yt_dlp/extractor/stacommu.py

   1 import time
   2
   3 from .wrestleuniverse import WrestleUniverseBaseIE
   4 from ..utils import (
   5     int_or_none,
   6     traverse_obj,
   7     url_or_none,
   8 )
   9
  10
  11 class StacommuBaseIE(WrestleUniverseBaseIE):
  12     _NETRC_MACHINE = 'stacommu'
  13     _API_HOST = 'api.stacommu.jp'
  14     _LOGIN_QUERY = {'key': 'AIzaSyCR9czxhH2eWuijEhTNWBZ5MCcOYEUTAhg'}
  15     _LOGIN_HEADERS = {
  16         'Accept': '*/*',
  17         'Content-Type': 'application/json',
  18         'X-Client-Version': 'Chrome/JsCore/9.9.4/FirebaseCore-web',
  19         'Referer': 'https://www.stacommu.jp/',
  20         'Origin': 'https://www.stacommu.jp',
  21     }
  22
  23     @WrestleUniverseBaseIE._TOKEN.getter
  24     def _TOKEN(self):
  25         if self._REAL_TOKEN and self._TOKEN_EXPIRY <= int(time.time()):
  26             self._refresh_token()
  27
  28         return self._REAL_TOKEN
  29
  30     def _get_formats(self, data, path, video_id=None):
  31         if not traverse_obj(data, path) and not data.get('canWatch') and not self._TOKEN:
  32             self.raise_login_required(method='password')
  33         return super()._get_formats(data, path, video_id)
  34
  35     def _extract_hls_key(self, data, path, decrypt):
  36         encryption_data = traverse_obj(data, path)
  37         if traverse_obj(encryption_data, ('encryptType', {int})) == 0:
  38             return None
  39         return traverse_obj(encryption_data, {'key': ('key', {decrypt}), 'iv': ('iv', {decrypt})})
  40
  41
  42 class StacommuVODIE(StacommuBaseIE):
  43     _VALID_URL = r'https?://www\.stacommu\.jp/videos/episodes/(?P<id>[\da-zA-Z]+)'
  44     _TESTS = [{
  45         # not encrypted
  46         'url': 'https://www.stacommu.jp/videos/episodes/aXcVKjHyAENEjard61soZZ',
  47         'info_dict': {
  48             'id': 'aXcVKjHyAENEjard61soZZ',
  49             'ext': 'mp4',
  50             'title': 'スタコミュAWARDの裏側、ほぼ全部見せます！〜晴れ舞台の直前ドキドキ編〜',
  51             'description': 'md5:6400275c57ae75c06da36b06f96beb1c',
  52             'timestamp': 1679652000,
  53             'upload_date': '20230324',
  54             'thumbnail': 'https://image.stacommu.jp/6eLobQan8PFtBoU4RL4uGg/6eLobQan8PFtBoU4RL4uGg',
  55             'cast': 'count:11',
  56             'duration': 250,
  57         },
  58         'params': {
  59             'skip_download': 'm3u8',
  60         },
  61     }, {
  62         # encrypted; requires a premium account
  63         'url': 'https://www.stacommu.jp/videos/episodes/3hybMByUvzMEqndSeu5LpD',
  64         'info_dict': {
  65             'id': '3hybMByUvzMEqndSeu5LpD',
  66             'ext': 'mp4',
  67             'title': 'スタプラフェス2023〜裏側ほぼ全部見せます〜＃10',
  68             'description': 'md5:85494488ccf1dfa1934accdeadd7b340',
  69             'timestamp': 1682506800,
  70             'upload_date': '20230426',
  71             'thumbnail': 'https://image.stacommu.jp/eMdXtEefR4kEyJJMpAFi7x/eMdXtEefR4kEyJJMpAFi7x',
  72             'cast': 'count:55',
  73             'duration': 312,
  74             'hls_aes': {
  75                 'key': '6bbaf241b8e1fd9f59ecf546a70e4ae7',
  76                 'iv': '1fc9002a23166c3bb1d240b953d09de9',
  77             },
  78         },
  79         'params': {
  80             'skip_download': 'm3u8',
  81         },
  82     }]
  83
  84     _API_PATH = 'videoEpisodes'
  85
  86     def _real_extract(self, url):
  87         video_id = self._match_id(url)
  88         video_info = self._download_metadata(
  89             url, video_id, 'ja', ('dehydratedState', 'queries', 0, 'state', 'data'))
  90         hls_info, decrypt = self._call_encrypted_api(
  91             video_id, ':watch', 'stream information', data={'method': 1})
  92
  93         return {
  94             'id': video_id,
  95             'formats': self._get_formats(hls_info, ('protocolHls', 'url', {url_or_none}), video_id),
  96             'hls_aes': self._extract_hls_key(hls_info, 'protocolHls', decrypt),
  97             **traverse_obj(video_info, {
  98                 'title': ('displayName', {str}),
  99                 'description': ('description', {str}),
 100                 'timestamp': ('watchStartTime', {int_or_none}),
 101                 'thumbnail': ('keyVisualUrl', {url_or_none}),
 102                 'cast': ('casts', ..., 'displayName', {str}),
 103                 'duration': ('duration', {int}),
 104             }),
 105         }
 106
 107
 108 class StacommuLiveIE(StacommuBaseIE):
 109     _VALID_URL = r'https?://www\.stacommu\.jp/live/(?P<id>[\da-zA-Z]+)'
 110     _TESTS = [{
 111         'url': 'https://www.stacommu.jp/live/d2FJ3zLnndegZJCAEzGM3m',
 112         'info_dict': {
 113             'id': 'd2FJ3zLnndegZJCAEzGM3m',
 114             'ext': 'mp4',
 115             'title': '仲村悠菜 2023/05/04',
 116             'timestamp': 1683195647,
 117             'upload_date': '20230504',
 118             'thumbnail': 'https://image.stacommu.jp/pHGF57SPEHE2ke83FS92FN/pHGF57SPEHE2ke83FS92FN',
 119             'duration': 5322,
 120             'hls_aes': {
 121                 'key': 'efbb3ec0b8246f61adf1764c5a51213a',
 122                 'iv': '80621d19a1f19167b64cedb415b05d1c',
 123             },
 124         },
 125         'params': {
 126             'skip_download': 'm3u8',
 127         },
 128     }]
 129
 130     _API_PATH = 'events'
 131
 132     def _real_extract(self, url):
 133         video_id = self._match_id(url)
 134         video_info = self._call_api(video_id, msg='video information', query={'al': 'ja'}, auth=False)
 135         hls_info, decrypt = self._call_encrypted_api(
 136             video_id, ':watchArchive', 'stream information', data={'method': 1})
 137
 138         return {
 139             'id': video_id,
 140             'formats': self._get_formats(hls_info, ('hls', 'urls', ..., {url_or_none}), video_id),
 141             'hls_aes': self._extract_hls_key(hls_info, 'hls', decrypt),
 142             **traverse_obj(video_info, {
 143                 'title': ('displayName', {str}),
 144                 'timestamp': ('startTime', {int_or_none}),
 145                 'thumbnail': ('keyVisualUrl', {url_or_none}),
 146                 'duration': ('duration', {int_or_none}),
 147             }),
 148         }