_ Git - youtube-dl/blob - youtube_dl/extractor/cbs.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4
   5
   6 class CBSIE(InfoExtractor):
   7     _VALID_URL = r'https?://(?:www\.)?(?:cbs\.com/shows/[^/]+/(?:video|artist)|colbertlateshow\.com/(?:video|podcasts))/[^/]+/(?P<id>[^/]+)'
   8
   9     _TESTS = [{
  10         'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/',
  11         'info_dict': {
  12             'id': '4JUVEwq3wUT7',
  13             'display_id': 'connect-chat-feat-garth-brooks',
  14             'ext': 'flv',
  15             'title': 'Connect Chat feat. Garth Brooks',
  16             'description': 'Connect with country music singer Garth Brooks, as he chats with fans on Wednesday November 27, 2013. Be sure to tune in to Garth Brooks: Live from Las Vegas, Friday November 29, at 9/8c on CBS!',
  17             'duration': 1495,
  18         },
  19         'params': {
  20             # rtmp download
  21             'skip_download': True,
  22         },
  23         '_skip': 'Blocked outside the US',
  24     }, {
  25         'url': 'http://www.cbs.com/shows/liveonletterman/artist/221752/st-vincent/',
  26         'info_dict': {
  27             'id': 'WWF_5KqY3PK1',
  28             'display_id': 'st-vincent',
  29             'ext': 'flv',
  30             'title': 'Live on Letterman - St. Vincent',
  31             'description': 'Live On Letterman: St. Vincent in concert from New York\'s Ed Sullivan Theater on Tuesday, July 16, 2014.',
  32             'duration': 3221,
  33         },
  34         'params': {
  35             # rtmp download
  36             'skip_download': True,
  37         },
  38         '_skip': 'Blocked outside the US',
  39     }, {
  40         'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/',
  41         'only_matching': True,
  42     }, {
  43         'url': 'http://www.colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/',
  44         'only_matching': True,
  45     }]
  46
  47     def _real_extract(self, url):
  48         display_id = self._match_id(url)
  49         webpage = self._download_webpage(url, display_id)
  50         real_id = self._search_regex(
  51             [r"video\.settings\.pid\s*=\s*'([^']+)';", r"cbsplayer\.pid\s*=\s*'([^']+)';"],
  52             webpage, 'real video ID')
  53         return {
  54             '_type': 'url_transparent',
  55             'ie_key': 'ThePlatform',
  56             'url': 'theplatform:%s' % real_id,
  57             'display_id': display_id,
  58         }