yt_dlp/extractor/tennistv.py

   1 import urllib.parse
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     ExtractorError,
   6     random_uuidv4,
   7     unified_timestamp,
   8     urlencode_postdata,
   9 )
  10
  11
  12 class TennisTVIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?tennistv\.com/videos/(?P<id>[-a-z0-9]+)'
  14     _TESTS = [{
  15         'url': 'https://www.tennistv.com/videos/indian-wells-2018-verdasco-fritz',
  16         'info_dict': {
  17             'id': 'indian-wells-2018-verdasco-fritz',
  18             'ext': 'mp4',
  19             'title': 'Fernando Verdasco v Taylor Fritz',
  20             'description': 're:^After his stunning victory.{174}$',
  21             'thumbnail': 'https://atp-prod.akamaized.net/api/images/v1/images/112831/landscape/1242/0',
  22             'timestamp': 1521017381,
  23             'upload_date': '20180314',
  24         },
  25         'params': {
  26             'skip_download': True,
  27         },
  28         'skip': 'Requires email and password of a subscribed account',
  29     }, {
  30         'url': 'https://www.tennistv.com/videos/2650480/best-matches-of-2022-part-5',
  31         'info_dict': {
  32             'id': '2650480',
  33             'ext': 'mp4',
  34             'title': 'Best Matches of 2022 - Part 5',
  35             'description': 'md5:36dec3bfae7ed74bd79e48045b17264c',
  36             'thumbnail': 'https://open.http.mp.streamamg.com/p/3001482/sp/300148200/thumbnail/entry_id/0_myef18pd/version/100001/height/1920',
  37         },
  38         'params': {'skip_download': 'm3u8'},
  39         'skip': 'Requires email and password of a subscribed account',
  40     }]
  41     _NETRC_MACHINE = 'tennistv'
  42
  43     access_token, refresh_token = None, None
  44     _PARTNER_ID = 3001482
  45     _FORMAT_URL = 'https://open.http.mp.streamamg.com/p/{partner}/sp/{partner}00/playManifest/entryId/{entry}/format/applehttp/protocol/https/a.m3u8?ks={session}'
  46     _AUTH_BASE_URL = 'https://sso.tennistv.com/auth/realms/TennisTV/protocol/openid-connect'
  47     _HEADERS = {
  48         'origin': 'https://www.tennistv.com',
  49         'referer': 'https://www.tennistv.com/',
  50         'content-Type': 'application/x-www-form-urlencoded',
  51     }
  52
  53     def _perform_login(self, username, password):
  54         login_page = self._download_webpage(
  55             f'{self._AUTH_BASE_URL}/auth', None, 'Downloading login page',
  56             query={
  57                 'client_id': 'tennis-tv-web',
  58                 'redirect_uri': 'https://tennistv.com',
  59                 'response_mode': 'fragment',
  60                 'response_type': 'code',
  61                 'scope': 'openid',
  62             })
  63
  64         post_url = self._html_search_regex(r'action=["\']([^"\']+?)["\']\s+method=["\']post["\']', login_page, 'login POST url')
  65         temp_page = self._download_webpage(
  66             post_url, None, 'Sending login data', 'Unable to send login data',
  67             headers=self._HEADERS, data=urlencode_postdata({
  68                 'username': username,
  69                 'password': password,
  70                 'submitAction': 'Log In',
  71             }))
  72         if 'Your username or password was incorrect' in temp_page:
  73             raise ExtractorError('Your username or password was incorrect', expected=True)
  74
  75         handle = self._request_webpage(
  76             f'{self._AUTH_BASE_URL}/auth', None, 'Logging in', headers=self._HEADERS,
  77             query={
  78                 'client_id': 'tennis-tv-web',
  79                 'redirect_uri': 'https://www.tennistv.com/resources/v1.1.10/html/silent-check-sso.html',
  80                 'state': random_uuidv4(),
  81                 'response_mode': 'fragment',
  82                 'response_type': 'code',
  83                 'scope': 'openid',
  84                 'nonce': random_uuidv4(),
  85                 'prompt': 'none',
  86             })
  87
  88         self.get_token(None, {
  89             'code': urllib.parse.parse_qs(handle.url)['code'][-1],
  90             'grant_type': 'authorization_code',
  91             'client_id': 'tennis-tv-web',
  92             'redirect_uri': 'https://www.tennistv.com/resources/v1.1.10/html/silent-check-sso.html',
  93         })
  94
  95     def get_token(self, video_id, payload):
  96         res = self._download_json(
  97             f'{self._AUTH_BASE_URL}/token', video_id, 'Fetching tokens',
  98             'Unable to fetch tokens', headers=self._HEADERS, data=urlencode_postdata(payload))
  99
 100         self.access_token = res.get('access_token') or self.access_token
 101         self.refresh_token = res.get('refresh_token') or self.refresh_token
 102
 103     def _real_initialize(self):
 104         if self.access_token and self.refresh_token:
 105             return
 106
 107         cookies = self._get_cookies('https://www.tennistv.com/')
 108         if not cookies.get('access_token') or not cookies.get('refresh_token'):
 109             self.raise_login_required()
 110         self.access_token, self.refresh_token = cookies['access_token'].value, cookies['refresh_token'].value
 111
 112     def _download_session_json(self, video_id, entryid):
 113         return self._download_json(
 114             f'https://atppayments.streamamg.com/api/v1/session/ksession/?lang=en&apijwttoken={self.access_token}&entryId={entryid}',
 115             video_id, 'Downloading ksession token', 'Failed to download ksession token', headers=self._HEADERS)
 116
 117     def _real_extract(self, url):
 118         video_id = self._match_id(url)
 119         webpage = self._download_webpage(url, video_id)
 120
 121         entryid = self._search_regex(r'data-entry-id=["\']([^"\']+)', webpage, 'entryID')
 122         session_json = self._download_session_json(video_id, entryid)
 123
 124         k_session = session_json.get('KSession')
 125         if k_session is None:
 126             self.get_token(video_id, {
 127                 'grant_type': 'refresh_token',
 128                 'refresh_token': self.refresh_token,
 129                 'client_id': 'tennis-tv-web',
 130             })
 131             k_session = self._download_session_json(video_id, entryid).get('KSession')
 132             if k_session is None:
 133                 raise ExtractorError('Failed to get KSession, possibly a premium video', expected=True)
 134
 135         if session_json.get('ErrorMessage'):
 136             self.report_warning(session_json['ErrorMessage'])
 137
 138         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
 139             self._FORMAT_URL.format(partner=self._PARTNER_ID, entry=entryid, session=k_session), video_id)
 140
 141         return {
 142             'id': video_id,
 143             'title': self._generic_title('', webpage),
 144             'description': self._html_search_regex(
 145                 (r'<span itemprop="description" content=["\']([^"\']+)["\']>', *self._og_regexes('description')),
 146                 webpage, 'description', fatal=False),
 147             'thumbnail': f'https://open.http.mp.streamamg.com/p/{self._PARTNER_ID}/sp/{self._PARTNER_ID}00/thumbnail/entry_id/{entryid}/version/100001/height/1920',
 148             'timestamp': unified_timestamp(self._html_search_regex(
 149                 r'<span itemprop="uploadDate" content=["\']([^"\']+)["\']>', webpage, 'upload time', fatal=False)),
 150             'series': self._html_search_regex(r'data-series\s*?=\s*?"(.*?)"', webpage, 'series', fatal=False) or None,
 151             'season': self._html_search_regex(r'data-tournament-city\s*?=\s*?"(.*?)"', webpage, 'season', fatal=False) or None,
 152             'episode': self._html_search_regex(r'data-round\s*?=\s*?"(.*?)"', webpage, 'round', fatal=False) or None,
 153             'formats': formats,
 154             'subtitles': subtitles,
 155         }