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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .theplatform import ThePlatformIE
   5 from ..utils import (
   6     int_or_none,
   7     parse_age_limit,
   8     try_get,
   9     update_url_query,
  10 )
  11
  12
  13 class AMCNetworksIE(ThePlatformIE):
  14     _VALID_URL = r'https?://(?:www\.)?(?:amc|bbcamerica|ifc|wetv)\.com/(?:movies|shows(?:/[^/]+)+)/(?P<id>[^/?#]+)'
  15     _TESTS = [{
  16         'url': 'http://www.ifc.com/shows/maron/season-04/episode-01/step-1',
  17         'md5': '',
  18         'info_dict': {
  19             'id': 's3MX01Nl4vPH',
  20             'ext': 'mp4',
  21             'title': 'Maron - Season 4 - Step 1',
  22             'description': 'In denial about his current situation, Marc is reluctantly convinced by his friends to enter rehab. Starring Marc Maron and Constance Zimmer.',
  23             'age_limit': 17,
  24             'upload_date': '20160505',
  25             'timestamp': 1462468831,
  26             'uploader': 'AMCN',
  27         },
  28         'params': {
  29             # m3u8 download
  30             'skip_download': True,
  31         },
  32         'skip': 'Requires TV provider accounts',
  33     }, {
  34         'url': 'http://www.bbcamerica.com/shows/the-hunt/full-episodes/season-1/episode-01-the-hardest-challenge',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://www.amc.com/shows/preacher/full-episodes/season-01/episode-00/pilot',
  38         'only_matching': True,
  39     }, {
  40         'url': 'http://www.wetv.com/shows/million-dollar-matchmaker/season-01/episode-06-the-dumped-dj-and-shallow-hal',
  41         'only_matching': True,
  42     }, {
  43         'url': 'http://www.ifc.com/movies/chaos',
  44         'only_matching': True,
  45     }, {
  46         'url': 'http://www.bbcamerica.com/shows/doctor-who/full-episodes/the-power-of-the-daleks/episode-01-episode-1-color-version',
  47         'only_matching': True,
  48     }, {
  49         'url': 'http://www.wetv.com/shows/mama-june-from-not-to-hot/full-episode/season-01/thin-tervention',
  50         'only_matching': True,
  51     }, {
  52         'url': 'http://www.wetv.com/shows/la-hair/videos/season-05/episode-09-episode-9-2/episode-9-sneak-peek-3',
  53         'only_matching': True,
  54     }]
  55
  56     def _real_extract(self, url):
  57         display_id = self._match_id(url)
  58         webpage = self._download_webpage(url, display_id)
  59         query = {
  60             'mbr': 'true',
  61             'manifest': 'm3u',
  62         }
  63         media_url = self._search_regex(
  64             r'window\.platformLinkURL\s*=\s*[\'"]([^\'"]+)',
  65             webpage, 'media url')
  66         theplatform_metadata = self._download_theplatform_metadata(self._search_regex(
  67             r'link\.theplatform\.com/s/([^?]+)',
  68             media_url, 'theplatform_path'), display_id)
  69         info = self._parse_theplatform_metadata(theplatform_metadata)
  70         video_id = theplatform_metadata['pid']
  71         title = theplatform_metadata['title']
  72         rating = try_get(
  73             theplatform_metadata, lambda x: x['ratings'][0]['rating'])
  74         auth_required = self._search_regex(
  75             r'window\.authRequired\s*=\s*(true|false);',
  76             webpage, 'auth required')
  77         if auth_required == 'true':
  78             requestor_id = self._search_regex(
  79                 r'window\.requestor_id\s*=\s*[\'"]([^\'"]+)',
  80                 webpage, 'requestor id')
  81             resource = self._get_mvpd_resource(
  82                 requestor_id, title, video_id, rating)
  83             query['auth'] = self._extract_mvpd_auth(
  84                 url, video_id, requestor_id, resource)
  85         media_url = update_url_query(media_url, query)
  86         formats, subtitles = self._extract_theplatform_smil(
  87             media_url, video_id)
  88         self._sort_formats(formats)
  89         info.update({
  90             'id': video_id,
  91             'subtitles': subtitles,
  92             'formats': formats,
  93             'age_limit': parse_age_limit(parse_age_limit(rating)),
  94         })
  95         ns_keys = theplatform_metadata.get('$xmlns', {}).keys()
  96         if ns_keys:
  97             ns = list(ns_keys)[0]
  98             series = theplatform_metadata.get(ns + '$show')
  99             season_number = int_or_none(
 100                 theplatform_metadata.get(ns + '$season'))
 101             episode = theplatform_metadata.get(ns + '$episodeTitle')
 102             episode_number = int_or_none(
 103                 theplatform_metadata.get(ns + '$episode'))
 104             if season_number:
 105                 title = 'Season %d - %s' % (season_number, title)
 106             if series:
 107                 title = '%s - %s' % (series, title)
 108             info.update({
 109                 'title': title,
 110                 'series': series,
 111                 'season_number': season_number,
 112                 'episode': episode,
 113                 'episode_number': episode_number,
 114             })
 115         return info