yt_dlp/extractor/skyit.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import (
   6     compat_parse_qs,
   7     compat_urllib_parse_urlparse,
   8 )
   9 from ..utils import (
  10     dict_get,
  11     int_or_none,
  12     parse_duration,
  13     unified_timestamp,
  14 )
  15
  16
  17 class SkyItPlayerIE(InfoExtractor):
  18     IE_NAME = 'player.sky.it'
  19     _VALID_URL = r'https?://player\.sky\.it/player/(?:external|social)\.html\?.*?\bid=(?P<id>\d+)'
  20     _GEO_BYPASS = False
  21     _DOMAIN = 'sky'
  22     _PLAYER_TMPL = 'https://player.sky.it/player/external.html?id=%s&domain=%s'
  23     # http://static.sky.it/static/skyplayer/conf.json
  24     _TOKEN_MAP = {
  25         'cielo': 'Hh9O7M8ks5yi6nSROL7bKYz933rdf3GhwZlTLMgvy4Q',
  26         'hotclub': 'kW020K2jq2lk2eKRJD2vWEg832ncx2EivZlTLQput2C',
  27         'mtv8': 'A5Nn9GGb326CI7vP5e27d7E4PIaQjota',
  28         'salesforce': 'C6D585FD1615272C98DE38235F38BD86',
  29         'sitocommerciale': 'VJwfFuSGnLKnd9Phe9y96WkXgYDCguPMJ2dLhGMb2RE',
  30         'sky': 'F96WlOd8yoFmLQgiqv6fNQRvHZcsWk5jDaYnDvhbiJk',
  31         'skyacademy': 'A6LAn7EkO2Q26FRy0IAMBekX6jzDXYL3',
  32         'skyarte': 'LWk29hfiU39NNdq87ePeRach3nzTSV20o0lTv2001Cd',
  33         'theupfront': 'PRSGmDMsg6QMGc04Obpoy7Vsbn7i2Whp',
  34     }
  35
  36     def _player_url_result(self, video_id):
  37         return self.url_result(
  38             self._PLAYER_TMPL % (video_id, self._DOMAIN),
  39             SkyItPlayerIE.ie_key(), video_id)
  40
  41     def _parse_video(self, video, video_id):
  42         title = video['title']
  43         is_live = video.get('type') == 'live'
  44         hls_url = video.get(('streaming' if is_live else 'hls') + '_url')
  45         if not hls_url and video.get('geoblock' if is_live else 'geob'):
  46             self.raise_geo_restricted(countries=['IT'])
  47
  48         if is_live:
  49             formats = self._extract_m3u8_formats(hls_url, video_id, 'mp4')
  50         else:
  51             formats = self._extract_akamai_formats(
  52                 hls_url, video_id, {'http': 'videoplatform.sky.it'})
  53         self._sort_formats(formats)
  54
  55         return {
  56             'id': video_id,
  57             'title': title,
  58             'formats': formats,
  59             'thumbnail': dict_get(video, ('video_still', 'video_still_medium', 'thumb')),
  60             'description': video.get('short_desc') or None,
  61             'timestamp': unified_timestamp(video.get('create_date')),
  62             'duration': int_or_none(video.get('duration_sec')) or parse_duration(video.get('duration')),
  63             'is_live': is_live,
  64         }
  65
  66     def _real_extract(self, url):
  67         video_id = self._match_id(url)
  68         domain = compat_parse_qs(compat_urllib_parse_urlparse(
  69             url).query).get('domain', [None])[0]
  70         token = dict_get(self._TOKEN_MAP, (domain, 'sky'))
  71         video = self._download_json(
  72             'https://apid.sky.it/vdp/v1/getVideoData',
  73             video_id, query={
  74                 'caller': 'sky',
  75                 'id': video_id,
  76                 'token': token
  77             }, headers=self.geo_verification_headers())
  78         return self._parse_video(video, video_id)
  79
  80
  81 class SkyItVideoIE(SkyItPlayerIE):
  82     IE_NAME = 'video.sky.it'
  83     _VALID_URL = r'https?://(?:masterchef|video|xfactor)\.sky\.it(?:/[^/]+)*/video/[0-9a-z-]+-(?P<id>\d+)'
  84     _TESTS = [{
  85         'url': 'https://video.sky.it/news/mondo/video/uomo-ucciso-da-uno-squalo-in-australia-631227',
  86         'md5': 'fe5c91e59a84a3437eaa0bca6e134ccd',
  87         'info_dict': {
  88             'id': '631227',
  89             'ext': 'mp4',
  90             'title': 'Uomo ucciso da uno squalo in Australia',
  91             'timestamp': 1606036192,
  92             'upload_date': '20201122',
  93         }
  94     }, {
  95         'url': 'https://xfactor.sky.it/video/x-factor-2020-replay-audizioni-1-615820',
  96         'only_matching': True,
  97     }, {
  98         'url': 'https://masterchef.sky.it/video/masterchef-9-cosa-e-successo-nella-prima-puntata-562831',
  99         'only_matching': True,
 100     }]
 101
 102     def _real_extract(self, url):
 103         video_id = self._match_id(url)
 104         return self._player_url_result(video_id)
 105
 106
 107 class SkyItVideoLiveIE(SkyItPlayerIE):
 108     IE_NAME = 'video.sky.it:live'
 109     _VALID_URL = r'https?://video\.sky\.it/diretta/(?P<id>[^/?&#]+)'
 110     _TEST = {
 111         'url': 'https://video.sky.it/diretta/tg24',
 112         'info_dict': {
 113             'id': '1',
 114             'ext': 'mp4',
 115             'title': r're:Diretta TG24 \d{4}-\d{2}-\d{2} \d{2}:\d{2}',
 116             'description': 'Guarda la diretta streaming di SkyTg24, segui con Sky tutti gli appuntamenti e gli speciali di Tg24.',
 117         },
 118         'params': {
 119             # m3u8 download
 120             'skip_download': True,
 121         },
 122     }
 123
 124     def _real_extract(self, url):
 125         display_id = self._match_id(url)
 126         webpage = self._download_webpage(url, display_id)
 127         asset_id = str(self._search_nextjs_data(webpage, display_id)['props']['initialState']['livePage']['content']['asset_id'])
 128         livestream = self._download_json(
 129             'https://apid.sky.it/vdp/v1/getLivestream',
 130             asset_id, query={'id': asset_id})
 131         return self._parse_video(livestream, asset_id)
 132
 133
 134 class SkyItIE(SkyItPlayerIE):
 135     IE_NAME = 'sky.it'
 136     _VALID_URL = r'https?://(?:sport|tg24)\.sky\.it(?:/[^/]+)*/\d{4}/\d{2}/\d{2}/(?P<id>[^/?&#]+)'
 137     _TESTS = [{
 138         'url': 'https://sport.sky.it/calcio/serie-a/2020/11/21/juventus-cagliari-risultato-gol',
 139         'info_dict': {
 140             'id': '631201',
 141             'ext': 'mp4',
 142             'title': 'Un rosso alla violenza: in campo per i diritti delle donne',
 143             'upload_date': '20201121',
 144             'timestamp': 1605995753,
 145         },
 146         'expected_warnings': ['Unable to download f4m manifest'],
 147     }, {
 148         'url': 'https://tg24.sky.it/mondo/2020/11/22/australia-squalo-uccide-uomo',
 149         'md5': 'fe5c91e59a84a3437eaa0bca6e134ccd',
 150         'info_dict': {
 151             'id': '631227',
 152             'ext': 'mp4',
 153             'title': 'Uomo ucciso da uno squalo in Australia',
 154             'timestamp': 1606036192,
 155             'upload_date': '20201122',
 156         },
 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 SkyItAcademyIE(SkyItIE):
 169     IE_NAME = 'skyacademy.it'
 170     _VALID_URL = r'https?://(?:www\.)?skyacademy\.it(?:/[^/]+)*/\d{4}/\d{2}/\d{2}/(?P<id>[^/?&#]+)'
 171     _TESTS = [{
 172         'url': 'https://www.skyacademy.it/eventi-speciali/2019/07/05/a-lezione-di-cinema-con-sky-academy-/',
 173         'md5': 'ced5c26638b7863190cbc44dd6f6ba08',
 174         'info_dict': {
 175             'id': '523458',
 176             'ext': 'mp4',
 177             'title': 'Sky Academy "The Best CineCamp 2019"',
 178             'timestamp': 1562843784,
 179             'upload_date': '20190711',
 180         }
 181     }]
 182     _DOMAIN = 'skyacademy'
 183     _VIDEO_ID_REGEX = r'id="news-videoId_(\d+)"'
 184
 185
 186 class SkyItArteIE(SkyItIE):
 187     IE_NAME = 'arte.sky.it'
 188     _VALID_URL = r'https?://arte\.sky\.it/video/(?P<id>[^/?&#]+)'
 189     _TESTS = [{
 190         'url': 'https://arte.sky.it/video/serie-musei-venezia-collezionismo-12-novembre/',
 191         'md5': '515aee97b87d7a018b6c80727d3e7e17',
 192         'info_dict': {
 193             'id': '627926',
 194             'ext': 'mp4',
 195             'title': "Musei Galleria Franchetti alla Ca' d'Oro Palazzo Grimani",
 196             'upload_date': '20201106',
 197             'timestamp': 1604664493,
 198         }
 199     }]
 200     _DOMAIN = 'skyarte'
 201     _VIDEO_ID_REGEX = r'(?s)<iframe[^>]+src="(?:https:)?//player\.sky\.it/player/external\.html\?[^"]*\bid=(\d+)'
 202
 203
 204 class CieloTVItIE(SkyItIE):
 205     IE_NAME = 'cielotv.it'
 206     _VALID_URL = r'https?://(?:www\.)?cielotv\.it/video/(?P<id>[^.]+)\.html'
 207     _TESTS = [{
 208         'url': 'https://www.cielotv.it/video/Il-lunedi-e-sempre-un-dramma.html',
 209         'md5': 'c4deed77552ba901c2a0d9258320304b',
 210         'info_dict': {
 211             'id': '499240',
 212             'ext': 'mp4',
 213             'title': 'Il lunedì è sempre un dramma',
 214             'upload_date': '20190329',
 215             'timestamp': 1553862178,
 216         }
 217     }]
 218     _DOMAIN = 'cielo'
 219     _VIDEO_ID_REGEX = r'videoId\s*=\s*"(\d+)"'
 220
 221
 222 class TV8ItIE(SkyItVideoIE):
 223     IE_NAME = 'tv8.it'
 224     _VALID_URL = r'https?://tv8\.it/showvideo/(?P<id>\d+)'
 225     _TESTS = [{
 226         'url': 'https://tv8.it/showvideo/630529/ogni-mattina-ucciso-asino-di-andrea-lo-cicero/18-11-2020/',
 227         'md5': '9ab906a3f75ea342ed928442f9dabd21',
 228         'info_dict': {
 229             'id': '630529',
 230             'ext': 'mp4',
 231             'title': 'Ogni mattina - Ucciso asino di Andrea Lo Cicero',
 232             'timestamp': 1605721374,
 233             'upload_date': '20201118',
 234         }
 235     }]
 236     _DOMAIN = 'mtv8'