kWidget\.(?:thumb)?[Ee]mbed\(
\{.*?
(?P<q1>['\"])wid(?P=q1)\s*:\s*
- (?P<q2>['\"])_?(?P<partner_id>[^'\"]+)(?P=q2),.*?
+ (?P<q2>['\"])_?(?P<partner_id>(?:(?!(?P=q2)).)+)(?P=q2),.*?
(?P<q3>['\"])entry_?[Ii]d(?P=q3)\s*:\s*
- (?P<q4>['\"])(?P<id>[^'\"]+)(?P=q4),
+ (?P<q4>['\"])(?P<id>(?:(?!(?P=q4)).)+)(?P=q4)(?:,|\s*\})
""", webpage) or
re.search(
r'''(?xs)
(?P<q1>["\'])
- (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com/.*?(?:p|partner_id)/(?P<partner_id>\d+).*?
+ (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com/(?:(?!(?P=q1)).)*(?:p|partner_id)/(?P<partner_id>\d+)(?:(?!(?P=q1)).)*
(?P=q1).*?
(?:
entry_?[Ii]d|
(?P<q2>["\'])entry_?[Ii]d(?P=q2)
)\s*:\s*
- (?P<q3>["\'])(?P<id>.+?)(?P=q3)
+ (?P<q3>["\'])(?P<id>(?:(?!(?P=q3)).)+)(?P=q3)
''', webpage))
if mobj:
embed_info = mobj.groupdict()
# skip for now.
if f.get('fileExt') == 'chun':
continue
+ if not f.get('fileExt'):
+ # QT indicates QuickTime; some videos have broken fileExt
+ if f.get('containerFormat') == 'qt':
+ f['fileExt'] = 'mov'
+ else:
+ f['fileExt'] = 'mp4'
video_url = sign_url(
'%s/flavorId/%s' % (data_url, f['id']))
# audio-only has no videoCodecId (e.g. kaltura:1926081:0_c03e1b5g
'thumbnail': info.get('thumbnailUrl'),
'duration': info.get('duration'),
'timestamp': info.get('createdAt'),
- 'uploader_id': info.get('userId'),
+ 'uploader_id': info.get('userId') if info.get('userId') != 'None' else None,
'view_count': info.get('plays'),
}