youtube_dlc/extractor/bravotv.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .adobepass import AdobePassIE
   7 from ..utils import (
   8     smuggle_url,
   9     update_url_query,
  10     int_or_none,
  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         }
  28     }, {
  29         'url': 'http://www.bravotv.com/below-deck/season-3/ep-14-reunion-part-1',
  30         'only_matching': True,
  31     }, {
  32         'url': 'https://www.oxygen.com/in-ice-cold-blood/season-2/episode-16/videos/handling-the-horwitz-house-after-the-murder-season-2',
  33         'only_matching': True,
  34     }]
  35
  36     def _real_extract(self, url):
  37         site, display_id = re.match(self._VALID_URL, url).groups()
  38         webpage = self._download_webpage(url, display_id)
  39         settings = self._parse_json(self._search_regex(
  40             r'<script[^>]+data-drupal-selector="drupal-settings-json"[^>]*>({.+?})</script>', webpage, 'drupal settings'),
  41             display_id)
  42         info = {}
  43         query = {
  44             'mbr': 'true',
  45         }
  46         account_pid, release_pid = [None] * 2
  47         tve = settings.get('ls_tve')
  48         if tve:
  49             query['manifest'] = 'm3u'
  50             mobj = re.search(r'<[^>]+id="pdk-player"[^>]+data-url=["\']?(?:https?:)?//player\.theplatform\.com/p/([^/]+)/(?:[^/]+/)*select/([^?#&"\']+)', webpage)
  51             if mobj:
  52                 account_pid, tp_path = mobj.groups()
  53                 release_pid = tp_path.strip('/').split('/')[-1]
  54             else:
  55                 account_pid = 'HNK2IC'
  56                 tp_path = release_pid = tve['release_pid']
  57             if tve.get('entitlement') == 'auth':
  58                 adobe_pass = settings.get('tve_adobe_auth', {})
  59                 if site == 'bravotv':
  60                     site = 'bravo'
  61                 resource = self._get_mvpd_resource(
  62                     adobe_pass.get('adobePassResourceId') or site,
  63                     tve['title'], release_pid, tve.get('rating'))
  64                 query['auth'] = self._extract_mvpd_auth(
  65                     url, release_pid,
  66                     adobe_pass.get('adobePassRequestorId') or site, resource)
  67         else:
  68             shared_playlist = settings['ls_playlist']
  69             account_pid = shared_playlist['account_pid']
  70             metadata = shared_playlist['video_metadata'][shared_playlist['default_clip']]
  71             tp_path = release_pid = metadata.get('release_pid')
  72             if not release_pid:
  73                 release_pid = metadata['guid']
  74                 tp_path = 'media/guid/2140479951/' + release_pid
  75             info.update({
  76                 'title': metadata['title'],
  77                 'description': metadata.get('description'),
  78                 'season_number': int_or_none(metadata.get('season_num')),
  79                 'episode_number': int_or_none(metadata.get('episode_num')),
  80             })
  81             query['switch'] = 'progressive'
  82         info.update({
  83             '_type': 'url_transparent',
  84             'id': release_pid,
  85             'url': smuggle_url(update_url_query(
  86                 'http://link.theplatform.com/s/%s/%s' % (account_pid, tp_path),
  87                 query), {'force_smil_url': True}),
  88             'ie_key': 'ThePlatform',
  89         })
  90         return info