tokens outside exp
   assert.equal(shouldExcludeTokenFromSubtitleAnnotations(token), false);
 });
 
+test('shouldExcludeTokenFromSubtitleAnnotations still excludes lexical non-independent kanji nouns from non-known annotations', () => {
+  const token = makeToken({
+    surface: '以外',
+    headword: '以外',
+    reading: 'イガイ',
+    partOfSpeech: PartOfSpeech.noun,
+    pos1: '名詞',
+    pos2: '非自立',
+    pos3: '副詞可能',
+  });
+
+  assert.equal(shouldExcludeTokenFromSubtitleAnnotations(token), true);
+  assert.equal(shouldExcludeTokenFromVocabularyPersistence(token), true);
+});
+
 test('shouldExcludeTokenFromSubtitleAnnotations excludes standalone particles auxiliaries and adnominals', () => {
   const tokens = [
     makeToken({
@@ -971,8 +986,8 @@ test('annotateTokens N+1 minimum sentence words counts only eligible word tokens
   );
 
   assert.equal(result[0]?.isKnown, false);
-  assert.equal(result[1]?.isKnown, false);
-  assert.equal(result[2]?.isKnown, false);
+  assert.equal(result[1]?.isKnown, true);
+  assert.equal(result[2]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
 });
 
@@ -1186,7 +1201,7 @@ test('annotateTokens excludes default non-independent pos2 from frequency and N+
   assert.equal(result[0]?.isNPlusOneTarget, false);
 });
 
-test('annotateTokens clears known-word status for non-independent kanji noun tokens', () => {
+test('annotateTokens keeps known-word status for non-independent kanji noun tokens', () => {
   const tokens = [
     makeToken({
       surface: '点',
@@ -1211,12 +1226,41 @@ test('annotateTokens clears known-word status for non-independent kanji noun tok
     { minSentenceWordsForNPlusOne: 1 },
   );
 
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
+test('annotateTokens keeps known-word status for lexical non-independent kanji nouns', () => {
+  const tokens = [
+    makeToken({
+      surface: '以外',
+      reading: 'イガイ',
+      headword: '以外',
+      partOfSpeech: PartOfSpeech.noun,
+      pos1: '名詞',
+      pos2: '非自立',
+      pos3: '副詞可能',
+      startPos: 2,
+      endPos: 4,
+      frequencyRank: 437,
+    }),
+  ];
+
+  const result = annotateTokens(
+    tokens,
+    makeDeps({
+      isKnownWord: (text) => text === '以外',
+    }),
+    { minSentenceWordsForNPlusOne: 1 },
+  );
+
+  assert.equal(result[0]?.isKnown, true);
+  assert.equal(result[0]?.frequencyRank, undefined);
+  assert.equal(result[0]?.isNPlusOneTarget, false);
+});
+
 test('annotateTokens clears all annotations for non-independent kanji noun tokens under unified gate', () => {
   const tokens = [
     makeToken({
@@ -1401,7 +1445,7 @@ test('annotateTokens excludes composite tokens when all component pos tags are e
   assert.equal(result[0]?.isNPlusOneTarget, false);
 });
 
-test('annotateTokens applies one shared exclusion gate across known N+1 frequency and JLPT', () => {
+test('annotateTokens lets known words bypass the shared exclusion gate for known status only', () => {
   const tokens = [
     makeToken({
       surface: 'これで',
@@ -1425,13 +1469,13 @@ test('annotateTokens applies one shared exclusion gate across known N+1 frequenc
     { minSentenceWordsForNPlusOne: 1 },
   );
 
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears known status and other annotations for kana-only non-independent noun helper merges', () => {
+test('annotateTokens keeps known status while clearing other annotations for kana-only non-independent noun helper merges', () => {
   const tokens = [
     makeToken({
       surface: 'ことに',
@@ -1455,13 +1499,13 @@ test('annotateTokens clears known status and other annotations for kana-only non
     { minSentenceWordsForNPlusOne: 1 },
   );
 
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears known status and other annotations for standalone auxiliary inflection fragments', () => {
+test('annotateTokens keeps known status while clearing other annotations for standalone auxiliary inflection fragments', () => {
   const tokens = [
     makeToken({
       surface: 'れる',
@@ -1497,14 +1541,14 @@ test('annotateTokens clears known status and other annotations for standalone au
   );
 
   for (const token of result) {
-    assert.equal(token.isKnown, false, token.surface);
+    assert.equal(token.isKnown, true, token.surface);
     assert.equal(token.isNPlusOneTarget, false, token.surface);
     assert.equal(token.frequencyRank, undefined, token.surface);
     assert.equal(token.jlptLevel, undefined, token.surface);
   }
 });
 
-test('annotateTokens clears known status and other annotations for auxiliary-only te-kureru helper spans', () => {
+test('annotateTokens keeps known status while clearing other annotations for auxiliary-only te-kureru helper spans', () => {
   const tokens = [
     makeToken({
       surface: 'てく',
@@ -1540,7 +1584,7 @@ test('annotateTokens clears known status and other annotations for auxiliary-onl
   );
 
   for (const token of result) {
-    assert.equal(token.isKnown, false, token.surface);
+    assert.equal(token.isKnown, true, token.surface);
     assert.equal(token.isNPlusOneTarget, false, token.surface);
     assert.equal(token.frequencyRank, undefined, token.surface);
     assert.equal(token.jlptLevel, undefined, token.surface);
@@ -1576,7 +1620,7 @@ test('annotateTokens keeps lexical くれる forms eligible for annotation', ()
   assert.equal(result[0]?.jlptLevel, 'N4');
 });
 
-test('annotateTokens clears known status and other annotations for standalone して helper fragments', () => {
+test('annotateTokens keeps known status while clearing other annotations for standalone して helper fragments', () => {
   const tokens = [
     makeToken({
       surface: 'してる',
@@ -1600,13 +1644,13 @@ test('annotateTokens clears known status and other annotations for standalone
     { minSentenceWordsForNPlusOne: 1 },
   );
 
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears known status and other annotations for standalone particle fragments without POS tags', () => {
+test('annotateTokens keeps known status while clearing other annotations for standalone particle fragments without POS tags', () => {
   const tokens = [
     makeToken({
       surface: 'と',
@@ -1630,13 +1674,13 @@ test('annotateTokens clears known status and other annotations for standalone pa
     { minSentenceWordsForNPlusOne: 1 },
   );
 
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears known status from standalone particles even when the known-word cache contains them', () => {
+test('annotateTokens keeps known status on standalone particles when the known-word cache contains them', () => {
   const tokens = [
     makeToken({
       surface: 'に',
@@ -1671,7 +1715,7 @@ test('annotateTokens clears known status from standalone particles even when the
     { minSentenceWordsForNPlusOne: 1 },
   );
 
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
@@ -1728,7 +1772,7 @@ test('annotateTokens does not mark standalone connective particles as N+1', () =
   assert.equal(result[1]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears known status and other annotations for rhetorical もんか grammar particle phrases', () => {
+test('annotateTokens keeps known status while clearing other annotations for rhetorical もんか grammar particle phrases', () => {
   const tokens = [
     makeToken({
       surface: 'もんか',
@@ -1752,13 +1796,13 @@ test('annotateTokens clears known status and other annotations for rhetorical
     { minSentenceWordsForNPlusOne: 1 },
   );
 
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears known status and other annotations for bare くれ auxiliary fragments', () => {
+test('annotateTokens keeps known status while clearing other annotations for bare くれ auxiliary fragments', () => {
   const tokens = [
     makeToken({
       surface: 'くれ',
@@ -1782,13 +1826,13 @@ test('annotateTokens clears known status and other annotations for bare くれ a
     { minSentenceWordsForNPlusOne: 1 },
   );
 
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears known status and other annotations for aru existence verbs', () => {
+test('annotateTokens keeps known status while clearing other annotations for aru existence verbs', () => {
   const tokens = [
     makeToken({
       surface: '有る',
@@ -1818,14 +1862,14 @@ test('annotateTokens clears known status and other annotations for aru existence
 
   assert.equal(result[0]?.surface, '有る');
   assert.equal(result[0]?.headword, '有る');
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.isNameMatch, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears known status and other annotations for standalone quote particle and auxiliary grammar terms', () => {
+test('annotateTokens keeps known status while clearing other annotations for standalone quote particle and auxiliary grammar terms', () => {
   const tokens = [
     makeToken({
       surface: 'って',
@@ -1861,14 +1905,14 @@ test('annotateTokens clears known status and other annotations for standalone qu
   );
 
   for (const token of result) {
-    assert.equal(token.isKnown, false, token.surface);
+    assert.equal(token.isKnown, true, token.surface);
     assert.equal(token.isNPlusOneTarget, false, token.surface);
     assert.equal(token.frequencyRank, undefined, token.surface);
     assert.equal(token.jlptLevel, undefined, token.surface);
   }
 });
 
-test('annotateTokens clears known status and other annotations from standalone あ interjections without POS tags', () => {
+test('annotateTokens keeps known status while clearing other annotations from standalone あ interjections without POS tags', () => {
   const tokens = [
     makeToken({
       surface: 'あ',
@@ -1898,13 +1942,13 @@ test('annotateTokens clears known status and other annotations from standalone
   assert.equal(result[0]?.surface, 'あ');
   assert.equal(result[0]?.headword, 'あ');
   assert.equal(result[0]?.reading, 'あ');
-  assert.equal(result[0]?.isKnown, false);
+  assert.equal(result[0]?.isKnown, true);
   assert.equal(result[0]?.isNPlusOneTarget, false);
   assert.equal(result[0]?.frequencyRank, undefined);
   assert.equal(result[0]?.jlptLevel, undefined);
 });
 
-test('annotateTokens clears all annotations from expressive subtitle interjections without POS tags', () => {
+test('annotateTokens keeps known status while clearing other annotations from expressive subtitle interjections without POS tags', () => {
   const tokens = [
     makeToken({
       surface: 'ハァ',
@@ -1960,7 +2004,7 @@ test('annotateTokens clears all annotations from expressive subtitle interjectio
   );
 
   for (const token of result.slice(0, 2)) {
-    assert.equal(token.isKnown, false, token.surface);
+    assert.equal(token.isKnown, true, token.surface);
     assert.equal(token.isNPlusOneTarget, false, token.surface);
     assert.equal(token.frequencyRank, undefined, token.surface);
     assert.equal(token.jlptLevel, undefined, token.surface);

src/core/services/tokenizer/annotation-stage.ts

@@ -680,6 +680,11 @@ export function annotateTokens(
 
   // Single pass: compute known word status, frequency filtering, and JLPT level together
   const annotated = tokens.map((token, index) => {
+    const isKnownForMatching = shouldComputeKnownStatus
+      ? computeTokenKnownStatus(token, deps.isKnownWord, deps.knownWordMatchMode)
+      : false;
+    nPlusOneKnownStatuses[index] = isKnownForMatching;
+
     if (
       sharedShouldExcludeTokenFromSubtitleAnnotations(token, {
         pos1Exclusions,
@@ -690,18 +695,13 @@ export function annotateTokens(
         pos1Exclusions,
         pos2Exclusions,
       });
-      nPlusOneKnownStatuses[index] = false;
       return {
         ...strippedToken,
-        isKnown: false,
+        isKnown: knownWordsEnabled ? isKnownForMatching : false,
       };
     }
 
     const prioritizedNameMatch = nameMatchEnabled && token.isNameMatch === true;
-    const isKnownForMatching = shouldComputeKnownStatus
-      ? computeTokenKnownStatus(token, deps.isKnownWord, deps.knownWordMatchMode)
-      : false;
-    nPlusOneKnownStatuses[index] = isKnownForMatching;
 
     const frequencyRank =
       frequencyEnabled && !prioritizedNameMatch