[ccc] Add language information to formats

youtube_dl/YoutubeDL.py

@@ -1791,6 +1791,10 @@ def _format_note(self, fdict):
         res = ''
         if fdict.get('ext') in ['f4f', 'f4m']:
             res += '(unsupported) '
+        if fdict.get('language'):
+            if res:
+                res += ' '
+            res += '[%s]' % fdict['language']
         if fdict.get('format_note') is not None:
             res += fdict['format_note'] + ' '
         if fdict.get('tbr') is not None:

youtube_dl/extractor/ccc.py

@@ -58,11 +58,12 @@ def _real_extract(self, url):
             webpage, 'duration', fatal=False, group='duration'))
 
         matches = re.finditer(r'''(?xs)
-            <(?:span|div)\s+class='label\s+filetype'>(?P<format>.*?)</(?:span|div)>\s*
+            <(?:span|div)\s+class='label\s+filetype'>(?P<format>[^<]*)</(?:span|div)>\s*
+            <(?:span|div)\s+class='label\s+filetype'>(?P<lang>[^<]*)</(?:span|div)>\s*
             <a\s+download\s+href='(?P<http_url>[^']+)'>\s*
             (?:
                 .*?
-                <a\s+href='(?P<torrent_url>[^']+\.torrent)'
+                <a\s+(?:download\s+)?href='(?P<torrent_url>[^']+\.torrent)'
             )?''', webpage)
         formats = []
         for m in matches:
@@ -70,12 +71,15 @@ def _real_extract(self, url):
             format_id = self._search_regex(
                 r'.*/([a-z0-9_-]+)/[^/]*$',
                 m.group('http_url'), 'format id', default=None)
+            if format_id:
+                format_id = m.group('lang') + '-' + format_id
             vcodec = 'h264' if 'h264' in format_id else (
                 'none' if format_id in ('mp3', 'opus') else None
             )
             formats.append({
                 'format_id': format_id,
                 'format': format,
+                'language': m.group('lang'),
                 'url': m.group('http_url'),
                 'vcodec': vcodec,
                 'preference': preference(format_id),

youtube_dl/extractor/common.py

@@ -108,8 +108,9 @@ class InfoExtractor(object):
                                  -2 or smaller for less than default.
                                  < -1000 to hide the format (if there is
                                     another one which is strictly better)
-                    * language_preference  Is this in the correct requested
-                                 language?
+                    * language   Language code, e.g. "de" or "en-US".
+                    * language_preference  Is this in the language mentioned in
+                                 the URL?
                                  10 if it's what the URL is about,
                                  -1 for default (don't know),
                                  -10 otherwise, other values reserved for now.