[vimeo] fix unlisted video extraction(closes #28414)
parent
605e7b5e47
commit
c785911870
|
@ -24,6 +24,7 @@ from ..utils import (
|
|||
merge_dicts,
|
||||
OnDemandPagedList,
|
||||
parse_filesize,
|
||||
parse_iso8601,
|
||||
RegexNotFoundError,
|
||||
sanitized_Request,
|
||||
smuggle_url,
|
||||
|
@ -278,7 +279,7 @@ class VimeoIE(VimeoBaseInfoExtractor):
|
|||
)?
|
||||
(?:videos?/)?
|
||||
(?P<id>[0-9]+)
|
||||
(?:/[\da-f]+)?
|
||||
(?:/(?P<unlisted_hash>[\da-f]{10}))?
|
||||
/?(?:[?&].*)?(?:[#].*)?$
|
||||
'''
|
||||
IE_NAME = 'vimeo'
|
||||
|
@ -577,11 +578,37 @@ class VimeoIE(VimeoBaseInfoExtractor):
|
|||
if 'Referer' not in headers:
|
||||
headers['Referer'] = url
|
||||
|
||||
channel_id = self._search_regex(
|
||||
r'vimeo\.com/channels/([^/]+)', url, 'channel id', default=None)
|
||||
|
||||
# Extract ID from URL
|
||||
video_id = self._match_id(url)
|
||||
video_id, unlisted_hash = re.match(self._VALID_URL, url).groups()
|
||||
if unlisted_hash:
|
||||
token = self._download_json(
|
||||
'https://vimeo.com/_rv/jwt', video_id, headers={
|
||||
'X-Requested-With': 'XMLHttpRequest'
|
||||
})['token']
|
||||
video = self._download_json(
|
||||
'https://api.vimeo.com/videos/%s:%s' % (video_id, unlisted_hash),
|
||||
video_id, headers={
|
||||
'Authorization': 'jwt ' + token,
|
||||
}, query={
|
||||
'fields': 'config_url,created_time,description,license,metadata.connections.comments.total,metadata.connections.likes.total,release_time,stats.plays',
|
||||
})
|
||||
info = self._parse_config(self._download_json(
|
||||
video['config_url'], video_id), video_id)
|
||||
self._vimeo_sort_formats(info['formats'])
|
||||
get_timestamp = lambda x: parse_iso8601(video.get(x + '_time'))
|
||||
info.update({
|
||||
'description': video.get('description'),
|
||||
'license': video.get('license'),
|
||||
'release_timestamp': get_timestamp('release'),
|
||||
'timestamp': get_timestamp('created'),
|
||||
'view_count': int_or_none(try_get(video, lambda x: x['stats']['plays'])),
|
||||
})
|
||||
connections = try_get(
|
||||
video, lambda x: x['metadata']['connections'], dict) or {}
|
||||
for k in ('comment', 'like'):
|
||||
info[k + '_count'] = int_or_none(try_get(connections, lambda x: x[k + 's']['total']))
|
||||
return info
|
||||
|
||||
orig_url = url
|
||||
is_pro = 'vimeopro.com/' in url
|
||||
is_player = '://player.vimeo.com/video/' in url
|
||||
|
@ -756,6 +783,8 @@ class VimeoIE(VimeoBaseInfoExtractor):
|
|||
r'<link[^>]+rel=["\']license["\'][^>]+href=(["\'])(?P<license>(?:(?!\1).)+)\1',
|
||||
webpage, 'license', default=None, group='license')
|
||||
|
||||
channel_id = self._search_regex(
|
||||
r'vimeo\.com/channels/([^/]+)', url, 'channel id', default=None)
|
||||
channel_url = 'https://vimeo.com/channels/%s' % channel_id if channel_id else None
|
||||
|
||||
info_dict = {
|
||||
|
|
Loading…
Reference in New Issue