[youtube:playlist] Fix mixes extraction (fixes #3051)

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

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index f7bfa9f886358c10672df2b2bc4da4974fb02000..2c51a0b47f134ddf0478210b97731893f76c998c 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -1140,7 +1140,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
         mobj = re.search(r'(?s)id="eow-date.*?>(.*?)</span>', video_webpage)
         if mobj is None:
             mobj = re.search(
-                r'(?s)id="watch-uploader-info".*?>.*?Published on (.*?)</strong>',
+                r'(?s)id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live) on (.*?)</strong>',
                 video_webpage)
         if mobj is not None:
             upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split())
@@ -1414,11 +1414,9 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor):
         title_span = (search_title('playlist-title') or
             search_title('title long-title') or search_title('title'))
         title = clean_html(title_span)
-        video_re = r'''(?x)data-video-username="(.*?)".*?
+        video_re = r'''(?x)data-video-username=".*?".*?
                        href="/watch\?v=([0-9A-Za-z_-]{11})&amp;[^"]*?list=%s''' % re.escape(playlist_id)
-        matches = orderedSet(re.findall(video_re, webpage, flags=re.DOTALL))
-        # Some of the videos may have been deleted, their username field is empty
-        ids = [video_id for (username, video_id) in matches if username]
+        ids = orderedSet(re.findall(video_re, webpage, flags=re.DOTALL))
         url_results = self._ids_to_results(ids)
 
         return self.playlist_result(url_results, playlist_id, title)