requirements.txt

@@ -3,3 +3,4 @@ tvdb_api==2.0
 hashids==1.2.0
 enzyme>=0.4.1
 click>=6.7
+langdetect>=1.0.7

src/core.py

@@ -14,6 +14,7 @@ import tvdb_api
 import click
 from pprint import pprint
 from titlecase import titlecase
+import langdetect
 
 import env_variables as env
 
@@ -49,6 +50,16 @@ def search_external_subtitles(path, directory=None):
             except (ValueError, LanguageReverseError):
                 logger.error('Cannot parse language code %r', language_code)
 
+                f = open(p, 'r', encoding='ISO-8859-15')
+
+                pattern = re.compile('[0-9:\,-<>]+')
+                # head = list(islice(f.read(), 10))
+                filecontent = pattern.sub('', f.read())
+                filecontent = filecontent[0:1000]
+                language = langdetect.detect(filecontent)
+                print(language)
+                f.close()
+
         subtitles[p] = language
     logger.debug('Found subtitles %r', subtitles)