[youtube] Use 'orderedSet' instead of 'set' to preserve the order

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

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index a24c7358455057db646e20cca9b83eacd6678931..bd87c75b6b6e19ba562c041e0140b9b964cce82b 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -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')
 
@@ -1602,7 +1602,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         }
 
 
-class YoutubePlaylistIE(YoutubeBaseInfoExtractor, YoutubePlaylistBaseInfoExtractor):
+class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
     IE_DESC = 'YouTube.com playlists'
     _VALID_URL = r"""(?x)(?:
                         (?:https?://)?