[inc] Add support for another embed schema (closes #16666)

This commit is contained in:
Sergey M․ 2018-06-09 02:53:04 +07:00
parent 6ae36035d9
commit ff2e486221
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -21,6 +21,21 @@ class IncIE(InfoExtractor):
'params': { 'params': {
'skip_download': True, 'skip_download': True,
}, },
}, {
# div with id=kaltura_player_1_kqs38cgm
'url': 'https://www.inc.com/oscar-raymundo/richard-branson-young-entrepeneurs.html',
'info_dict': {
'id': '1_kqs38cgm',
'ext': 'mp4',
'title': 'Branson: "In the end, you have to say, Screw it. Just do it."',
'description': 'md5:21b832d034f9af5191ca5959da5e9cb6',
'timestamp': 1364403232,
'upload_date': '20130327',
'uploader_id': 'incdigital@inc.com',
},
'params': {
'skip_download': True,
},
}, { }, {
'url': 'http://www.inc.com/video/david-whitford/founders-forum-tripadvisor-steve-kaufer-most-enjoyable-moment-for-entrepreneur.html', 'url': 'http://www.inc.com/video/david-whitford/founders-forum-tripadvisor-steve-kaufer-most-enjoyable-moment-for-entrepreneur.html',
'only_matching': True, 'only_matching': True,
@ -31,9 +46,12 @@ def _real_extract(self, url):
webpage = self._download_webpage(url, display_id) webpage = self._download_webpage(url, display_id)
partner_id = self._search_regex( partner_id = self._search_regex(
r'var\s+_?bizo_data_partner_id\s*=\s*["\'](\d+)', webpage, 'partner id') r'var\s+_?bizo_data_partner_id\s*=\s*["\'](\d+)', webpage,
'partner id', default='1034971')
kaltura_id = self._parse_json(self._search_regex( kaltura_id = self._search_regex(
r'id=(["\'])kaltura_player_(?P<id>.+?)\1', webpage, 'kaltura id',
default=None, group='id') or self._parse_json(self._search_regex(
r'pageInfo\.videos\s*=\s*\[(.+)\];', webpage, 'kaltura id'), r'pageInfo\.videos\s*=\s*\[(.+)\];', webpage, 'kaltura id'),
display_id)['vid_kaltura_id'] display_id)['vid_kaltura_id']