yt_dlp/extractor/sevenplus.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5 import re
   6
   7 from .brightcove import BrightcoveNewIE
   8 from ..compat import (
   9     compat_HTTPError,
  10     compat_str,
  11 )
  12 from ..utils import (
  13     ExtractorError,
  14     try_get,
  15     update_url_query,
  16 )
  17
  18
  19 class SevenPlusIE(BrightcoveNewIE):
  20     IE_NAME = '7plus'
  21     _VALID_URL = r'https?://(?:www\.)?7plus\.com\.au/(?P<path>[^?]+\?.*?\bepisode-id=(?P<id>[^&#]+))'
  22     _TESTS = [{
  23         'url': 'https://7plus.com.au/MTYS?episode-id=MTYS7-003',
  24         'info_dict': {
  25             'id': 'MTYS7-003',
  26             'ext': 'mp4',
  27             'title': 'S7 E3 - Wind Surf',
  28             'description': 'md5:29c6a69f21accda7601278f81b46483d',
  29             'uploader_id': '5303576322001',
  30             'upload_date': '20171201',
  31             'timestamp': 1512106377,
  32             'series': 'Mighty Ships',
  33             'season_number': 7,
  34             'episode_number': 3,
  35             'episode': 'Wind Surf',
  36         },
  37         'params': {
  38             'format': 'bestvideo',
  39             'skip_download': True,
  40         }
  41     }, {
  42         'url': 'https://7plus.com.au/UUUU?episode-id=AUMS43-001',
  43         'only_matching': True,
  44     }]
  45
  46     def _real_initialize(self):
  47         self.token = None
  48
  49         cookies = self._get_cookies('https://7plus.com.au')
  50         api_key = next((x for x in cookies if x.startswith('glt_')), '')[4:]
  51         if not api_key:  # Cookies are signed out, skip login
  52             return
  53
  54         login_resp = self._download_json(
  55             'https://login.7plus.com.au/accounts.getJWT', None, 'Logging in', fatal=False,
  56             query={
  57                 'APIKey': api_key,
  58                 'sdk': 'js_latest',
  59                 'login_token': cookies[f'glt_{api_key}'].value,
  60                 'authMode': 'cookie',
  61                 'pageURL': 'https://7plus.com.au/',
  62                 'sdkBuild': '12471',
  63                 'format': 'json',
  64             }) or {}
  65
  66         if 'errorMessage' in login_resp:
  67             self.report_warning(f'Unable to login: 7plus said: {login_resp["errorMessage"]}')
  68             return
  69         id_token = login_resp.get('id_token')
  70         if not id_token:
  71             self.report_warning('Unable to login: Could not extract id token')
  72             return
  73
  74         token_resp = self._download_json(
  75             'https://7plus.com.au/auth/token', None, 'Getting auth token', fatal=False,
  76             headers={'Content-Type': 'application/json'}, data=json.dumps({
  77                 'idToken': id_token,
  78                 'platformId': 'web',
  79                 'regSource': '7plus',
  80             }).encode('utf-8')) or {}
  81         self.token = token_resp.get('token')
  82         if not self.token:
  83             self.report_warning('Unable to log in: Could not extract auth token')
  84
  85     def _real_extract(self, url):
  86         path, episode_id = self._match_valid_url(url).groups()
  87
  88         headers = {}
  89         if self.token:
  90             headers['Authorization'] = f'Bearer {self.token}'
  91
  92         try:
  93             media = self._download_json(
  94                 'https://videoservice.swm.digital/playback', episode_id, query={
  95                     'appId': '7plus',
  96                     'deviceType': 'web',
  97                     'platformType': 'web',
  98                     'accountId': 5303576322001,
  99                     'referenceId': 'ref:' + episode_id,
 100                     'deliveryId': 'csai',
 101                     'videoType': 'vod',
 102                 }, headers=headers)['media']
 103         except ExtractorError as e:
 104             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
 105                 raise ExtractorError(self._parse_json(
 106                     e.cause.read().decode(), episode_id)[0]['error_code'], expected=True)
 107             raise
 108
 109         for source in media.get('sources', {}):
 110             src = source.get('src')
 111             if not src:
 112                 continue
 113             source['src'] = update_url_query(src, {'rule': ''})
 114
 115         info = self._parse_brightcove_metadata(media, episode_id)
 116
 117         content = self._download_json(
 118             'https://component-cdn.swm.digital/content/' + path,
 119             episode_id, headers={
 120                 'market-id': 4,
 121             }, fatal=False) or {}
 122         for item in content.get('items', {}):
 123             if item.get('componentData', {}).get('componentType') == 'infoPanel':
 124                 for src_key, dst_key in [('title', 'title'), ('shortSynopsis', 'description')]:
 125                     value = item.get(src_key)
 126                     if value:
 127                         info[dst_key] = value
 128                 info['series'] = try_get(
 129                     item, lambda x: x['seriesLogo']['name'], compat_str)
 130                 mobj = re.search(r'^S(\d+)\s+E(\d+)\s+-\s+(.+)$', info['title'])
 131                 if mobj:
 132                     info.update({
 133                         'season_number': int(mobj.group(1)),
 134                         'episode_number': int(mobj.group(2)),
 135                         'episode': mobj.group(3),
 136                     })
 137
 138         return info