yt_dlp/extractor/zee5.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_str
   5 from ..utils import (
   6     ExtractorError,
   7     int_or_none,
   8     parse_age_limit,
   9     str_or_none,
  10     try_get,
  11     unified_strdate,
  12     unified_timestamp,
  13     url_or_none,
  14 )
  15
  16
  17 class Zee5IE(InfoExtractor):
  18     _VALID_URL = r'''(?x)
  19                      (?:
  20                         zee5:|
  21                         https?://(?:www\.)?zee5\.com/(?:[^#?]+/)?
  22                         (?:
  23                             (?:tv-shows|kids|web-series|zee5originals)(?:/[^#/?]+){3}
  24                             |movies/[^#/?]+
  25                         )/(?P<display_id>[^#/?]+)/
  26                      )
  27                      (?P<id>[^#/?]+)/?(?:$|[?#])
  28                      '''
  29     _TESTS = [{
  30         'url': 'https://www.zee5.com/movies/details/adavari-matalaku-ardhale-verule/0-0-movie_1143162669',
  31         'info_dict': {
  32             'id': '0-0-movie_1143162669',
  33             'ext': 'mp4',
  34             'display_id': 'adavari-matalaku-ardhale-verule',
  35             'title': 'Adavari Matalaku Ardhale Verule',
  36             'duration': 9360,
  37             'description': compat_str,
  38             'alt_title': 'Adavari Matalaku Ardhale Verule',
  39             'uploader': 'Zee Entertainment Enterprises Ltd',
  40             'release_date': '20070427',
  41             'upload_date': '20070427',
  42             'timestamp': 1177632000,
  43             'thumbnail': r're:^https?://.*\.jpg$',
  44             'episode_number': 0,
  45             'episode': 'Episode 0',
  46             'tags': list
  47         },
  48         'params': {
  49             'format': 'bv',
  50         },
  51     }, {
  52         'url': 'https://www.zee5.com/kids/kids-shows/bandbudh-aur-budbak/0-6-1899/yoga-se-hoga-bandbudh-aur-budbak/0-1-239839',
  53         'info_dict': {
  54             'id': '0-1-239839',
  55             'ext': 'mp4',
  56             'display_id': 'yoga-se-hoga-bandbudh-aur-budbak',
  57             'title': 'Yoga Se Hoga-Bandbudh aur Budbak',
  58             'duration': 659,
  59             'description': compat_str,
  60             'alt_title': 'Yoga Se Hoga-Bandbudh aur Budbak',
  61             'uploader': 'Zee Entertainment Enterprises Ltd',
  62             'release_date': '20150101',
  63             'upload_date': '20150101',
  64             'timestamp': 1420070400,
  65             'thumbnail': r're:^https?://.*\.jpg$',
  66             'series': 'Bandbudh Aur Budbak',
  67             'season_number': 1,
  68             'episode_number': 1,
  69             'episode': 'Episode 1',
  70             'season': 'Season 1',
  71             'tags': list,
  72         },
  73         'params': {
  74             'format': 'bv',
  75         },
  76     }, {
  77         'url': 'https://www.zee5.com/hi/tv-shows/details/kundali-bhagya/0-6-366/kundali-bhagya-march-08-2021/0-1-manual_7g9jv1os7730?country=IN',
  78         'only_matching': True
  79     }, {
  80         'url': 'https://www.zee5.com/global/hi/tv-shows/details/kundali-bhagya/0-6-366/kundali-bhagya-march-08-2021/0-1-manual_7g9jv1os7730',
  81         'only_matching': True
  82     }, {
  83         'url': 'https://www.zee5.com/web-series/details/mithya/0-6-4z587408/maine-dekhi-hai-uski-mrityu/0-1-6z587412',
  84         'only_matching': True
  85     }]
  86     _DETAIL_API_URL = 'https://spapi.zee5.com/singlePlayback/getDetails/secure?content_id={}&device_id={}&platform_name=desktop_web&country=IN&check_parental_control=false'
  87     _DEVICE_ID = '1q70TH8Wz0wTyw4buVgg000000000000'
  88     _USER_TOKEN = None
  89     _LOGIN_HINT = 'Use "--username <mobile_number>" to login using otp or "--username token" and "--password <user_token>" to login using user token.'
  90     _NETRC_MACHINE = 'zee5'
  91     _GEO_COUNTRIES = ['IN']
  92
  93     def _perform_login(self, username, password):
  94         if len(username) == 10 and username.isdigit() and self._USER_TOKEN is None:
  95             self.report_login()
  96             otp_request_json = self._download_json(f'https://b2bapi.zee5.com/device/sendotp_v1.php?phoneno=91{username}',
  97                                                    None, note='Sending OTP')
  98             if otp_request_json['code'] == 0:
  99                 self.to_screen(otp_request_json['message'])
 100             else:
 101                 raise ExtractorError(otp_request_json['message'], expected=True)
 102             otp_code = self._get_tfa_info('OTP')
 103             otp_verify_json = self._download_json(f'https://b2bapi.zee5.com/device/verifyotp_v1.php?phoneno=91{username}&otp={otp_code}&guest_token={self._DEVICE_ID}&platform=web',
 104                                                   None, note='Verifying OTP', fatal=False)
 105             if not otp_verify_json:
 106                 raise ExtractorError('Unable to verify OTP.', expected=True)
 107             self._USER_TOKEN = otp_verify_json.get('token')
 108             if not self._USER_TOKEN:
 109                 raise ExtractorError(otp_request_json['message'], expected=True)
 110         elif username.lower() == 'token' and len(password) > 1198:
 111             self._USER_TOKEN = password
 112         else:
 113             raise ExtractorError(self._LOGIN_HINT, expected=True)
 114
 115     def _real_extract(self, url):
 116         video_id, display_id = self._match_valid_url(url).group('id', 'display_id')
 117         access_token_request = self._download_json(
 118             'https://useraction.zee5.com/token/platform_tokens.php?platform_name=web_app',
 119             video_id, note='Downloading access token')
 120         data = {
 121             'x-access-token': access_token_request['token']
 122         }
 123         if self._USER_TOKEN:
 124             data['Authorization'] = 'bearer %s' % self._USER_TOKEN
 125         else:
 126             data['X-Z5-Guest-Token'] = self._DEVICE_ID
 127
 128         json_data = self._download_json(
 129             self._DETAIL_API_URL.format(video_id, self._DEVICE_ID),
 130             video_id, headers={'content-type': 'application/json'}, data=json.dumps(data).encode('utf-8'))
 131         asset_data = json_data['assetDetails']
 132         show_data = json_data.get('showDetails', {})
 133         if 'premium' in asset_data['business_type']:
 134             raise ExtractorError('Premium content is DRM protected.', expected=True)
 135         if not asset_data.get('hls_url'):
 136             self.raise_login_required(self._LOGIN_HINT, metadata_available=True, method=None)
 137         formats, m3u8_subs = self._extract_m3u8_formats_and_subtitles(asset_data['hls_url'], video_id, 'mp4', fatal=False)
 138         self._sort_formats(formats)
 139
 140         subtitles = {}
 141         for sub in asset_data.get('subtitle_url', []):
 142             sub_url = sub.get('url')
 143             if not sub_url:
 144                 continue
 145             subtitles.setdefault(sub.get('language', 'en'), []).append({
 146                 'url': self._proto_relative_url(sub_url),
 147             })
 148         subtitles = self._merge_subtitles(subtitles, m3u8_subs)
 149         return {
 150             'id': video_id,
 151             'display_id': display_id,
 152             'title': asset_data['title'],
 153             'formats': formats,
 154             'subtitles': subtitles,
 155             'duration': int_or_none(asset_data.get('duration')),
 156             'description': str_or_none(asset_data.get('description')),
 157             'alt_title': str_or_none(asset_data.get('original_title')),
 158             'uploader': str_or_none(asset_data.get('content_owner')),
 159             'age_limit': parse_age_limit(asset_data.get('age_rating')),
 160             'release_date': unified_strdate(asset_data.get('release_date')),
 161             'timestamp': unified_timestamp(asset_data.get('release_date')),
 162             'thumbnail': url_or_none(asset_data.get('image_url')),
 163             'series': str_or_none(asset_data.get('tvshow_name')),
 164             'season': try_get(show_data, lambda x: x['seasons']['title'], str),
 165             'season_number': int_or_none(try_get(show_data, lambda x: x['seasons'][0]['orderid'])),
 166             'episode_number': int_or_none(try_get(asset_data, lambda x: x['orderid'])),
 167             'tags': try_get(asset_data, lambda x: x['tags'], list)
 168         }
 169
 170
 171 class Zee5SeriesIE(InfoExtractor):
 172     IE_NAME = 'zee5:series'
 173     _VALID_URL = r'''(?x)
 174                      (?:
 175                         zee5:series:|
 176                         https?://(?:www\.)?zee5\.com/(?:[^#?]+/)?
 177                         (?:tv-shows|web-series|kids|zee5originals)(?:/[^#/?]+){2}/
 178                      )
 179                      (?P<id>[^#/?]+)(?:/episodes)?/?(?:$|[?#])
 180                      '''
 181     _TESTS = [{
 182         'url': 'https://www.zee5.com/kids/kids-shows/bandbudh-aur-budbak/0-6-1899',
 183         'playlist_mincount': 156,
 184         'info_dict': {
 185             'id': '0-6-1899',
 186         },
 187     }, {
 188         'url': 'https://www.zee5.com/tv-shows/details/bhabi-ji-ghar-par-hai/0-6-199',
 189         'playlist_mincount': 1500,
 190         'info_dict': {
 191             'id': '0-6-199',
 192         },
 193     }, {
 194         'url': 'https://www.zee5.com/tv-shows/details/agent-raghav-crime-branch/0-6-965',
 195         'playlist_mincount': 24,
 196         'info_dict': {
 197             'id': '0-6-965',
 198         },
 199     }, {
 200         'url': 'https://www.zee5.com/ta/tv-shows/details/nagabhairavi/0-6-3201',
 201         'playlist_mincount': 3,
 202         'info_dict': {
 203             'id': '0-6-3201',
 204         },
 205     }, {
 206         'url': 'https://www.zee5.com/global/hi/tv-shows/details/khwaabon-ki-zamin-par/0-6-270',
 207         'playlist_mincount': 150,
 208         'info_dict': {
 209             'id': '0-6-270',
 210         },
 211     }, {
 212         'url': 'https://www.zee5.com/tv-shows/details/chala-hawa-yeu-dya-ladies-zindabaad/0-6-2943/episodes',
 213         'only_matching': True,
 214     }, {
 215         'url': 'https://www.zee5.com/web-series/details/mithya/0-6-4z587408',
 216         'only_matching': True,
 217     }]
 218
 219     def _entries(self, show_id):
 220         access_token_request = self._download_json(
 221             'https://useraction.zee5.com/token/platform_tokens.php?platform_name=web_app',
 222             show_id, note='Downloading access token')
 223         headers = {
 224             'X-Access-Token': access_token_request['token'],
 225             'Referer': 'https://www.zee5.com/',
 226         }
 227         show_url = f'https://gwapi.zee5.com/content/tvshow/{show_id}?translation=en&country=IN'
 228
 229         page_num = 0
 230         show_json = self._download_json(show_url, video_id=show_id, headers=headers)
 231         for season in show_json.get('seasons') or []:
 232             season_id = try_get(season, lambda x: x['id'], compat_str)
 233             next_url = f'https://gwapi.zee5.com/content/tvshow/?season_id={season_id}&type=episode&translation=en&country=IN&on_air=false&asset_subtype=tvshow&page=1&limit=100'
 234             while next_url:
 235                 page_num += 1
 236                 episodes_json = self._download_json(
 237                     next_url, video_id=show_id, headers=headers,
 238                     note='Downloading JSON metadata page %d' % page_num)
 239                 for episode in try_get(episodes_json, lambda x: x['episode'], list) or []:
 240                     video_id = episode.get('id')
 241                     yield self.url_result(
 242                         'zee5:%s' % video_id,
 243                         ie=Zee5IE.ie_key(), video_id=video_id)
 244                 next_url = url_or_none(episodes_json.get('next_episode_api'))
 245
 246     def _real_extract(self, url):
 247         show_id = self._match_id(url)
 248         return self.playlist_result(self._entries(show_id), playlist_id=show_id)