fix(tokenizer): disambiguate Yomitan frequency lookup by reading

src/core/services/tokenizer.test.ts

@@ -297,7 +297,7 @@ test('tokenizeSubtitle starts Yomitan frequency lookup and MeCab enrichment in p
   assert.equal(result.tokens?.[0]?.frequencyRank, 77);
 });
 
-test('tokenizeSubtitle queries headword frequencies without forcing surface reading', async () => {
+test('tokenizeSubtitle queries headword frequencies with token reading for disambiguation', async () => {
   const result = await tokenizeSubtitle(
     '鍛えた',
     makeDeps({
@@ -309,7 +309,7 @@ test('tokenizeSubtitle queries headword frequencies without forcing surface read
           webContents: {
             executeJavaScript: async (script: string) => {
               if (script.includes('getTermFrequencies')) {
-                if (!script.includes('"term":"鍛える","reading":null')) {
+                if (!script.includes('"term":"鍛える","reading":"きた"')) {
                   return [];
                 }
                 return [
@@ -351,6 +351,68 @@ test('tokenizeSubtitle queries headword frequencies without forcing surface read
   assert.equal(result.tokens?.[0]?.frequencyRank, 2847);
 });
 
+test('tokenizeSubtitle avoids headword term-only fallback rank when reading-specific frequency exists', async () => {
+  const result = await tokenizeSubtitle(
+    '無人',
+    makeDeps({
+      getFrequencyDictionaryEnabled: () => true,
+      getYomitanExt: () => ({ id: 'dummy-ext' }) as any,
+      getYomitanParserWindow: () =>
+        ({
+          isDestroyed: () => false,
+          webContents: {
+            executeJavaScript: async (script: string) => {
+              if (script.includes('getTermFrequencies')) {
+                if (!script.includes('"term":"無人","reading":"むじん"')) {
+                  return [];
+                }
+                return [
+                  {
+                    term: '無人',
+                    reading: null,
+                    dictionary: 'CC100',
+                    dictionaryPriority: 0,
+                    frequency: 157632,
+                    displayValue: null,
+                    displayValueParsed: false,
+                  },
+                  {
+                    term: '無人',
+                    reading: 'むじん',
+                    dictionary: 'CC100',
+                    dictionaryPriority: 0,
+                    frequency: 7141,
+                    displayValue: null,
+                    displayValueParsed: false,
+                  },
+                ];
+              }
+
+              return [
+                {
+                  source: 'scanning-parser',
+                  index: 0,
+                  content: [
+                    [
+                      {
+                        text: '無人',
+                        reading: 'むじん',
+                        headwords: [[{ term: '無人' }]],
+                      },
+                    ],
+                  ],
+                },
+              ];
+            },
+          },
+        }) as unknown as Electron.BrowserWindow,
+    }),
+  );
+
+  assert.equal(result.tokens?.length, 1);
+  assert.equal(result.tokens?.[0]?.frequencyRank, 7141);
+});
+
 test('tokenizeSubtitle prefers Yomitan frequency from highest-priority dictionary', async () => {
   const result = await tokenizeSubtitle(
     '猫',

src/core/services/tokenizer.ts

@@ -284,8 +284,7 @@ function buildYomitanFrequencyTermReadingList(
       }
       const readingRaw =
         token.reading && token.reading.trim().length > 0 ? token.reading.trim() : null;
-      const reading = matchMode === 'headword' ? null : readingRaw;
-      return { term, reading };
+      return { term, reading: readingRaw };
     })
     .filter((pair): pair is { term: string; reading: string | null } => pair !== null);
 }

src/core/services/tokenizer/yomitan-parser-runtime.test.ts

@@ -130,6 +130,28 @@ test('requestYomitanTermFrequencies returns normalized frequency entries', async
   assert.match(scriptValue, /optionsGetFull/);
 });
 
+test('requestYomitanTermFrequencies prefers primary rank from displayValue array pair', async () => {
+  const deps = createDeps(async () => [
+    {
+      term: '無人',
+      reading: 'むじん',
+      dictionary: 'freq-dict',
+      dictionaryPriority: 0,
+      frequency: 157632,
+      displayValue: [7141, 157632],
+      displayValueParsed: true,
+    },
+  ]);
+
+  const result = await requestYomitanTermFrequencies([{ term: '無人', reading: 'むじん' }], deps, {
+    error: () => undefined,
+  });
+
+  assert.equal(result.length, 1);
+  assert.equal(result[0]?.term, '無人');
+  assert.equal(result[0]?.frequency, 7141);
+});
+
 test('requestYomitanTermFrequencies caches profile metadata between calls', async () => {
   const scripts: string[] = [];
   const deps = createDeps(async (script) => {

src/core/services/tokenizer/yomitan-parser-runtime.ts

@@ -96,6 +96,28 @@ function parsePositiveFrequencyString(value: string): number | null {
   return parsed;
 }
 
+function parsePositiveFrequencyValue(value: unknown): number | null {
+  const numeric = asPositiveInteger(value);
+  if (numeric !== null) {
+    return numeric;
+  }
+
+  if (typeof value === 'string') {
+    return parsePositiveFrequencyString(value);
+  }
+
+  if (Array.isArray(value)) {
+    for (const item of value) {
+      const parsed = parsePositiveFrequencyValue(item);
+      if (parsed !== null) {
+        return parsed;
+      }
+    }
+  }
+
+  return null;
+}
+
 function toYomitanTermFrequency(value: unknown): YomitanTermFrequency | null {
   if (!isObject(value)) {
     return null;
@@ -103,15 +125,12 @@ function toYomitanTermFrequency(value: unknown): YomitanTermFrequency | null {
 
   const term = typeof value.term === 'string' ? value.term.trim() : '';
   const dictionary = typeof value.dictionary === 'string' ? value.dictionary.trim() : '';
-  const rawFrequency = asPositiveInteger(value.frequency);
-  const displayValueRaw =
-    value.displayValue === null
-      ? null
-      : typeof value.displayValue === 'string'
-        ? value.displayValue
-        : null;
+  const rawFrequency = parsePositiveFrequencyValue(value.frequency);
+  const displayValueRaw = value.displayValue;
   const parsedDisplayFrequency =
-    displayValueRaw !== null ? parsePositiveFrequencyString(displayValueRaw) : null;
+    displayValueRaw !== null && displayValueRaw !== undefined
+      ? parsePositiveFrequencyValue(displayValueRaw)
+      : null;
   const frequency = parsedDisplayFrequency ?? rawFrequency;
   if (!term || !dictionary || frequency === null) {
     return null;
@@ -128,7 +147,7 @@ function toYomitanTermFrequency(value: unknown): YomitanTermFrequency | null {
       : typeof value.reading === 'string'
         ? value.reading
         : null;
-  const displayValue = displayValueRaw;
+  const displayValue = typeof displayValueRaw === 'string' ? displayValueRaw : null;
   const displayValueParsed = value.displayValueParsed === true;
 
   return {