midgard/youtube-dl
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

[voxmedia] fix volume embed extraction(closes #28338)

Browse source
This commit is contained in:
Remita Amine 2021-03-04 12:41:49 +01:00
parent f68692b004
commit ec64ec9651
1 changed files with 18 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

26
youtube_dl/extractor/voxmedia.py
View file

@ -7,6 +7,8 @@ from ..compat import compat_urllib_parse_unquote
from ..utils import ( from ..utils import (
ExtractorError, ExtractorError,
int_or_none, int_or_none,
try_get,
unified_timestamp,
) )
@ -19,14 +21,17 @@ class VoxMediaVolumeIE(OnceIE):
setup = self._parse_json(self._search_regex( setup = self._parse_json(self._search_regex(
r'setup\s*=\s*({.+});', webpage, 'setup'), video_id) r'setup\s*=\s*({.+});', webpage, 'setup'), video_id)
video_data = setup.get('video') or {} player_setup = setup.get('player_setup') or setup
video_data = player_setup.get('video') or {}
formatted_metadata = video_data.get('formatted_metadata') or {}
info = { info = {
'id': video_id, 'id': video_id,
'title': video_data.get('title_short'), 'title': player_setup.get('title') or video_data.get('title_short'),
'description': video_data.get('description_long') or video_data.get('description_short'), 'description': video_data.get('description_long') or video_data.get('description_short'),
'thumbnail': video_data.get('brightcove_thumbnail') 'thumbnail': formatted_metadata.get('thumbnail') or video_data.get('brightcove_thumbnail'),
'timestamp': unified_timestamp(formatted_metadata.get('video_publish_date')),
} }
asset = setup.get('asset') or setup.get('params') or {} asset = try_get(setup, lambda x: x['embed_assets']['chorus'], dict) or {}
formats = [] formats = []
hls_url = asset.get('hls_url') hls_url = asset.get('hls_url')
@ -47,6 +52,7 @@ class VoxMediaVolumeIE(OnceIE):
if formats: if formats:
self._sort_formats(formats) self._sort_formats(formats)
info['formats'] = formats info['formats'] = formats
info['duration'] = int_or_none(asset.get('duration'))
return info return info
for provider_video_type in ('ooyala', 'youtube', 'brightcove'): for provider_video_type in ('ooyala', 'youtube', 'brightcove'):
@ -84,7 +90,7 @@ class VoxMediaIE(InfoExtractor):
}, { }, {
# Volume embed, Youtube # Volume embed, Youtube
'url': 'http://www.theverge.com/2014/10/21/7025853/google-nexus-6-hands-on-photos-video-android-phablet', 'url': 'http://www.theverge.com/2014/10/21/7025853/google-nexus-6-hands-on-photos-video-android-phablet',
'md5': '4c8f4a0937752b437c3ebc0ed24802b5', 'md5': 'fd19aa0cf3a0eea515d4fd5c8c0e9d68',
'info_dict': { 'info_dict': {
'id': 'Gy8Md3Eky38', 'id': 'Gy8Md3Eky38',
'ext': 'mp4', 'ext': 'mp4',
@ -93,6 +99,7 @@ class VoxMediaIE(InfoExtractor):
'uploader_id': 'TheVerge', 'uploader_id': 'TheVerge',
'upload_date': '20141021', 'upload_date': '20141021',
'uploader': 'The Verge', 'uploader': 'The Verge',
'timestamp': 1413907200,
}, },
'add_ie': ['Youtube'], 'add_ie': ['Youtube'],
'skip': 'similar to the previous test', 'skip': 'similar to the previous test',
@ -100,13 +107,13 @@ class VoxMediaIE(InfoExtractor):
# Volume embed, Youtube # Volume embed, Youtube
'url': 'http://www.vox.com/2016/3/31/11336640/mississippi-lgbt-religious-freedom-bill', 'url': 'http://www.vox.com/2016/3/31/11336640/mississippi-lgbt-religious-freedom-bill',
'info_dict': { 'info_dict': {
'id': 'YCjDnX-Xzhg', 'id': '22986359b',
'ext': 'mp4', 'ext': 'mp4',
'title': "Mississippi's laws are so bad that its anti-LGBTQ law isn't needed to allow discrimination", 'title': "Mississippi's laws are so bad that its anti-LGBTQ law isn't needed to allow discrimination",
'description': 'md5:fc1317922057de31cd74bce91eb1c66c', 'description': 'md5:fc1317922057de31cd74bce91eb1c66c',
'uploader_id': 'voxdotcom',
'upload_date': '20150915', 'upload_date': '20150915',
'uploader': 'Vox', 'timestamp': 1442332800,
'duration': 285,
}, },
'add_ie': ['Youtube'], 'add_ie': ['Youtube'],
'skip': 'similar to the previous test', 'skip': 'similar to the previous test',
@ -160,6 +167,9 @@ class VoxMediaIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Post-Post-PC CEO: The Full Code Conference Video of Microsoft\'s Satya Nadella', 'title': 'Post-Post-PC CEO: The Full Code Conference Video of Microsoft\'s Satya Nadella',
'description': 'The longtime veteran was chosen earlier this year as the software giant\'s third leader in its history.', 'description': 'The longtime veteran was chosen earlier this year as the software giant\'s third leader in its history.',
'timestamp': 1402938000,
'upload_date': '20140616',
'duration': 4114,
}, },
'add_ie': ['VoxMediaVolume'], 'add_ie': ['VoxMediaVolume'],
}] }]