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

# coding: utf-8
from __future__ import unicode_literals


from .common import InfoExtractor
from ..compat import compat_HTTPError
from ..utils import (
    float_or_none,
    ExtractorError,
)


class RedBullTVIE(InfoExtractor):
    _VALID_URL = r'https?://(?:www\.)?redbull(?:\.tv|\.com(?:/[^/]+)?(?:/tv)?)(?:/events/[^/]+)?/(?:videos?|live|(?:film|episode)s)/(?P<id>AP-\w+)'
    _TESTS = [{
        # film
        'url': 'https://www.redbull.tv/video/AP-1Q6XCDTAN1W11',
        'md5': 'fb0445b98aa4394e504b413d98031d1f',
        'info_dict': {
            'id': 'AP-1Q6XCDTAN1W11',
            'ext': 'mp4',
            'title': 'ABC of... WRC - ABC of... S1E6',
            'description': 'md5:5c7ed8f4015c8492ecf64b6ab31e7d31',
            'duration': 1582.04,
        },
    }, {
        # episode
        'url': 'https://www.redbull.tv/video/AP-1PMHKJFCW1W11',
        'info_dict': {
            'id': 'AP-1PMHKJFCW1W11',
            'ext': 'mp4',
            'title': 'Grime - Hashtags S2E4',
            'description': 'md5:5546aa612958c08a98faaad4abce484d',
            'duration': 904,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'https://www.redbull.com/int-en/tv/video/AP-1UWHCAR9S1W11/rob-meets-sam-gaze?playlist=playlists::3f81040a-2f31-4832-8e2e-545b1d39d173',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/us-en/videos/AP-1YM9QCYE52111',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/us-en/events/AP-1XV2K61Q51W11/live/AP-1XUJ86FDH1W11',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/int-en/films/AP-1ZSMAW8FH2111',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/int-en/episodes/AP-1TQWK7XE11W11',
        'only_matching': True,
    }]

    def extract_info(self, video_id):
        session = self._download_json(
            'https://api.redbull.tv/v3/session', video_id,
            note='Downloading access token', query={
                'category': 'personal_computer',
                'os_family': 'http',
            })
        if session.get('code') == 'error':
            raise ExtractorError('%s said: %s' % (
                self.IE_NAME, session['message']))
        token = session['token']

        try:
            video = self._download_json(
                'https://api.redbull.tv/v3/products/' + video_id,
                video_id, note='Downloading video information',
                headers={'Authorization': token}
            )
        except ExtractorError as e:
            if isinstance(e.cause, compat_HTTPError) and e.cause.code == 404:
                error_message = self._parse_json(
                    e.cause.read().decode(), video_id)['error']
                raise ExtractorError('%s said: %s' % (
                    self.IE_NAME, error_message), expected=True)
            raise

        title = video['title'].strip()

        formats = self._extract_m3u8_formats(
            'https://dms.redbull.tv/v3/%s/%s/playlist.m3u8' % (video_id, token),
            video_id, 'mp4', entry_protocol='m3u8_native', m3u8_id='hls')
        self._sort_formats(formats)

        subtitles = {}
        for resource in video.get('resources', []):
            if resource.startswith('closed_caption_'):
                splitted_resource = resource.split('_')
                if splitted_resource[2]:
                    subtitles.setdefault('en', []).append({
                        'url': 'https://resources.redbull.tv/%s/%s' % (video_id, resource),
                        'ext': splitted_resource[2],
                    })

        subheading = video.get('subheading')
        if subheading:
            title += ' - %s' % subheading

        return {
            'id': video_id,
            'title': title,
            'description': video.get('long_description') or video.get(
                'short_description'),
            'duration': float_or_none(video.get('duration'), scale=1000),
            'formats': formats,
            'subtitles': subtitles,
        }

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


class RedBullEmbedIE(RedBullTVIE):
    _VALID_URL = r'https?://(?:www\.)?redbull\.com/embed/(?P<id>rrn:content:[^:]+:[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}:[a-z]{2}-[A-Z]{2,3})'
    _TESTS = [{
        # HLS manifest accessible only using assetId
        'url': 'https://www.redbull.com/embed/rrn:content:episode-videos:f3021f4f-3ed4-51ac-915a-11987126e405:en-INT',
        'only_matching': True,
    }]
    _VIDEO_ESSENSE_TMPL = '''... on %s {
      videoEssence {
        attributes
      }
    }'''

    def _real_extract(self, url):
        rrn_id = self._match_id(url)
        asset_id = self._download_json(
            'https://edge-graphql.crepo-production.redbullaws.com/v1/graphql',
            rrn_id, headers={
                'Accept': 'application/json',
                'API-KEY': 'e90a1ff11335423998b100c929ecc866',
            }, query={
                'query': '''{
  resource(id: "%s", enforceGeoBlocking: false) {
    %s
    %s
  }
}''' % (rrn_id, self._VIDEO_ESSENSE_TMPL % 'LiveVideo', self._VIDEO_ESSENSE_TMPL % 'VideoResource'),
            })['data']['resource']['videoEssence']['attributes']['assetId']
        return self.extract_info(asset_id)


class RedBullTVRrnContentIE(InfoExtractor):
    _VALID_URL = r'https?://(?:www\.)?redbull\.com/(?P<region>[a-z]{2,3})-(?P<lang>[a-z]{2})/tv/(?:video|live|film)/(?P<id>rrn:content:[^:]+:[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'
    _TESTS = [{
        'url': 'https://www.redbull.com/int-en/tv/video/rrn:content:live-videos:e3e6feb4-e95f-50b7-962a-c70f8fd13c73/mens-dh-finals-fort-william',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/int-en/tv/video/rrn:content:videos:a36a0f36-ff1b-5db8-a69d-ee11a14bf48b/tn-ts-style?playlist=rrn:content:event-profiles:83f05926-5de8-5389-b5e4-9bb312d715e8:extras',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/int-en/tv/film/rrn:content:films:d1f4d00e-4c04-5d19-b510-a805ffa2ab83/follow-me',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        region, lang, rrn_id = self._match_valid_url(url).groups()
        rrn_id += ':%s-%s' % (lang, region.upper())
        return self.url_result(
            'https://www.redbull.com/embed/' + rrn_id,
            RedBullEmbedIE.ie_key(), rrn_id)


class RedBullIE(InfoExtractor):
    _VALID_URL = r'https?://(?:www\.)?redbull\.com/(?P<region>[a-z]{2,3})-(?P<lang>[a-z]{2})/(?P<type>(?:episode|film|(?:(?:recap|trailer)-)?video)s|live)/(?!AP-|rrn:content:)(?P<id>[^/?#&]+)'
    _TESTS = [{
        'url': 'https://www.redbull.com/int-en/episodes/grime-hashtags-s02-e04',
        'md5': 'db8271a7200d40053a1809ed0dd574ff',
        'info_dict': {
            'id': 'AA-1MT8DQWA91W14',
            'ext': 'mp4',
            'title': 'Grime - Hashtags S2E4',
            'description': 'md5:5546aa612958c08a98faaad4abce484d',
        },
    }, {
        'url': 'https://www.redbull.com/int-en/films/kilimanjaro-mountain-of-greatness',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/int-en/recap-videos/uci-mountain-bike-world-cup-2017-mens-xco-finals-from-vallnord',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/int-en/trailer-videos/kings-of-content',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/int-en/videos/tnts-style-red-bull-dance-your-style-s1-e12',
        'only_matching': True,
    }, {
        'url': 'https://www.redbull.com/int-en/live/mens-dh-finals-fort-william',
        'only_matching': True,
    }, {
        # only available on the int-en website so a fallback is need for the API
        # https://www.redbull.com/v3/api/graphql/v1/v3/query/en-GB>en-INT?filter[uriSlug]=fia-wrc-saturday-recap-estonia&rb3Schema=v1:hero
        'url': 'https://www.redbull.com/gb-en/live/fia-wrc-saturday-recap-estonia',
        'only_matching': True,
    }]
    _INT_FALLBACK_LIST = ['de', 'en', 'es', 'fr']
    _LAT_FALLBACK_MAP = ['ar', 'bo', 'car', 'cl', 'co', 'mx', 'pe']

    def _real_extract(self, url):
        region, lang, filter_type, display_id = self._match_valid_url(url).groups()
        if filter_type == 'episodes':
            filter_type = 'episode-videos'
        elif filter_type == 'live':
            filter_type = 'live-videos'

        regions = [region.upper()]
        if region != 'int':
            if region in self._LAT_FALLBACK_MAP:
                regions.append('LAT')
            if lang in self._INT_FALLBACK_LIST:
                regions.append('INT')
        locale = '>'.join(['%s-%s' % (lang, reg) for reg in regions])

        rrn_id = self._download_json(
            'https://www.redbull.com/v3/api/graphql/v1/v3/query/' + locale,
            display_id, query={
                'filter[type]': filter_type,
                'filter[uriSlug]': display_id,
                'rb3Schema': 'v1:hero',
            })['data']['id']

        return self.url_result(
            'https://www.redbull.com/embed/' + rrn_id,
            RedBullEmbedIE.ie_key(), rrn_id)
Commit	Line	Data
054a587d S	1	# coding: utf-8
	2	from __future__ import unicode_literals
	3
1f767545	4
054a587d	5	from .common import InfoExtractor
ba448445	6	from ..compat import compat_HTTPError
054a587d S	7	from ..utils import (
054a587d S	8	float_or_none,
ba448445	9	ExtractorError,
054a587d S	10	)
	11
	12
	13	class RedBullTVIE(InfoExtractor):
1f767545	14	_VALID_URL = r'https?://(?:www\.)?redbull(?:\.tv\|\.com(?:/[^/]+)?(?:/tv)?)(?:/events/[^/]+)?/(?:videos?\|live\|(?:film\|episode)s)/(?P<id>AP-\w+)'
054a587d S	15	_TESTS = [{
054a587d S	16	# film
e2e18694	17	'url': 'https://www.redbull.tv/video/AP-1Q6XCDTAN1W11',
ba448445	18	'md5': 'fb0445b98aa4394e504b413d98031d1f',
054a587d	19	'info_dict': {
e2e18694	20	'id': 'AP-1Q6XCDTAN1W11',
054a587d	21	'ext': 'mp4',
e2e18694	22	'title': 'ABC of... WRC - ABC of... S1E6',
054a587d S	23	'description': 'md5:5c7ed8f4015c8492ecf64b6ab31e7d31',
054a587d S	24	'duration': 1582.04,
054a587d S	25	},
	26	}, {
	27	# episode
e2e18694	28	'url': 'https://www.redbull.tv/video/AP-1PMHKJFCW1W11',
054a587d	29	'info_dict': {
e2e18694	30	'id': 'AP-1PMHKJFCW1W11',
054a587d	31	'ext': 'mp4',
e2e18694	32	'title': 'Grime - Hashtags S2E4',
1f767545 RA	33	'description': 'md5:5546aa612958c08a98faaad4abce484d',
1f767545 RA	34	'duration': 904,
054a587d	35	},
27449ad8 S	36	'params': {
	37	'skip_download': True,
	38	},
4779420c RA	39	}, {
	40	'url': 'https://www.redbull.com/int-en/tv/video/AP-1UWHCAR9S1W11/rob-meets-sam-gaze?playlist=playlists::3f81040a-2f31-4832-8e2e-545b1d39d173',
	41	'only_matching': True,
274519dd S	42	}, {
	43	'url': 'https://www.redbull.com/us-en/videos/AP-1YM9QCYE52111',
	44	'only_matching': True,
	45	}, {
	46	'url': 'https://www.redbull.com/us-en/events/AP-1XV2K61Q51W11/live/AP-1XUJ86FDH1W11',
	47	'only_matching': True,
1f767545 RA	48	}, {
	49	'url': 'https://www.redbull.com/int-en/films/AP-1ZSMAW8FH2111',
	50	'only_matching': True,
	51	}, {
	52	'url': 'https://www.redbull.com/int-en/episodes/AP-1TQWK7XE11W11',
	53	'only_matching': True,
054a587d S	54	}]
054a587d S	55
1f767545	56	def extract_info(self, video_id):
ba448445	57	session = self._download_json(
e2e18694	58	'https://api.redbull.tv/v3/session', video_id,
054a587d	59	note='Downloading access token', query={
ba448445	60	'category': 'personal_computer',
ba448445 RA	61	'os_family': 'http',
	62	})
	63	if session.get('code') == 'error':
	64	raise ExtractorError('%s said: %s' % (
	65	self.IE_NAME, session['message']))
e2e18694	66	token = session['token']
054a587d	67
ba448445	68	try:
e2e18694 RA	69	video = self._download_json(
e2e18694 RA	70	'https://api.redbull.tv/v3/products/' + video_id,
ba448445	71	video_id, note='Downloading video information',
e2e18694	72	headers={'Authorization': token}
ba448445 RA	73	)
	74	except ExtractorError as e:
	75	if isinstance(e.cause, compat_HTTPError) and e.cause.code == 404:
	76	error_message = self._parse_json(
e2e18694	77	e.cause.read().decode(), video_id)['error']
ba448445 RA	78	raise ExtractorError('%s said: %s' % (
	79	self.IE_NAME, error_message), expected=True)
	80	raise
054a587d	81
e2e18694	82	title = video['title'].strip()
054a587d S	83
054a587d S	84	formats = self._extract_m3u8_formats(
e2e18694 RA	85	'https://dms.redbull.tv/v3/%s/%s/playlist.m3u8' % (video_id, token),
e2e18694 RA	86	video_id, 'mp4', entry_protocol='m3u8_native', m3u8_id='hls')
ba448445	87	self._sort_formats(formats)
054a587d S	88
054a587d S	89	subtitles = {}
e2e18694 RA	90	for resource in video.get('resources', []):
	91	if resource.startswith('closed_caption_'):
	92	splitted_resource = resource.split('_')
	93	if splitted_resource[2]:
	94	subtitles.setdefault('en', []).append({
	95	'url': 'https://resources.redbull.tv/%s/%s' % (video_id, resource),
	96	'ext': splitted_resource[2],
	97	})
054a587d	98
e2e18694	99	subheading = video.get('subheading')
054a587d S	100	if subheading:
	101	title += ' - %s' % subheading
	102
	103	return {
	104	'id': video_id,
	105	'title': title,
e2e18694	106	'description': video.get('long_description') or video.get(
054a587d S	107	'short_description'),
054a587d S	108	'duration': float_or_none(video.get('duration'), scale=1000),
054a587d S	109	'formats': formats,
	110	'subtitles': subtitles,
	111	}
2efefdda	112
1f767545 RA	113	def _real_extract(self, url):
	114	video_id = self._match_id(url)
	115	return self.extract_info(video_id)
	116
	117
	118	class RedBullEmbedIE(RedBullTVIE):
	119	_VALID_URL = r'https?://(?:www\.)?redbull\.com/embed/(?P<id>rrn:content:[^:]+:[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}:[a-z]{2}-[A-Z]{2,3})'
	120	_TESTS = [{
	121	# HLS manifest accessible only using assetId
	122	'url': 'https://www.redbull.com/embed/rrn:content:episode-videos:f3021f4f-3ed4-51ac-915a-11987126e405:en-INT',
	123	'only_matching': True,
	124	}]
	125	_VIDEO_ESSENSE_TMPL = '''... on %s {
	126	videoEssence {
	127	attributes
	128	}
	129	}'''
	130
	131	def _real_extract(self, url):
	132	rrn_id = self._match_id(url)
	133	asset_id = self._download_json(
	134	'https://edge-graphql.crepo-production.redbullaws.com/v1/graphql',
b73612a2	135	rrn_id, headers={
	136	'Accept': 'application/json',
	137	'API-KEY': 'e90a1ff11335423998b100c929ecc866',
	138	}, query={
1f767545 RA	139	'query': '''{
	140	resource(id: "%s", enforceGeoBlocking: false) {
	141	%s
	142	%s
	143	}
	144	}''' % (rrn_id, self._VIDEO_ESSENSE_TMPL % 'LiveVideo', self._VIDEO_ESSENSE_TMPL % 'VideoResource'),
	145	})['data']['resource']['videoEssence']['attributes']['assetId']
	146	return self.extract_info(asset_id)
	147
2efefdda S	148
2efefdda S	149	class RedBullTVRrnContentIE(InfoExtractor):
1f767545	150	_VALID_URL = r'https?://(?:www\.)?redbull\.com/(?P<region>[a-z]{2,3})-(?P<lang>[a-z]{2})/tv/(?:video\|live\|film)/(?P<id>rrn:content:[^:]+:[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'
2efefdda S	151	_TESTS = [{
	152	'url': 'https://www.redbull.com/int-en/tv/video/rrn:content:live-videos:e3e6feb4-e95f-50b7-962a-c70f8fd13c73/mens-dh-finals-fort-william',
	153	'only_matching': True,
	154	}, {
	155	'url': 'https://www.redbull.com/int-en/tv/video/rrn:content:videos:a36a0f36-ff1b-5db8-a69d-ee11a14bf48b/tn-ts-style?playlist=rrn:content:event-profiles:83f05926-5de8-5389-b5e4-9bb312d715e8:extras',
	156	'only_matching': True,
1f767545 RA	157	}, {
	158	'url': 'https://www.redbull.com/int-en/tv/film/rrn:content:films:d1f4d00e-4c04-5d19-b510-a805ffa2ab83/follow-me',
	159	'only_matching': True,
2efefdda S	160	}]
	161
	162	def _real_extract(self, url):
5ad28e7f	163	region, lang, rrn_id = self._match_valid_url(url).groups()
1f767545 RA	164	rrn_id += ':%s-%s' % (lang, region.upper())
	165	return self.url_result(
	166	'https://www.redbull.com/embed/' + rrn_id,
	167	RedBullEmbedIE.ie_key(), rrn_id)
2efefdda	168
2efefdda	169
1f767545 RA	170	class RedBullIE(InfoExtractor):
	171	_VALID_URL = r'https?://(?:www\.)?redbull\.com/(?P<region>[a-z]{2,3})-(?P<lang>[a-z]{2})/(?P<type>(?:episode\|film\|(?:(?:recap\|trailer)-)?video)s\|live)/(?!AP-\|rrn:content:)(?P<id>[^/?#&]+)'
	172	_TESTS = [{
	173	'url': 'https://www.redbull.com/int-en/episodes/grime-hashtags-s02-e04',
	174	'md5': 'db8271a7200d40053a1809ed0dd574ff',
	175	'info_dict': {
	176	'id': 'AA-1MT8DQWA91W14',
	177	'ext': 'mp4',
	178	'title': 'Grime - Hashtags S2E4',
	179	'description': 'md5:5546aa612958c08a98faaad4abce484d',
	180	},
	181	}, {
	182	'url': 'https://www.redbull.com/int-en/films/kilimanjaro-mountain-of-greatness',
	183	'only_matching': True,
	184	}, {
	185	'url': 'https://www.redbull.com/int-en/recap-videos/uci-mountain-bike-world-cup-2017-mens-xco-finals-from-vallnord',
	186	'only_matching': True,
	187	}, {
	188	'url': 'https://www.redbull.com/int-en/trailer-videos/kings-of-content',
	189	'only_matching': True,
	190	}, {
	191	'url': 'https://www.redbull.com/int-en/videos/tnts-style-red-bull-dance-your-style-s1-e12',
	192	'only_matching': True,
	193	}, {
	194	'url': 'https://www.redbull.com/int-en/live/mens-dh-finals-fort-william',
	195	'only_matching': True,
b03eebdb RA	196	}, {
	197	# only available on the int-en website so a fallback is need for the API
	198	# https://www.redbull.com/v3/api/graphql/v1/v3/query/en-GB>en-INT?filter[uriSlug]=fia-wrc-saturday-recap-estonia&rb3Schema=v1:hero
	199	'url': 'https://www.redbull.com/gb-en/live/fia-wrc-saturday-recap-estonia',
	200	'only_matching': True,
1f767545	201	}]
b03eebdb RA	202	_INT_FALLBACK_LIST = ['de', 'en', 'es', 'fr']
b03eebdb RA	203	_LAT_FALLBACK_MAP = ['ar', 'bo', 'car', 'cl', 'co', 'mx', 'pe']
1f767545 RA	204
1f767545 RA	205	def _real_extract(self, url):
5ad28e7f	206	region, lang, filter_type, display_id = self._match_valid_url(url).groups()
1f767545 RA	207	if filter_type == 'episodes':
	208	filter_type = 'episode-videos'
	209	elif filter_type == 'live':
	210	filter_type = 'live-videos'
	211
b03eebdb RA	212	regions = [region.upper()]
	213	if region != 'int':
	214	if region in self._LAT_FALLBACK_MAP:
	215	regions.append('LAT')
	216	if lang in self._INT_FALLBACK_LIST:
	217	regions.append('INT')
	218	locale = '>'.join(['%s-%s' % (lang, reg) for reg in regions])
	219
1f767545	220	rrn_id = self._download_json(
b03eebdb	221	'https://www.redbull.com/v3/api/graphql/v1/v3/query/' + locale,
1f767545 RA	222	display_id, query={
	223	'filter[type]': filter_type,
	224	'filter[uriSlug]': display_id,
	225	'rb3Schema': 'v1:hero',
	226	})['data']['id']
2efefdda S	227
2efefdda S	228	return self.url_result(
1f767545 RA	229	'https://www.redbull.com/embed/' + rrn_id,
1f767545 RA	230	RedBullEmbedIE.ie_key(), rrn_id)