♻️ refactor(lint): Update language family sets and improve severity logic for accuracy

canisminor1990 · canisminor1990 · commit 028ac5a09bde · 2025-06-24T15:34:51.000+08:00
diff --git a/packages/lobe-i18n/src/commands/Lint/constants.ts b/packages/lobe-i18n/src/commands/Lint/constants.ts
@@ -176,7 +176,7 @@ export const HYPHENATED_LANGUAGE_MAP: { [key: string]: string } = {
 export const ARABIC_SIMILAR_LANGUAGES = new Set(['ar', 'fa', 'ur', 'ku']);
 
 export const LANGUAGE_FAMILIES = {
-  EAST_ASIAN: new Set(['zh', 'ja', 'ko', 'vi', 'th', 'my', 'km', 'lo', 'mn', 'bo', 'ug']),
+  EAST_ASIAN: new Set(['ja', 'ko', 'vi', 'th', 'my', 'km', 'lo', 'mn', 'bo', 'ug']),
   GERMANIC: new Set(['de', 'nl', 'da', 'no', 'sv', 'is', 'fo', 'af']),
   MALAY: new Set(['id', 'ms']),
   ROMANCE: new Set(['es', 'pt', 'ca', 'it', 'fr', 'ro', 'gl']),
diff --git a/packages/lobe-i18n/src/commands/Lint/utils/languageValidation.ts b/packages/lobe-i18n/src/commands/Lint/utils/languageValidation.ts
@@ -63,32 +63,43 @@ export function determineSeverity(
   mainLanguage: string,
   text: string,
 ): 'error' | 'warning' {
-  // 对于短文本（<=3），置信度低于0.7时只报 warning
-  let severity: 'error' | 'warning' =
-    textLength <= 3
-      ? confidence > CONFIDENCE_THRESHOLDS.SHORT_TEXT
-        ? 'error'
-        : 'warning'
-      : confidence > CONFIDENCE_THRESHOLDS.LONG_TEXT
-        ? 'error'
-        : 'warning';
-
-  // 如果是东亚语言和其他语系的混用，判定为 error
+  // 默认为 warning
+  let severity: 'error' | 'warning' = 'warning';
+
+  // 规则1：对于长文本，如果置信度足够高，则为 error
+  if (textLength > 3 && confidence > CONFIDENCE_THRESHOLDS.LONG_TEXT) {
+    severity = 'error';
+  }
+
+  // 规则2：对于短文本，如果置信度足够高，则为 error
+  if (textLength <= 3 && confidence > CONFIDENCE_THRESHOLDS.SHORT_TEXT) {
+    severity = 'error';
+  }
+
+  // 规则3：检测到东亚语言和其他语系混用，则为 error
   const isDetectedEastAsian = LANGUAGE_FAMILIES.EAST_ASIAN.has(detectedLanguage);
   const isMainEastAsian = LANGUAGE_FAMILIES.EAST_ASIAN.has(mainLanguage);
-
   if (isDetectedEastAsian !== isMainEastAsian && confidence > 0.5) {
-    return 'error';
+    severity = 'error';
+  }
+
+  // 规则4：如果主语言和检测语言不属于同一个语系，则为 error
+  const mainFamily = Object.values(LANGUAGE_FAMILIES).find((family) => family.has(mainLanguage));
+  const detectedFamily = Object.values(LANGUAGE_FAMILIES).find((family) =>
+    family.has(detectedLanguage),
+  );
+  if (mainFamily && detectedFamily && mainFamily !== detectedFamily && confidence > 0.6) {
+    severity = 'error';
   }
 
-  // 检测明显英文内容的逻辑
+  // 规则5：检测到明显英文内容，则为 error
   if (
     detectedLanguage === 'en' &&
     mainLanguage !== 'en' &&
     containsEnglishTerms(text) &&
     confidence > 0.5
   ) {
-    severity = 'error'; // 明显英文内容直接设为error
+    severity = 'error';
   }
 
   return severity;
