yt_dlp/extractor/skyit.py

   1 from .common import InfoExtractor
   2 from ..compat import (
   3     compat_parse_qs,
   4     compat_urllib_parse_urlparse,
   5 )
   6 from ..utils import (
   7     dict_get,
   8     int_or_none,
   9     parse_duration,
  10     unified_timestamp,
  11 )
  12
  13
  14 class SkyItPlayerIE(InfoExtractor):
  15     IE_NAME = 'player.sky.it'
  16     _VALID_URL = r'https?://player\.sky\.it/player/(?:external|social)\.html\?.*?\bid=(?P<id>\d+)'
  17     _GEO_BYPASS = False
  18     _DOMAIN = 'sky'
  19     _PLAYER_TMPL = 'https://player.sky.it/player/external.html?id=%s&domain=%s'
  20     # http://static.sky.it/static/skyplayer/conf.json
  21     _TOKEN_MAP = {
  22         'cielo': 'Hh9O7M8ks5yi6nSROL7bKYz933rdf3GhwZlTLMgvy4Q',
  23         'hotclub': 'kW020K2jq2lk2eKRJD2vWEg832ncx2EivZlTLQput2C',
  24         'mtv8': 'A5Nn9GGb326CI7vP5e27d7E4PIaQjota',
  25         'salesforce': 'C6D585FD1615272C98DE38235F38BD86',
  26         'sitocommerciale': 'VJwfFuSGnLKnd9Phe9y96WkXgYDCguPMJ2dLhGMb2RE',
  27         'sky': 'F96WlOd8yoFmLQgiqv6fNQRvHZcsWk5jDaYnDvhbiJk',
  28         'skyarte': 'LWk29hfiU39NNdq87ePeRach3nzTSV20o0lTv2001Cd',
  29         'theupfront': 'PRSGmDMsg6QMGc04Obpoy7Vsbn7i2Whp',
  30     }
  31
  32     def _player_url_result(self, video_id):
  33         return self.url_result(
  34             self._PLAYER_TMPL % (video_id, self._DOMAIN),
  35             SkyItPlayerIE.ie_key(), video_id)
  36
  37     def _parse_video(self, video, video_id):
  38         title = video['title']
  39         is_live = video.get('type') == 'live'
  40         hls_url = video.get(('streaming' if is_live else 'hls') + '_url')
  41         if not hls_url and video.get('geoblock' if is_live else 'geob'):
  42             self.raise_geo_restricted(countries=['IT'])
  43
  44         formats = self._extract_m3u8_formats(hls_url, video_id, 'mp4')
  45
  46         return {
  47             'id': video_id,
  48             'title': title,
  49             'formats': formats,
  50             'thumbnail': dict_get(video, ('video_still', 'video_still_medium', 'thumb')),
  51             'description': video.get('short_desc') or None,
  52             'timestamp': unified_timestamp(video.get('create_date')),
  53             'duration': int_or_none(video.get('duration_sec')) or parse_duration(video.get('duration')),
  54             'is_live': is_live,
  55         }
  56
  57     def _real_extract(self, url):
  58         video_id = self._match_id(url)
  59         domain = compat_parse_qs(compat_urllib_parse_urlparse(
  60             url).query).get('domain', [None])[0]
  61         token = dict_get(self._TOKEN_MAP, (domain, 'sky'))
  62         video = self._download_json(
  63             'https://apid.sky.it/vdp/v1/getVideoData',
  64             video_id, query={
  65                 'caller': 'sky',
  66                 'id': video_id,
  67                 'token': token
  68             }, headers=self.geo_verification_headers())
  69         return self._parse_video(video, video_id)
  70
  71
  72 class SkyItVideoIE(SkyItPlayerIE):  # XXX: Do not subclass from concrete IE
  73     IE_NAME = 'video.sky.it'
  74     _VALID_URL = r'https?://(?:masterchef|video|xfactor)\.sky\.it(?:/[^/]+)*/video/[0-9a-z-]+-(?P<id>\d+)'
  75     _TESTS = [{
  76         'url': 'https://video.sky.it/news/mondo/video/uomo-ucciso-da-uno-squalo-in-australia-631227',
  77         'md5': '5b858a62d9ffe2ab77b397553024184a',
  78         'info_dict': {
  79             'id': '631227',
  80             'ext': 'mp4',
  81             'title': 'Uomo ucciso da uno squalo in Australia',
  82             'timestamp': 1606036192,
  83             'upload_date': '20201122',
  84             'duration': 26,
  85             'thumbnail': 'https://video.sky.it/captures/thumbs/631227/631227_thumb_880x494.jpg',
  86         },
  87         'params': {'skip_download': 'm3u8'},
  88     }, {
  89         'url': 'https://xfactor.sky.it/video/x-factor-2020-replay-audizioni-1-615820',
  90         'only_matching': True,
  91     }, {
  92         'url': 'https://masterchef.sky.it/video/masterchef-9-cosa-e-successo-nella-prima-puntata-562831',
  93         'only_matching': True,
  94     }]
  95
  96     def _real_extract(self, url):
  97         video_id = self._match_id(url)
  98         return self._player_url_result(video_id)
  99
 100
 101 class SkyItVideoLiveIE(SkyItPlayerIE):  # XXX: Do not subclass from concrete IE
 102     IE_NAME = 'video.sky.it:live'
 103     _VALID_URL = r'https?://video\.sky\.it/diretta/(?P<id>[^/?&#]+)'
 104     _TEST = {
 105         'url': 'https://video.sky.it/diretta/tg24',
 106         'info_dict': {
 107             'id': '1',
 108             'ext': 'mp4',
 109             'title': r're:Diretta TG24 \d{4}-\d{2}-\d{2} \d{2}:\d{2}',
 110             'description': r're:(?:Clicca play e )?[Gg]uarda la diretta streaming di SkyTg24, segui con Sky tutti gli appuntamenti e gli speciali di Tg24\.',
 111             'live_status': 'is_live',
 112         },
 113         'params': {
 114             # m3u8 download
 115             'skip_download': True,
 116         },
 117     }
 118
 119     def _real_extract(self, url):
 120         display_id = self._match_id(url)
 121         webpage = self._download_webpage(url, display_id)
 122         asset_id = str(self._search_nextjs_data(webpage, display_id)['props']['initialState']['livePage']['content']['asset_id'])
 123         livestream = self._download_json(
 124             'https://apid.sky.it/vdp/v1/getLivestream',
 125             asset_id, query={'id': asset_id})
 126         return self._parse_video(livestream, asset_id)
 127
 128
 129 class SkyItIE(SkyItPlayerIE):  # XXX: Do not subclass from concrete IE
 130     IE_NAME = 'sky.it'
 131     _VALID_URL = r'https?://(?:sport|tg24)\.sky\.it(?:/[^/]+)*/\d{4}/\d{2}/\d{2}/(?P<id>[^/?&#]+)'
 132     _TESTS = [{
 133         'url': 'https://sport.sky.it/calcio/serie-a/2022/11/03/brozovic-inter-news',
 134         'info_dict': {
 135             'id': '789222',
 136             'ext': 'mp4',
 137             'title': 'Brozovic con il gruppo: verso convocazione per Juve-Inter',
 138             'upload_date': '20221103',
 139             'timestamp': 1667484130,
 140             'duration': 22,
 141             'thumbnail': 'https://videoplatform.sky.it/still/2022/11/03/1667480526353_brozovic_videostill_1.jpg',
 142         },
 143         'params': {'skip_download': 'm3u8'},
 144     }, {
 145         'url': 'https://tg24.sky.it/mondo/2020/11/22/australia-squalo-uccide-uomo',
 146         'md5': 'fe5c91e59a84a3437eaa0bca6e134ccd',
 147         'info_dict': {
 148             'id': '631227',
 149             'ext': 'mp4',
 150             'title': 'Uomo ucciso da uno squalo in Australia',
 151             'timestamp': 1606036192,
 152             'upload_date': '20201122',
 153             'duration': 26,
 154             'thumbnail': 'https://video.sky.it/captures/thumbs/631227/631227_thumb_880x494.jpg',
 155         },
 156         'params': {'skip_download': 'm3u8'},
 157     }]
 158     _VIDEO_ID_REGEX = r'data-videoid="(\d+)"'
 159
 160     def _real_extract(self, url):
 161         display_id = self._match_id(url)
 162         webpage = self._download_webpage(url, display_id)
 163         video_id = self._search_regex(
 164             self._VIDEO_ID_REGEX, webpage, 'video id')
 165         return self._player_url_result(video_id)
 166
 167
 168 class SkyItArteIE(SkyItIE):  # XXX: Do not subclass from concrete IE
 169     IE_NAME = 'arte.sky.it'
 170     _VALID_URL = r'https?://arte\.sky\.it/video/(?P<id>[^/?&#]+)'
 171     _TESTS = [{
 172         'url': 'https://arte.sky.it/video/oliviero-toscani-torino-galleria-mazzoleni-788962',
 173         'md5': '515aee97b87d7a018b6c80727d3e7e17',
 174         'info_dict': {
 175             'id': '788962',
 176             'ext': 'mp4',
 177             'title': 'La fotografia di Oliviero Toscani conquista Torino',
 178             'upload_date': '20221102',
 179             'timestamp': 1667399996,
 180             'duration': 12,
 181             'thumbnail': 'https://videoplatform.sky.it/still/2022/11/02/1667396388552_oliviero-toscani-torino-galleria-mazzoleni_videostill_1.jpg',
 182         },
 183         'params': {'skip_download': 'm3u8'},
 184     }]
 185     _DOMAIN = 'skyarte'
 186     _VIDEO_ID_REGEX = r'"embedUrl"\s*:\s*"(?:https:)?//player\.sky\.it/player/external\.html\?[^"]*\bid=(\d+)'
 187
 188
 189 class CieloTVItIE(SkyItIE):  # XXX: Do not subclass from concrete IE
 190     IE_NAME = 'cielotv.it'
 191     _VALID_URL = r'https?://(?:www\.)?cielotv\.it/video/(?P<id>[^.]+)\.html'
 192     _TESTS = [{
 193         'url': 'https://www.cielotv.it/video/Il-lunedi-e-sempre-un-dramma.html',
 194         'md5': 'c4deed77552ba901c2a0d9258320304b',
 195         'info_dict': {
 196             'id': '499240',
 197             'ext': 'mp4',
 198             'title': 'Il lunedì è sempre un dramma',
 199             'upload_date': '20190329',
 200             'timestamp': 1553862178,
 201             'duration': 30,
 202             'thumbnail': 'https://videoplatform.sky.it/still/2019/03/29/1553858575610_lunedi_dramma_mant_videostill_1.jpg',
 203         },
 204         'params': {'skip_download': 'm3u8'},
 205     }]
 206     _DOMAIN = 'cielo'
 207     _VIDEO_ID_REGEX = r'videoId\s*=\s*"(\d+)"'
 208
 209
 210 class TV8ItIE(SkyItVideoIE):  # XXX: Do not subclass from concrete IE
 211     IE_NAME = 'tv8.it'
 212     _VALID_URL = r'https?://(?:www\.)?tv8\.it/(?:show)?video/[0-9a-z-]+-(?P<id>\d+)'
 213     _TESTS = [{
 214         'url': 'https://www.tv8.it/video/ogni-mattina-ucciso-asino-di-andrea-lo-cicero-630529',
 215         'md5': '9ab906a3f75ea342ed928442f9dabd21',
 216         'info_dict': {
 217             'id': '630529',
 218             'ext': 'mp4',
 219             'title': 'Ogni mattina - Ucciso asino di Andrea Lo Cicero',
 220             'timestamp': 1605721374,
 221             'upload_date': '20201118',
 222             'duration': 114,
 223             'thumbnail': 'https://videoplatform.sky.it/still/2020/11/18/1605717753954_ogni-mattina-ucciso-asino-di-andrea-lo-cicero_videostill_1.jpg',
 224         },
 225         'params': {'skip_download': 'm3u8'},
 226     }]
 227     _DOMAIN = 'mtv8'