fix: include cyrillic for detection

beinarovic · beinarovic · commit 36ea0d5fe5da · 2021-09-08T22:37:58.000+01:00
diff --git a/src/extraction/rules/cjk.ts b/src/extraction/rules/cjk.ts
@@ -13,5 +13,8 @@ export class CJKExtrationRule extends ExtractionRule {
     // Korean
     if (str.match(/\p{Script=Hangul}/u))
       return ExtractionScore.MustInclude
+    // Russian/Cyrillic
+    if (str.match(/\p{Script=Cyrillic}/u))
+      return ExtractionScore.MustInclude
   }
 }
diff --git a/test/unit/extraction/shouldExtract/cjk.test.ts b/test/unit/extraction/shouldExtract/cjk.test.ts
@@ -12,6 +12,7 @@ const includes = [
   '안녕하십니까',
   '再見',
   '再见',
+  'Привет',
 ]
 
 const excludes = [

Original file line number	Diff line number	Diff line change
`@@ -13,5 +13,8 @@ export class CJKExtrationRule extends ExtractionRule {`
`13`	`13`	`// Korean`
`14`	`14`	`if (str.match(/\p{Script=Hangul}/u))`
`15`	`15`	`return ExtractionScore.MustInclude`
	`16`	`+ // Russian/Cyrillic`
	`17`	`+ if (str.match(/\p{Script=Cyrillic}/u))`
	`18`	`+ return ExtractionScore.MustInclude`
`16`	`19`	`}`
`17`	`20`	`}`