yt_dlp/extractor/bravotv.py

   1 import re
   2
   3 from .adobepass import AdobePassIE
   4 from ..utils import (
   5     smuggle_url,
   6     update_url_query,
   7     int_or_none,
   8     float_or_none,
   9     try_get,
  10     dict_get,
  11 )
  12
  13
  14 class BravoTVIE(AdobePassIE):
  15     _VALID_URL = r'https?://(?:www\.)?(?P<req_id>bravotv|oxygen)\.com/(?:[^/]+/)+(?P<id>[^/?#]+)'
  16     _TESTS = [{
  17         'url': 'https://www.bravotv.com/top-chef/season-16/episode-15/videos/the-top-chef-season-16-winner-is',
  18         'md5': 'e34684cfea2a96cd2ee1ef3a60909de9',
  19         'info_dict': {
  20             'id': 'epL0pmK1kQlT',
  21             'ext': 'mp4',
  22             'title': 'The Top Chef Season 16 Winner Is...',
  23             'description': 'Find out who takes the title of Top Chef!',
  24             'uploader': 'NBCU-BRAV',
  25             'upload_date': '20190314',
  26             'timestamp': 1552591860,
  27             'season_number': 16,
  28             'episode_number': 15,
  29             'series': 'Top Chef',
  30             'episode': 'The Top Chef Season 16 Winner Is...',
  31             'duration': 190.0,
  32         }
  33     }, {
  34         'url': 'http://www.bravotv.com/below-deck/season-3/ep-14-reunion-part-1',
  35         'only_matching': True,
  36     }, {
  37         'url': 'https://www.oxygen.com/in-ice-cold-blood/season-2/episode-16/videos/handling-the-horwitz-house-after-the-murder-season-2',
  38         'only_matching': True,
  39     }]
  40
  41     def _real_extract(self, url):
  42         site, display_id = self._match_valid_url(url).groups()
  43         webpage = self._download_webpage(url, display_id)
  44         settings = self._parse_json(self._search_regex(
  45             r'<script[^>]+data-drupal-selector="drupal-settings-json"[^>]*>({.+?})</script>', webpage, 'drupal settings'),
  46             display_id)
  47         info = {}
  48         query = {
  49             'mbr': 'true',
  50         }
  51         account_pid, release_pid = [None] * 2
  52         tve = settings.get('ls_tve')
  53         if tve:
  54             query['manifest'] = 'm3u'
  55             mobj = re.search(r'<[^>]+id="pdk-player"[^>]+data-url=["\']?(?:https?:)?//player\.theplatform\.com/p/([^/]+)/(?:[^/]+/)*select/([^?#&"\']+)', webpage)
  56             if mobj:
  57                 account_pid, tp_path = mobj.groups()
  58                 release_pid = tp_path.strip('/').split('/')[-1]
  59             else:
  60                 account_pid = 'HNK2IC'
  61                 tp_path = release_pid = tve['release_pid']
  62             if tve.get('entitlement') == 'auth':
  63                 adobe_pass = settings.get('tve_adobe_auth', {})
  64                 if site == 'bravotv':
  65                     site = 'bravo'
  66                 resource = self._get_mvpd_resource(
  67                     adobe_pass.get('adobePassResourceId') or site,
  68                     tve['title'], release_pid, tve.get('rating'))
  69                 query['auth'] = self._extract_mvpd_auth(
  70                     url, release_pid,
  71                     adobe_pass.get('adobePassRequestorId') or site, resource)
  72         else:
  73             shared_playlist = settings['ls_playlist']
  74             account_pid = shared_playlist['account_pid']
  75             metadata = shared_playlist['video_metadata'][shared_playlist['default_clip']]
  76             tp_path = release_pid = metadata.get('release_pid')
  77             if not release_pid:
  78                 release_pid = metadata['guid']
  79                 tp_path = 'media/guid/2140479951/' + release_pid
  80             info.update({
  81                 'title': metadata['title'],
  82                 'description': metadata.get('description'),
  83                 'season_number': int_or_none(metadata.get('season_num')),
  84                 'episode_number': int_or_none(metadata.get('episode_num')),
  85             })
  86             query['switch'] = 'progressive'
  87
  88         tp_url = 'http://link.theplatform.com/s/%s/%s' % (account_pid, tp_path)
  89
  90         tp_metadata = self._download_json(
  91             update_url_query(tp_url, {'format': 'preview'}),
  92             display_id, fatal=False)
  93         if tp_metadata:
  94             info.update({
  95                 'title': tp_metadata.get('title'),
  96                 'description': tp_metadata.get('description'),
  97                 'duration': float_or_none(tp_metadata.get('duration'), 1000),
  98                 'season_number': int_or_none(
  99                     dict_get(tp_metadata, ('pl1$seasonNumber', 'nbcu$seasonNumber'))),
 100                 'episode_number': int_or_none(
 101                     dict_get(tp_metadata, ('pl1$episodeNumber', 'nbcu$episodeNumber'))),
 102                 # For some reason the series is sometimes wrapped into a single element array.
 103                 'series': try_get(
 104                     dict_get(tp_metadata, ('pl1$show', 'nbcu$show')),
 105                     lambda x: x[0] if isinstance(x, list) else x,
 106                     expected_type=str),
 107                 'episode': dict_get(
 108                     tp_metadata, ('pl1$episodeName', 'nbcu$episodeName', 'title')),
 109             })
 110
 111         info.update({
 112             '_type': 'url_transparent',
 113             'id': release_pid,
 114             'url': smuggle_url(update_url_query(tp_url, query), {'force_smil_url': True}),
 115             'ie_key': 'ThePlatform',
 116         })
 117         return info