diff options
| author | dirkf <fieldhouse@gmx.net> | 2025-09-28 07:09:31 +0100 |
|---|---|---|
| committer | dirkf <fieldhouse@gmx.net> | 2025-09-29 02:15:05 +0100 |
| commit | 4222c6d78b63440849ad6a886a2ff5b607f3feae (patch) | |
| tree | 3fcc54e2aabf37b9ecfbe3987a35e48971bda458 | |
| parent | 2735d1bf1d1f947891776330cb58d792d03cf436 (diff) | |
[YouTube] Extract fallback title and description from initial data
Based on yt-dlp/yt-dlp#14078, thx bashonly
| -rw-r--r-- | youtube_dl/extractor/youtube.py | 31 |
1 files changed, 25 insertions, 6 deletions
diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index d32c9df99..dea109eae 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -2849,6 +2849,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): initial_data = self._call_api( 'next', {'videoId': video_id}, video_id, fatal=False) + initial_sdcr = None if initial_data: chapters = self._extract_chapters_from_json( initial_data, video_id, duration) @@ -2976,12 +2977,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor): info['track'] = mrr_contents_text # this is not extraction but spelunking! - carousel_lockups = traverse_obj( - initial_data, - ('engagementPanels', Ellipsis, 'engagementPanelSectionListRenderer', - 'content', 'structuredDescriptionContentRenderer', 'items', Ellipsis, - 'videoDescriptionMusicSectionRenderer', 'carouselLockups', Ellipsis), - expected_type=dict) or [] + initial_sdcr = traverse_obj(initial_data, ( + 'engagementPanels', Ellipsis, 'engagementPanelSectionListRenderer', + 'content', 'structuredDescriptionContentRenderer', T(dict)), + get_all=False) + carousel_lockups = traverse_obj(initial_sdcr, ( + 'items', Ellipsis, 'videoDescriptionMusicSectionRenderer', + 'carouselLockups', Ellipsis, T(dict))) or [] # try to reproduce logic from metadataRowContainerRenderer above (if it still is) fields = (('ALBUM', 'album'), ('ARTIST', 'artist'), ('SONG', 'track'), ('LICENSES', 'license')) # multiple_songs ? @@ -3006,6 +3008,23 @@ class YoutubeIE(YoutubeBaseInfoExtractor): self.mark_watched(video_id, player_response) + # Fallbacks for missing metadata + if initial_sdcr: + if info.get('description') is None: + info['description'] = traverse_obj(initial_sdcr, ( + 'items', Ellipsis, 'expandableVideoDescriptionBodyRenderer', + 'attributedDescriptionBodyText', 'content', T(compat_str)), + get_all=False) + # videoDescriptionHeaderRenderer also has publishDate/channel/handle/ucid, but not needed + if info.get('title') is None: + info['title'] = traverse_obj( + (initial_sdcr, initial_data), + (0, 'items', Ellipsis, 'videoDescriptionHeaderRenderer', T(dict)), + (1, 'playerOverlays', 'playerOverlayRenderer', 'videoDetails', + 'playerOverlayVideoDetailsRenderer', T(dict)), + expected_type=lambda x: self._get_text(x, 'title'), + get_all=False) + return merge_dicts( info, { 'uploader_id': self._extract_uploader_id(owner_profile_url), |