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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .brightcove import BrightcoveNewIE
   7 from ..utils import update_url_query
   8
   9
  10 class SevenPlusIE(BrightcoveNewIE):
  11     IE_NAME = '7plus'
  12     _VALID_URL = r'https?://(?:www\.)?7plus\.com\.au/(?P<path>[^?]+\?.*?\bepisode-id=(?P<id>[^&#]+))'
  13     _TESTS = [{
  14         'url': 'https://7plus.com.au/BEAT?episode-id=BEAT-001',
  15         'info_dict': {
  16             'id': 'BEAT-001',
  17             'ext': 'mp4',
  18             'title': 'S1 E1 - Help / Lucy In The Sky With Diamonds',
  19             'description': 'md5:37718bea20a8eedaca7f7361af566131',
  20             'uploader_id': '5303576322001',
  21             'upload_date': '20171031',
  22             'timestamp': 1509440068,
  23         },
  24         'params': {
  25             'format': 'bestvideo',
  26             'skip_download': True,
  27         }
  28     }, {
  29         'url': 'https://7plus.com.au/UUUU?episode-id=AUMS43-001',
  30         'only_matching': True,
  31     }]
  32
  33     def _real_extract(self, url):
  34         path, episode_id = re.match(self._VALID_URL, url).groups()
  35
  36         media = self._download_json(
  37             'https://videoservice.swm.digital/playback', episode_id, query={
  38                 'appId': '7plus',
  39                 'deviceType': 'web',
  40                 'platformType': 'web',
  41                 'accountId': 5303576322001,
  42                 'referenceId': 'ref:' + episode_id,
  43                 'deliveryId': 'csai',
  44                 'videoType': 'vod',
  45             })['media']
  46
  47         for source in media.get('sources', {}):
  48             src = source.get('src')
  49             if not src:
  50                 continue
  51             source['src'] = update_url_query(src, {'rule': ''})
  52
  53         info = self._parse_brightcove_metadata(media, episode_id)
  54
  55         content = self._download_json(
  56             'https://component-cdn.swm.digital/content/' + path,
  57             episode_id, headers={
  58                 'market-id': 4,
  59             }, fatal=False) or {}
  60         for item in content.get('items', {}):
  61             if item.get('componentData', {}).get('componentType') == 'infoPanel':
  62                 for src_key, dst_key in [('title', 'title'), ('shortSynopsis', 'description')]:
  63                     value = item.get(src_key)
  64                     if value:
  65                         info[dst_key] = value
  66
  67         return info