Merge pull request #8374 from yan12125/facebook-dash
[youtube-dl] / youtube_dl / extractor / youtube.py
index a2f7760502d3771335fdc88496e443ed67f6ea2b..a7f8c968ef2eb870355be1898bafe123e59dc294 100644 (file)
@@ -181,7 +181,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
             return
 
 
-class YoutubeEntryListBaseInfoExtractor(InfoExtractor):
+class YoutubeEntryListBaseInfoExtractor(YoutubeBaseInfoExtractor):
     # Extract entries from page with "Load more" button
     def _entries(self, page, playlist_id):
         more_widget_html = content_html = page
@@ -233,7 +233,7 @@ class YoutubePlaylistBaseInfoExtractor(YoutubeEntryListBaseInfoExtractor):
 
 class YoutubePlaylistsBaseInfoExtractor(YoutubeEntryListBaseInfoExtractor):
     def _process_page(self, content):
-        for playlist_id in re.findall(r'href="/?playlist\?list=(.+?)"', content):
+        for playlist_id in orderedSet(re.findall(r'href="/?playlist\?list=([0-9A-Za-z-_]{10,})"', content)):
             yield self.url_result(
                 'https://www.youtube.com/playlist?list=%s' % playlist_id, 'YoutubePlaylist')
 
@@ -1472,14 +1472,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                         return '/signature/%s' % dec_s
 
                     dash_manifest_url = re.sub(r'/s/([a-fA-F0-9\.]+)', decrypt_sig, dash_manifest_url)
-                    dash_doc = self._download_xml(
-                        dash_manifest_url, video_id,
-                        note='Downloading DASH manifest',
-                        errnote='Could not download DASH manifest',
-                        fatal=dash_mpd_fatal)
-
-                    for df in self._parse_dash_manifest(
-                            dash_doc, namespace='urn:mpeg:DASH:schema:MPD:2011', formats_dict=self._formats):
+
+                    for df in self._extract_dash_manifest_formats(
+                            dash_manifest_url, video_id, fatal=dash_mpd_fatal,
+                            namespace='urn:mpeg:DASH:schema:MPD:2011', formats_dict=self._formats):
                         # Do not overwrite DASH format found in some previous DASH manifest
                         if df['format_id'] not in dash_formats:
                             dash_formats[df['format_id']] = df
@@ -1547,7 +1543,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         }
 
 
-class YoutubePlaylistIE(YoutubeBaseInfoExtractor, YoutubePlaylistBaseInfoExtractor):
+class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
     IE_DESC = 'YouTube.com playlists'
     _VALID_URL = r"""(?x)(?:
                         (?:https?://)?