yt_dlp/extractor/kick.py

   1 from .common import InfoExtractor
   2 from ..networking import HEADRequest
   3 from ..utils import (
   4     UserNotLive,
   5     float_or_none,
   6     merge_dicts,
   7     str_or_none,
   8     traverse_obj,
   9     unified_timestamp,
  10     url_or_none,
  11 )
  12
  13
  14 class KickBaseIE(InfoExtractor):
  15     def _real_initialize(self):
  16         self._request_webpage(
  17             HEADRequest('https://kick.com/'), None, 'Setting up session', fatal=False, impersonate=True)
  18         xsrf_token = self._get_cookies('https://kick.com/').get('XSRF-TOKEN')
  19         if not xsrf_token:
  20             self.write_debug('kick.com did not set XSRF-TOKEN cookie')
  21         KickBaseIE._API_HEADERS = {
  22             'Authorization': f'Bearer {xsrf_token.value}',
  23             'X-XSRF-TOKEN': xsrf_token.value,
  24         } if xsrf_token else {}
  25
  26     def _call_api(self, path, display_id, note='Downloading API JSON', headers={}, **kwargs):
  27         return self._download_json(
  28             f'https://kick.com/api/v1/{path}', display_id, note=note,
  29             headers=merge_dicts(headers, self._API_HEADERS), impersonate=True, **kwargs)
  30
  31
  32 class KickIE(KickBaseIE):
  33     _VALID_URL = r'https?://(?:www\.)?kick\.com/(?!(?:video|categories|search|auth)(?:[/?#]|$))(?P<id>[\w-]+)'
  34     _TESTS = [{
  35         'url': 'https://kick.com/yuppy',
  36         'info_dict': {
  37             'id': '6cde1-kickrp-joe-flemmingskick-info-heremust-knowmust-see21',
  38             'ext': 'mp4',
  39             'title': str,
  40             'description': str,
  41             'channel': 'yuppy',
  42             'channel_id': '33538',
  43             'uploader': 'Yuppy',
  44             'uploader_id': '33793',
  45             'upload_date': str,
  46             'live_status': 'is_live',
  47             'timestamp': int,
  48             'thumbnail': r're:^https?://.*\.jpg',
  49             'categories': list,
  50         },
  51         'skip': 'livestream',
  52     }, {
  53         'url': 'https://kick.com/kmack710',
  54         'only_matching': True,
  55     }]
  56
  57     def _real_extract(self, url):
  58         channel = self._match_id(url)
  59         response = self._call_api(f'channels/{channel}', channel)
  60         if not traverse_obj(response, 'livestream', expected_type=dict):
  61             raise UserNotLive(video_id=channel)
  62
  63         return {
  64             'id': str(traverse_obj(
  65                 response, ('livestream', ('slug', 'id')), get_all=False, default=channel)),
  66             'formats': self._extract_m3u8_formats(
  67                 response['playback_url'], channel, 'mp4', live=True),
  68             'title': traverse_obj(
  69                 response, ('livestream', ('session_title', 'slug')), get_all=False, default=''),
  70             'description': traverse_obj(response, ('user', 'bio')),
  71             'channel': channel,
  72             'channel_id': str_or_none(traverse_obj(response, 'id', ('livestream', 'channel_id'))),
  73             'uploader': traverse_obj(response, 'name', ('user', 'username')),
  74             'uploader_id': str_or_none(traverse_obj(response, 'user_id', ('user', 'id'))),
  75             'is_live': True,
  76             'timestamp': unified_timestamp(traverse_obj(response, ('livestream', 'created_at'))),
  77             'thumbnail': traverse_obj(
  78                 response, ('livestream', 'thumbnail', 'url'), expected_type=url_or_none),
  79             'categories': traverse_obj(response, ('recent_categories', ..., 'name')),
  80         }
  81
  82
  83 class KickVODIE(KickBaseIE):
  84     _VALID_URL = r'https?://(?:www\.)?kick\.com/video/(?P<id>[\da-f]{8}-(?:[\da-f]{4}-){3}[\da-f]{12})'
  85     _TESTS = [{
  86         'url': 'https://kick.com/video/58bac65b-e641-4476-a7ba-3707a35e60e3',
  87         'md5': '3870f94153e40e7121a6e46c068b70cb',
  88         'info_dict': {
  89             'id': '58bac65b-e641-4476-a7ba-3707a35e60e3',
  90             'ext': 'mp4',
  91             'title': '🤠REBIRTH IS BACK!!!!🤠!stake CODE JAREDFPS 🤠',
  92             'description': 'md5:02b0c46f9b4197fb545ab09dddb85b1d',
  93             'channel': 'jaredfps',
  94             'channel_id': '26608',
  95             'uploader': 'JaredFPS',
  96             'uploader_id': '26799',
  97             'upload_date': '20240402',
  98             'timestamp': 1712097108,
  99             'duration': 33859.0,
 100             'thumbnail': r're:^https?://.*\.jpg',
 101             'categories': ['Call of Duty: Warzone'],
 102         },
 103         'params': {
 104             'skip_download': 'm3u8',
 105         },
 106         'expected_warnings': [r'impersonation'],
 107     }]
 108
 109     def _real_extract(self, url):
 110         video_id = self._match_id(url)
 111         response = self._call_api(f'video/{video_id}', video_id)
 112
 113         return {
 114             'id': video_id,
 115             'formats': self._extract_m3u8_formats(response['source'], video_id, 'mp4'),
 116             'title': traverse_obj(
 117                 response, ('livestream', ('session_title', 'slug')), get_all=False, default=''),
 118             'description': traverse_obj(response, ('livestream', 'channel', 'user', 'bio')),
 119             'channel': traverse_obj(response, ('livestream', 'channel', 'slug')),
 120             'channel_id': str_or_none(traverse_obj(response, ('livestream', 'channel', 'id'))),
 121             'uploader': traverse_obj(response, ('livestream', 'channel', 'user', 'username')),
 122             'uploader_id': str_or_none(traverse_obj(response, ('livestream', 'channel', 'user_id'))),
 123             'timestamp': unified_timestamp(response.get('created_at')),
 124             'duration': float_or_none(traverse_obj(response, ('livestream', 'duration')), scale=1000),
 125             'thumbnail': traverse_obj(
 126                 response, ('livestream', 'thumbnail'), expected_type=url_or_none),
 127             'categories': traverse_obj(response, ('livestream', 'categories', ..., 'name')),
 128         }