yt_dlp/extractor/sevenplus.py

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