yt_dlp/extractor/sevenplus.py

   1 import json
   2 import re
   3
   4 from .brightcove import BrightcoveNewIE
   5 from ..compat import (
   6     compat_HTTPError,
   7     compat_str,
   8 )
   9 from ..utils import (
  10     ExtractorError,
  11     try_get,
  12     update_url_query,
  13 )
  14
  15
  16 class SevenPlusIE(BrightcoveNewIE):  # XXX: Do not subclass from concrete IE
  17     IE_NAME = '7plus'
  18     _VALID_URL = r'https?://(?:www\.)?7plus\.com\.au/(?P<path>[^?]+\?.*?\bepisode-id=(?P<id>[^&#]+))'
  19     _TESTS = [{
  20         'url': 'https://7plus.com.au/MTYS?episode-id=MTYS7-003',
  21         'info_dict': {
  22             'id': 'MTYS7-003',
  23             'ext': 'mp4',
  24             'title': 'S7 E3 - Wind Surf',
  25             'description': 'md5:29c6a69f21accda7601278f81b46483d',
  26             'uploader_id': '5303576322001',
  27             'upload_date': '20171201',
  28             'timestamp': 1512106377,
  29             'series': 'Mighty Ships',
  30             'season_number': 7,
  31             'episode_number': 3,
  32             'episode': 'Wind Surf',
  33         },
  34         'params': {
  35             'skip_download': True,
  36         }
  37     }, {
  38         'url': 'https://7plus.com.au/UUUU?episode-id=AUMS43-001',
  39         'only_matching': True,
  40     }]
  41
  42     def _real_initialize(self):
  43         self.token = None
  44
  45         cookies = self._get_cookies('https://7plus.com.au')
  46         api_key = next((x for x in cookies if x.startswith('glt_')), '')[4:]
  47         if not api_key:  # Cookies are signed out, skip login
  48             return
  49
  50         login_resp = self._download_json(
  51             'https://login.7plus.com.au/accounts.getJWT', None, 'Logging in', fatal=False,
  52             query={
  53                 'APIKey': api_key,
  54                 'sdk': 'js_latest',
  55                 'login_token': cookies[f'glt_{api_key}'].value,
  56                 'authMode': 'cookie',
  57                 'pageURL': 'https://7plus.com.au/',
  58                 'sdkBuild': '12471',
  59                 'format': 'json',
  60             }) or {}
  61
  62         if 'errorMessage' in login_resp:
  63             self.report_warning(f'Unable to login: 7plus said: {login_resp["errorMessage"]}')
  64             return
  65         id_token = login_resp.get('id_token')
  66         if not id_token:
  67             self.report_warning('Unable to login: Could not extract id token')
  68             return
  69
  70         token_resp = self._download_json(
  71             'https://7plus.com.au/auth/token', None, 'Getting auth token', fatal=False,
  72             headers={'Content-Type': 'application/json'}, data=json.dumps({
  73                 'idToken': id_token,
  74                 'platformId': 'web',
  75                 'regSource': '7plus',
  76             }).encode('utf-8')) or {}
  77         self.token = token_resp.get('token')
  78         if not self.token:
  79             self.report_warning('Unable to log in: Could not extract auth token')
  80
  81     def _real_extract(self, url):
  82         path, episode_id = self._match_valid_url(url).groups()
  83
  84         headers = {}
  85         if self.token:
  86             headers['Authorization'] = f'Bearer {self.token}'
  87
  88         try:
  89             media = self._download_json(
  90                 'https://videoservice.swm.digital/playback', episode_id, query={
  91                     'appId': '7plus',
  92                     'deviceType': 'web',
  93                     'platformType': 'web',
  94                     'accountId': 5303576322001,
  95                     'referenceId': 'ref:' + episode_id,
  96                     'deliveryId': 'csai',
  97                     'videoType': 'vod',
  98                 }, headers=headers)['media']
  99         except ExtractorError as e:
 100             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
 101                 raise ExtractorError(self._parse_json(
 102                     e.cause.read().decode(), episode_id)[0]['error_code'], expected=True)
 103             raise
 104
 105         for source in media.get('sources', {}):
 106             src = source.get('src')
 107             if not src:
 108                 continue
 109             source['src'] = update_url_query(src, {'rule': ''})
 110
 111         info = self._parse_brightcove_metadata(media, episode_id)
 112
 113         content = self._download_json(
 114             'https://component-cdn.swm.digital/content/' + path,
 115             episode_id, headers={
 116                 'market-id': 4,
 117             }, fatal=False) or {}
 118         for item in content.get('items', {}):
 119             if item.get('componentData', {}).get('componentType') == 'infoPanel':
 120                 for src_key, dst_key in [('title', 'title'), ('shortSynopsis', 'description')]:
 121                     value = item.get(src_key)
 122                     if value:
 123                         info[dst_key] = value
 124                 info['series'] = try_get(
 125                     item, lambda x: x['seriesLogo']['name'], compat_str)
 126                 mobj = re.search(r'^S(\d+)\s+E(\d+)\s+-\s+(.+)$', info['title'])
 127                 if mobj:
 128                     info.update({
 129                         'season_number': int(mobj.group(1)),
 130                         'episode_number': int(mobj.group(2)),
 131                         'episode': mobj.group(3),
 132                     })
 133
 134         return info