1 from __future__ import unicode_literals
3 from .theplatform import ThePlatformIE
13 class CBSBaseIE(ThePlatformIE):
14 def _parse_smil_subtitles(self, smil, namespace=None, subtitles_lang='en'):
15 closed_caption_e = find_xpath_attr(smil, self._xpath_ns('.//param', namespace), 'name', 'ClosedCaptionURL')
19 'url': closed_caption_e.attrib['value'],
21 } if closed_caption_e is not None and closed_caption_e.attrib.get('value') else []
24 class CBSIE(CBSBaseIE):
25 _VALID_URL = r'https?://(?:www\.)?(?:cbs\.com/shows/[^/]+/(?:video|artist)|colbertlateshow\.com/(?:video|podcasts))/[^/]+/(?P<id>[^/]+)'
28 'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/',
30 'id': '_u7W953k6la293J7EPTd9oHkSPs6Xn6_',
31 'display_id': 'connect-chat-feat-garth-brooks',
33 'title': 'Connect Chat feat. Garth Brooks',
34 '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!',
36 'timestamp': 1385585425,
37 'upload_date': '20131127',
38 'uploader': 'CBSI-NEW',
42 'skip_download': True,
44 '_skip': 'Blocked outside the US',
46 'url': 'http://www.cbs.com/shows/liveonletterman/artist/221752/st-vincent/',
49 'display_id': 'st-vincent',
51 'title': 'Live on Letterman - St. Vincent',
52 'description': 'Live On Letterman: St. Vincent in concert from New York\'s Ed Sullivan Theater on Tuesday, July 16, 2014.',
57 'skip_download': True,
59 '_skip': 'Blocked outside the US',
61 'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/',
62 'only_matching': True,
64 'url': 'http://www.colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/',
65 'only_matching': True,
67 TP_RELEASE_URL_TEMPLATE = 'http://link.theplatform.com/s/dJ5BDC/%s?manifest=m3u&mbr=true'
69 def _real_extract(self, url):
70 display_id = self._match_id(url)
71 webpage = self._download_webpage(url, display_id)
72 content_id = self._search_regex(
73 [r"video\.settings\.content_id\s*=\s*'([^']+)';", r"cbsplayer\.contentId\s*=\s*'([^']+)';"],
74 webpage, 'content id')
75 items_data = self._download_xml(
76 'http://can.cbs.com/thunder/player/videoPlayerService.php',
77 content_id, query={'partner': 'cbs', 'contentId': content_id})
78 video_data = xpath_element(items_data, './/item')
79 title = xpath_text(video_data, 'videoTitle', 'title', True)
83 for item in items_data.findall('.//item'):
84 pid = xpath_text(item, 'pid')
88 tp_formats, tp_subtitles = self._extract_theplatform_smil(
89 self.TP_RELEASE_URL_TEMPLATE % pid, content_id, 'Downloading %s SMIL data' % pid)
90 except ExtractorError:
92 formats.extend(tp_formats)
93 subtitles = self._merge_subtitles(subtitles, tp_subtitles)
94 self._sort_formats(formats)
96 info = self.get_metadata('dJ5BDC/media/guid/2198311517/%s' % content_id, content_id)
99 'display_id': display_id,
101 'series': xpath_text(video_data, 'seriesTitle'),
102 'season_number': int_or_none(xpath_text(video_data, 'seasonNumber')),
103 'episode_number': int_or_none(xpath_text(video_data, 'episodeNumber')),
104 'duration': int_or_none(xpath_text(video_data, 'videoLength'), 1000),
105 'thumbnail': xpath_text(video_data, 'previewImageURL'),
107 'subtitles': subtitles,