yt_dlp/extractor/sevenplus.py

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