[youtube] Fix history, trending and mix playlists (#136)

Co-authored-by: pukkandan <pukkandan.ytdlp@gmail.com>
Co-authored-by: Matthew <colethedj@protonmail.com>

test/test_all_urls.py

@@ -37,7 +37,6 @@ def test_youtube_playlist_matching(self):
         assertPlaylist('PL63F0C78739B09958')
         assertTab('https://www.youtube.com/AsapSCIENCE')
         assertTab('https://www.youtube.com/embedded')
-        assertTab('https://www.youtube.com/feed')  # Own channel's home page
         assertTab('https://www.youtube.com/playlist?list=UUBABnxM4Ar9ten8Mdjj1j0Q')
         assertTab('https://www.youtube.com/course?list=ECUl4u3cNGP61MdtwGTqZA0MreSaDybji8')
         assertTab('https://www.youtube.com/playlist?list=PLwP_SiAcdui0KVebT0mU9Apz359a4ubsC')

yt_dlp/extractor/youtube.py

@@ -59,9 +59,9 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
     _TFA_URL = 'https://accounts.google.com/_/signin/challenge?hl=en&TL={0}'
 
     _RESERVED_NAMES = (
-        r'embed|e|watch_popup|channel|c|user|playlist|watch|w|v|movies|results|shared|hashtag|'
+        r'channel|c|user|playlist|watch|w|v|embed|e|watch_popup|'
-        r'storefront|oops|index|account|reporthistory|t/terms|about|upload|signin|logout|'
+        r'movies|results|shared|hashtag|trending|feed|feeds|'
-        r'feed/(?:watch_later|history|subscriptions|library|trending|recommended)')
+        r'storefront|oops|index|account|reporthistory|t/terms|about|upload|signin|logout')
 
     _NETRC_MACHINE = 'youtube'
     # If True it will raise an error if no login info is provided
@@ -2520,17 +2520,22 @@ def _extract_channel_id(self, webpage):
             channel_url, 'channel id')
 
     @staticmethod
-    def _extract_grid_item_renderer(item):
+    def _extract_basic_item_renderer(item):
-        for item_kind in ('Playlist', 'Video', 'Channel'):
+        # Modified from _extract_grid_item_renderer
-            renderer = item.get('grid%sRenderer' % item_kind)
+        known_renderers = (
-            if renderer:
+            'playlistRenderer', 'videoRenderer', 'channelRenderer'
+            'gridPlaylistRenderer', 'gridVideoRenderer', 'gridChannelRenderer'
+        )
+        for key, renderer in item.items():
+            if key not in known_renderers:
+                continue
             return renderer
 
     def _grid_entries(self, grid_renderer):
         for item in grid_renderer['items']:
             if not isinstance(item, dict):
                 continue
-            renderer = self._extract_grid_item_renderer(item)
+            renderer = self._extract_basic_item_renderer(item)
             if not isinstance(renderer, dict):
                 continue
             title = try_get(
@@ -2559,7 +2564,7 @@ def _shelf_entries_from_content(self, shelf_renderer):
         content = shelf_renderer.get('content')
         if not isinstance(content, dict):
             return
-        renderer = content.get('gridRenderer')
+        renderer = content.get('gridRenderer') or content.get('expandedShelfContentsRenderer')
         if renderer:
             # TODO: add support for nested playlists so each shelf is processed
             # as separate playlist
@@ -2601,20 +2606,6 @@ def _playlist_entries(self, video_list_renderer):
                 continue
             yield self._extract_video(renderer)
 
-    r""" # Not needed in the new implementation
-    def _itemSection_entries(self, item_sect_renderer):
-        for content in item_sect_renderer['contents']:
-            if not isinstance(content, dict):
-                continue
-            renderer = content.get('videoRenderer', {})
-            if not isinstance(renderer, dict):
-                continue
-            video_id = renderer.get('videoId')
-            if not video_id:
-                continue
-            yield self._extract_video(renderer)
-    """
-
     def _rich_entries(self, rich_grid_renderer):
         renderer = try_get(
             rich_grid_renderer, lambda x: x['content']['videoRenderer'], dict) or {}
@@ -2837,7 +2828,7 @@ def extract_entries(parent_renderer):  # this needs to called again for continua
                 'gridPlaylistRenderer': (self._grid_entries, 'items'),
                 'gridVideoRenderer': (self._grid_entries, 'items'),
                 'playlistVideoRenderer': (self._playlist_entries, 'contents'),
-                'itemSectionRenderer': (self._playlist_entries, 'contents'),
+                'itemSectionRenderer': (extract_entries, 'contents'),  # for feeds
                 'richItemRenderer': (extract_entries, 'contents'),  # for hashtag
             }
             continuation_items = try_get(
@@ -2955,13 +2946,30 @@ def _extract_from_tabs(self, item_id, webpage, data, tabs, identity_token):
             self._entries(selected_tab, identity_token, playlist_id),
             **metadata)
 
+    def _extract_mix_playlist(self, playlist, playlist_id):
+        page_num = 0
+        while True:
+            videos = list(self._playlist_entries(playlist))
+            if not videos:
+                return
+            video_count = len(videos)
+            start = min(video_count - 24, 26) if video_count > 25 else 0
+            for item in videos[start:]:
+                yield item
+
+            page_num += 1
+            _, data = self._extract_webpage(
+                'https://www.youtube.com/watch?list=%s&v=%s' % (playlist_id, videos[-1]['id']),
+                '%s page %d' % (playlist_id, page_num))
+            playlist = try_get(
+                data, lambda x: x['contents']['twoColumnWatchNextResults']['playlist']['playlist'], dict)
+
     def _extract_from_playlist(self, item_id, url, data, playlist):
         title = playlist.get('title') or try_get(
             data, lambda x: x['titleText']['simpleText'], compat_str)
         playlist_id = playlist.get('playlistId') or item_id
-        # Inline playlist rendition continuation does not always work
+
-        # at Youtube side, so delegating regular tab-based playlist URL
+        # Delegating everything except mix playlists to regular tab-based playlist URL
-        # processing whenever possible.
         playlist_url = urljoin(url, try_get(
             playlist, lambda x: x['endpoint']['commandMetadata']['webCommandMetadata']['url'],
             compat_str))
@@ -2969,9 +2977,10 @@ def _extract_from_playlist(self, item_id, url, data, playlist):
             return self.url_result(
                 playlist_url, ie=YoutubeTabIE.ie_key(), video_id=playlist_id,
                 video_title=title)
+
         return self.playlist_result(
-            self._playlist_entries(playlist), playlist_id=playlist_id,
+            self._extract_mix_playlist(playlist, playlist_id),
-            playlist_title=title)
+            playlist_id=playlist_id, playlist_title=title)
 
     @staticmethod
     def _extract_alerts(data):
@@ -3001,35 +3010,7 @@ def _extract_identity_token(self, webpage, item_id):
             r'\bID_TOKEN["\']\s*:\s*["\'](.+?)["\']', webpage,
             'identity token', default=None)
 
-    def _real_extract(self, url):
+    def _extract_webpage(self, url, item_id):
-        item_id = self._match_id(url)
-        url = compat_urlparse.urlunparse(
-            compat_urlparse.urlparse(url)._replace(netloc='www.youtube.com'))
-        is_home = re.match(r'(?P<pre>%s)(?P<post>/?(?![^#?]).*$)' % self._VALID_URL, url)
-        if is_home is not None and is_home.group('not_channel') is None and item_id != 'feed':
-            self._downloader.report_warning(
-                'A channel/user page was given. All the channel\'s videos will be downloaded. '
-                'To download only the videos in the home page, add a "/featured" to the URL')
-            url = '%s/videos%s' % (is_home.group('pre'), is_home.group('post') or '')
-
-        # Handle both video/playlist URLs
-        qs = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query)
-        video_id = qs.get('v', [None])[0]
-        playlist_id = qs.get('list', [None])[0]
-
-        if is_home is not None and is_home.group('not_channel') is not None and is_home.group('not_channel').startswith('watch') and not video_id:
-            if playlist_id:
-                self._downloader.report_warning('%s is not a valid Youtube URL. Trying to download playlist %s' % (url, playlist_id))
-                url = 'https://www.youtube.com/playlist?list=%s' % playlist_id
-                # return self.url_result(playlist_id, ie=YoutubePlaylistIE.ie_key())
-            else:
-                raise ExtractorError('Unable to recognize tab page')
-        if video_id and playlist_id:
-            if self._downloader.params.get('noplaylist'):
-                self.to_screen('Downloading just video %s because of --no-playlist' % video_id)
-                return self.url_result(video_id, ie=YoutubeIE.ie_key(), video_id=video_id)
-            self.to_screen('Downloading playlist %s - add --no-playlist to just download video %s' % (playlist_id, video_id))
-
         retries = self._downloader.params.get('extractor_retries', 3)
         count = -1
         last_error = 'Incomplete yt initial data recieved'
@@ -3041,8 +3022,7 @@ def _real_extract(self, url):
                 self.report_warning('%s. Retrying ...' % last_error)
             webpage = self._download_webpage(
                 url, item_id,
-                'Downloading webpage%s' % ' (retry #%d)' % count if count else '')
+                'Downloading webpage%s' % (' (retry #%d)' % count if count else ''))
-            identity_token = self._extract_identity_token(webpage, item_id)
             data = self._extract_yt_initial_data(item_id, webpage)
             err_msg = None
             for alert_type, alert_message in self._extract_alerts(data):
@@ -3058,23 +3038,61 @@ def _real_extract(self, url):
                 break
             if count >= retries:
                 self._downloader.report_error(last_error)
+        return webpage, data
+
+    def _real_extract(self, url):
+        item_id = self._match_id(url)
+        url = compat_urlparse.urlunparse(
+            compat_urlparse.urlparse(url)._replace(netloc='www.youtube.com'))
+
+        # This is not matched in a channel page with a tab selected
+        mobj = re.match(r'(?P<pre>%s)(?P<post>/?(?![^#?]).*$)' % self._VALID_URL, url)
+        mobj = mobj.groupdict() if mobj else {}
+        if mobj and not mobj.get('not_channel'):
+            self._downloader.report_warning(
+                'A channel/user page was given. All the channel\'s videos will be downloaded. '
+                'To download only the videos in the home page, add a "/featured" to the URL')
+            url = '%s/videos%s' % (mobj.get('pre'), mobj.get('post') or '')
+
+        # Handle both video/playlist URLs
+        qs = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query)
+        video_id = qs.get('v', [None])[0]
+        playlist_id = qs.get('list', [None])[0]
+
+        if not video_id and (mobj.get('not_channel') or '').startswith('watch'):
+            if not playlist_id:
+                # If there is neither video or playlist ids,
+                # youtube redirects to home page, which is undesirable
+                raise ExtractorError('Unable to recognize tab page')
+            self._downloader.report_warning('A video URL was given without video ID. Trying to download playlist %s' % playlist_id)
+            url = 'https://www.youtube.com/playlist?list=%s' % playlist_id
+
+        if video_id and playlist_id:
+            if self._downloader.params.get('noplaylist'):
+                self.to_screen('Downloading just video %s because of --no-playlist' % video_id)
+                return self.url_result(video_id, ie=YoutubeIE.ie_key(), video_id=video_id)
+            self.to_screen('Downloading playlist %s; add --no-playlist to just download video %s' % (playlist_id, video_id))
+
+        webpage, data = self._extract_webpage(url, item_id)
 
         tabs = try_get(
             data, lambda x: x['contents']['twoColumnBrowseResultsRenderer']['tabs'], list)
         if tabs:
+            identity_token = self._extract_identity_token(webpage, item_id)
             return self._extract_from_tabs(item_id, webpage, data, tabs, identity_token)
+
         playlist = try_get(
             data, lambda x: x['contents']['twoColumnWatchNextResults']['playlist']['playlist'], dict)
         if playlist:
             return self._extract_from_playlist(item_id, url, data, playlist)
-        # Fallback to video extraction if no playlist alike page is recognized.
+
-        # First check for the current video then try the v attribute of URL query.
         video_id = try_get(
             data, lambda x: x['currentVideoEndpoint']['watchEndpoint']['videoId'],
             compat_str) or video_id
         if video_id:
+            self._downloader.report_warning('Unable to recognize playlist. Downloading just video %s' % video_id)
             return self.url_result(video_id, ie=YoutubeIE.ie_key(), video_id=video_id)
-        # Failed to recognize
+
         raise ExtractorError('Unable to recognize tab page')
 
 
@@ -3338,7 +3356,6 @@ class YoutubeFeedsInfoExtractor(YoutubeTabIE):
     Subclasses must define the _FEED_NAME property.
     """
     _LOGIN_REQUIRED = True
-    # _MAX_PAGES = 5
     _TESTS = []
 
     @property