pepepper/youtube-dl
1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl synced 2025-10-02 06:18:37 +09:00
Code Issues Projects Releases Wiki Activity

[YouTube] Extract fallback title and description from initial data

Browse Source 
Based on yt-dlp/yt-dlp#14078, thx bashonly
This commit is contained in:
dirkf
2025-09-28 07:09:31 +01:00
parent 2735d1bf1d
commit 4222c6d78b
1 changed files with 25 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
youtube_dl/extractor/youtube.py
View File

@@ -2849,6 +2849,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
initial_data = self._call_api( initial_data = self._call_api(
'next', {'videoId': video_id}, video_id, fatal=False) 'next', {'videoId': video_id}, video_id, fatal=False)
initial_sdcr = None
if initial_data: if initial_data:
chapters = self._extract_chapters_from_json( chapters = self._extract_chapters_from_json(
initial_data, video_id, duration) initial_data, video_id, duration)
@@ -2976,12 +2977,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
info['track'] = mrr_contents_text info['track'] = mrr_contents_text
# this is not extraction but spelunking! # this is not extraction but spelunking!
carousel_lockups = traverse_obj( initial_sdcr = traverse_obj(initial_data, (
initial_data, 'engagementPanels', Ellipsis, 'engagementPanelSectionListRenderer',
('engagementPanels', Ellipsis, 'engagementPanelSectionListRenderer', 'content', 'structuredDescriptionContentRenderer', T(dict)),
'content', 'structuredDescriptionContentRenderer', 'items', Ellipsis, get_all=False)
'videoDescriptionMusicSectionRenderer', 'carouselLockups', Ellipsis), carousel_lockups = traverse_obj(initial_sdcr, (
expected_type=dict) or [] 'items', Ellipsis, 'videoDescriptionMusicSectionRenderer',
'carouselLockups', Ellipsis, T(dict))) or []
# try to reproduce logic from metadataRowContainerRenderer above (if it still is) # try to reproduce logic from metadataRowContainerRenderer above (if it still is)
fields = (('ALBUM', 'album'), ('ARTIST', 'artist'), ('SONG', 'track'), ('LICENSES', 'license')) fields = (('ALBUM', 'album'), ('ARTIST', 'artist'), ('SONG', 'track'), ('LICENSES', 'license'))
# multiple_songs ? # multiple_songs ?
@@ -3006,6 +3008,23 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
self.mark_watched(video_id, player_response) 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( return merge_dicts(
info, { info, {
'uploader_id': self._extract_uploader_id(owner_profile_url), 'uploader_id': self._extract_uploader_id(owner_profile_url),