Merge branch 'master' into rtmpdump

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

diff --git a/youtube_dl/extractor/arte.py b/youtube_dl/extractor/arte.py
index a636c6be56395e71764d0522526988bf0288b3cb..44d0b5d708aa31aef6d9997321142cb8638ab56b 100644 (file)
--- a/youtube_dl/extractor/arte.py
+++ b/youtube_dl/extractor/arte.py
@@ -69,7 +69,7 @@ class ArteTvIE(InfoExtractor):
             lang = mobj.group('lang')
             return self._extract_liveweb(url, name, lang)
 
-        if re.search(self._LIVE_URL, video_id) is not None:
+        if re.search(self._LIVE_URL, url) is not None:
             raise ExtractorError(u'Arte live streams are not yet supported, sorry')
             # self.extractLiveStream(url)
             # return
@@ -115,7 +115,7 @@ class ArteTvIE(InfoExtractor):
         event_doc = config_doc.find('event')
         url_node = event_doc.find('video').find('urlHd')
         if url_node is None:
-            url_node = video_doc.find('urlSd')
+            url_node = event_doc.find('urlSd')
 
         return {'id': video_id,
                 'title': event_doc.find('name%s' % lang.capitalize()).text,
@@ -182,15 +182,22 @@ class ArteTVPlus7IE(InfoExtractor):
                 formats = all_formats
             else:
                 raise ExtractorError(u'The formats list is empty')
-        # We order the formats by quality
+
         if re.match(r'[A-Z]Q', formats[0]['quality']) is not None:
-            sort_key = lambda f: ['HQ', 'MQ', 'EQ', 'SQ'].index(f['quality'])
+            def sort_key(f):
+                return ['HQ', 'MQ', 'EQ', 'SQ'].index(f['quality'])
         else:
-            sort_key = lambda f: int(f.get('height',-1))
+            def sort_key(f):
+                return (
+                    # Sort first by quality
+                    int(f.get('height',-1)),
+                    int(f.get('bitrate',-1)),
+                    # The original version with subtitles has lower relevance
+                    re.match(r'VO-ST(F|A)', f.get('versionCode', '')) is None,
+                    # The version with sourds/mal subtitles has also lower relevance
+                    re.match(r'VO?(F|A)-STM\1', f.get('versionCode', '')) is None,
+                )
         formats = sorted(formats, key=sort_key)
-        # Prefer videos without subtitles in the same language
-        formats = sorted(formats, key=lambda f: re.match(r'VO(F|A)-STM\1', f.get('versionCode', '')) is None)
-        # Pick the best quality
         def _format(format_info):
             quality = ''
             height = format_info.get('height')