yt_dlp/extractor/zee5.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5
   6 from .common import InfoExtractor
   7 from ..compat import compat_str
   8 from ..utils import (
   9     ExtractorError,
  10     int_or_none,
  11     parse_age_limit,
  12     str_or_none,
  13     try_get,
  14     unified_strdate,
  15     unified_timestamp,
  16     url_or_none,
  17 )
  18
  19
  20 class Zee5IE(InfoExtractor):
  21     _VALID_URL = r'''(?x)
  22                      (?:
  23                         zee5:|
  24                         https?://(?:www\.)?zee5\.com/(?:[^#?]+/)?
  25                         (?:
  26                             (?:tv-shows|kids|zee5originals)(?:/[^#/?]+){3}
  27                             |movies/[^#/?]+
  28                         )/(?P<display_id>[^#/?]+)/
  29                      )
  30                      (?P<id>[^#/?]+)/?(?:$|[?#])
  31                      '''
  32     _TESTS = [{
  33         'url': 'https://www.zee5.com/movies/details/krishna-the-birth/0-0-63098',
  34         'info_dict': {
  35             'id': '0-0-63098',
  36             'ext': 'mp4',
  37             'display_id': 'krishna-the-birth',
  38             'title': 'Krishna - The Birth',
  39             'duration': 4368,
  40             'description': compat_str,
  41             'alt_title': 'Krishna - The Birth',
  42             'uploader': 'Zee Entertainment Enterprises Ltd',
  43             'release_date': '20060101',
  44             'upload_date': '20060101',
  45             'timestamp': 1136073600,
  46             'thumbnail': r're:^https?://.*\.jpg$',
  47             'episode_number': 0,
  48             'episode': 'Episode 0',
  49             'tags': list
  50         },
  51         'params': {
  52             'format': 'bv',
  53         },
  54     }, {
  55         'url': 'https://www.zee5.com/kids/kids-shows/bandbudh-aur-budbak/0-6-1899/yoga-se-hoga-bandbudh-aur-budbak/0-1-239839',
  56         'info_dict': {
  57             'id': '0-1-239839',
  58             'ext': 'mp4',
  59             'display_id': 'yoga-se-hoga-bandbudh-aur-budbak',
  60             'title': 'Yoga Se Hoga-Bandbudh aur Budbak',
  61             'duration': 659,
  62             'description': compat_str,
  63             'alt_title': 'Yoga Se Hoga-Bandbudh aur Budbak',
  64             'uploader': 'Zee Entertainment Enterprises Ltd',
  65             'release_date': '20150101',
  66             'upload_date': '20150101',
  67             'timestamp': 1420070400,
  68             'thumbnail': r're:^https?://.*\.jpg$',
  69             'series': 'Bandbudh Aur Budbak',
  70             'season_number': 1,
  71             'episode_number': 1,
  72             'episode': 'Episode 1',
  73             'season': 'Season 1',
  74             'tags': list,
  75         },
  76         'params': {
  77             'format': 'bv',
  78         },
  79     }, {
  80         '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',
  81         'only_matching': True
  82     }, {
  83         'url': 'https://www.zee5.com/global/hi/tv-shows/details/kundali-bhagya/0-6-366/kundali-bhagya-march-08-2021/0-1-manual_7g9jv1os7730',
  84         'only_matching': True
  85     }]
  86     _DETAIL_API_URL = 'https://spapi.zee5.com/singlePlayback/getDetails?content_id={}&device_id={}&platform_name=desktop_web&country=IN&check_parental_control=false'
  87     _DEVICE_ID = 'iIxsxYf40cqO3koIkwzKHZhnJzHN13zb'
  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 _login(self):
  94         username, password = self._get_login_info()
  95         if username:
  96             if len(username) == 10 and username.isdigit() and self._USER_TOKEN is None:
  97                 self.report_login()
  98                 otp_request_json = self._download_json('https://b2bapi.zee5.com/device/sendotp_v1.php?phoneno=91{}'.format(username),
  99                                                        None, note='Sending OTP')
 100                 if otp_request_json['code'] == 0:
 101                     self.to_screen(otp_request_json['message'])
 102                 else:
 103                     raise ExtractorError(otp_request_json['message'], expected=True)
 104                 otp_code = self._get_tfa_info('OTP')
 105                 otp_verify_json = self._download_json('https://b2bapi.zee5.com/device/verifyotp_v1.php?phoneno=91{}&otp={}&guest_token={}&platform=web'.format(username, otp_code, self._DEVICE_ID),
 106                                                       None, note='Verifying OTP', fatal=False)
 107                 if not otp_verify_json:
 108                     raise ExtractorError('Unable to verify OTP.', expected=True)
 109                 self._USER_TOKEN = otp_verify_json.get('token')
 110                 if not self._USER_TOKEN:
 111                     raise ExtractorError(otp_request_json['message'], expected=True)
 112             elif username.lower() == 'token' and len(password) > 1198:
 113                 self._USER_TOKEN = password
 114             else:
 115                 raise ExtractorError(self._LOGIN_HINT, expected=True)
 116
 117     def _real_initialize(self):
 118         self._login()
 119
 120     def _real_extract(self, url):
 121         video_id, display_id = self._match_valid_url(url).group('id', 'display_id')
 122         access_token_request = self._download_json(
 123             'https://useraction.zee5.com/token/platform_tokens.php?platform_name=web_app',
 124             video_id, note='Downloading access token')
 125         data = {
 126             'x-access-token': access_token_request['token']
 127         }
 128         if self._USER_TOKEN:
 129             data['Authorization'] = 'bearer %s' % self._USER_TOKEN
 130         else:
 131             data['X-Z5-Guest-Token'] = self._DEVICE_ID
 132
 133         json_data = self._download_json(
 134             self._DETAIL_API_URL.format(video_id, self._DEVICE_ID),
 135             video_id, headers={'content-type': 'application/json'}, data=json.dumps(data).encode('utf-8'))
 136         asset_data = json_data['assetDetails']
 137         show_data = json_data.get('showDetails', {})
 138         if 'premium' in asset_data['business_type']:
 139             raise ExtractorError('Premium content is DRM protected.', expected=True)
 140         if not asset_data.get('hls_url'):
 141             self.raise_login_required(self._LOGIN_HINT, metadata_available=True, method=None)
 142         formats, m3u8_subs = self._extract_m3u8_formats_and_subtitles(asset_data['hls_url'], video_id, 'mp4', fatal=False)
 143         self._sort_formats(formats)
 144
 145         subtitles = {}
 146         for sub in asset_data.get('subtitle_url', []):
 147             sub_url = sub.get('url')
 148             if not sub_url:
 149                 continue
 150             subtitles.setdefault(sub.get('language', 'en'), []).append({
 151                 'url': self._proto_relative_url(sub_url),
 152             })
 153         subtitles = self._merge_subtitles(subtitles, m3u8_subs)
 154         return {
 155             'id': video_id,
 156             'display_id': display_id,
 157             'title': asset_data['title'],
 158             'formats': formats,
 159             'subtitles': subtitles,
 160             'duration': int_or_none(asset_data.get('duration')),
 161             'description': str_or_none(asset_data.get('description')),
 162             'alt_title': str_or_none(asset_data.get('original_title')),
 163             'uploader': str_or_none(asset_data.get('content_owner')),
 164             'age_limit': parse_age_limit(asset_data.get('age_rating')),
 165             'release_date': unified_strdate(asset_data.get('release_date')),
 166             'timestamp': unified_timestamp(asset_data.get('release_date')),
 167             'thumbnail': url_or_none(asset_data.get('image_url')),
 168             'series': str_or_none(asset_data.get('tvshow_name')),
 169             'season': try_get(show_data, lambda x: x['seasons']['title'], str),
 170             'season_number': int_or_none(try_get(show_data, lambda x: x['seasons'][0]['orderid'])),
 171             'episode_number': int_or_none(try_get(asset_data, lambda x: x['orderid'])),
 172             'tags': try_get(asset_data, lambda x: x['tags'], list)
 173         }
 174
 175
 176 class Zee5SeriesIE(InfoExtractor):
 177     IE_NAME = 'zee5:series'
 178     _VALID_URL = r'''(?x)
 179                      (?:
 180                         zee5:series:|
 181                         https?://(?:www\.)?zee5\.com/(?:[^#?]+/)?
 182                         (?:tv-shows|kids|zee5originals)(?:/[^#/?]+){2}/
 183                      )
 184                      (?P<id>[^#/?]+)(?:/episodes)?/?(?:$|[?#])
 185                      '''
 186     _TESTS = [{
 187         'url': 'https://www.zee5.com/kids/kids-shows/bandbudh-aur-budbak/0-6-1899',
 188         'playlist_mincount': 156,
 189         'info_dict': {
 190             'id': '0-6-1899',
 191         },
 192     }, {
 193         'url': 'https://www.zee5.com/tv-shows/details/bhabi-ji-ghar-par-hai/0-6-199',
 194         'playlist_mincount': 1500,
 195         'info_dict': {
 196             'id': '0-6-199',
 197         },
 198     }, {
 199         'url': 'https://www.zee5.com/tv-shows/details/agent-raghav-crime-branch/0-6-965',
 200         'playlist_mincount': 24,
 201         'info_dict': {
 202             'id': '0-6-965',
 203         },
 204     }, {
 205         'url': 'https://www.zee5.com/ta/tv-shows/details/nagabhairavi/0-6-3201',
 206         'playlist_mincount': 3,
 207         'info_dict': {
 208             'id': '0-6-3201',
 209         },
 210     }, {
 211         'url': 'https://www.zee5.com/global/hi/tv-shows/details/khwaabon-ki-zamin-par/0-6-270',
 212         'playlist_mincount': 150,
 213         'info_dict': {
 214             'id': '0-6-270',
 215         },
 216     }, {
 217         'url': 'https://www.zee5.com/tv-shows/details/chala-hawa-yeu-dya-ladies-zindabaad/0-6-2943/episodes',
 218         'only_matching': True,
 219     }]
 220
 221     def _entries(self, show_id):
 222         access_token_request = self._download_json(
 223             'https://useraction.zee5.com/token/platform_tokens.php?platform_name=web_app',
 224             show_id, note='Downloading access token')
 225         headers = {
 226             'X-Access-Token': access_token_request['token'],
 227             'Referer': 'https://www.zee5.com/',
 228         }
 229         show_url = 'https://gwapi.zee5.com/content/tvshow/{}?translation=en&country=IN'.format(show_id)
 230
 231         page_num = 0
 232         show_json = self._download_json(show_url, video_id=show_id, headers=headers)
 233         for season in show_json.get('seasons') or []:
 234             season_id = try_get(season, lambda x: x['id'], compat_str)
 235             next_url = 'https://gwapi.zee5.com/content/tvshow/?season_id={}&type=episode&translation=en&country=IN&on_air=false&asset_subtype=tvshow&page=1&limit=100'.format(season_id)
 236             while next_url:
 237                 page_num += 1
 238                 episodes_json = self._download_json(
 239                     next_url, video_id=show_id, headers=headers,
 240                     note='Downloading JSON metadata page %d' % page_num)
 241                 for episode in try_get(episodes_json, lambda x: x['episode'], list) or []:
 242                     video_id = episode.get('id')
 243                     yield self.url_result(
 244                         'zee5:%s' % video_id,
 245                         ie=Zee5IE.ie_key(), video_id=video_id)
 246                 next_url = url_or_none(episodes_json.get('next_episode_api'))
 247
 248     def _real_extract(self, url):
 249         show_id = self._match_id(url)
 250         return self.playlist_result(self._entries(show_id), playlist_id=show_id)