[youtube] Fix extraction.
[youtube-dl] / youtube_dl / 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\.)?bravotv\.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
33     def _real_extract(self, url):
34         display_id = self._match_id(url)
35         webpage = self._download_webpage(url, display_id)
36         settings = self._parse_json(self._search_regex(
37             r'<script[^>]+data-drupal-selector="drupal-settings-json"[^>]*>({.+?})</script>', webpage, 'drupal settings'),
38             display_id)
39         info = {}
40         query = {
41             'mbr': 'true',
42         }
43         account_pid, release_pid = [None] * 2
44         tve = settings.get('ls_tve')
45         if tve:
46             query['manifest'] = 'm3u'
47             mobj = re.search(r'<[^>]+id="pdk-player"[^>]+data-url=["\']?(?:https?:)?//player\.theplatform\.com/p/([^/]+)/(?:[^/]+/)*select/([^?#&"\']+)', webpage)
48             if mobj:
49                 account_pid, tp_path = mobj.groups()
50                 release_pid = tp_path.strip('/').split('/')[-1]
51             else:
52                 account_pid = 'HNK2IC'
53                 tp_path = release_pid = tve['release_pid']
54             if tve.get('entitlement') == 'auth':
55                 adobe_pass = settings.get('tve_adobe_auth', {})
56                 resource = self._get_mvpd_resource(
57                     adobe_pass.get('adobePassResourceId', 'bravo'),
58                     tve['title'], release_pid, tve.get('rating'))
59                 query['auth'] = self._extract_mvpd_auth(
60                     url, release_pid, adobe_pass.get('adobePassRequestorId', 'bravo'), resource)
61         else:
62             shared_playlist = settings['ls_playlist']
63             account_pid = shared_playlist['account_pid']
64             metadata = shared_playlist['video_metadata'][shared_playlist['default_clip']]
65             tp_path = release_pid = metadata.get('release_pid')
66             if not release_pid:
67                 release_pid = metadata['guid']
68                 tp_path = 'media/guid/2140479951/' + release_pid
69             info.update({
70                 'title': metadata['title'],
71                 'description': metadata.get('description'),
72                 'season_number': int_or_none(metadata.get('season_num')),
73                 'episode_number': int_or_none(metadata.get('episode_num')),
74             })
75             query['switch'] = 'progressive'
76         info.update({
77             '_type': 'url_transparent',
78             'id': release_pid,
79             'url': smuggle_url(update_url_query(
80                 'http://link.theplatform.com/s/%s/%s' % (account_pid, tp_path),
81                 query), {'force_smil_url': True}),
82             'ie_key': 'ThePlatform',
83         })
84         return info