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_str,
   7     compat_parse_qs,
   8     compat_urllib_parse_urlparse,
   9 )
  10 from ..utils import (
  11     dict_get,
  12     int_or_none,
  13     parse_duration,
  14     unified_timestamp,
  15 )
  16
  17
  18 class SkyItPlayerIE(InfoExtractor):
  19     IE_NAME = 'player.sky.it'
  20     _VALID_URL = r'https?://player\.sky\.it/player/(?:external|social)\.html\?.*?\bid=(?P<id>\d+)'
  21     _GEO_BYPASS = False
  22     _DOMAIN = 'sky'
  23     _PLAYER_TMPL = 'https://player.sky.it/player/external.html?id=%s&domain=%s'
  24     # http://static.sky.it/static/skyplayer/conf.json
  25     _TOKEN_MAP = {
  26         'cielo': 'Hh9O7M8ks5yi6nSROL7bKYz933rdf3GhwZlTLMgvy4Q',
  27         'hotclub': 'kW020K2jq2lk2eKRJD2vWEg832ncx2EivZlTLQput2C',
  28         'mtv8': 'A5Nn9GGb326CI7vP5e27d7E4PIaQjota',
  29         'salesforce': 'C6D585FD1615272C98DE38235F38BD86',
  30         'sitocommerciale': 'VJwfFuSGnLKnd9Phe9y96WkXgYDCguPMJ2dLhGMb2RE',
  31         'sky': 'F96WlOd8yoFmLQgiqv6fNQRvHZcsWk5jDaYnDvhbiJk',
  32         'skyacademy': 'A6LAn7EkO2Q26FRy0IAMBekX6jzDXYL3',
  33         'skyarte': 'LWk29hfiU39NNdq87ePeRach3nzTSV20o0lTv2001Cd',
  34         'theupfront': 'PRSGmDMsg6QMGc04Obpoy7Vsbn7i2Whp',
  35     }
  36
  37     def _player_url_result(self, video_id):
  38         return self.url_result(
  39             self._PLAYER_TMPL % (video_id, self._DOMAIN),
  40             SkyItPlayerIE.ie_key(), video_id)
  41
  42     def _parse_video(self, video, video_id):
  43         title = video['title']
  44         is_live = video.get('type') == 'live'
  45         hls_url = video.get(('streaming' if is_live else 'hls') + '_url')
  46         if not hls_url and video.get('geoblock' if is_live else 'geob'):
  47             self.raise_geo_restricted(countries=['IT'])
  48
  49         if is_live:
  50             formats = self._extract_m3u8_formats(hls_url, video_id, 'mp4')
  51         else:
  52             formats = self._extract_akamai_formats(
  53                 hls_url, video_id, {'http': 'videoplatform.sky.it'})
  54         self._sort_formats(formats)
  55
  56         return {
  57             'id': video_id,
  58             'title': self._live_title(title) if is_live else title,
  59             'formats': formats,
  60             'thumbnail': dict_get(video, ('video_still', 'video_still_medium', 'thumb')),
  61             'description': video.get('short_desc') or None,
  62             'timestamp': unified_timestamp(video.get('create_date')),
  63             'duration': int_or_none(video.get('duration_sec')) or parse_duration(video.get('duration')),
  64             'is_live': is_live,
  65         }
  66
  67     def _real_extract(self, url):
  68         video_id = self._match_id(url)
  69         domain = compat_parse_qs(compat_urllib_parse_urlparse(
  70             url).query).get('domain', [None])[0]
  71         token = dict_get(self._TOKEN_MAP, (domain, 'sky'))
  72         video = self._download_json(
  73             'https://apid.sky.it/vdp/v1/getVideoData',
  74             video_id, query={
  75                 'caller': 'sky',
  76                 'id': video_id,
  77                 'token': token
  78             }, headers=self.geo_verification_headers())
  79         return self._parse_video(video, video_id)
  80
  81
  82 class SkyItVideoIE(SkyItPlayerIE):
  83     IE_NAME = 'video.sky.it'
  84     _VALID_URL = r'https?://(?:masterchef|video|xfactor)\.sky\.it(?:/[^/]+)*/video/[0-9a-z-]+-(?P<id>\d+)'
  85     _TESTS = [{
  86         'url': 'https://video.sky.it/news/mondo/video/uomo-ucciso-da-uno-squalo-in-australia-631227',
  87         'md5': 'fe5c91e59a84a3437eaa0bca6e134ccd',
  88         'info_dict': {
  89             'id': '631227',
  90             'ext': 'mp4',
  91             'title': 'Uomo ucciso da uno squalo in Australia',
  92             'timestamp': 1606036192,
  93             'upload_date': '20201122',
  94         }
  95     }, {
  96         'url': 'https://xfactor.sky.it/video/x-factor-2020-replay-audizioni-1-615820',
  97         'only_matching': True,
  98     }, {
  99         'url': 'https://masterchef.sky.it/video/masterchef-9-cosa-e-successo-nella-prima-puntata-562831',
 100         'only_matching': True,
 101     }]
 102
 103     def _real_extract(self, url):
 104         video_id = self._match_id(url)
 105         return self._player_url_result(video_id)
 106
 107
 108 class SkyItVideoLiveIE(SkyItPlayerIE):
 109     IE_NAME = 'video.sky.it:live'
 110     _VALID_URL = r'https?://video\.sky\.it/diretta/(?P<id>[^/?&#]+)'
 111     _TEST = {
 112         'url': 'https://video.sky.it/diretta/tg24',
 113         'info_dict': {
 114             'id': '1',
 115             'ext': 'mp4',
 116             'title': r're:Diretta TG24 \d{4}-\d{2}-\d{2} \d{2}:\d{2}',
 117             'description': 'Guarda la diretta streaming di SkyTg24, segui con Sky tutti gli appuntamenti e gli speciali di Tg24.',
 118         },
 119         'params': {
 120             # m3u8 download
 121             'skip_download': True,
 122         },
 123     }
 124
 125     def _real_extract(self, url):
 126         display_id = self._match_id(url)
 127         webpage = self._download_webpage(url, display_id)
 128         asset_id = compat_str(self._parse_json(self._search_regex(
 129             r'<script[^>]+id="__NEXT_DATA__"[^>]*>({.+?})</script>',
 130             webpage, 'next data'), display_id)['props']['initialState']['livePage']['content']['asset_id'])
 131         livestream = self._download_json(
 132             'https://apid.sky.it/vdp/v1/getLivestream',
 133             asset_id, query={'id': asset_id})
 134         return self._parse_video(livestream, asset_id)
 135
 136
 137 class SkyItIE(SkyItPlayerIE):
 138     IE_NAME = 'sky.it'
 139     _VALID_URL = r'https?://(?:sport|tg24)\.sky\.it(?:/[^/]+)*/\d{4}/\d{2}/\d{2}/(?P<id>[^/?&#]+)'
 140     _TESTS = [{
 141         'url': 'https://sport.sky.it/calcio/serie-a/2020/11/21/juventus-cagliari-risultato-gol',
 142         'info_dict': {
 143             'id': '631201',
 144             'ext': 'mp4',
 145             'title': 'Un rosso alla violenza: in campo per i diritti delle donne',
 146             'upload_date': '20201121',
 147             'timestamp': 1605995753,
 148         },
 149         'expected_warnings': ['Unable to download f4m manifest'],
 150     }, {
 151         'url': 'https://tg24.sky.it/mondo/2020/11/22/australia-squalo-uccide-uomo',
 152         'md5': 'fe5c91e59a84a3437eaa0bca6e134ccd',
 153         'info_dict': {
 154             'id': '631227',
 155             'ext': 'mp4',
 156             'title': 'Uomo ucciso da uno squalo in Australia',
 157             'timestamp': 1606036192,
 158             'upload_date': '20201122',
 159         },
 160     }]
 161     _VIDEO_ID_REGEX = r'data-videoid="(\d+)"'
 162
 163     def _real_extract(self, url):
 164         display_id = self._match_id(url)
 165         webpage = self._download_webpage(url, display_id)
 166         video_id = self._search_regex(
 167             self._VIDEO_ID_REGEX, webpage, 'video id')
 168         return self._player_url_result(video_id)
 169
 170
 171 class SkyItAcademyIE(SkyItIE):
 172     IE_NAME = 'skyacademy.it'
 173     _VALID_URL = r'https?://(?:www\.)?skyacademy\.it(?:/[^/]+)*/\d{4}/\d{2}/\d{2}/(?P<id>[^/?&#]+)'
 174     _TESTS = [{
 175         'url': 'https://www.skyacademy.it/eventi-speciali/2019/07/05/a-lezione-di-cinema-con-sky-academy-/',
 176         'md5': 'ced5c26638b7863190cbc44dd6f6ba08',
 177         'info_dict': {
 178             'id': '523458',
 179             'ext': 'mp4',
 180             'title': 'Sky Academy "The Best CineCamp 2019"',
 181             'timestamp': 1562843784,
 182             'upload_date': '20190711',
 183         }
 184     }]
 185     _DOMAIN = 'skyacademy'
 186     _VIDEO_ID_REGEX = r'id="news-videoId_(\d+)"'
 187
 188
 189 class SkyItArteIE(SkyItIE):
 190     IE_NAME = 'arte.sky.it'
 191     _VALID_URL = r'https?://arte\.sky\.it/video/(?P<id>[^/?&#]+)'
 192     _TESTS = [{
 193         'url': 'https://arte.sky.it/video/serie-musei-venezia-collezionismo-12-novembre/',
 194         'md5': '515aee97b87d7a018b6c80727d3e7e17',
 195         'info_dict': {
 196             'id': '627926',
 197             'ext': 'mp4',
 198             'title': "Musei Galleria Franchetti alla Ca' d'Oro Palazzo Grimani",
 199             'upload_date': '20201106',
 200             'timestamp': 1604664493,
 201         }
 202     }]
 203     _DOMAIN = 'skyarte'
 204     _VIDEO_ID_REGEX = r'(?s)<iframe[^>]+src="(?:https:)?//player\.sky\.it/player/external\.html\?[^"]*\bid=(\d+)'
 205
 206
 207 class CieloTVItIE(SkyItIE):
 208     IE_NAME = 'cielotv.it'
 209     _VALID_URL = r'https?://(?:www\.)?cielotv\.it/video/(?P<id>[^.]+)\.html'
 210     _TESTS = [{
 211         'url': 'https://www.cielotv.it/video/Il-lunedi-e-sempre-un-dramma.html',
 212         'md5': 'c4deed77552ba901c2a0d9258320304b',
 213         'info_dict': {
 214             'id': '499240',
 215             'ext': 'mp4',
 216             'title': 'Il lunedì è sempre un dramma',
 217             'upload_date': '20190329',
 218             'timestamp': 1553862178,
 219         }
 220     }]
 221     _DOMAIN = 'cielo'
 222     _VIDEO_ID_REGEX = r'videoId\s*=\s*"(\d+)"'
 223
 224
 225 class TV8ItIE(SkyItVideoIE):
 226     IE_NAME = 'tv8.it'
 227     _VALID_URL = r'https?://tv8\.it/showvideo/(?P<id>\d+)'
 228     _TESTS = [{
 229         'url': 'https://tv8.it/showvideo/630529/ogni-mattina-ucciso-asino-di-andrea-lo-cicero/18-11-2020/',
 230         'md5': '9ab906a3f75ea342ed928442f9dabd21',
 231         'info_dict': {
 232             'id': '630529',
 233             'ext': 'mp4',
 234             'title': 'Ogni mattina - Ucciso asino di Andrea Lo Cicero',
 235             'timestamp': 1605721374,
 236             'upload_date': '20201118',
 237         }
 238     }]
 239     _DOMAIN = 'mtv8'