[yt-dlp.git] / yt_dlp / extractor / laola1tv.py

import json
import re

from .common import InfoExtractor
from ..utils import (
    ExtractorError,
    unified_strdate,
    urlencode_postdata,
    xpath_element,
    xpath_text,
    update_url_query,
    js_to_json,
)


class Laola1TvEmbedIE(InfoExtractor):
    IE_NAME = 'laola1tv:embed'
    _VALID_URL = r'https?://(?:www\.)?laola1\.tv/titanplayer\.php\?.*?\bvideoid=(?P<id>\d+)'
    _TESTS = [{
        # flashvars.premium = "false";
        'url': 'https://www.laola1.tv/titanplayer.php?videoid=708065&type=V&lang=en&portal=int&customer=1024',
        'info_dict': {
            'id': '708065',
            'ext': 'mp4',
            'title': 'MA Long CHN - FAN Zhendong CHN',
            'uploader': 'ITTF - International Table Tennis Federation',
            'upload_date': '20161211',
        },
    }]

    def _extract_token_url(self, stream_access_url, video_id, data):
        return self._download_json(
            self._proto_relative_url(stream_access_url, 'https:'), video_id,
            headers={
                'Content-Type': 'application/json',
            }, data=json.dumps(data).encode())['data']['stream-access'][0]

    def _extract_formats(self, token_url, video_id):
        token_doc = self._download_xml(
            token_url, video_id, 'Downloading token',
            headers=self.geo_verification_headers())

        token_attrib = xpath_element(token_doc, './/token').attrib

        if token_attrib['status'] != '0':
            raise ExtractorError(
                'Token error: %s' % token_attrib['comment'], expected=True)

        formats = self._extract_akamai_formats(
            '%s?hdnea=%s' % (token_attrib['url'], token_attrib['auth']),
            video_id)
        self._sort_formats(formats)
        return formats

    def _real_extract(self, url):
        video_id = self._match_id(url)
        webpage = self._download_webpage(url, video_id)
        flash_vars = self._search_regex(
            r'(?s)flashvars\s*=\s*({.+?});', webpage, 'flash vars')

        def get_flashvar(x, *args, **kwargs):
            flash_var = self._search_regex(
                r'%s\s*:\s*"([^"]+)"' % x,
                flash_vars, x, default=None)
            if not flash_var:
                flash_var = self._search_regex([
                    r'flashvars\.%s\s*=\s*"([^"]+)"' % x,
                    r'%s\s*=\s*"([^"]+)"' % x],
                    webpage, x, *args, **kwargs)
            return flash_var

        hd_doc = self._download_xml(
            'http://www.laola1.tv/server/hd_video.php', video_id, query={
                'play': get_flashvar('streamid'),
                'partner': get_flashvar('partnerid'),
                'portal': get_flashvar('portalid'),
                'lang': get_flashvar('sprache'),
                'v5ident': '',
            })

        _v = lambda x, **k: xpath_text(hd_doc, './/video/' + x, **k)
        title = _v('title', fatal=True)

        token_url = None
        premium = get_flashvar('premium', default=None)
        if premium:
            token_url = update_url_query(
                _v('url', fatal=True), {
                    'timestamp': get_flashvar('timestamp'),
                    'auth': get_flashvar('auth'),
                })
        else:
            data_abo = urlencode_postdata(
                dict((i, v) for i, v in enumerate(_v('req_liga_abos').split(','))))
            stream_access_url = update_url_query(
                'https://club.laola1.tv/sp/laola1/api/v3/user/session/premium/player/stream-access', {
                    'videoId': _v('id'),
                    'target': self._search_regex(r'vs_target = (\d+);', webpage, 'vs target'),
                    'label': _v('label'),
                    'area': _v('area'),
                })
            token_url = self._extract_token_url(stream_access_url, video_id, data_abo)

        formats = self._extract_formats(token_url, video_id)

        categories_str = _v('meta_sports')
        categories = categories_str.split(',') if categories_str else []
        is_live = _v('islive') == 'true'

        return {
            'id': video_id,
            'title': title,
            'upload_date': unified_strdate(_v('time_date')),
            'uploader': _v('meta_organisation'),
            'categories': categories,
            'is_live': is_live,
            'formats': formats,
        }


class Laola1TvBaseIE(Laola1TvEmbedIE):  # XXX: Do not subclass from concrete IE
    def _extract_video(self, url):
        display_id = self._match_id(url)
        webpage = self._download_webpage(url, display_id)

        if 'Dieser Livestream ist bereits beendet.' in webpage:
            raise ExtractorError('This live stream has already finished.', expected=True)

        conf = self._parse_json(self._search_regex(
            r'(?s)conf\s*=\s*({.+?});', webpage, 'conf'),
            display_id,
            transform_source=lambda s: js_to_json(re.sub(r'shareurl:.+,', '', s)))
        video_id = conf['videoid']

        config = self._download_json(conf['configUrl'], video_id, query={
            'videoid': video_id,
            'partnerid': conf['partnerid'],
            'language': conf.get('language', ''),
            'portal': conf.get('portalid', ''),
        })
        error = config.get('error')
        if error:
            raise ExtractorError('%s said: %s' % (self.IE_NAME, error), expected=True)

        video_data = config['video']
        title = video_data['title']
        is_live = video_data.get('isLivestream') and video_data.get('isLive')
        meta = video_data.get('metaInformation')
        sports = meta.get('sports')
        categories = sports.split(',') if sports else []

        token_url = self._extract_token_url(
            video_data['streamAccess'], video_id,
            video_data['abo']['required'])

        formats = self._extract_formats(token_url, video_id)

        return {
            'id': video_id,
            'display_id': display_id,
            'title': title,
            'description': video_data.get('description'),
            'thumbnail': video_data.get('image'),
            'categories': categories,
            'formats': formats,
            'is_live': is_live,
        }


class Laola1TvIE(Laola1TvBaseIE):
    IE_NAME = 'laola1tv'
    _VALID_URL = r'https?://(?:www\.)?laola1\.tv/[a-z]+-[a-z]+/[^/]+/(?P<id>[^/?#&]+)'

    _TESTS = [{
        'url': 'http://www.laola1.tv/de-de/video/straubing-tigers-koelner-haie/227883.html',
        'info_dict': {
            'id': '227883',
            'display_id': 'straubing-tigers-koelner-haie',
            'ext': 'flv',
            'title': 'Straubing Tigers - Kölner Haie',
            'upload_date': '20140912',
            'is_live': False,
            'categories': ['Eishockey'],
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://www.laola1.tv/de-de/video/straubing-tigers-koelner-haie',
        'info_dict': {
            'id': '464602',
            'display_id': 'straubing-tigers-koelner-haie',
            'ext': 'flv',
            'title': 'Straubing Tigers - Kölner Haie',
            'upload_date': '20160129',
            'is_live': False,
            'categories': ['Eishockey'],
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://www.laola1.tv/de-de/livestream/2016-03-22-belogorie-belgorod-trentino-diatec-lde',
        'info_dict': {
            'id': '487850',
            'display_id': '2016-03-22-belogorie-belgorod-trentino-diatec-lde',
            'ext': 'flv',
            'title': 'Belogorie BELGOROD - TRENTINO Diatec',
            'upload_date': '20160322',
            'uploader': 'CEV - Europäischer Volleyball Verband',
            'is_live': True,
            'categories': ['Volleyball'],
        },
        'params': {
            'skip_download': True,
        },
        'skip': 'This live stream has already finished.',
    }]

    def _real_extract(self, url):
        return self._extract_video(url)


class EHFTVIE(Laola1TvBaseIE):
    IE_NAME = 'ehftv'
    _VALID_URL = r'https?://(?:www\.)?ehftv\.com/[a-z]+(?:-[a-z]+)?/[^/]+/(?P<id>[^/?#&]+)'

    _TESTS = [{
        'url': 'https://www.ehftv.com/int/video/paris-saint-germain-handball-pge-vive-kielce/1166761',
        'info_dict': {
            'id': '1166761',
            'display_id': 'paris-saint-germain-handball-pge-vive-kielce',
            'ext': 'mp4',
            'title': 'Paris Saint-Germain Handball - PGE Vive Kielce',
            'is_live': False,
            'categories': ['Handball'],
        },
        'params': {
            'skip_download': True,
        },
    }]

    def _real_extract(self, url):
        return self._extract_video(url)


class ITTFIE(InfoExtractor):
    _VALID_URL = r'https?://tv\.ittf\.com/video/[^/]+/(?P<id>\d+)'
    _TEST = {
        'url': 'https://tv.ittf.com/video/peng-wang-wei-matsudaira-kenta/951802',
        'only_matching': True,
    }

    def _real_extract(self, url):
        return self.url_result(
            update_url_query('https://www.laola1.tv/titanplayer.php', {
                'videoid': self._match_id(url),
                'type': 'V',
                'lang': 'en',
                'portal': 'int',
                'customer': 1024,
            }), Laola1TvEmbedIE.ie_key())
Commit	Line	Data
4fe14732	1	import json
faac1c1f	2	import re
4fe14732	3
e5193599	4	from .common import InfoExtractor
9dcefb23	5	from ..utils import (
3037b91e	6	ExtractorError,
5e19323e JW	7	unified_strdate,
5e19323e JW	8	urlencode_postdata,
9dcefb23 S	9	xpath_element,
9dcefb23 S	10	xpath_text,
e029c43b	11	update_url_query,
4fe14732	12	js_to_json,
3037b91e	13	)
e5193599 PH	14
e5193599 PH	15
b42a0bf3	16	class Laola1TvEmbedIE(InfoExtractor):
e029c43b	17	IE_NAME = 'laola1tv:embed'
b42a0bf3	18	_VALID_URL = r'https?://(?:www\.)?laola1\.tv/titanplayer\.php\?.*?\bvideoid=(?P<id>\d+)'
7f09e523	19	_TESTS = [{
e029c43b RA	20	# flashvars.premium = "false";
	21	'url': 'https://www.laola1.tv/titanplayer.php?videoid=708065&type=V&lang=en&portal=int&customer=1024',
	22	'info_dict': {
	23	'id': '708065',
	24	'ext': 'mp4',
	25	'title': 'MA Long CHN - FAN Zhendong CHN',
	26	'uploader': 'ITTF - International Table Tennis Federation',
	27	'upload_date': '20161211',
	28	},
7f09e523	29	}]
b42a0bf3	30
4fe14732 RA	31	def _extract_token_url(self, stream_access_url, video_id, data):
4fe14732 RA	32	return self._download_json(
6895ea4d S	33	self._proto_relative_url(stream_access_url, 'https:'), video_id,
6895ea4d S	34	headers={
4fe14732 RA	35	'Content-Type': 'application/json',
	36	}, data=json.dumps(data).encode())['data']['stream-access'][0]
	37
	38	def _extract_formats(self, token_url, video_id):
	39	token_doc = self._download_xml(
	40	token_url, video_id, 'Downloading token',
	41	headers=self.geo_verification_headers())
	42
	43	token_attrib = xpath_element(token_doc, './/token').attrib
	44
	45	if token_attrib['status'] != '0':
	46	raise ExtractorError(
	47	'Token error: %s' % token_attrib['comment'], expected=True)
	48
	49	formats = self._extract_akamai_formats(
	50	'%s?hdnea=%s' % (token_attrib['url'], token_attrib['auth']),
	51	video_id)
	52	self._sort_formats(formats)
	53	return formats
	54
b42a0bf3 RA	55	def _real_extract(self, url):
	56	video_id = self._match_id(url)
	57	webpage = self._download_webpage(url, video_id)
	58	flash_vars = self._search_regex(
	59	r'(?s)flashvars\s=\s({.+?});', webpage, 'flash vars')
e029c43b RA	60
	61	def get_flashvar(x, args, *kwargs):
	62	flash_var = self._search_regex(
	63	r'%s\s:\s"([^"]+)"' % x,
	64	flash_vars, x, default=None)
	65	if not flash_var:
	66	flash_var = self._search_regex([
	67	r'flashvars\.%s\s=\s"([^"]+)"' % x,
	68	r'%s\s=\s"([^"]+)"' % x],
	69	webpage, x, args, *kwargs)
	70	return flash_var
b42a0bf3 RA	71
	72	hd_doc = self._download_xml(
	73	'http://www.laola1.tv/server/hd_video.php', video_id, query={
	74	'play': get_flashvar('streamid'),
	75	'partner': get_flashvar('partnerid'),
	76	'portal': get_flashvar('portalid'),
	77	'lang': get_flashvar('sprache'),
	78	'v5ident': '',
	79	})
	80
	81	_v = lambda x, k: xpath_text(hd_doc, './/video/' + x, k)
	82	title = _v('title', fatal=True)
	83
e029c43b RA	84	token_url = None
	85	premium = get_flashvar('premium', default=None)
	86	if premium:
	87	token_url = update_url_query(
	88	_v('url', fatal=True), {
	89	'timestamp': get_flashvar('timestamp'),
	90	'auth': get_flashvar('auth'),
	91	})
	92	else:
	93	data_abo = urlencode_postdata(
	94	dict((i, v) for i, v in enumerate(_v('req_liga_abos').split(','))))
4fe14732 RA	95	stream_access_url = update_url_query(
4fe14732 RA	96	'https://club.laola1.tv/sp/laola1/api/v3/user/session/premium/player/stream-access', {
e029c43b RA	97	'videoId': _v('id'),
	98	'target': self._search_regex(r'vs_target = (\d+);', webpage, 'vs target'),
	99	'label': _v('label'),
	100	'area': _v('area'),
4fe14732 RA	101	})
4fe14732 RA	102	token_url = self._extract_token_url(stream_access_url, video_id, data_abo)
b42a0bf3	103
4fe14732	104	formats = self._extract_formats(token_url, video_id)
b42a0bf3 RA	105
	106	categories_str = _v('meta_sports')
	107	categories = categories_str.split(',') if categories_str else []
	108	is_live = _v('islive') == 'true'
	109
	110	return {
	111	'id': video_id,
39ca3b5c	112	'title': title,
b42a0bf3 RA	113	'upload_date': unified_strdate(_v('time_date')),
	114	'uploader': _v('meta_organisation'),
	115	'categories': categories,
	116	'is_live': is_live,
	117	'formats': formats,
	118	}
	119
	120
6368e2e6	121	class Laola1TvBaseIE(Laola1TvEmbedIE): # XXX: Do not subclass from concrete IE
faac1c1f AS	122	def _extract_video(self, url):
	123	display_id = self._match_id(url)
	124	webpage = self._download_webpage(url, display_id)
	125
	126	if 'Dieser Livestream ist bereits beendet.' in webpage:
	127	raise ExtractorError('This live stream has already finished.', expected=True)
	128
	129	conf = self._parse_json(self._search_regex(
	130	r'(?s)conf\s=\s({.+?});', webpage, 'conf'),
	131	display_id,
	132	transform_source=lambda s: js_to_json(re.sub(r'shareurl:.+,', '', s)))
	133	video_id = conf['videoid']
	134
	135	config = self._download_json(conf['configUrl'], video_id, query={
	136	'videoid': video_id,
	137	'partnerid': conf['partnerid'],
	138	'language': conf.get('language', ''),
	139	'portal': conf.get('portalid', ''),
	140	})
	141	error = config.get('error')
	142	if error:
	143	raise ExtractorError('%s said: %s' % (self.IE_NAME, error), expected=True)
	144
	145	video_data = config['video']
	146	title = video_data['title']
	147	is_live = video_data.get('isLivestream') and video_data.get('isLive')
	148	meta = video_data.get('metaInformation')
	149	sports = meta.get('sports')
	150	categories = sports.split(',') if sports else []
	151
	152	token_url = self._extract_token_url(
	153	video_data['streamAccess'], video_id,
	154	video_data['abo']['required'])
	155
	156	formats = self._extract_formats(token_url, video_id)
	157
	158	return {
	159	'id': video_id,
	160	'display_id': display_id,
39ca3b5c	161	'title': title,
faac1c1f AS	162	'description': video_data.get('description'),
	163	'thumbnail': video_data.get('image'),
	164	'categories': categories,
	165	'formats': formats,
	166	'is_live': is_live,
	167	}
	168
	169
	170	class Laola1TvIE(Laola1TvBaseIE):
e029c43b	171	IE_NAME = 'laola1tv'
b42a0bf3	172	_VALID_URL = r'https?://(?:www\.)?laola1\.tv/[a-z]+-[a-z]+/[^/]+/(?P<id>[^/?#&]+)'
faac1c1f	173
5e19323e	174	_TESTS = [{
3037b91e	175	'url': 'http://www.laola1.tv/de-de/video/straubing-tigers-koelner-haie/227883.html',
e5193599	176	'info_dict': {
3037b91e	177	'id': '227883',
9dcefb23 S	178	'display_id': 'straubing-tigers-koelner-haie',
9dcefb23 S	179	'ext': 'flv',
3037b91e	180	'title': 'Straubing Tigers - Kölner Haie',
5e19323e	181	'upload_date': '20140912',
9dcefb23 S	182	'is_live': False,
9dcefb23 S	183	'categories': ['Eishockey'],
5e19323e JW	184	},
	185	'params': {
	186	'skip_download': True,
2beeb286	187	},
5e19323e JW	188	}, {
	189	'url': 'http://www.laola1.tv/de-de/video/straubing-tigers-koelner-haie',
	190	'info_dict': {
5e19323e	191	'id': '464602',
9dcefb23 S	192	'display_id': 'straubing-tigers-koelner-haie',
9dcefb23 S	193	'ext': 'flv',
5e19323e JW	194	'title': 'Straubing Tigers - Kölner Haie',
5e19323e JW	195	'upload_date': '20160129',
9dcefb23 S	196	'is_live': False,
9dcefb23 S	197	'categories': ['Eishockey'],
e5193599 PH	198	},
	199	'params': {
	200	'skip_download': True,
2beeb286 S	201	},
	202	}, {
	203	'url': 'http://www.laola1.tv/de-de/livestream/2016-03-22-belogorie-belgorod-trentino-diatec-lde',
	204	'info_dict': {
	205	'id': '487850',
	206	'display_id': '2016-03-22-belogorie-belgorod-trentino-diatec-lde',
	207	'ext': 'flv',
	208	'title': 'Belogorie BELGOROD - TRENTINO Diatec',
	209	'upload_date': '20160322',
	210	'uploader': 'CEV - Europäischer Volleyball Verband',
	211	'is_live': True,
	212	'categories': ['Volleyball'],
	213	},
	214	'params': {
	215	'skip_download': True,
	216	},
d1c4e4ba	217	'skip': 'This live stream has already finished.',
5e19323e	218	}]
e5193599 PH	219
e5193599 PH	220	def _real_extract(self, url):
faac1c1f	221	return self._extract_video(url)
e5193599	222
9dcefb23	223
faac1c1f AS	224	class EHFTVIE(Laola1TvBaseIE):
faac1c1f AS	225	IE_NAME = 'ehftv'
6895ea4d	226	_VALID_URL = r'https?://(?:www\.)?ehftv\.com/[a-z]+(?:-[a-z]+)?/[^/]+/(?P<id>[^/?#&]+)'
4fe14732	227
faac1c1f AS	228	_TESTS = [{
	229	'url': 'https://www.ehftv.com/int/video/paris-saint-germain-handball-pge-vive-kielce/1166761',
	230	'info_dict': {
	231	'id': '1166761',
	232	'display_id': 'paris-saint-germain-handball-pge-vive-kielce',
	233	'ext': 'mp4',
	234	'title': 'Paris Saint-Germain Handball - PGE Vive Kielce',
	235	'is_live': False,
	236	'categories': ['Handball'],
	237	},
	238	'params': {
	239	'skip_download': True,
	240	},
	241	}]
e5193599	242
faac1c1f AS	243	def _real_extract(self, url):
faac1c1f AS	244	return self._extract_video(url)
f5469da9 S	245
	246
	247	class ITTFIE(InfoExtractor):
	248	_VALID_URL = r'https?://tv\.ittf\.com/video/[^/]+/(?P<id>\d+)'
	249	_TEST = {
	250	'url': 'https://tv.ittf.com/video/peng-wang-wei-matsudaira-kenta/951802',
	251	'only_matching': True,
	252	}
	253
	254	def _real_extract(self, url):
	255	return self.url_result(
	256	update_url_query('https://www.laola1.tv/titanplayer.php', {
	257	'videoid': self._match_id(url),
	258	'type': 'V',
	259	'lang': 'en',
	260	'portal': 'int',
	261	'customer': 1024,
	262	}), Laola1TvEmbedIE.ie_key())