yt_dlp/extractor/jiocinema.py

   1 import base64
   2 import itertools
   3 import json
   4 import random
   5 import re
   6 import string
   7 import time
   8
   9 from .common import InfoExtractor
  10 from ..utils import (
  11     ExtractorError,
  12     float_or_none,
  13     int_or_none,
  14     jwt_decode_hs256,
  15     parse_age_limit,
  16     try_call,
  17     url_or_none,
  18 )
  19 from ..utils.traversal import traverse_obj
  20
  21
  22 class JioCinemaBaseIE(InfoExtractor):
  23     _NETRC_MACHINE = 'jiocinema'
  24     _GEO_BYPASS = False
  25     _ACCESS_TOKEN = None
  26     _REFRESH_TOKEN = None
  27     _GUEST_TOKEN = None
  28     _USER_ID = None
  29     _DEVICE_ID = None
  30     _API_HEADERS = {'Origin': 'https://www.jiocinema.com', 'Referer': 'https://www.jiocinema.com/'}
  31     _APP_NAME = {'appName': 'RJIL_JioCinema'}
  32     _APP_VERSION = {'appVersion': '5.0.0'}
  33     _API_SIGNATURES = 'o668nxgzwff'
  34     _METADATA_API_BASE = 'https://content-jiovoot.voot.com/psapi'
  35     _ACCESS_HINT = 'the `accessToken` from your browser local storage'
  36     _LOGIN_HINT = (
  37         'Log in with "-u phone -p <PHONE_NUMBER>" to authenticate with OTP, '
  38         f'or use "-u token -p <ACCESS_TOKEN>" to log in with {_ACCESS_HINT}. '
  39         'If you have previously logged in with yt-dlp and your session '
  40         'has been cached, you can use "-u device -p <DEVICE_ID>"')
  41
  42     def _cache_token(self, token_type):
  43         assert token_type in ('access', 'refresh', 'all')
  44         if token_type in ('access', 'all'):
  45             self.cache.store(
  46                 JioCinemaBaseIE._NETRC_MACHINE, f'{JioCinemaBaseIE._DEVICE_ID}-access', JioCinemaBaseIE._ACCESS_TOKEN)
  47         if token_type in ('refresh', 'all'):
  48             self.cache.store(
  49                 JioCinemaBaseIE._NETRC_MACHINE, f'{JioCinemaBaseIE._DEVICE_ID}-refresh', JioCinemaBaseIE._REFRESH_TOKEN)
  50
  51     def _call_api(self, url, video_id, note='Downloading API JSON', headers={}, data={}):
  52         return self._download_json(
  53             url, video_id, note, data=json.dumps(data, separators=(',', ':')).encode(), headers={
  54                 'Content-Type': 'application/json',
  55                 'Accept': 'application/json',
  56                 **self._API_HEADERS,
  57                 **headers,
  58             }, expected_status=(400, 403, 474))
  59
  60     def _call_auth_api(self, service, endpoint, note, headers={}, data={}):
  61         return self._call_api(
  62             f'https://auth-jiocinema.voot.com/{service}service/apis/v4/{endpoint}',
  63             None, note=note, headers=headers, data=data)
  64
  65     def _refresh_token(self):
  66         if not JioCinemaBaseIE._REFRESH_TOKEN or not JioCinemaBaseIE._DEVICE_ID:
  67             raise ExtractorError('User token has expired', expected=True)
  68         response = self._call_auth_api(
  69             'token', 'refreshtoken', 'Refreshing token',
  70             headers={'accesstoken': self._ACCESS_TOKEN}, data={
  71                 **self._APP_NAME,
  72                 'deviceId': self._DEVICE_ID,
  73                 'refreshToken': self._REFRESH_TOKEN,
  74                 **self._APP_VERSION,
  75             })
  76         refresh_token = response.get('refreshTokenId')
  77         if refresh_token and refresh_token != JioCinemaBaseIE._REFRESH_TOKEN:
  78             JioCinemaBaseIE._REFRESH_TOKEN = refresh_token
  79             self._cache_token('refresh')
  80         JioCinemaBaseIE._ACCESS_TOKEN = response['authToken']
  81         self._cache_token('access')
  82
  83     def _fetch_guest_token(self):
  84         JioCinemaBaseIE._DEVICE_ID = ''.join(random.choices(string.digits, k=10))
  85         guest_token = self._call_auth_api(
  86             'token', 'guest', 'Downloading guest token', data={
  87                 **self._APP_NAME,
  88                 'deviceType': 'phone',
  89                 'os': 'ios',
  90                 'deviceId': self._DEVICE_ID,
  91                 'freshLaunch': False,
  92                 'adId': self._DEVICE_ID,
  93                 **self._APP_VERSION,
  94             })
  95         self._GUEST_TOKEN = guest_token['authToken']
  96         self._USER_ID = guest_token['userId']
  97
  98     def _call_login_api(self, endpoint, guest_token, data, note):
  99         return self._call_auth_api(
 100             'user', f'loginotp/{endpoint}', note, headers={
 101                 **self.geo_verification_headers(),
 102                 'accesstoken': self._GUEST_TOKEN,
 103                 **self._APP_NAME,
 104                 **traverse_obj(guest_token, 'data', {
 105                     'deviceType': ('deviceType', {str}),
 106                     'os': ('os', {str}),
 107                 })}, data=data)
 108
 109     def _is_token_expired(self, token):
 110         return (try_call(lambda: jwt_decode_hs256(token)['exp']) or 0) <= int(time.time() - 180)
 111
 112     def _perform_login(self, username, password):
 113         if self._ACCESS_TOKEN and not self._is_token_expired(self._ACCESS_TOKEN):
 114             return
 115
 116         UUID_RE = r'[\da-f]{8}-(?:[\da-f]{4}-){3}[\da-f]{12}'
 117
 118         if username.lower() == 'token':
 119             if try_call(lambda: jwt_decode_hs256(password)):
 120                 JioCinemaBaseIE._ACCESS_TOKEN = password
 121                 refresh_hint = 'the `refreshToken` UUID from your browser local storage'
 122                 refresh_token = self._configuration_arg('refresh_token', [''], ie_key=JioCinemaIE)[0]
 123                 if not refresh_token:
 124                     self.to_screen(
 125                         'To extend the life of your login session, in addition to your access token, '
 126                         'you can pass --extractor-args "jiocinema:refresh_token=REFRESH_TOKEN" '
 127                         f'where REFRESH_TOKEN is {refresh_hint}')
 128                 elif re.fullmatch(UUID_RE, refresh_token):
 129                     JioCinemaBaseIE._REFRESH_TOKEN = refresh_token
 130                 else:
 131                     self.report_warning(f'Invalid refresh_token value. Use {refresh_hint}')
 132             else:
 133                 raise ExtractorError(
 134                     f'The password given could not be decoded as a token; use {self._ACCESS_HINT}', expected=True)
 135
 136         elif username.lower() == 'device' and re.fullmatch(rf'(?:{UUID_RE}|\d+)', password):
 137             JioCinemaBaseIE._REFRESH_TOKEN = self.cache.load(JioCinemaBaseIE._NETRC_MACHINE, f'{password}-refresh')
 138             JioCinemaBaseIE._ACCESS_TOKEN = self.cache.load(JioCinemaBaseIE._NETRC_MACHINE, f'{password}-access')
 139             if not JioCinemaBaseIE._REFRESH_TOKEN or not JioCinemaBaseIE._ACCESS_TOKEN:
 140                 raise ExtractorError(f'Failed to load cached tokens for device ID "{password}"', expected=True)
 141
 142         elif username.lower() == 'phone' and re.fullmatch(r'\+?\d+', password):
 143             self._fetch_guest_token()
 144             guest_token = jwt_decode_hs256(self._GUEST_TOKEN)
 145             initial_data = {
 146                 'number': base64.b64encode(password.encode()).decode(),
 147                 **self._APP_VERSION,
 148             }
 149             response = self._call_login_api('send', guest_token, initial_data, 'Requesting OTP')
 150             if not traverse_obj(response, ('OTPInfo', {dict})):
 151                 raise ExtractorError('There was a problem with the phone number login attempt')
 152
 153             is_iphone = guest_token.get('os') == 'ios'
 154             response = self._call_login_api('verify', guest_token, {
 155                 'deviceInfo': {
 156                     'consumptionDeviceName': 'iPhone' if is_iphone else 'Android',
 157                     'info': {
 158                         'platform': {'name': 'iPhone OS' if is_iphone else 'Android'},
 159                         'androidId': self._DEVICE_ID,
 160                         'type': 'iOS' if is_iphone else 'Android',
 161                     },
 162                 },
 163                 **initial_data,
 164                 'otp': self._get_tfa_info('the one-time password sent to your phone'),
 165             }, 'Submitting OTP')
 166             if traverse_obj(response, 'code') == 1043:
 167                 raise ExtractorError('Wrong OTP', expected=True)
 168             JioCinemaBaseIE._REFRESH_TOKEN = response['refreshToken']
 169             JioCinemaBaseIE._ACCESS_TOKEN = response['authToken']
 170
 171         else:
 172             raise ExtractorError(self._LOGIN_HINT, expected=True)
 173
 174         user_token = jwt_decode_hs256(JioCinemaBaseIE._ACCESS_TOKEN)['data']
 175         JioCinemaBaseIE._USER_ID = user_token['userId']
 176         JioCinemaBaseIE._DEVICE_ID = user_token['deviceId']
 177         if JioCinemaBaseIE._REFRESH_TOKEN and username != 'device':
 178             self._cache_token('all')
 179             if self.get_param('cachedir') is not False:
 180                 self.to_screen(
 181                     f'NOTE: For subsequent logins you can use "-u device -p {JioCinemaBaseIE._DEVICE_ID}"')
 182         elif not JioCinemaBaseIE._REFRESH_TOKEN:
 183             JioCinemaBaseIE._REFRESH_TOKEN = self.cache.load(
 184                 JioCinemaBaseIE._NETRC_MACHINE, f'{JioCinemaBaseIE._DEVICE_ID}-refresh')
 185             if JioCinemaBaseIE._REFRESH_TOKEN:
 186                 self._cache_token('access')
 187         self.to_screen(f'Logging in as device ID "{JioCinemaBaseIE._DEVICE_ID}"')
 188         if self._is_token_expired(JioCinemaBaseIE._ACCESS_TOKEN):
 189             self._refresh_token()
 190
 191
 192 class JioCinemaIE(JioCinemaBaseIE):
 193     IE_NAME = 'jiocinema'
 194     _VALID_URL = r'https?://(?:www\.)?jiocinema\.com/?(?:movies?/[^/?#]+/|tv-shows/(?:[^/?#]+/){3})(?P<id>\d{3,})'
 195     _TESTS = [{
 196         'url': 'https://www.jiocinema.com/tv-shows/agnisakshi-ek-samjhauta/1/pradeep-to-stop-the-wedding/3759931',
 197         'info_dict': {
 198             'id': '3759931',
 199             'ext': 'mp4',
 200             'title': 'Pradeep to stop the wedding?',
 201             'description': 'md5:75f72d1d1a66976633345a3de6d672b1',
 202             'episode': 'Pradeep to stop the wedding?',
 203             'episode_number': 89,
 204             'season': 'Agnisakshi…Ek Samjhauta-S1',
 205             'season_number': 1,
 206             'series': 'Agnisakshi Ek Samjhauta',
 207             'duration': 1238.0,
 208             'thumbnail': r're:https?://.+\.jpg',
 209             'age_limit': 13,
 210             'season_id': '3698031',
 211             'upload_date': '20230606',
 212             'timestamp': 1686009600,
 213             'release_date': '20230607',
 214             'genres': ['Drama'],
 215         },
 216         'params': {'skip_download': 'm3u8'},
 217     }, {
 218         'url': 'https://www.jiocinema.com/movies/bhediya/3754021/watch',
 219         'info_dict': {
 220             'id': '3754021',
 221             'ext': 'mp4',
 222             'title': 'Bhediya',
 223             'description': 'md5:a6bf2900371ac2fc3f1447401a9f7bb0',
 224             'episode': 'Bhediya',
 225             'duration': 8500.0,
 226             'thumbnail': r're:https?://.+\.jpg',
 227             'age_limit': 13,
 228             'upload_date': '20230525',
 229             'timestamp': 1685026200,
 230             'release_date': '20230524',
 231             'genres': ['Comedy'],
 232         },
 233         'params': {'skip_download': 'm3u8'},
 234     }]
 235
 236     def _extract_formats_and_subtitles(self, playback, video_id):
 237         m3u8_url = traverse_obj(playback, (
 238             'data', 'playbackUrls', lambda _, v: v['streamtype'] == 'hls', 'url', {url_or_none}, any))
 239         if not m3u8_url:  # DRM-only content only serves dash urls
 240             self.report_drm(video_id)
 241         formats, subtitles = self._extract_m3u8_formats_and_subtitles(m3u8_url, video_id, m3u8_id='hls')
 242         self._remove_duplicate_formats(formats)
 243
 244         return {
 245             # '/_definst_/smil:vod/' m3u8 manifests claim to have 720p+ formats but max out at 480p
 246             'formats': traverse_obj(formats, (
 247                 lambda _, v: '/_definst_/smil:vod/' not in v['url'] or v['height'] <= 480)),
 248             'subtitles': subtitles,
 249         }
 250
 251     def _real_extract(self, url):
 252         video_id = self._match_id(url)
 253         if not self._ACCESS_TOKEN and self._is_token_expired(self._GUEST_TOKEN):
 254             self._fetch_guest_token()
 255         elif self._ACCESS_TOKEN and self._is_token_expired(self._ACCESS_TOKEN):
 256             self._refresh_token()
 257
 258         playback = self._call_api(
 259             f'https://apis-jiovoot.voot.com/playbackjv/v3/{video_id}', video_id,
 260             'Downloading playback JSON', headers={
 261                 **self.geo_verification_headers(),
 262                 'accesstoken': self._ACCESS_TOKEN or self._GUEST_TOKEN,
 263                 **self._APP_NAME,
 264                 'deviceid': self._DEVICE_ID,
 265                 'uniqueid': self._USER_ID,
 266                 'x-apisignatures': self._API_SIGNATURES,
 267                 'x-platform': 'androidweb',
 268                 'x-platform-token': 'web',
 269             }, data={
 270                 '4k': False,
 271                 'ageGroup': '18+',
 272                 'appVersion': '3.4.0',
 273                 'bitrateProfile': 'xhdpi',
 274                 'capability': {
 275                     'drmCapability': {
 276                         'aesSupport': 'yes',
 277                         'fairPlayDrmSupport': 'none',
 278                         'playreadyDrmSupport': 'none',
 279                         'widevineDRMSupport': 'none',
 280                     },
 281                     'frameRateCapability': [{
 282                         'frameRateSupport': '30fps',
 283                         'videoQuality': '1440p',
 284                     }],
 285                 },
 286                 'continueWatchingRequired': False,
 287                 'dolby': False,
 288                 'downloadRequest': False,
 289                 'hevc': False,
 290                 'kidsSafe': False,
 291                 'manufacturer': 'Windows',
 292                 'model': 'Windows',
 293                 'multiAudioRequired': True,
 294                 'osVersion': '10',
 295                 'parentalPinValid': True,
 296                 'x-apisignatures': self._API_SIGNATURES,
 297             })
 298
 299         status_code = traverse_obj(playback, ('code', {int}))
 300         if status_code == 474:
 301             self.raise_geo_restricted(countries=['IN'])
 302         elif status_code == 1008:
 303             error_msg = 'This content is only available for premium users'
 304             if self._ACCESS_TOKEN:
 305                 raise ExtractorError(error_msg, expected=True)
 306             self.raise_login_required(f'{error_msg}. {self._LOGIN_HINT}', method=None)
 307         elif status_code == 400:
 308             raise ExtractorError('The requested content is not available', expected=True)
 309         elif status_code is not None and status_code != 200:
 310             raise ExtractorError(
 311                 f'JioCinema says: {traverse_obj(playback, ("message", {str})) or status_code}')
 312
 313         metadata = self._download_json(
 314             f'{self._METADATA_API_BASE}/voot/v1/voot-web/content/query/asset-details',
 315             video_id, fatal=False, query={
 316                 'ids': f'include:{video_id}',
 317                 'responseType': 'common',
 318                 'devicePlatformType': 'desktop',
 319             })
 320
 321         return {
 322             'id': video_id,
 323             'http_headers': self._API_HEADERS,
 324             **self._extract_formats_and_subtitles(playback, video_id),
 325             **traverse_obj(playback, ('data', {
 326                 # fallback metadata
 327                 'title': ('name', {str}),
 328                 'description': ('fullSynopsis', {str}),
 329                 'series': ('show', 'name', {str}, filter),
 330                 'season': ('tournamentName', {str}, {lambda x: x if x != 'Season 0' else None}),
 331                 'season_number': ('episode', 'season', {int_or_none}, filter),
 332                 'episode': ('fullTitle', {str}),
 333                 'episode_number': ('episode', 'episodeNo', {int_or_none}, filter),
 334                 'age_limit': ('ageNemonic', {parse_age_limit}),
 335                 'duration': ('totalDuration', {float_or_none}),
 336                 'thumbnail': ('images', {url_or_none}),
 337             })),
 338             **traverse_obj(metadata, ('result', 0, {
 339                 'title': ('fullTitle', {str}),
 340                 'description': ('fullSynopsis', {str}),
 341                 'series': ('showName', {str}, filter),
 342                 'season': ('seasonName', {str}, filter),
 343                 'season_number': ('season', {int_or_none}),
 344                 'season_id': ('seasonId', {str}, filter),
 345                 'episode': ('fullTitle', {str}),
 346                 'episode_number': ('episode', {int_or_none}),
 347                 'timestamp': ('uploadTime', {int_or_none}),
 348                 'release_date': ('telecastDate', {str}),
 349                 'age_limit': ('ageNemonic', {parse_age_limit}),
 350                 'duration': ('duration', {float_or_none}),
 351                 'genres': ('genres', ..., {str}),
 352                 'thumbnail': ('seo', 'ogImage', {url_or_none}),
 353             })),
 354         }
 355
 356
 357 class JioCinemaSeriesIE(JioCinemaBaseIE):
 358     IE_NAME = 'jiocinema:series'
 359     _VALID_URL = r'https?://(?:www\.)?jiocinema\.com/tv-shows/(?P<slug>[\w-]+)/(?P<id>\d{3,})'
 360     _TESTS = [{
 361         'url': 'https://www.jiocinema.com/tv-shows/naagin/3499917',
 362         'info_dict': {
 363             'id': '3499917',
 364             'title': 'naagin',
 365         },
 366         'playlist_mincount': 120,
 367     }, {
 368         'url': 'https://www.jiocinema.com/tv-shows/mtv-splitsvilla-x5/3499820',
 369         'info_dict': {
 370             'id': '3499820',
 371             'title': 'mtv-splitsvilla-x5',
 372         },
 373         'playlist_mincount': 310,
 374     }]
 375
 376     def _entries(self, series_id):
 377         seasons = traverse_obj(self._download_json(
 378             f'{self._METADATA_API_BASE}/voot/v1/voot-web/view/show/{series_id}', series_id,
 379             'Downloading series metadata JSON', query={'responseType': 'common'}), (
 380             'trays', lambda _, v: v['trayId'] == 'season-by-show-multifilter',
 381             'trayTabs', lambda _, v: v['id']))
 382
 383         for season_num, season in enumerate(seasons, start=1):
 384             season_id = season['id']
 385             label = season.get('label') or season_num
 386             for page_num in itertools.count(1):
 387                 episodes = traverse_obj(self._download_json(
 388                     f'{self._METADATA_API_BASE}/voot/v1/voot-web/content/generic/series-wise-episode',
 389                     season_id, f'Downloading season {label} page {page_num} JSON', query={
 390                         'sort': 'episode:asc',
 391                         'id': season_id,
 392                         'responseType': 'common',
 393                         'page': page_num,
 394                     }), ('result', lambda _, v: v['id'] and url_or_none(v['slug'])))
 395                 if not episodes:
 396                     break
 397                 for episode in episodes:
 398                     yield self.url_result(
 399                         episode['slug'], JioCinemaIE, **traverse_obj(episode, {
 400                             'video_id': 'id',
 401                             'video_title': ('fullTitle', {str}),
 402                             'season_number': ('season', {int_or_none}),
 403                             'episode_number': ('episode', {int_or_none}),
 404                         }))
 405
 406     def _real_extract(self, url):
 407         slug, series_id = self._match_valid_url(url).group('slug', 'id')
 408         return self.playlist_result(self._entries(series_id), series_id, slug)