2 from __future__ import unicode_literals
4 from .common import InfoExtractor
5 from .jwplatform import JWPlatformIE
8 class BusinessInsiderIE(InfoExtractor):
9 _VALID_URL = r'https?://(?:[^/]+\.)?businessinsider\.(?:com|nl)/(?:[^/]+/)*(?P<id>[^/?#&]+)'
11 'url': 'http://uk.businessinsider.com/how-much-radiation-youre-exposed-to-in-everyday-life-2016-6',
12 'md5': 'ffed3e1e12a6f950aa2f7d83851b497a',
16 'title': "Bananas give you more radiation exposure than living next to a nuclear power plant",
17 'description': 'md5:0175a3baf200dd8fa658f94cade841b3',
18 'upload_date': '20160611',
19 'timestamp': 1465675620,
22 'url': 'https://www.businessinsider.nl/5-scientifically-proven-things-make-you-less-attractive-2017-7/',
23 'md5': '43f438dbc6da0b89f5ac42f68529d84a',
27 'title': 'Deze dingen zorgen ervoor dat je minder snel een date scoort',
28 'description': 'md5:2af8975825d38a4fed24717bbe51db49',
29 'upload_date': '20170705',
30 'timestamp': 1499270528,
33 'url': 'http://www.businessinsider.com/excel-index-match-vlookup-video-how-to-2015-2?IR=T',
34 'only_matching': True,
37 def _real_extract(self, url):
38 video_id = self._match_id(url)
39 webpage = self._download_webpage(url, video_id)
40 jwplatform_id = self._search_regex(
41 (r'data-media-id=["\']([a-zA-Z0-9]{8})',
42 r'id=["\']jwplayer_([a-zA-Z0-9]{8})',
43 r'id["\']?\s*:\s*["\']?([a-zA-Z0-9]{8})',
44 r'(?:jwplatform\.com/players/|jwplayer_)([a-zA-Z0-9]{8})'),
45 webpage, 'jwplatform id')
46 return self.url_result(
47 'jwplatform:%s' % jwplatform_id, ie=JWPlatformIE.ie_key(),