[youtube:playlist] Work around buggy playlists (fixes #4449)

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

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 823d6aaf3a4f38ad5696e67a23b14e6fe690fabd..aaa07b5eb31f1c15a36a5d05c66dbd29dd85c0e9 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -478,7 +478,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
 
     def _extract_signature_function(self, video_id, player_url, example_sig):
         id_m = re.match(
-            r'.*-(?P<id>[a-zA-Z0-9_-]+)(?:/watch_as3|/html5player)?\.(?P<ext>[a-z]+)$',
+            r'.*?-(?P<id>[a-zA-Z0-9_-]+)(?:/watch_as3|/html5player)?\.(?P<ext>[a-z]+)$',
             player_url)
         if not id_m:
             raise ExtractorError('Cannot identify player %r' % player_url)
@@ -1128,6 +1128,13 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor):
         'info_dict': {
             'title': 'JODA7',
         }
+    }, {
+        'note': 'Buggy playlist: the webpage has a "Load more" button but it doesn\'t have more videos',
+        'url': 'https://www.youtube.com/playlist?list=UUXw-G3eDE9trcvY2sBMM_aA',
+        'info_dict': {
+                'title': 'Uploads from Interstellar Movie',
+        },
+        'playlist_mincout': 21,
     }]
 
     def _real_initialize(self):
@@ -1212,6 +1219,10 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor):
                 'Downloading page #%s' % page_num,
                 transform_source=uppercase_escape)
             content_html = more['content_html']
+            if not content_html.strip():
+                # Some webpages show a "Load more" button but they don't
+                # have more videos
+                break
             more_widget_html = more['load_more_widget_html']
 
         playlist_title = self._html_search_regex(