Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/usanetwork.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     extract_attributes,
   9     smuggle_url,
  10     update_url_query,
  11 )
  12
  13
  14 class USANetworkIE(AdobePassIE):
  15     _VALID_URL = r'https?://(?:www\.)?usanetwork\.com/(?:[^/]+/videos|movies)/(?P<id>[^/?#]+)'
  16     _TEST = {
  17         'url': 'http://www.usanetwork.com/mrrobot/videos/hpe-cybersecurity',
  18         'md5': '33c0d2ba381571b414024440d08d57fd',
  19         'info_dict': {
  20             'id': '3086229',
  21             'ext': 'mp4',
  22             'title': 'HPE Cybersecurity',
  23             'description': 'The more we digitize our world, the more vulnerable we are.',
  24             'upload_date': '20160818',
  25             'timestamp': 1471535460,
  26             'uploader': 'NBCU-USA',
  27         },
  28     }
  29
  30     def _real_extract(self, url):
  31         display_id = self._match_id(url)
  32         webpage = self._download_webpage(url, display_id)
  33
  34         player_params = extract_attributes(self._search_regex(
  35             r'(<div[^>]+data-usa-tve-player-container[^>]*>)', webpage, 'player params'))
  36         video_id = player_params['data-mpx-guid']
  37         title = player_params['data-episode-title']
  38
  39         account_pid, path = re.search(
  40             r'data-src="(?:https?)?//player\.theplatform\.com/p/([^/]+)/.*?/(media/guid/\d+/\d+)',
  41             webpage).groups()
  42
  43         query = {
  44             'mbr': 'true',
  45         }
  46         if player_params.get('data-is-full-episode') == '1':
  47             query['manifest'] = 'm3u'
  48
  49         if player_params.get('data-entitlement') == 'auth':
  50             adobe_pass = {}
  51             drupal_settings = self._search_regex(
  52                 r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);',
  53                 webpage, 'drupal settings', fatal=False)
  54             if drupal_settings:
  55                 drupal_settings = self._parse_json(drupal_settings, video_id, fatal=False)
  56                 if drupal_settings:
  57                     adobe_pass = drupal_settings.get('adobePass', {})
  58             resource = self._get_mvpd_resource(
  59                 adobe_pass.get('adobePassResourceId', 'usa'),
  60                 title, video_id, player_params.get('data-episode-rating', 'TV-14'))
  61             query['auth'] = self._extract_mvpd_auth(
  62                 url, video_id, adobe_pass.get('adobePassRequestorId', 'usa'), resource)
  63
  64         info = self._search_json_ld(webpage, video_id, default={})
  65         info.update({
  66             '_type': 'url_transparent',
  67             'url': smuggle_url(update_url_query(
  68                 'http://link.theplatform.com/s/%s/%s' % (account_pid, path),
  69                 query), {'force_smil_url': True}),
  70             'id': video_id,
  71             'title': title,
  72             'series': player_params.get('data-show-title'),
  73             'episode': title,
  74             'ie_key': 'ThePlatform',
  75         })
  76         return info