[refactor] Single quotes consistency

[youtube-dl] / youtube_dl / extractor / ccc.py

diff --git a/youtube_dl/extractor/ccc.py b/youtube_dl/extractor/ccc.py
index 710e5919ce52b853046fa8bf7f5a5bb15dae5085..dda2c0959882c3cd3c5de56b817ccd7815ef0068 100644 (file)
--- a/youtube_dl/extractor/ccc.py
+++ b/youtube_dl/extractor/ccc.py
@@ -45,7 +45,7 @@ class CCCIE(InfoExtractor):
         title = self._html_search_regex(
             r'(?s)<h1>(.*?)</h1>', webpage, 'title')
         description = self._html_search_regex(
-            r"(?s)<h3>About</h3>(.+?)<h3>",
+            r'(?s)<h3>About</h3>(.+?)<h3>',
             webpage, 'description', fatal=False)
         upload_date = unified_strdate(self._html_search_regex(
             r"(?s)<span[^>]+class='[^']*fa-calendar-o'[^>]*>(.+?)</span>",
@@ -58,11 +58,12 @@ class CCCIE(InfoExtractor):
             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 @@ class CCCIE(InfoExtractor):
             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),