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

   1 from __future__ import unicode_literals
   2
   3 from .mtv import MTVServicesInfoExtractor
   4 from .common import InfoExtractor
   5
   6
   7 class ComedyCentralIE(MTVServicesInfoExtractor):
   8     _VALID_URL = r'''(?x)https?://(?:www\.)?cc\.com/
   9         (video-clips|episodes|cc-studios|video-collections|shows(?=/[^/]+/(?!full-episodes)))
  10         /(?P<title>.*)'''
  11     _FEED_URL = 'http://comedycentral.com/feeds/mrss/'
  12
  13     _TESTS = [{
  14         'url': 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother',
  15         'md5': 'c4f48e9eda1b16dd10add0744344b6d8',
  16         'info_dict': {
  17             'id': 'cef0cbb3-e776-4bc9-b62e-8016deccb354',
  18             'ext': 'mp4',
  19             'title': 'CC:Stand-Up|August 18, 2013|1|0101|Uncensored - Too Good of a Mother',
  20             'description': 'After a certain point, breastfeeding becomes c**kblocking.',
  21             'timestamp': 1376798400,
  22             'upload_date': '20130818',
  23         },
  24     }, {
  25         'url': 'http://www.cc.com/shows/the-daily-show-with-trevor-noah/interviews/6yx39d/exclusive-rand-paul-extended-interview',
  26         'only_matching': True,
  27     }]
  28
  29
  30 class ComedyCentralFullEpisodesIE(MTVServicesInfoExtractor):
  31     _VALID_URL = r'''(?x)https?://(?:www\.)?cc\.com/
  32         (?:full-episodes|shows(?=/[^/]+/full-episodes))
  33         /(?P<id>[^?]+)'''
  34     _FEED_URL = 'http://comedycentral.com/feeds/mrss/'
  35
  36     _TESTS = [{
  37         'url': 'http://www.cc.com/full-episodes/pv391a/the-daily-show-with-trevor-noah-november-28--2016---ryan-speedo-green-season-22-ep-22028',
  38         'info_dict': {
  39             'description': 'Donald Trump is accused of exploiting his president-elect status for personal gain, Cuban leader Fidel Castro dies, and Ryan Speedo Green discusses "Sing for Your Life."',
  40             'title': 'November 28, 2016 - Ryan Speedo Green',
  41         },
  42         'playlist_count': 4,
  43     }, {
  44         'url': 'http://www.cc.com/shows/the-daily-show-with-trevor-noah/full-episodes',
  45         'only_matching': True,
  46     }]
  47
  48     def _real_extract(self, url):
  49         playlist_id = self._match_id(url)
  50         webpage = self._download_webpage(url, playlist_id)
  51         mgid = self._extract_triforce_mgid(webpage, data_zone='t2_lc_promo1')
  52         videos_info = self._get_videos_info(mgid)
  53         return videos_info
  54
  55
  56 class ToshIE(MTVServicesInfoExtractor):
  57     IE_DESC = 'Tosh.0'
  58     _VALID_URL = r'^https?://tosh\.cc\.com/video-(?:clips|collections)/[^/]+/(?P<videotitle>[^/?#]+)'
  59     _FEED_URL = 'http://tosh.cc.com/feeds/mrss'
  60
  61     _TESTS = [{
  62         'url': 'http://tosh.cc.com/video-clips/68g93d/twitter-users-share-summer-plans',
  63         'info_dict': {
  64             'description': 'Tosh asked fans to share their summer plans.',
  65             'title': 'Twitter Users Share Summer Plans',
  66         },
  67         'playlist': [{
  68             'md5': 'f269e88114c1805bb6d7653fecea9e06',
  69             'info_dict': {
  70                 'id': '90498ec2-ed00-11e0-aca6-0026b9414f30',
  71                 'ext': 'mp4',
  72                 'title': 'Tosh.0|June 9, 2077|2|211|Twitter Users Share Summer Plans',
  73                 'description': 'Tosh asked fans to share their summer plans.',
  74                 'thumbnail': r're:^https?://.*\.jpg',
  75                 # It's really reported to be published on year 2077
  76                 'upload_date': '20770610',
  77                 'timestamp': 3390510600,
  78                 'subtitles': {
  79                     'en': 'mincount:3',
  80                 },
  81             },
  82         }]
  83     }, {
  84         'url': 'http://tosh.cc.com/video-collections/x2iz7k/just-plain-foul/m5q4fp',
  85         'only_matching': True,
  86     }]
  87
  88
  89 class ComedyCentralTVIE(MTVServicesInfoExtractor):
  90     _VALID_URL = r'https?://(?:www\.)?comedycentral\.tv/(?:staffeln|shows)/(?P<id>[^/?#&]+)'
  91     _TESTS = [{
  92         'url': 'http://www.comedycentral.tv/staffeln/7436-the-mindy-project-staffel-4',
  93         'info_dict': {
  94             'id': 'local_playlist-f99b626bdfe13568579a',
  95             'ext': 'flv',
  96             'title': 'Episode_the-mindy-project_shows_season-4_episode-3_full-episode_part1',
  97         },
  98         'params': {
  99             # rtmp download
 100             'skip_download': True,
 101         },
 102     }, {
 103         'url': 'http://www.comedycentral.tv/shows/1074-workaholics',
 104         'only_matching': True,
 105     }, {
 106         'url': 'http://www.comedycentral.tv/shows/1727-the-mindy-project/bonus',
 107         'only_matching': True,
 108     }]
 109
 110     def _real_extract(self, url):
 111         video_id = self._match_id(url)
 112
 113         webpage = self._download_webpage(url, video_id)
 114
 115         mrss_url = self._search_regex(
 116             r'data-mrss=(["\'])(?P<url>(?:(?!\1).)+)\1',
 117             webpage, 'mrss url', group='url')
 118
 119         return self._get_videos_info_from_url(mrss_url, video_id)
 120
 121
 122 class ComedyCentralShortnameIE(InfoExtractor):
 123     _VALID_URL = r'^:(?P<id>tds|thedailyshow)$'
 124     _TESTS = [{
 125         'url': ':tds',
 126         'only_matching': True,
 127     }, {
 128         'url': ':thedailyshow',
 129         'only_matching': True,
 130     }]
 131
 132     def _real_extract(self, url):
 133         video_id = self._match_id(url)
 134         shortcut_map = {
 135             'tds': 'http://www.cc.com/shows/the-daily-show-with-trevor-noah/full-episodes',
 136             'thedailyshow': 'http://www.cc.com/shows/the-daily-show-with-trevor-noah/full-episodes',
 137         }
 138         return self.url_result(shortcut_map[video_id])