[yt-dlp.git] / yt_dlp / extractor / skyit.py

from .common import InfoExtractor
from ..compat import (
    compat_parse_qs,
    compat_urllib_parse_urlparse,
)
from ..utils import (
    dict_get,
    int_or_none,
    parse_duration,
    unified_timestamp,
)


class SkyItPlayerIE(InfoExtractor):
    IE_NAME = 'player.sky.it'
    _VALID_URL = r'https?://player\.sky\.it/player/(?:external|social)\.html\?.*?\bid=(?P<id>\d+)'
    _GEO_BYPASS = False
    _DOMAIN = 'sky'
    _PLAYER_TMPL = 'https://player.sky.it/player/external.html?id=%s&domain=%s'
    # http://static.sky.it/static/skyplayer/conf.json
    _TOKEN_MAP = {
        'cielo': 'Hh9O7M8ks5yi6nSROL7bKYz933rdf3GhwZlTLMgvy4Q',
        'hotclub': 'kW020K2jq2lk2eKRJD2vWEg832ncx2EivZlTLQput2C',
        'mtv8': 'A5Nn9GGb326CI7vP5e27d7E4PIaQjota',
        'salesforce': 'C6D585FD1615272C98DE38235F38BD86',
        'sitocommerciale': 'VJwfFuSGnLKnd9Phe9y96WkXgYDCguPMJ2dLhGMb2RE',
        'sky': 'F96WlOd8yoFmLQgiqv6fNQRvHZcsWk5jDaYnDvhbiJk',
        'skyarte': 'LWk29hfiU39NNdq87ePeRach3nzTSV20o0lTv2001Cd',
        'theupfront': 'PRSGmDMsg6QMGc04Obpoy7Vsbn7i2Whp',
    }

    def _player_url_result(self, video_id):
        return self.url_result(
            self._PLAYER_TMPL % (video_id, self._DOMAIN),
            SkyItPlayerIE.ie_key(), video_id)

    def _parse_video(self, video, video_id):
        title = video['title']
        is_live = video.get('type') == 'live'
        hls_url = video.get(('streaming' if is_live else 'hls') + '_url')
        if not hls_url and video.get('geoblock' if is_live else 'geob'):
            self.raise_geo_restricted(countries=['IT'])

        formats = self._extract_m3u8_formats(hls_url, video_id, 'mp4')
        self._sort_formats(formats)

        return {
            'id': video_id,
            'title': title,
            'formats': formats,
            'thumbnail': dict_get(video, ('video_still', 'video_still_medium', 'thumb')),
            'description': video.get('short_desc') or None,
            'timestamp': unified_timestamp(video.get('create_date')),
            'duration': int_or_none(video.get('duration_sec')) or parse_duration(video.get('duration')),
            'is_live': is_live,
        }

    def _real_extract(self, url):
        video_id = self._match_id(url)
        domain = compat_parse_qs(compat_urllib_parse_urlparse(
            url).query).get('domain', [None])[0]
        token = dict_get(self._TOKEN_MAP, (domain, 'sky'))
        video = self._download_json(
            'https://apid.sky.it/vdp/v1/getVideoData',
            video_id, query={
                'caller': 'sky',
                'id': video_id,
                'token': token
            }, headers=self.geo_verification_headers())
        return self._parse_video(video, video_id)


class SkyItVideoIE(SkyItPlayerIE):  # XXX: Do not subclass from concrete IE
    IE_NAME = 'video.sky.it'
    _VALID_URL = r'https?://(?:masterchef|video|xfactor)\.sky\.it(?:/[^/]+)*/video/[0-9a-z-]+-(?P<id>\d+)'
    _TESTS = [{
        'url': 'https://video.sky.it/news/mondo/video/uomo-ucciso-da-uno-squalo-in-australia-631227',
        'md5': '5b858a62d9ffe2ab77b397553024184a',
        'info_dict': {
            'id': '631227',
            'ext': 'mp4',
            'title': 'Uomo ucciso da uno squalo in Australia',
            'timestamp': 1606036192,
            'upload_date': '20201122',
            'duration': 26,
            'thumbnail': 'https://video.sky.it/captures/thumbs/631227/631227_thumb_880x494.jpg',
        },
        'params': {'skip_download': 'm3u8'},
    }, {
        'url': 'https://xfactor.sky.it/video/x-factor-2020-replay-audizioni-1-615820',
        'only_matching': True,
    }, {
        'url': 'https://masterchef.sky.it/video/masterchef-9-cosa-e-successo-nella-prima-puntata-562831',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        video_id = self._match_id(url)
        return self._player_url_result(video_id)


class SkyItVideoLiveIE(SkyItPlayerIE):  # XXX: Do not subclass from concrete IE
    IE_NAME = 'video.sky.it:live'
    _VALID_URL = r'https?://video\.sky\.it/diretta/(?P<id>[^/?&#]+)'
    _TEST = {
        'url': 'https://video.sky.it/diretta/tg24',
        'info_dict': {
            'id': '1',
            'ext': 'mp4',
            'title': r're:Diretta TG24 \d{4}-\d{2}-\d{2} \d{2}:\d{2}',
            'description': r're:(?:Clicca play e )?[Gg]uarda la diretta streaming di SkyTg24, segui con Sky tutti gli appuntamenti e gli speciali di Tg24\.',
            'live_status': 'is_live',
        },
        'params': {
            # m3u8 download
            'skip_download': True,
        },
    }

    def _real_extract(self, url):
        display_id = self._match_id(url)
        webpage = self._download_webpage(url, display_id)
        asset_id = str(self._search_nextjs_data(webpage, display_id)['props']['initialState']['livePage']['content']['asset_id'])
        livestream = self._download_json(
            'https://apid.sky.it/vdp/v1/getLivestream',
            asset_id, query={'id': asset_id})
        return self._parse_video(livestream, asset_id)


class SkyItIE(SkyItPlayerIE):  # XXX: Do not subclass from concrete IE
    IE_NAME = 'sky.it'
    _VALID_URL = r'https?://(?:sport|tg24)\.sky\.it(?:/[^/]+)*/\d{4}/\d{2}/\d{2}/(?P<id>[^/?&#]+)'
    _TESTS = [{
        'url': 'https://sport.sky.it/calcio/serie-a/2022/11/03/brozovic-inter-news',
        'info_dict': {
            'id': '789222',
            'ext': 'mp4',
            'title': 'Brozovic con il gruppo: verso convocazione per Juve-Inter',
            'upload_date': '20221103',
            'timestamp': 1667484130,
            'duration': 22,
            'thumbnail': 'https://videoplatform.sky.it/still/2022/11/03/1667480526353_brozovic_videostill_1.jpg',
        },
        'params': {'skip_download': 'm3u8'},
    }, {
        'url': 'https://tg24.sky.it/mondo/2020/11/22/australia-squalo-uccide-uomo',
        'md5': 'fe5c91e59a84a3437eaa0bca6e134ccd',
        'info_dict': {
            'id': '631227',
            'ext': 'mp4',
            'title': 'Uomo ucciso da uno squalo in Australia',
            'timestamp': 1606036192,
            'upload_date': '20201122',
            'duration': 26,
            'thumbnail': 'https://video.sky.it/captures/thumbs/631227/631227_thumb_880x494.jpg',
        },
        'params': {'skip_download': 'm3u8'},
    }]
    _VIDEO_ID_REGEX = r'data-videoid="(\d+)"'

    def _real_extract(self, url):
        display_id = self._match_id(url)
        webpage = self._download_webpage(url, display_id)
        video_id = self._search_regex(
            self._VIDEO_ID_REGEX, webpage, 'video id')
        return self._player_url_result(video_id)


class SkyItArteIE(SkyItIE):  # XXX: Do not subclass from concrete IE
    IE_NAME = 'arte.sky.it'
    _VALID_URL = r'https?://arte\.sky\.it/video/(?P<id>[^/?&#]+)'
    _TESTS = [{
        'url': 'https://arte.sky.it/video/oliviero-toscani-torino-galleria-mazzoleni-788962',
        'md5': '515aee97b87d7a018b6c80727d3e7e17',
        'info_dict': {
            'id': '788962',
            'ext': 'mp4',
            'title': 'La fotografia di Oliviero Toscani conquista Torino',
            'upload_date': '20221102',
            'timestamp': 1667399996,
            'duration': 12,
            'thumbnail': 'https://videoplatform.sky.it/still/2022/11/02/1667396388552_oliviero-toscani-torino-galleria-mazzoleni_videostill_1.jpg',
        },
        'params': {'skip_download': 'm3u8'},
    }]
    _DOMAIN = 'skyarte'
    _VIDEO_ID_REGEX = r'"embedUrl"\s*:\s*"(?:https:)?//player\.sky\.it/player/external\.html\?[^"]*\bid=(\d+)'


class CieloTVItIE(SkyItIE):  # XXX: Do not subclass from concrete IE
    IE_NAME = 'cielotv.it'
    _VALID_URL = r'https?://(?:www\.)?cielotv\.it/video/(?P<id>[^.]+)\.html'
    _TESTS = [{
        'url': 'https://www.cielotv.it/video/Il-lunedi-e-sempre-un-dramma.html',
        'md5': 'c4deed77552ba901c2a0d9258320304b',
        'info_dict': {
            'id': '499240',
            'ext': 'mp4',
            'title': 'Il lunedì è sempre un dramma',
            'upload_date': '20190329',
            'timestamp': 1553862178,
            'duration': 30,
            'thumbnail': 'https://videoplatform.sky.it/still/2019/03/29/1553858575610_lunedi_dramma_mant_videostill_1.jpg',
        },
        'params': {'skip_download': 'm3u8'},
    }]
    _DOMAIN = 'cielo'
    _VIDEO_ID_REGEX = r'videoId\s*=\s*"(\d+)"'


class TV8ItIE(SkyItVideoIE):  # XXX: Do not subclass from concrete IE
    IE_NAME = 'tv8.it'
    _VALID_URL = r'https?://(?:www\.)?tv8\.it/(?:show)?video/[0-9a-z-]+-(?P<id>\d+)'
    _TESTS = [{
        'url': 'https://www.tv8.it/video/ogni-mattina-ucciso-asino-di-andrea-lo-cicero-630529',
        'md5': '9ab906a3f75ea342ed928442f9dabd21',
        'info_dict': {
            'id': '630529',
            'ext': 'mp4',
            'title': 'Ogni mattina - Ucciso asino di Andrea Lo Cicero',
            'timestamp': 1605721374,
            'upload_date': '20201118',
            'duration': 114,
            'thumbnail': 'https://videoplatform.sky.it/still/2020/11/18/1605717753954_ogni-mattina-ucciso-asino-di-andrea-lo-cicero_videostill_1.jpg',
        },
        'params': {'skip_download': 'm3u8'},
    }]
    _DOMAIN = 'mtv8'
Commit	Line	Data
12300fa4 RA	1	from .common import InfoExtractor
12300fa4 RA	2	from ..compat import (
12300fa4 RA	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',
12300fa4 RA	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
d715b0e4	44	formats = self._extract_m3u8_formats(hls_url, video_id, 'mp4')
12300fa4 RA	45	self._sort_formats(formats)
	46
	47	return {
	48	'id': video_id,
39ca3b5c	49	'title': title,
12300fa4 RA	50	'formats': formats,
	51	'thumbnail': dict_get(video, ('video_still', 'video_still_medium', 'thumb')),
	52	'description': video.get('short_desc') or None,
	53	'timestamp': unified_timestamp(video.get('create_date')),
	54	'duration': int_or_none(video.get('duration_sec')) or parse_duration(video.get('duration')),
	55	'is_live': is_live,
	56	}
	57
	58	def _real_extract(self, url):
	59	video_id = self._match_id(url)
	60	domain = compat_parse_qs(compat_urllib_parse_urlparse(
	61	url).query).get('domain', [None])[0]
	62	token = dict_get(self._TOKEN_MAP, (domain, 'sky'))
	63	video = self._download_json(
	64	'https://apid.sky.it/vdp/v1/getVideoData',
	65	video_id, query={
	66	'caller': 'sky',
	67	'id': video_id,
	68	'token': token
	69	}, headers=self.geo_verification_headers())
	70	return self._parse_video(video, video_id)
	71
	72
6368e2e6	73	class SkyItVideoIE(SkyItPlayerIE): # XXX: Do not subclass from concrete IE
12300fa4 RA	74	IE_NAME = 'video.sky.it'
	75	_VALID_URL = r'https?://(?:masterchef\|video\|xfactor)\.sky\.it(?:/[^/]+)*/video/[0-9a-z-]+-(?P<id>\d+)'
	76	_TESTS = [{
	77	'url': 'https://video.sky.it/news/mondo/video/uomo-ucciso-da-uno-squalo-in-australia-631227',
d715b0e4	78	'md5': '5b858a62d9ffe2ab77b397553024184a',
12300fa4 RA	79	'info_dict': {
	80	'id': '631227',
	81	'ext': 'mp4',
	82	'title': 'Uomo ucciso da uno squalo in Australia',
	83	'timestamp': 1606036192,
	84	'upload_date': '20201122',
d715b0e4	85	'duration': 26,
	86	'thumbnail': 'https://video.sky.it/captures/thumbs/631227/631227_thumb_880x494.jpg',
	87	},
	88	'params': {'skip_download': 'm3u8'},
12300fa4 RA	89	}, {
	90	'url': 'https://xfactor.sky.it/video/x-factor-2020-replay-audizioni-1-615820',
	91	'only_matching': True,
	92	}, {
	93	'url': 'https://masterchef.sky.it/video/masterchef-9-cosa-e-successo-nella-prima-puntata-562831',
	94	'only_matching': True,
	95	}]
	96
	97	def _real_extract(self, url):
	98	video_id = self._match_id(url)
	99	return self._player_url_result(video_id)
	100
	101
6368e2e6	102	class SkyItVideoLiveIE(SkyItPlayerIE): # XXX: Do not subclass from concrete IE
12300fa4 RA	103	IE_NAME = 'video.sky.it:live'
	104	_VALID_URL = r'https?://video\.sky\.it/diretta/(?P<id>[^/?&#]+)'
	105	_TEST = {
	106	'url': 'https://video.sky.it/diretta/tg24',
	107	'info_dict': {
	108	'id': '1',
	109	'ext': 'mp4',
	110	'title': r're:Diretta TG24 \d{4}-\d{2}-\d{2} \d{2}:\d{2}',
d715b0e4	111	'description': r're:(?:Clicca play e )?[Gg]uarda la diretta streaming di SkyTg24, segui con Sky tutti gli appuntamenti e gli speciali di Tg24\.',
d715b0e4	112	'live_status': 'is_live',
12300fa4 RA	113	},
	114	'params': {
	115	# m3u8 download
	116	'skip_download': True,
	117	},
	118	}
	119
	120	def _real_extract(self, url):
	121	display_id = self._match_id(url)
	122	webpage = self._download_webpage(url, display_id)
135dfa2c	123	asset_id = str(self._search_nextjs_data(webpage, display_id)['props']['initialState']['livePage']['content']['asset_id'])
12300fa4 RA	124	livestream = self._download_json(
	125	'https://apid.sky.it/vdp/v1/getLivestream',
	126	asset_id, query={'id': asset_id})
	127	return self._parse_video(livestream, asset_id)
	128
	129
6368e2e6	130	class SkyItIE(SkyItPlayerIE): # XXX: Do not subclass from concrete IE
12300fa4 RA	131	IE_NAME = 'sky.it'
	132	_VALID_URL = r'https?://(?:sport\|tg24)\.sky\.it(?:/[^/]+)*/\d{4}/\d{2}/\d{2}/(?P<id>[^/?&#]+)'
	133	_TESTS = [{
d715b0e4	134	'url': 'https://sport.sky.it/calcio/serie-a/2022/11/03/brozovic-inter-news',
12300fa4	135	'info_dict': {
d715b0e4	136	'id': '789222',
12300fa4	137	'ext': 'mp4',
d715b0e4	138	'title': 'Brozovic con il gruppo: verso convocazione per Juve-Inter',
	139	'upload_date': '20221103',
	140	'timestamp': 1667484130,
	141	'duration': 22,
	142	'thumbnail': 'https://videoplatform.sky.it/still/2022/11/03/1667480526353_brozovic_videostill_1.jpg',
12300fa4	143	},
d715b0e4	144	'params': {'skip_download': 'm3u8'},
12300fa4 RA	145	}, {
	146	'url': 'https://tg24.sky.it/mondo/2020/11/22/australia-squalo-uccide-uomo',
	147	'md5': 'fe5c91e59a84a3437eaa0bca6e134ccd',
	148	'info_dict': {
	149	'id': '631227',
	150	'ext': 'mp4',
	151	'title': 'Uomo ucciso da uno squalo in Australia',
	152	'timestamp': 1606036192,
	153	'upload_date': '20201122',
d715b0e4	154	'duration': 26,
d715b0e4	155	'thumbnail': 'https://video.sky.it/captures/thumbs/631227/631227_thumb_880x494.jpg',
12300fa4	156	},
d715b0e4	157	'params': {'skip_download': 'm3u8'},
12300fa4 RA	158	}]
	159	_VIDEO_ID_REGEX = r'data-videoid="(\d+)"'
	160
	161	def _real_extract(self, url):
	162	display_id = self._match_id(url)
	163	webpage = self._download_webpage(url, display_id)
	164	video_id = self._search_regex(
	165	self._VIDEO_ID_REGEX, webpage, 'video id')
	166	return self._player_url_result(video_id)
	167
	168
6368e2e6	169	class SkyItArteIE(SkyItIE): # XXX: Do not subclass from concrete IE
12300fa4 RA	170	IE_NAME = 'arte.sky.it'
	171	_VALID_URL = r'https?://arte\.sky\.it/video/(?P<id>[^/?&#]+)'
	172	_TESTS = [{
d715b0e4	173	'url': 'https://arte.sky.it/video/oliviero-toscani-torino-galleria-mazzoleni-788962',
12300fa4 RA	174	'md5': '515aee97b87d7a018b6c80727d3e7e17',
12300fa4 RA	175	'info_dict': {
d715b0e4	176	'id': '788962',
12300fa4	177	'ext': 'mp4',
d715b0e4	178	'title': 'La fotografia di Oliviero Toscani conquista Torino',
	179	'upload_date': '20221102',
	180	'timestamp': 1667399996,
	181	'duration': 12,
	182	'thumbnail': 'https://videoplatform.sky.it/still/2022/11/02/1667396388552_oliviero-toscani-torino-galleria-mazzoleni_videostill_1.jpg',
	183	},
	184	'params': {'skip_download': 'm3u8'},
12300fa4 RA	185	}]
12300fa4 RA	186	_DOMAIN = 'skyarte'
d715b0e4	187	_VIDEO_ID_REGEX = r'"embedUrl"\s:\s"(?:https:)?//player\.sky\.it/player/external\.html\?[^"]*\bid=(\d+)'
12300fa4 RA	188
12300fa4 RA	189
6368e2e6	190	class CieloTVItIE(SkyItIE): # XXX: Do not subclass from concrete IE
12300fa4 RA	191	IE_NAME = 'cielotv.it'
	192	_VALID_URL = r'https?://(?:www\.)?cielotv\.it/video/(?P<id>[^.]+)\.html'
	193	_TESTS = [{
	194	'url': 'https://www.cielotv.it/video/Il-lunedi-e-sempre-un-dramma.html',
	195	'md5': 'c4deed77552ba901c2a0d9258320304b',
	196	'info_dict': {
	197	'id': '499240',
	198	'ext': 'mp4',
	199	'title': 'Il lunedì è sempre un dramma',
	200	'upload_date': '20190329',
	201	'timestamp': 1553862178,
d715b0e4	202	'duration': 30,
	203	'thumbnail': 'https://videoplatform.sky.it/still/2019/03/29/1553858575610_lunedi_dramma_mant_videostill_1.jpg',
	204	},
	205	'params': {'skip_download': 'm3u8'},
12300fa4 RA	206	}]
	207	_DOMAIN = 'cielo'
	208	_VIDEO_ID_REGEX = r'videoId\s=\s"(\d+)"'
	209
	210
6368e2e6	211	class TV8ItIE(SkyItVideoIE): # XXX: Do not subclass from concrete IE
12300fa4	212	IE_NAME = 'tv8.it'
d715b0e4	213	_VALID_URL = r'https?://(?:www\.)?tv8\.it/(?:show)?video/[0-9a-z-]+-(?P<id>\d+)'
12300fa4	214	_TESTS = [{
d715b0e4	215	'url': 'https://www.tv8.it/video/ogni-mattina-ucciso-asino-di-andrea-lo-cicero-630529',
12300fa4 RA	216	'md5': '9ab906a3f75ea342ed928442f9dabd21',
	217	'info_dict': {
	218	'id': '630529',
	219	'ext': 'mp4',
	220	'title': 'Ogni mattina - Ucciso asino di Andrea Lo Cicero',
	221	'timestamp': 1605721374,
	222	'upload_date': '20201118',
d715b0e4	223	'duration': 114,
	224	'thumbnail': 'https://videoplatform.sky.it/still/2020/11/18/1605717753954_ogni-mattina-ucciso-asino-di-andrea-lo-cicero_videostill_1.jpg',
	225	},
	226	'params': {'skip_download': 'm3u8'},
12300fa4 RA	227	}]
12300fa4 RA	228	_DOMAIN = 'mtv8'