diff options
Diffstat (limited to 'yt_dlp/extractor/youtube.py')
| -rw-r--r-- | yt_dlp/extractor/youtube.py | 42 |
1 files changed, 33 insertions, 9 deletions
diff --git a/yt_dlp/extractor/youtube.py b/yt_dlp/extractor/youtube.py index 19b4985f6..017554c88 100644 --- a/yt_dlp/extractor/youtube.py +++ b/yt_dlp/extractor/youtube.py @@ -837,17 +837,20 @@ class YoutubeBaseInfoExtractor(InfoExtractor): uploader = self._get_text(renderer, 'ownerText', 'shortBylineText') channel_id = traverse_obj( - renderer, ('shortBylineText', 'runs', ..., 'navigationEndpoint', 'browseEndpoint', 'browseId'), expected_type=str, get_all=False) + renderer, ('shortBylineText', 'runs', ..., 'navigationEndpoint', 'browseEndpoint', 'browseId'), + expected_type=str, get_all=False) timestamp, time_text = self._extract_time_text(renderer, 'publishedTimeText') scheduled_timestamp = str_to_int(traverse_obj(renderer, ('upcomingEventData', 'startTime'), get_all=False)) overlay_style = traverse_obj( - renderer, ('thumbnailOverlays', ..., 'thumbnailOverlayTimeStatusRenderer', 'style'), get_all=False, expected_type=str) + renderer, ('thumbnailOverlays', ..., 'thumbnailOverlayTimeStatusRenderer', 'style'), + get_all=False, expected_type=str) badges = self._extract_badges(renderer) thumbnails = self._extract_thumbnails(renderer, 'thumbnail') navigation_url = urljoin('https://www.youtube.com/', traverse_obj( - renderer, ('navigationEndpoint', 'commandMetadata', 'webCommandMetadata', 'url'), expected_type=str)) + renderer, ('navigationEndpoint', 'commandMetadata', 'webCommandMetadata', 'url'), + expected_type=str)) or '' url = f'https://www.youtube.com/watch?v={video_id}' - if overlay_style == 'SHORTS' or (navigation_url and '/shorts/' in navigation_url): + if overlay_style == 'SHORTS' or '/shorts/' in navigation_url: url = f'https://www.youtube.com/shorts/{video_id}' return { @@ -862,7 +865,9 @@ class YoutubeBaseInfoExtractor(InfoExtractor): 'uploader': uploader, 'channel_id': channel_id, 'thumbnails': thumbnails, - 'upload_date': strftime_or_none(timestamp, '%Y%m%d') if self._configuration_arg('approximate_date', ie_key='youtubetab') else None, + 'upload_date': (strftime_or_none(timestamp, '%Y%m%d') + if self._configuration_arg('approximate_date', ie_key='youtubetab') + else None), 'live_status': ('is_upcoming' if scheduled_timestamp is not None else 'was_live' if 'streamed' in time_text.lower() else 'is_live' if overlay_style is not None and overlay_style == 'LIVE' or 'live now' in badges @@ -3777,7 +3782,7 @@ class YoutubeTabBaseInfoExtractor(YoutubeBaseInfoExtractor): def _extract_basic_item_renderer(item): # Modified from _extract_grid_item_renderer known_basic_renderers = ( - 'playlistRenderer', 'videoRenderer', 'channelRenderer', 'showRenderer' + 'playlistRenderer', 'videoRenderer', 'channelRenderer', 'showRenderer', 'reelItemRenderer' ) for key, renderer in item.items(): if not isinstance(renderer, dict): @@ -3903,6 +3908,13 @@ class YoutubeTabBaseInfoExtractor(YoutubeBaseInfoExtractor): if video_id: return self._extract_video(video_renderer) + def _hashtag_tile_entry(self, hashtag_tile_renderer): + url = urljoin('https://youtube.com', traverse_obj( + hashtag_tile_renderer, ('onTapCommand', 'commandMetadata', 'webCommandMetadata', 'url'))) + if url: + return self.url_result( + url, ie=YoutubeTabIE.ie_key(), title=self._get_text(hashtag_tile_renderer, 'hashtag')) + def _post_thread_entries(self, post_thread_renderer): post_renderer = try_get( post_thread_renderer, lambda x: x['post']['backstagePostRenderer'], dict) @@ -3985,12 +3997,14 @@ class YoutubeTabBaseInfoExtractor(YoutubeBaseInfoExtractor): known_renderers = { 'playlistVideoListRenderer': self._playlist_entries, 'gridRenderer': self._grid_entries, - 'shelfRenderer': lambda x: self._shelf_entries(x), + 'reelShelfRenderer': self._grid_entries, + 'shelfRenderer': self._shelf_entries, 'musicResponsiveListItemRenderer': lambda x: [self._music_reponsive_list_entry(x)], 'backstagePostThreadRenderer': self._post_thread_entries, 'videoRenderer': lambda x: [self._video_entry(x)], 'playlistRenderer': lambda x: self._grid_entries({'items': [{'playlistRenderer': x}]}), 'channelRenderer': lambda x: self._grid_entries({'items': [{'channelRenderer': x}]}), + 'hashtagTileRenderer': lambda x: [self._hashtag_tile_entry(x)] } for key, renderer in isr_content.items(): if key not in known_renderers: @@ -4162,7 +4176,7 @@ class YoutubeTabBaseInfoExtractor(YoutubeBaseInfoExtractor): }) primary_thumbnails = self._extract_thumbnails( - primary_sidebar_renderer, ('thumbnailRenderer', 'playlistVideoThumbnailRenderer', 'thumbnail')) + primary_sidebar_renderer, ('thumbnailRenderer', ('playlistVideoThumbnailRenderer', 'playlistCustomThumbnailRenderer'), 'thumbnail')) if playlist_id is None: playlist_id = item_id @@ -5520,7 +5534,17 @@ class YoutubeSearchURLIE(YoutubeTabBaseInfoExtractor): 'id': 'python', 'title': 'python', } - + }, { + 'url': 'https://www.youtube.com/results?search_query=%23cats', + 'playlist_mincount': 1, + 'info_dict': { + 'id': '#cats', + 'title': '#cats', + 'entries': [{ + 'url': r're:https://(www\.)?youtube\.com/hashtag/cats', + 'title': '#cats', + }], + }, }, { 'url': 'https://www.youtube.com/results?q=test&sp=EgQIBBgB', 'only_matching': True, |