Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/cbs.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     sanitized_Request,
   6     smuggle_url,
   7 )
   8
   9
  10 class CBSIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?(?:cbs\.com/shows/[^/]+/(?:video|artist)|colbertlateshow\.com/(?:video|podcasts))/[^/]+/(?P<id>[^/]+)'
  12
  13     _TESTS = [{
  14         'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/',
  15         'info_dict': {
  16             'id': '4JUVEwq3wUT7',
  17             'display_id': 'connect-chat-feat-garth-brooks',
  18             'ext': 'flv',
  19             'title': 'Connect Chat feat. Garth Brooks',
  20             '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!',
  21             'duration': 1495,
  22         },
  23         'params': {
  24             # rtmp download
  25             'skip_download': True,
  26         },
  27         '_skip': 'Blocked outside the US',
  28     }, {
  29         'url': 'http://www.cbs.com/shows/liveonletterman/artist/221752/st-vincent/',
  30         'info_dict': {
  31             'id': 'WWF_5KqY3PK1',
  32             'display_id': 'st-vincent',
  33             'ext': 'flv',
  34             'title': 'Live on Letterman - St. Vincent',
  35             'description': 'Live On Letterman: St. Vincent in concert from New York\'s Ed Sullivan Theater on Tuesday, July 16, 2014.',
  36             'duration': 3221,
  37         },
  38         'params': {
  39             # rtmp download
  40             'skip_download': True,
  41         },
  42         '_skip': 'Blocked outside the US',
  43     }, {
  44         'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/',
  45         'only_matching': True,
  46     }, {
  47         'url': 'http://www.colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/',
  48         'only_matching': True,
  49     }]
  50
  51     def _real_extract(self, url):
  52         display_id = self._match_id(url)
  53         request = sanitized_Request(url)
  54         # Android UA is served with higher quality (720p) streams (see
  55         # https://github.com/rg3/youtube-dl/issues/7490)
  56         request.add_header('User-Agent', 'Mozilla/5.0 (Linux; Android 4.4; Nexus 5)')
  57         webpage = self._download_webpage(request, display_id)
  58         real_id = self._search_regex(
  59             [r"video\.settings\.pid\s*=\s*'([^']+)';", r"cbsplayer\.pid\s*=\s*'([^']+)';"],
  60             webpage, 'real video ID')
  61         return {
  62             '_type': 'url_transparent',
  63             'ie_key': 'ThePlatform',
  64             'url': smuggle_url(
  65                 'http://link.theplatform.com/s/dJ5BDC/%s?mbr=true&manifest=m3u' % real_id,
  66                 {'force_smil_url': True}),
  67             'display_id': display_id,
  68         }