fix: suppress N+1 for kana-only candidates and fix minSentenceWords coun

- Treat kana-only tokens with surrounding subtitle punctuation (…, ―, etc.) as kana-only so they are not promoted to N+1 targets
- Exclude unknown tokens filtered from N+1 targeting from the minSentenceWords count so filtered kana-only unknowns cannot satisfy sentence length threshold
- Add regression tests for kana-only candidate suppression and filtered-unknown padding cases

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

@@ -709,6 +709,63 @@ test('annotateTokens N+1 handoff marks expected target when threshold is satisfi
   assert.equal(result[2]?.isNPlusOneTarget, false);
 });
 
+test('annotateTokens does not mark kana-only unknown target with subtitle punctuation as N+1', () => {
+  const tokens = [
+    makeToken({
+      surface: '何やら',
+      headword: '何やら',
+      reading: 'ナニヤラ',
+      pos1: '副詞',
+      startPos: 0,
+      endPos: 3,
+    }),
+    makeToken({
+      surface: 'ボタン',
+      headword: 'ボタン',
+      reading: 'ボタン',
+      pos1: '名詞',
+      startPos: 3,
+      endPos: 6,
+    }),
+    makeToken({
+      surface: 'スイッチ…',
+      headword: 'スイッチ',
+      reading: 'スイッチ',
+      pos1: '名詞',
+      startPos: 6,
+      endPos: 11,
+    }),
+  ];
+
+  const result = annotateTokens(
+    tokens,
+    makeDeps({
+      isKnownWord: (text) => text === '何やら' || text === 'ボタン',
+    }),
+    { minSentenceWordsForNPlusOne: 3 },
+  );
+
+  assert.equal(result[2]?.isNPlusOneTarget, false);
+});
+
+test('annotateTokens still marks kanji unknown target in otherwise eligible sentence as N+1', () => {
+  const tokens = [
+    makeToken({ surface: '私', headword: '私', pos1: '名詞', startPos: 0, endPos: 1 }),
+    makeToken({ surface: '猫', headword: '猫', pos1: '名詞', startPos: 1, endPos: 2 }),
+    makeToken({ surface: '装置…', headword: '装置', pos1: '名詞', startPos: 2, endPos: 5 }),
+  ];
+
+  const result = annotateTokens(
+    tokens,
+    makeDeps({
+      isKnownWord: (text) => text === '私' || text === '猫',
+    }),
+    { minSentenceWordsForNPlusOne: 3 },
+  );
+
+  assert.equal(result[2]?.isNPlusOneTarget, true);
+});
+
 test('annotateTokens N+1 minimum sentence words counts only eligible word tokens', () => {
   const tokens = [
     makeToken({ surface: '猫', headword: '猫', startPos: 0, endPos: 1 }),
@@ -744,6 +801,32 @@ test('annotateTokens N+1 minimum sentence words counts only eligible word tokens
   assert.equal(result[0]?.isNPlusOneTarget, false);
 });
 
+test('annotateTokens N+1 minimum sentence words excludes unknown tokens filtered from N+1 targeting', () => {
+  const tokens = [
+    makeToken({ surface: '私', headword: '私', pos1: '名詞', startPos: 0, endPos: 1 }),
+    makeToken({ surface: '猫', headword: '猫', pos1: '名詞', startPos: 1, endPos: 2 }),
+    makeToken({
+      surface: 'スイッチ',
+      headword: 'スイッチ',
+      reading: 'スイッチ',
+      pos1: '名詞',
+      startPos: 2,
+      endPos: 6,
+    }),
+    makeToken({ surface: '装置', headword: '装置', pos1: '名詞', startPos: 6, endPos: 8 }),
+  ];
+
+  const result = annotateTokens(
+    tokens,
+    makeDeps({
+      isKnownWord: (text) => text === '私' || text === '猫',
+    }),
+    { minSentenceWordsForNPlusOne: 4 },
+  );
+
+  assert.equal(result[3]?.isNPlusOneTarget, false);
+});
+
 test('annotateTokens N+1 sentence word count respects source punctuation gaps omitted by Yomitan', () => {
   const tokens = [
     makeToken({

src/token-merger.ts

@@ -298,9 +298,28 @@ function isKanaChar(char: string): boolean {
   );
 }
 
+function isKanaCandidateIgnorableChar(char: string): boolean {
+  return /^[\s.,!?;:()[\]{}"'`、。！？…‥・「」『』（）［］｛｝〈〉《》【】―-]$/u.test(char);
+}
+
 function isKanaOnlyText(text: string): boolean {
   const normalized = text.trim();
-  return normalized.length > 0 && Array.from(normalized).every((char) => isKanaChar(char));
+  if (normalized.length === 0) {
+    return false;
+  }
+
+  let hasKana = false;
+  for (const char of normalized) {
+    if (isKanaChar(char)) {
+      hasKana = true;
+      continue;
+    }
+    if (!isKanaCandidateIgnorableChar(char)) {
+      return false;
+    }
+  }
+
+  return hasKana;
 }
 
 function normalizeSourceTextForTokenOffsets(sourceText: string | undefined): string | undefined {
@@ -367,6 +386,18 @@ function isNPlusOneWordCountToken(
   return true;
 }
 
+function isNPlusOneSentenceLengthToken(
+  token: MergedToken,
+  pos1Exclusions: ReadonlySet<string> = N_PLUS_ONE_IGNORED_POS1,
+  pos2Exclusions: ReadonlySet<string> = N_PLUS_ONE_IGNORED_POS2,
+): boolean {
+  if (!isNPlusOneWordCountToken(token, pos1Exclusions, pos2Exclusions)) {
+    return false;
+  }
+
+  return token.isKnown || isNPlusOneCandidateToken(token, pos1Exclusions, pos2Exclusions);
+}
+
 function isSentenceBoundaryToken(token: MergedToken): boolean {
   if (token.partOfSpeech !== PartOfSpeech.symbol) {
     return false;
@@ -418,7 +449,7 @@ export function markNPlusOneTargets(
     for (let i = start; i < endExclusive; i++) {
       const token = markedTokens[i];
       if (!token) continue;
-      if (isNPlusOneWordCountToken(token, pos1Exclusions, pos2Exclusions)) {
+      if (isNPlusOneSentenceLengthToken(token, pos1Exclusions, pos2Exclusions)) {
         sentenceWordCount += 1;
       }