fix: add a negative lookahead for subtitles in language parsing

packages/parser/src/regex.ts

@@ -1,6 +1,9 @@
 const createRegex = (pattern: string): RegExp =>
   new RegExp(`(?<![^\\s\\[(_\\-.,])(${pattern})(?=[\\s\\)\\]_.\\-,]|$)`, 'i');
 
+const createLanguageRegex = (pattern: string): RegExp =>
+  createRegex(`${pattern}(?![ .\\-_]?sub(title)?s?)`);
+
 export const PARSE_REGEX = {
   resolutions: {
     '2160p': createRegex(
@@ -74,57 +77,57 @@ export const PARSE_REGEX = {
     'H-SBS': createRegex('h?(alf)?[ .\\-_]?(sbs|side[ .\\-_]?by[ .\\-_]?side)'),
   },
   languages: {
-    Multi: createRegex('multi|multi[ .\\-_]?audio'),
-    'Dual Audio': createRegex(
+    Multi: createLanguageRegex('multi'),
+    'Dual Audio': createLanguageRegex(
       'dual[ .\\-_]?(audio|lang(uage)?|flac|ac3|aac2?)'
     ),
-    Dubbed: createRegex('dub(bed)?'),
-    English: createRegex('english|eng'),
-    Japanese: createRegex('japanese|jap'),
-    Chinese: createRegex('chinese|chi'),
-    Russian: createRegex('russian|rus'),
-    Arabic: createRegex('arabic|ara'),
-    Portuguese: createRegex('portuguese|por'),
-    Spanish: createRegex('spanish|spa|esp'),
-    French: createRegex('french|fra'),
-    German: createRegex('german|ger'),
-    Italian: createRegex('italian|ita'),
-    Korean: createRegex('korean|kor'),
-    Hindi: createRegex('hindi|hin'),
-    Bengali: createRegex('bengali|ben'),
-    Punjabi: createRegex('punjabi|pan'),
-    Marathi: createRegex('marathi|mar'),
-    Gujarati: createRegex('gujarati|guj'),
-    Tamil: createRegex('tamil|tam'),
-    Telugu: createRegex('telugu|tel'),
-    Kannada: createRegex('kannada|kan'),
-    Malayalam: createRegex('malayalam|mal'),
-    Thai: createRegex('thai|tha'),
-    Vietnamese: createRegex('vietnamese|vie'),
-    Indonesian: createRegex('indonesian|ind'),
-    Turkish: createRegex('turkish|tur'),
-    Hebrew: createRegex('hebrew|heb'),
-    Persian: createRegex('persian|per'),
-    Ukrainian: createRegex('ukrainian|ukr'),
-    Greek: createRegex('greek|ell'),
-    Lithuanian: createRegex('lithuanian|lit'),
-    Latvian: createRegex('latvian|lav'),
-    Estonian: createRegex('estonian|est'),
-    Polish: createRegex('polish|pol'),
-    Czech: createRegex('czech|cze'),
-    Slovak: createRegex('slovak|slo'),
-    Hungarian: createRegex('hungarian|hun'),
-    Romanian: createRegex('romanian|rum'),
-    Bulgarian: createRegex('bulgarian|bul'),
-    Serbian: createRegex('serbian|srp'),
-    Croatian: createRegex('croatian|hrv'),
-    Slovenian: createRegex('slovenian|slv'),
-    Dutch: createRegex('dutch|dut'),
-    Danish: createRegex('danish|dan'),
-    Finnish: createRegex('finnish|fin'),
-    Swedish: createRegex('swedish|swe'),
-    Norwegian: createRegex('norwegian|nor'),
-    Malay: createRegex('malay'),
-    Latino: createRegex('latino|lat'),
+    Dubbed: createLanguageRegex('dub(bed)?'),
+    English: createLanguageRegex('english|eng'),
+    Japanese: createLanguageRegex('japanese|jap'),
+    Chinese: createLanguageRegex('chinese|chi'),
+    Russian: createLanguageRegex('russian|rus'),
+    Arabic: createLanguageRegex('arabic|ara'),
+    Portuguese: createLanguageRegex('portuguese|por'),
+    Spanish: createLanguageRegex('spanish|spa|esp'),
+    French: createLanguageRegex('french|fra'),
+    German: createLanguageRegex('german|ger'),
+    Italian: createLanguageRegex('italian|ita'),
+    Korean: createLanguageRegex('korean|kor'),
+    Hindi: createLanguageRegex('hindi|hin'),
+    Bengali: createLanguageRegex('bengali|ben'),
+    Punjabi: createLanguageRegex('punjabi|pan'),
+    Marathi: createLanguageRegex('marathi|mar'),
+    Gujarati: createLanguageRegex('gujarati|guj'),
+    Tamil: createLanguageRegex('tamil|tam'),
+    Telugu: createLanguageRegex('telugu|tel'),
+    Kannada: createLanguageRegex('kannada|kan'),
+    Malayalam: createLanguageRegex('malayalam|mal'),
+    Thai: createLanguageRegex('thai|tha'),
+    Vietnamese: createLanguageRegex('vietnamese|vie'),
+    Indonesian: createLanguageRegex('indonesian|ind'),
+    Turkish: createLanguageRegex('turkish|tur'),
+    Hebrew: createLanguageRegex('hebrew|heb'),
+    Persian: createLanguageRegex('persian|per'),
+    Ukrainian: createLanguageRegex('ukrainian|ukr'),
+    Greek: createLanguageRegex('greek|ell'),
+    Lithuanian: createLanguageRegex('lithuanian|lit'),
+    Latvian: createLanguageRegex('latvian|lav'),
+    Estonian: createLanguageRegex('estonian|est'),
+    Polish: createLanguageRegex('polish|pol'),
+    Czech: createLanguageRegex('czech|cze'),
+    Slovak: createLanguageRegex('slovak|slo'),
+    Hungarian: createLanguageRegex('hungarian|hun'),
+    Romanian: createLanguageRegex('romanian|rum'),
+    Bulgarian: createLanguageRegex('bulgarian|bul'),
+    Serbian: createLanguageRegex('serbian|srp'),
+    Croatian: createLanguageRegex('croatian|hrv'),
+    Slovenian: createLanguageRegex('slovenian|slv'),
+    Dutch: createLanguageRegex('dutch|dut'),
+    Danish: createLanguageRegex('danish|dan'),
+    Finnish: createLanguageRegex('finnish|fin'),
+    Swedish: createLanguageRegex('swedish|swe'),
+    Norwegian: createLanguageRegex('norwegian|nor'),
+    Malay: createLanguageRegex('malay'),
+    Latino: createLanguageRegex('latino|lat'),
   },
 };