fix: 康熙部首等不应被标注 kana

src/parse/krc.ts

@@ -39,16 +39,14 @@ export const parseKRC = (text: string, options: ParseOptions = {}): LyricResult
     cleanKangxi = false,
     applyOffset = false,
   } = options;
-  const content = cleanKangxi ? normalizeKangxi(text) : text;
-
   const metadata: LyricMetadata = extractMetadata ? { timingMode: "Word" } : {};
   const lines: LyricLine[] = [];
   let krcTranslations: string[] = [];
   let krcRomanizations: string[] = [];
   const mainLines: LyricLine[] = [];
   let kanaTag = "";
 
-  for (const raw of content.split("\n")) {
+  for (const raw of text.split("\n")) {
     const trimmed = raw.trim();
     if (!trimmed) continue;
 
@@ -163,6 +161,14 @@ export const parseKRC = (text: string, options: ParseOptions = {}): LyricResult
     applyKanaToLines(lines, kanaTag);
   }
 
+  if (cleanKangxi) {
+    for (const line of lines) {
+      for (const word of line.words) {
+        word.word = normalizeKangxi(word.word);
+      }
+    }
+  }
+
   if (applyOffset && metadata.offset) {
     applyTimestampOffset(lines, metadata.offset);
   }

src/parse/qrc.ts

@@ -92,8 +92,7 @@ export const parseQRC = (text: string, options: ParseOptions = {}): LyricResult
     cleanKangxi = false,
     applyOffset = false,
   } = options;
-  const normalized = cleanKangxi ? normalizeKangxi(text) : text;
-  const { content, xmlMeta } = extractFromXml(normalized);
+  const { content, xmlMeta } = extractFromXml(text);
   const metadata: LyricMetadata =
     extractMetadata && xmlMeta
       ? { ...xmlMeta, timingMode: "Word" }
@@ -149,6 +148,14 @@ export const parseQRC = (text: string, options: ParseOptions = {}): LyricResult
     applyKanaToLines(lines, kanaTag);
   }
 
+  if (cleanKangxi) {
+    for (const line of lines) {
+      for (const word of line.words) {
+        word.word = normalizeKangxi(word.word);
+      }
+    }
+  }
+
   if (applyOffset && metadata.offset) {
     applyTimestampOffset(lines, metadata.offset);
   }

tests/kana.spec.ts

@@ -120,6 +120,22 @@ describe("Kana / Ruby Furigana Support", () => {
       expect(result.lines).toHaveLength(1);
       expect(result.lines[0].words[0].ruby).toBeUndefined();
     });
+
+    it("应在康熙部首归一化前判断 kana 基础字符", () => {
+      const qrcText = `[offset:0]
+[kana:1は(150865,168)な(151033,252)1お(153326,189)ど(153515,243)]
+[150865,1633]離(150865,420)さ(151285,168)な(151453,260)い(151713,268)と(151981,249)⾔(152230,38)う(152268,230)
+[152498,2173]((152498,192)Memento (152690,328)mori)(153018,308)躍(153326,432)る(153758,177)よ(153935,78)う(154013,78)に(154091,580)`;
+
+      const result = parseQRC(qrcText, { cleanKangxi: true });
+      const [line0, line1] = result.lines;
+      expect(line0.words[0].word).toBe("離");
+      expect(line0.words[0].ruby?.map((ruby) => ruby.word)).toEqual(["は", "な"]);
+      expect(line0.words[5].word).toBe("言");
+      expect(line0.words[5].ruby).toBeUndefined();
+      expect(line1.words[3].word).toBe("躍");
+      expect(line1.words[3].ruby?.map((ruby) => ruby.word)).toEqual(["お", "ど"]);
+    });
   });
 
   describe("parseKRC with Kana", () => {