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(HEADRequest('https://kick.com/'), None, 'Setting up session', fatal=False)
  17         xsrf_token = self._get_cookies('https://kick.com/').get('XSRF-TOKEN')
  18         if not xsrf_token:
  19             self.write_debug('kick.com did not set XSRF-TOKEN cookie')
  20         KickBaseIE._API_HEADERS = {
  21             'Authorization': f'Bearer {xsrf_token.value}',
  22             'X-XSRF-TOKEN': xsrf_token.value,
  23         } if xsrf_token else {}
  24
  25     def _call_api(self, path, display_id, note='Downloading API JSON', headers={}, **kwargs):
  26         return self._download_json(
  27             f'https://kick.com/api/v1/{path}', display_id, note=note,
  28             headers=merge_dicts(headers, self._API_HEADERS), **kwargs)
  29
  30
  31 class KickIE(KickBaseIE):
  32     _VALID_URL = r'https?://(?:www\.)?kick\.com/(?!(?:video|categories|search|auth)(?:[/?#]|$))(?P<id>[\w-]+)'
  33     _TESTS = [{
  34         'url': 'https://kick.com/yuppy',
  35         'info_dict': {
  36             'id': '6cde1-kickrp-joe-flemmingskick-info-heremust-knowmust-see21',
  37             'ext': 'mp4',
  38             'title': str,
  39             'description': str,
  40             'channel': 'yuppy',
  41             'channel_id': '33538',
  42             'uploader': 'Yuppy',
  43             'uploader_id': '33793',
  44             'upload_date': str,
  45             'live_status': 'is_live',
  46             'timestamp': int,
  47             'thumbnail': r're:^https?://.*\.jpg',
  48             'categories': list,
  49         },
  50         'skip': 'livestream',
  51     }, {
  52         'url': 'https://kick.com/kmack710',
  53         'only_matching': True,
  54     }]
  55
  56     def _real_extract(self, url):
  57         channel = self._match_id(url)
  58         response = self._call_api(f'channels/{channel}', channel)
  59         if not traverse_obj(response, 'livestream', expected_type=dict):
  60             raise UserNotLive(video_id=channel)
  61
  62         return {
  63             'id': str(traverse_obj(
  64                 response, ('livestream', ('slug', 'id')), get_all=False, default=channel)),
  65             'formats': self._extract_m3u8_formats(
  66                 response['playback_url'], channel, 'mp4', live=True),
  67             'title': traverse_obj(
  68                 response, ('livestream', ('session_title', 'slug')), get_all=False, default=''),
  69             'description': traverse_obj(response, ('user', 'bio')),
  70             'channel': channel,
  71             'channel_id': str_or_none(traverse_obj(response, 'id', ('livestream', 'channel_id'))),
  72             'uploader': traverse_obj(response, 'name', ('user', 'username')),
  73             'uploader_id': str_or_none(traverse_obj(response, 'user_id', ('user', 'id'))),
  74             'is_live': True,
  75             'timestamp': unified_timestamp(traverse_obj(response, ('livestream', 'created_at'))),
  76             'thumbnail': traverse_obj(
  77                 response, ('livestream', 'thumbnail', 'url'), expected_type=url_or_none),
  78             'categories': traverse_obj(response, ('recent_categories', ..., 'name')),
  79         }
  80
  81
  82 class KickVODIE(KickBaseIE):
  83     _VALID_URL = r'https?://(?:www\.)?kick\.com/video/(?P<id>[\da-f]{8}-(?:[\da-f]{4}-){3}[\da-f]{12})'
  84     _TESTS = [{
  85         'url': 'https://kick.com/video/54244b5e-050a-4df4-a013-b2433dafbe35',
  86         'md5': '73691206a6a49db25c5aa1588e6538fc',
  87         'info_dict': {
  88             'id': '54244b5e-050a-4df4-a013-b2433dafbe35',
  89             'ext': 'mp4',
  90             'title': 'Making 710-carBoosting. Kinda No Pixel inspired.  !guilded  - !links',
  91             'description': 'md5:a0d3546bf7955d0a8252ffe0fd6f518f',
  92             'channel': 'kmack710',
  93             'channel_id': '16278',
  94             'uploader': 'Kmack710',
  95             'uploader_id': '16412',
  96             'upload_date': '20221206',
  97             'timestamp': 1670318289,
  98             'duration': 40104.0,
  99             'thumbnail': r're:^https?://.*\.jpg',
 100             'categories': ['Grand Theft Auto V'],
 101         },
 102         'params': {
 103             'skip_download': 'm3u8',
 104         },
 105     }]
 106
 107     def _real_extract(self, url):
 108         video_id = self._match_id(url)
 109         response = self._call_api(f'video/{video_id}', video_id)
 110
 111         return {
 112             'id': video_id,
 113             'formats': self._extract_m3u8_formats(response['source'], video_id, 'mp4'),
 114             'title': traverse_obj(
 115                 response, ('livestream', ('session_title', 'slug')), get_all=False, default=''),
 116             'description': traverse_obj(response, ('livestream', 'channel', 'user', 'bio')),
 117             'channel': traverse_obj(response, ('livestream', 'channel', 'slug')),
 118             'channel_id': str_or_none(traverse_obj(response, ('livestream', 'channel', 'id'))),
 119             'uploader': traverse_obj(response, ('livestream', 'channel', 'user', 'username')),
 120             'uploader_id': str_or_none(traverse_obj(response, ('livestream', 'channel', 'user_id'))),
 121             'timestamp': unified_timestamp(response.get('created_at')),
 122             'duration': float_or_none(traverse_obj(response, ('livestream', 'duration')), scale=1000),
 123             'thumbnail': traverse_obj(
 124                 response, ('livestream', 'thumbnail'), expected_type=url_or_none),
 125             'categories': traverse_obj(response, ('livestream', 'categories', ..., 'name')),
 126         }