[cbs,cbsnews,cbssports] reduce requests while extracting all formats
[youtube-dl] / youtube_dl / extractor / cbs.py
1 from __future__ import unicode_literals
2
3 import re
4
5 from .theplatform import ThePlatformFeedIE
6 from ..utils import (
7     int_or_none,
8     find_xpath_attr,
9 )
10
11
12 class CBSBaseIE(ThePlatformFeedIE):
13     def _parse_smil_subtitles(self, smil, namespace=None, subtitles_lang='en'):
14         closed_caption_e = find_xpath_attr(smil, self._xpath_ns('.//param', namespace), 'name', 'ClosedCaptionURL')
15         return {
16             'en': [{
17                 'ext': 'ttml',
18                 'url': closed_caption_e.attrib['value'],
19             }]
20         } if closed_caption_e is not None and closed_caption_e.attrib.get('value') else []
21
22     def _extract_video_info(self, filter_query, video_id):
23         return self._extract_feed_info(
24             'dJ5BDC', 'VxxJg8Ymh8sE', filter_query, video_id, lambda entry: {
25                 'series': entry.get('cbs$SeriesTitle'),
26                 'season_number': int_or_none(entry.get('cbs$SeasonNumber')),
27                 'episode': entry.get('cbs$EpisodeTitle'),
28                 'episode_number': int_or_none(entry.get('cbs$EpisodeNumber')),
29             }, {
30                 'StreamPack': {
31                     'manifest': 'm3u',
32                 }
33             })
34
35
36 class CBSIE(CBSBaseIE):
37     _VALID_URL = r'(?:cbs|https?://(?:www\.)?(?:cbs\.com/shows/[^/]+/video|colbertlateshow\.com/(?:video|podcasts))/)(?P<id>[\w-]+)'
38
39     _TESTS = [{
40         'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/',
41         'info_dict': {
42             'id': '_u7W953k6la293J7EPTd9oHkSPs6Xn6_',
43             'display_id': 'connect-chat-feat-garth-brooks',
44             'ext': 'mp4',
45             'title': 'Connect Chat feat. Garth Brooks',
46             '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!',
47             'duration': 1495,
48             'timestamp': 1385585425,
49             'upload_date': '20131127',
50             'uploader': 'CBSI-NEW',
51         },
52         'expected_warnings': ['Failed to download m3u8 information'],
53         '_skip': 'Blocked outside the US',
54     }, {
55         'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/',
56         'only_matching': True,
57     }, {
58         'url': 'http://www.colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/',
59         'only_matching': True,
60     }]
61     TP_RELEASE_URL_TEMPLATE = 'http://link.theplatform.com/s/dJ5BDC/%s?mbr=true'
62
63     def _real_extract(self, url):
64         content_id = self._match_id(url)
65         return self._extract_video_info('byGuid=%s' % content_id, content_id)