yt_dlp/extractor/voot.py

   1 import json
   2 import time
   3 import uuid
   4
   5 from .common import InfoExtractor
   6 from ..compat import compat_str
   7 from ..networking.exceptions import HTTPError
   8 from ..utils import (
   9     ExtractorError,
  10     float_or_none,
  11     int_or_none,
  12     jwt_decode_hs256,
  13     parse_age_limit,
  14     traverse_obj,
  15     try_call,
  16     try_get,
  17     unified_strdate,
  18 )
  19
  20
  21 class VootBaseIE(InfoExtractor):
  22     _NETRC_MACHINE = 'voot'
  23     _GEO_BYPASS = False
  24     _LOGIN_HINT = 'Log in with "-u <email_address> -p <password>", or use "-u token -p <auth_token>" to login with auth token.'
  25     _TOKEN = None
  26     _EXPIRY = 0
  27     _API_HEADERS = {'Origin': 'https://www.voot.com', 'Referer': 'https://www.voot.com/'}
  28
  29     def _perform_login(self, username, password):
  30         if self._TOKEN and self._EXPIRY:
  31             return
  32
  33         if username.lower() == 'token' and try_call(lambda: jwt_decode_hs256(password)):
  34             VootBaseIE._TOKEN = password
  35             VootBaseIE._EXPIRY = jwt_decode_hs256(password)['exp']
  36             self.report_login()
  37
  38         # Mobile number as username is not supported
  39         elif not username.isdigit():
  40             check_username = self._download_json(
  41                 'https://userauth.voot.com/usersV3/v3/checkUser', None, data=json.dumps({
  42                     'type': 'email',
  43                     'email': username
  44                 }, separators=(',', ':')).encode(), headers={
  45                     **self._API_HEADERS,
  46                     'Content-Type': 'application/json;charset=utf-8',
  47                 }, note='Checking username', expected_status=403)
  48             if not traverse_obj(check_username, ('isExist', {bool})):
  49                 if traverse_obj(check_username, ('status', 'code', {int})) == 9999:
  50                     self.raise_geo_restricted(countries=['IN'])
  51                 raise ExtractorError('Incorrect username', expected=True)
  52             auth_token = traverse_obj(self._download_json(
  53                 'https://userauth.voot.com/usersV3/v3/login', None, data=json.dumps({
  54                     'type': 'traditional',
  55                     'deviceId': str(uuid.uuid4()),
  56                     'deviceBrand': 'PC/MAC',
  57                     'data': {
  58                         'email': username,
  59                         'password': password
  60                     }
  61                 }, separators=(',', ':')).encode(), headers={
  62                     **self._API_HEADERS,
  63                     'Content-Type': 'application/json;charset=utf-8',
  64                 }, note='Logging in', expected_status=400), ('data', 'authToken', {dict}))
  65             if not auth_token:
  66                 raise ExtractorError('Incorrect password', expected=True)
  67             VootBaseIE._TOKEN = auth_token['accessToken']
  68             VootBaseIE._EXPIRY = auth_token['expirationTime']
  69
  70         else:
  71             raise ExtractorError(self._LOGIN_HINT, expected=True)
  72
  73     def _check_token_expiry(self):
  74         if int(time.time()) >= self._EXPIRY:
  75             raise ExtractorError('Access token has expired', expected=True)
  76
  77     def _real_initialize(self):
  78         if not self._TOKEN:
  79             self.raise_login_required(self._LOGIN_HINT, method=None)
  80         self._check_token_expiry()
  81
  82
  83 class VootIE(VootBaseIE):
  84     _VALID_URL = r'''(?x)
  85                     (?:
  86                         voot:|
  87                         https?://(?:www\.)?voot\.com/?
  88                         (?:
  89                             movies?/[^/]+/|
  90                             (?:shows|kids)/(?:[^/]+/){4}
  91                         )
  92                      )
  93                     (?P<id>\d{3,})
  94                     '''
  95     _TESTS = [{
  96         'url': 'https://www.voot.com/shows/ishq-ka-rang-safed/1/360558/is-this-the-end-of-kamini-/441353',
  97         'info_dict': {
  98             'id': '441353',
  99             'ext': 'mp4',
 100             'title': 'Is this the end of Kamini?',
 101             'description': 'md5:06291fbbbc4dcbe21235c40c262507c1',
 102             'timestamp': 1472103000,
 103             'upload_date': '20160825',
 104             'series': 'Ishq Ka Rang Safed',
 105             'season_number': 1,
 106             'episode': 'Is this the end of Kamini?',
 107             'episode_number': 340,
 108             'release_date': '20160825',
 109             'season': 'Season 1',
 110             'age_limit': 13,
 111             'duration': 1146.0,
 112         },
 113         'params': {'skip_download': 'm3u8'},
 114     }, {
 115         'url': 'https://www.voot.com/kids/characters/mighty-cat-masked-niyander-e-/400478/school-bag-disappears/440925',
 116         'only_matching': True,
 117     }, {
 118         'url': 'https://www.voot.com/movies/pandavas-5/424627',
 119         'only_matching': True,
 120     }, {
 121         'url': 'https://www.voot.com/movie/fight-club/621842',
 122         'only_matching': True,
 123     }]
 124
 125     def _real_extract(self, url):
 126         video_id = self._match_id(url)
 127         media_info = self._download_json(
 128             'https://psapi.voot.com/jio/voot/v1/voot-web/content/query/asset-details', video_id,
 129             query={'ids': f'include:{video_id}', 'responseType': 'common'}, headers={'accesstoken': self._TOKEN})
 130
 131         try:
 132             m3u8_url = self._download_json(
 133                 'https://vootapi.media.jio.com/playback/v1/playbackrights', video_id,
 134                 'Downloading playback JSON', data=b'{}', headers={
 135                     **self.geo_verification_headers(),
 136                     **self._API_HEADERS,
 137                     'Content-Type': 'application/json;charset=utf-8',
 138                     'platform': 'androidwebdesktop',
 139                     'vootid': video_id,
 140                     'voottoken': self._TOKEN,
 141                 })['m3u8']
 142         except ExtractorError as e:
 143             if isinstance(e.cause, HTTPError) and e.cause.status == 400:
 144                 self._check_token_expiry()
 145             raise
 146
 147         formats = self._extract_m3u8_formats(m3u8_url, video_id, 'mp4', m3u8_id='hls')
 148         self._remove_duplicate_formats(formats)
 149
 150         return {
 151             'id': video_id,
 152             # '/_definst_/smil:vod/' m3u8 manifests claim to have 720p+ formats but max out at 480p
 153             'formats': traverse_obj(formats, (
 154                 lambda _, v: '/_definst_/smil:vod/' not in v['url'] or v['height'] <= 480)),
 155             'http_headers': self._API_HEADERS,
 156             **traverse_obj(media_info, ('result', 0, {
 157                 'title': ('fullTitle', {str}),
 158                 'description': ('fullSynopsis', {str}),
 159                 'series': ('showName', {str}),
 160                 'season_number': ('season', {int_or_none}),
 161                 'episode': ('fullTitle', {str}),
 162                 'episode_number': ('episode', {int_or_none}),
 163                 'timestamp': ('uploadTime', {int_or_none}),
 164                 'release_date': ('telecastDate', {unified_strdate}),
 165                 'age_limit': ('ageNemonic', {parse_age_limit}),
 166                 'duration': ('duration', {float_or_none}),
 167             })),
 168         }
 169
 170
 171 class VootSeriesIE(VootBaseIE):
 172     _VALID_URL = r'https?://(?:www\.)?voot\.com/shows/[^/]+/(?P<id>\d{3,})'
 173     _TESTS = [{
 174         'url': 'https://www.voot.com/shows/chakravartin-ashoka-samrat/100002',
 175         'playlist_mincount': 442,
 176         'info_dict': {
 177             'id': '100002',
 178         },
 179     }, {
 180         'url': 'https://www.voot.com/shows/ishq-ka-rang-safed/100003',
 181         'playlist_mincount': 341,
 182         'info_dict': {
 183             'id': '100003',
 184         },
 185     }]
 186     _SHOW_API = 'https://psapi.voot.com/media/voot/v1/voot-web/content/generic/season-by-show?sort=season%3Aasc&id={}&responseType=common'
 187     _SEASON_API = 'https://psapi.voot.com/media/voot/v1/voot-web/content/generic/series-wise-episode?sort=episode%3Aasc&id={}&responseType=common&page={:d}'
 188
 189     def _entries(self, show_id):
 190         show_json = self._download_json(self._SHOW_API.format(show_id), video_id=show_id)
 191         for season in show_json.get('result', []):
 192             page_num = 1
 193             season_id = try_get(season, lambda x: x['id'], compat_str)
 194             season_json = self._download_json(self._SEASON_API.format(season_id, page_num),
 195                                               video_id=season_id,
 196                                               note='Downloading JSON metadata page %d' % page_num)
 197             episodes_json = season_json.get('result', [])
 198             while episodes_json:
 199                 page_num += 1
 200                 for episode in episodes_json:
 201                     video_id = episode.get('id')
 202                     yield self.url_result(
 203                         'voot:%s' % video_id, ie=VootIE.ie_key(), video_id=video_id)
 204                 episodes_json = self._download_json(self._SEASON_API.format(season_id, page_num),
 205                                                     video_id=season_id,
 206                                                     note='Downloading JSON metadata page %d' % page_num)['result']
 207
 208     def _real_extract(self, url):
 209         show_id = self._match_id(url)
 210         return self.playlist_result(self._entries(show_id), playlist_id=show_id)