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

   1 from __future__ import unicode_literals
   2
   3 from .theplatform import ThePlatformFeedIE
   4 from ..utils import (
   5     int_or_none,
   6     find_xpath_attr,
   7     ExtractorError,
   8 )
   9
  10
  11 class CBSBaseIE(ThePlatformFeedIE):
  12     def _parse_smil_subtitles(self, smil, namespace=None, subtitles_lang='en'):
  13         closed_caption_e = find_xpath_attr(smil, self._xpath_ns('.//param', namespace), 'name', 'ClosedCaptionURL')
  14         return {
  15             'en': [{
  16                 'ext': 'ttml',
  17                 'url': closed_caption_e.attrib['value'],
  18             }]
  19         } if closed_caption_e is not None and closed_caption_e.attrib.get('value') else []
  20
  21
  22 class CBSIE(CBSBaseIE):
  23     _VALID_URL = r'(?:cbs:|https?://(?:www\.)?(?:cbs\.com/shows/[^/]+/video|colbertlateshow\.com/(?:video|podcasts))/)(?P<id>[\w-]+)'
  24
  25     _TESTS = [{
  26         'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/',
  27         'info_dict': {
  28             'id': '_u7W953k6la293J7EPTd9oHkSPs6Xn6_',
  29             'ext': 'mp4',
  30             'title': 'Connect Chat feat. Garth Brooks',
  31             '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!',
  32             'duration': 1495,
  33             'timestamp': 1385585425,
  34             'upload_date': '20131127',
  35             'uploader': 'CBSI-NEW',
  36         },
  37         'params': {
  38             # m3u8 download
  39             'skip_download': True,
  40         },
  41         '_skip': 'Blocked outside the US',
  42     }, {
  43         'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/',
  44         'only_matching': True,
  45     }, {
  46         'url': 'http://www.colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/',
  47         'only_matching': True,
  48     }]
  49
  50     def _extract_video_info(self, guid):
  51         path = 'dJ5BDC/media/guid/2198311517/' + guid
  52         smil_url = 'http://link.theplatform.com/s/%s?mbr=true' % path
  53         formats, subtitles = self._extract_theplatform_smil(smil_url + '&manifest=m3u', guid)
  54         for r in ('HLS&formats=M3U', 'RTMP', 'WIFI', '3G'):
  55             try:
  56                 tp_formats, _ = self._extract_theplatform_smil(smil_url + '&assetTypes=' + r, guid, 'Downloading %s SMIL data' % r.split('&')[0])
  57                 formats.extend(tp_formats)
  58             except ExtractorError:
  59                 continue
  60         self._sort_formats(formats)
  61         metadata = self._download_theplatform_metadata(path, guid)
  62         info = self._parse_theplatform_metadata(metadata)
  63         info.update({
  64             'id': guid,
  65             'formats': formats,
  66             'subtitles': subtitles,
  67             'series': metadata.get('cbs$SeriesTitle'),
  68             'season_number': int_or_none(metadata.get('cbs$SeasonNumber')),
  69             'episode': metadata.get('cbs$EpisodeTitle'),
  70             'episode_number': int_or_none(metadata.get('cbs$EpisodeNumber')),
  71         })
  72         return info
  73
  74     def _real_extract(self, url):
  75         content_id = self._match_id(url)
  76         return self._extract_video_info(content_id)