yt_dlp/extractor/skyit.py

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