]> git.bitcoin.ninja Git - youtube-dl/blob - youtube_dl/extractor/businessinsider.py
[extractor/common] Relax interaction count extraction in _json_ld
[youtube-dl] / youtube_dl / extractor / businessinsider.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from .jwplatform import JWPlatformIE
6
7
8 class BusinessInsiderIE(InfoExtractor):
9     _VALID_URL = r'https?://(?:[^/]+\.)?businessinsider\.(?:com|nl)/(?:[^/]+/)*(?P<id>[^/?#&]+)'
10     _TESTS = [{
11         'url': 'http://uk.businessinsider.com/how-much-radiation-youre-exposed-to-in-everyday-life-2016-6',
12         'md5': 'ffed3e1e12a6f950aa2f7d83851b497a',
13         'info_dict': {
14             'id': 'cjGDb0X9',
15             'ext': 'mp4',
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,
20         },
21     }, {
22         'url': 'https://www.businessinsider.nl/5-scientifically-proven-things-make-you-less-attractive-2017-7/',
23         'md5': '43f438dbc6da0b89f5ac42f68529d84a',
24         'info_dict': {
25             'id': '5zJwd4FK',
26             'ext': 'mp4',
27             'title': 'Deze dingen zorgen ervoor dat je minder snel een date scoort',
28             'description': 'md5:2af8975825d38a4fed24717bbe51db49',
29             'upload_date': '20170705',
30             'timestamp': 1499270528,
31         },
32     }, {
33         'url': 'http://www.businessinsider.com/excel-index-match-vlookup-video-how-to-2015-2?IR=T',
34         'only_matching': True,
35     }]
36
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(),
48             video_id=video_id)