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

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_urllib_request
   5 from ..utils import smuggle_url
   6
   7
   8 class CBSIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?(?:cbs\.com/shows/[^/]+/(?:video|artist)|colbertlateshow\.com/(?:video|podcasts))/[^/]+/(?P<id>[^/]+)'
  10
  11     _TESTS = [{
  12         'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/',
  13         'info_dict': {
  14             'id': '4JUVEwq3wUT7',
  15             'display_id': 'connect-chat-feat-garth-brooks',
  16             'ext': 'flv',
  17             'title': 'Connect Chat feat. Garth Brooks',
  18             '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!',
  19             'duration': 1495,
  20         },
  21         'params': {
  22             # rtmp download
  23             'skip_download': True,
  24         },
  25         '_skip': 'Blocked outside the US',
  26     }, {
  27         'url': 'http://www.cbs.com/shows/liveonletterman/artist/221752/st-vincent/',
  28         'info_dict': {
  29             'id': 'WWF_5KqY3PK1',
  30             'display_id': 'st-vincent',
  31             'ext': 'flv',
  32             'title': 'Live on Letterman - St. Vincent',
  33             'description': 'Live On Letterman: St. Vincent in concert from New York\'s Ed Sullivan Theater on Tuesday, July 16, 2014.',
  34             'duration': 3221,
  35         },
  36         'params': {
  37             # rtmp download
  38             'skip_download': True,
  39         },
  40         '_skip': 'Blocked outside the US',
  41     }, {
  42         'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/',
  43         'only_matching': True,
  44     }, {
  45         'url': 'http://www.colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/',
  46         'only_matching': True,
  47     }]
  48
  49     def _real_extract(self, url):
  50         display_id = self._match_id(url)
  51         request = compat_urllib_request.Request(url)
  52         # Android UA is served with higher quality (720p) streams (see
  53         # https://github.com/rg3/youtube-dl/issues/7490)
  54         request.add_header('User-Agent', 'Mozilla/5.0 (Linux; Android 4.4; Nexus 5)')
  55         webpage = self._download_webpage(request, display_id)
  56         real_id = self._search_regex(
  57             [r"video\.settings\.pid\s*=\s*'([^']+)';", r"cbsplayer\.pid\s*=\s*'([^']+)';"],
  58             webpage, 'real video ID')
  59         return {
  60             '_type': 'url_transparent',
  61             'ie_key': 'ThePlatform',
  62             'url': smuggle_url(
  63                 'http://link.theplatform.com/s/dJ5BDC/%s?mbr=true&manifest=m3u' % real_id,
  64                 {'force_smil_url': True}),
  65             'display_id': display_id,
  66         }