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

from .common import InfoExtractor
from ..compat import (
    compat_str,
)
from ..utils import (
    int_or_none,
    parse_qs,
)


class VideomoreBaseIE(InfoExtractor):
    _API_BASE_URL = 'https://more.tv/api/v3/web/'
    _VALID_URL_BASE = r'https?://(?:videomore\.ru|more\.tv)/'

    def _download_page_data(self, display_id):
        return self._download_json(
            self._API_BASE_URL + 'PageData', display_id, query={
                'url': '/' + display_id,
            })['attributes']['response']['data']

    def _track_url_result(self, track):
        track_vod = track['trackVod']
        video_url = track_vod.get('playerLink') or track_vod['link']
        return self.url_result(
            video_url, VideomoreIE.ie_key(), track_vod.get('hubId'))


class VideomoreIE(InfoExtractor):
    IE_NAME = 'videomore'
    _VALID_URL = r'''(?x)
                    videomore:(?P<sid>\d+)$|
                    https?://
                        (?:
                            videomore\.ru/
                            (?:
                                embed|
                                [^/]+/[^/]+
                            )/|
                            (?:
                                (?:player\.)?videomore\.ru|
                                siren\.more\.tv/player
                            )/[^/]*\?.*?\btrack_id=|
                            odysseus\.more.tv/player/(?P<partner_id>\d+)/
                        )
                        (?P<id>\d+)
                        (?:[/?#&]|\.(?:xml|json)|$)
                    '''
    _EMBED_REGEX = [r'''(?x)
        (?:
            <iframe[^>]+src=([\'"])|
            <object[^>]+data=(["\'])https?://videomore\.ru/player\.swf\?.*config=
        )(?P<url>https?://videomore\.ru/[^?#"']+/\d+(?:\.xml)?)
    ''']
    _TESTS = [{
        'url': 'http://videomore.ru/kino_v_detalayah/5_sezon/367617',
        'md5': '44455a346edc0d509ac5b5a5b531dc35',
        'info_dict': {
            'id': '367617',
            'ext': 'flv',
            'title': 'Кино в деталях 5 сезон В гостях Алексей Чумаков и Юлия Ковальчук',
            'series': 'Кино в деталях',
            'episode': 'В гостях Алексей Чумаков и Юлия Ковальчук',
            'thumbnail': r're:^https?://.*\.jpg',
            'duration': 2910,
            'view_count': int,
            'comment_count': int,
            'age_limit': 16,
        },
        'skip': 'The video is not available for viewing.',
    }, {
        'url': 'http://videomore.ru/embed/259974',
        'info_dict': {
            'id': '259974',
            'ext': 'mp4',
            'title': 'Молодежка 2 сезон 40 серия',
            'series': 'Молодежка',
            'season': '2 сезон',
            'episode': '40 серия',
            'thumbnail': r're:^https?://.*\.jpg',
            'duration': 2789,
            'view_count': int,
            'age_limit': 16,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://videomore.ru/molodezhka/sezon_promo/341073',
        'info_dict': {
            'id': '341073',
            'ext': 'flv',
            'title': 'Промо Команда проиграла из-за Бакина?',
            'episode': 'Команда проиграла из-за Бакина?',
            'thumbnail': r're:^https?://.*\.jpg',
            'duration': 29,
            'age_limit': 16,
            'view_count': int,
        },
        'params': {
            'skip_download': True,
        },
        'skip': 'The video is not available for viewing.',
    }, {
        'url': 'http://videomore.ru/elki_3?track_id=364623',
        'only_matching': True,
    }, {
        'url': 'http://videomore.ru/embed/364623',
        'only_matching': True,
    }, {
        'url': 'http://videomore.ru/video/tracks/364623.xml',
        'only_matching': True,
    }, {
        'url': 'http://videomore.ru/video/tracks/364623.json',
        'only_matching': True,
    }, {
        'url': 'http://videomore.ru/video/tracks/158031/quotes/33248',
        'only_matching': True,
    }, {
        'url': 'videomore:367617',
        'only_matching': True,
    }, {
        'url': 'https://player.videomore.ru/?partner_id=97&track_id=736234&autoplay=0&userToken=',
        'only_matching': True,
    }, {
        'url': 'https://odysseus.more.tv/player/1788/352317',
        'only_matching': True,
    }, {
        'url': 'https://siren.more.tv/player/config?track_id=352317&partner_id=1788&user_token=',
        'only_matching': True,
    }]
    _GEO_BYPASS = False

    def _real_extract(self, url):
        mobj = self._match_valid_url(url)
        video_id = mobj.group('sid') or mobj.group('id')
        partner_id = mobj.group('partner_id') or parse_qs(url).get('partner_id', [None])[0] or '97'

        item = self._download_json(
            'https://siren.more.tv/player/config', video_id, query={
                'partner_id': partner_id,
                'track_id': video_id,
            })['data']['playlist']['items'][0]

        title = item.get('title')
        series = item.get('project_name')
        season = item.get('season_name')
        episode = item.get('episode_name')
        if not title:
            title = []
            for v in (series, season, episode):
                if v:
                    title.append(v)
            title = ' '.join(title)

        streams = item.get('streams') or []
        for protocol in ('DASH', 'HLS'):
            stream_url = item.get(protocol.lower() + '_url')
            if stream_url:
                streams.append({'protocol': protocol, 'url': stream_url})

        formats = []
        for stream in streams:
            stream_url = stream.get('url')
            if not stream_url:
                continue
            protocol = stream.get('protocol')
            if protocol == 'DASH':
                formats.extend(self._extract_mpd_formats(
                    stream_url, video_id, mpd_id='dash', fatal=False))
            elif protocol == 'HLS':
                formats.extend(self._extract_m3u8_formats(
                    stream_url, video_id, 'mp4', 'm3u8_native',
                    m3u8_id='hls', fatal=False))
            elif protocol == 'MSS':
                formats.extend(self._extract_ism_formats(
                    stream_url, video_id, ism_id='mss', fatal=False))

        if not formats:
            error = item.get('error')
            if error:
                if error in ('Данное видео недоступно для просмотра на территории этой страны', 'Данное видео доступно для просмотра только на территории России'):
                    self.raise_geo_restricted(countries=['RU'], metadata_available=True)
                self.raise_no_formats(error, expected=True)

        return {
            'id': video_id,
            'title': title,
            'series': series,
            'season': season,
            'episode': episode,
            'thumbnail': item.get('thumbnail_url'),
            'duration': int_or_none(item.get('duration')),
            'view_count': int_or_none(item.get('views')),
            'age_limit': int_or_none(item.get('min_age')),
            'formats': formats,
        }


class VideomoreVideoIE(VideomoreBaseIE):
    IE_NAME = 'videomore:video'
    _VALID_URL = VideomoreBaseIE._VALID_URL_BASE + r'(?P<id>(?:(?:[^/]+/){2})?[^/?#&]+)(?:/*|[?#&].*?)$'
    _TESTS = [{
        # single video with og:video:iframe
        'url': 'http://videomore.ru/elki_3',
        'info_dict': {
            'id': '364623',
            'ext': 'flv',
            'title': 'Ёлки 3',
            'description': '',
            'thumbnail': r're:^https?://.*\.jpg',
            'duration': 5579,
            'age_limit': 6,
            'view_count': int,
        },
        'params': {
            'skip_download': True,
        },
        'skip': 'Requires logging in',
    }, {
        # season single series with og:video:iframe
        'url': 'http://videomore.ru/poslednii_ment/1_sezon/14_seriya',
        'info_dict': {
            'id': '352317',
            'ext': 'mp4',
            'title': 'Последний мент 1 сезон 14 серия',
            'series': 'Последний мент',
            'season': '1 сезон',
            'episode': '14 серия',
            'thumbnail': r're:^https?://.*\.jpg',
            'duration': 2464,
            'age_limit': 16,
            'view_count': int,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://videomore.ru/sejchas_v_seti/serii_221-240/226_vypusk',
        'only_matching': True,
    }, {
        # single video without og:video:iframe
        'url': 'http://videomore.ru/marin_i_ego_druzya',
        'info_dict': {
            'id': '359073',
            'ext': 'flv',
            'title': '1 серия. Здравствуй, Аквавилль!',
            'description': 'md5:c6003179538b5d353e7bcd5b1372b2d7',
            'thumbnail': r're:^https?://.*\.jpg',
            'duration': 754,
            'age_limit': 6,
            'view_count': int,
        },
        'params': {
            'skip_download': True,
        },
        'skip': 'redirects to https://more.tv/'
    }, {
        'url': 'https://videomore.ru/molodezhka/6_sezon/29_seriya?utm_so',
        'only_matching': True,
    }, {
        'url': 'https://more.tv/poslednii_ment/1_sezon/14_seriya',
        'only_matching': True,
    }]

    @classmethod
    def suitable(cls, url):
        return False if VideomoreIE.suitable(url) else super(VideomoreVideoIE, cls).suitable(url)

    def _real_extract(self, url):
        display_id = self._match_id(url)
        return self._track_url_result(self._download_page_data(display_id))


class VideomoreSeasonIE(VideomoreBaseIE):
    IE_NAME = 'videomore:season'
    _VALID_URL = VideomoreBaseIE._VALID_URL_BASE + r'(?!embed)(?P<id>[^/]+/[^/?#&]+)(?:/*|[?#&].*?)$'
    _TESTS = [{
        'url': 'http://videomore.ru/molodezhka/film_o_filme',
        'info_dict': {
            'id': 'molodezhka/film_o_filme',
            'title': 'Фильм о фильме',
        },
        'playlist_mincount': 3,
    }, {
        'url': 'http://videomore.ru/molodezhka/sezon_promo?utm_so',
        'only_matching': True,
    }, {
        'url': 'https://more.tv/molodezhka/film_o_filme',
        'only_matching': True,
    }]

    @classmethod
    def suitable(cls, url):
        return (False if (VideomoreIE.suitable(url) or VideomoreVideoIE.suitable(url))
                else super(VideomoreSeasonIE, cls).suitable(url))

    def _real_extract(self, url):
        display_id = self._match_id(url)
        season = self._download_page_data(display_id)
        season_id = compat_str(season['id'])
        tracks = self._download_json(
            self._API_BASE_URL + 'seasons/%s/tracks' % season_id,
            season_id)['data']
        entries = []
        for track in tracks:
            entries.append(self._track_url_result(track))
        return self.playlist_result(entries, display_id, season.get('title'))
Commit	Line	Data
030dfb04	1	from .common import InfoExtractor
29f7c58a	2	from ..compat import (
29f7c58a	3	compat_str,
29f7c58a	4	)
030dfb04 S	5	from ..utils import (
030dfb04 S	6	int_or_none,
4dfbf869	7	parse_qs,
030dfb04 S	8	)
	9
	10
29f7c58a	11	class VideomoreBaseIE(InfoExtractor):
	12	_API_BASE_URL = 'https://more.tv/api/v3/web/'
	13	_VALID_URL_BASE = r'https?://(?:videomore\.ru\|more\.tv)/'
	14
	15	def _download_page_data(self, display_id):
	16	return self._download_json(
	17	self._API_BASE_URL + 'PageData', display_id, query={
	18	'url': '/' + display_id,
	19	})['attributes']['response']['data']
	20
	21	def _track_url_result(self, track):
	22	track_vod = track['trackVod']
	23	video_url = track_vod.get('playerLink') or track_vod['link']
	24	return self.url_result(
	25	video_url, VideomoreIE.ie_key(), track_vod.get('hubId'))
	26
	27
030dfb04	28	class VideomoreIE(InfoExtractor):
69759a59	29	IE_NAME = 'videomore'
19d69913 S	30	_VALID_URL = r'''(?x)
19d69913 S	31	videomore:(?P<sid>\d+)$\|
29f7c58a	32	https?://
19d69913	33	(?:
29f7c58a	34	videomore\.ru/
19d69913 S	35	(?:
	36	embed\|
	37	[^/]+/[^/]+
	38	)/\|
29f7c58a	39	(?:
	40	(?:player\.)?videomore\.ru\|
	41	siren\.more\.tv/player
	42	)/[^/]\?.?\btrack_id=\|
	43	odysseus\.more.tv/player/(?P<partner_id>\d+)/
19d69913 S	44	)
	45	(?P<id>\d+)
	46	(?:[/?#&]\|\.(?:xml\|json)\|$)
	47	'''
bfd973ec	48	_EMBED_REGEX = [r'''(?x)
	49	(?:
	50	<iframe[^>]+src=([\'"])\|
	51	<object[^>]+data=(["\'])https?://videomore\.ru/player\.swf\?.*config=
	52	)(?P<url>https?://videomore\.ru/[^?#"']+/\d+(?:\.xml)?)
	53	''']
030dfb04 S	54	_TESTS = [{
030dfb04 S	55	'url': 'http://videomore.ru/kino_v_detalayah/5_sezon/367617',
010d034f	56	'md5': '44455a346edc0d509ac5b5a5b531dc35',
030dfb04 S	57	'info_dict': {
	58	'id': '367617',
	59	'ext': 'flv',
010d034f	60	'title': 'Кино в деталях 5 сезон В гостях Алексей Чумаков и Юлия Ковальчук',
7a0b07c7 S	61	'series': 'Кино в деталях',
7a0b07c7 S	62	'episode': 'В гостях Алексей Чумаков и Юлия Ковальчук',
ec85ded8	63	'thumbnail': r're:^https?://.*\.jpg',
030dfb04	64	'duration': 2910,
030dfb04	65	'view_count': int,
010d034f S	66	'comment_count': int,
010d034f S	67	'age_limit': 16,
030dfb04	68	},
29f7c58a	69	'skip': 'The video is not available for viewing.',
7a0b07c7 S	70	}, {
	71	'url': 'http://videomore.ru/embed/259974',
	72	'info_dict': {
	73	'id': '259974',
29f7c58a	74	'ext': 'mp4',
010d034f	75	'title': 'Молодежка 2 сезон 40 серия',
7a0b07c7	76	'series': 'Молодежка',
29f7c58a	77	'season': '2 сезон',
010d034f	78	'episode': '40 серия',
ec85ded8	79	'thumbnail': r're:^https?://.*\.jpg',
29f7c58a	80	'duration': 2789,
7a0b07c7	81	'view_count': int,
010d034f	82	'age_limit': 16,
7a0b07c7 S	83	},
	84	'params': {
	85	'skip_download': True,
	86	},
	87	}, {
	88	'url': 'http://videomore.ru/molodezhka/sezon_promo/341073',
	89	'info_dict': {
	90	'id': '341073',
	91	'ext': 'flv',
010d034f	92	'title': 'Промо Команда проиграла из-за Бакина?',
7a0b07c7	93	'episode': 'Команда проиграла из-за Бакина?',
ec85ded8	94	'thumbnail': r're:^https?://.*\.jpg',
7a0b07c7 S	95	'duration': 29,
	96	'age_limit': 16,
	97	'view_count': int,
	98	},
	99	'params': {
	100	'skip_download': True,
	101	},
29f7c58a	102	'skip': 'The video is not available for viewing.',
030dfb04 S	103	}, {
	104	'url': 'http://videomore.ru/elki_3?track_id=364623',
	105	'only_matching': True,
	106	}, {
	107	'url': 'http://videomore.ru/embed/364623',
	108	'only_matching': True,
	109	}, {
	110	'url': 'http://videomore.ru/video/tracks/364623.xml',
	111	'only_matching': True,
	112	}, {
	113	'url': 'http://videomore.ru/video/tracks/364623.json',
	114	'only_matching': True,
	115	}, {
	116	'url': 'http://videomore.ru/video/tracks/158031/quotes/33248',
	117	'only_matching': True,
	118	}, {
	119	'url': 'videomore:367617',
	120	'only_matching': True,
19d69913 S	121	}, {
	122	'url': 'https://player.videomore.ru/?partner_id=97&track_id=736234&autoplay=0&userToken=',
	123	'only_matching': True,
29f7c58a	124	}, {
	125	'url': 'https://odysseus.more.tv/player/1788/352317',
	126	'only_matching': True,
	127	}, {
	128	'url': 'https://siren.more.tv/player/config?track_id=352317&partner_id=1788&user_token=',
	129	'only_matching': True,
030dfb04	130	}]
29f7c58a	131	_GEO_BYPASS = False
030dfb04	132
030dfb04	133	def _real_extract(self, url):
5ad28e7f	134	mobj = self._match_valid_url(url)
030dfb04	135	video_id = mobj.group('sid') or mobj.group('id')
4dfbf869	136	partner_id = mobj.group('partner_id') or parse_qs(url).get('partner_id', [None])[0] or '97'
29f7c58a	137
	138	item = self._download_json(
	139	'https://siren.more.tv/player/config', video_id, query={
	140	'partner_id': partner_id,
	141	'track_id': video_id,
	142	})['data']['playlist']['items'][0]
	143
	144	title = item.get('title')
	145	series = item.get('project_name')
	146	season = item.get('season_name')
	147	episode = item.get('episode_name')
	148	if not title:
	149	title = []
	150	for v in (series, season, episode):
	151	if v:
	152	title.append(v)
	153	title = ' '.join(title)
	154
	155	streams = item.get('streams') or []
	156	for protocol in ('DASH', 'HLS'):
	157	stream_url = item.get(protocol.lower() + '_url')
	158	if stream_url:
	159	streams.append({'protocol': protocol, 'url': stream_url})
	160
	161	formats = []
	162	for stream in streams:
	163	stream_url = stream.get('url')
	164	if not stream_url:
	165	continue
	166	protocol = stream.get('protocol')
	167	if protocol == 'DASH':
	168	formats.extend(self._extract_mpd_formats(
	169	stream_url, video_id, mpd_id='dash', fatal=False))
	170	elif protocol == 'HLS':
	171	formats.extend(self._extract_m3u8_formats(
	172	stream_url, video_id, 'mp4', 'm3u8_native',
	173	m3u8_id='hls', fatal=False))
	174	elif protocol == 'MSS':
	175	formats.extend(self._extract_ism_formats(
	176	stream_url, video_id, ism_id='mss', fatal=False))
	177
	178	if not formats:
	179	error = item.get('error')
	180	if error:
	181	if error in ('Данное видео недоступно для просмотра на территории этой страны', 'Данное видео доступно для просмотра только на территории России'):
b7da73eb	182	self.raise_geo_restricted(countries=['RU'], metadata_available=True)
b7da73eb	183	self.raise_no_formats(error, expected=True)
030dfb04	184
030dfb04 S	185	return {
	186	'id': video_id,
	187	'title': title,
7a0b07c7	188	'series': series,
29f7c58a	189	'season': season,
7a0b07c7	190	'episode': episode,
29f7c58a	191	'thumbnail': item.get('thumbnail_url'),
	192	'duration': int_or_none(item.get('duration')),
	193	'view_count': int_or_none(item.get('views')),
	194	'age_limit': int_or_none(item.get('min_age')),
030dfb04 S	195	'formats': formats,
	196	}
	197
	198
29f7c58a	199	class VideomoreVideoIE(VideomoreBaseIE):
030dfb04	200	IE_NAME = 'videomore:video'
29f7c58a	201	_VALID_URL = VideomoreBaseIE._VALID_URL_BASE + r'(?P<id>(?:(?:[^/]+/){2})?[^/?#&]+)(?:/\|[?#&].?)$'
030dfb04 S	202	_TESTS = [{
	203	# single video with og:video:iframe
	204	'url': 'http://videomore.ru/elki_3',
	205	'info_dict': {
	206	'id': '364623',
	207	'ext': 'flv',
	208	'title': 'Ёлки 3',
	209	'description': '',
ec85ded8	210	'thumbnail': r're:^https?://.*\.jpg',
030dfb04 S	211	'duration': 5579,
	212	'age_limit': 6,
	213	'view_count': int,
	214	},
	215	'params': {
	216	'skip_download': True,
	217	},
29f7c58a	218	'skip': 'Requires logging in',
030dfb04	219	}, {
dfb1b146	220	# season single series with og:video:iframe
030dfb04	221	'url': 'http://videomore.ru/poslednii_ment/1_sezon/14_seriya',
29f7c58a	222	'info_dict': {
	223	'id': '352317',
	224	'ext': 'mp4',
	225	'title': 'Последний мент 1 сезон 14 серия',
	226	'series': 'Последний мент',
	227	'season': '1 сезон',
	228	'episode': '14 серия',
	229	'thumbnail': r're:^https?://.*\.jpg',
	230	'duration': 2464,
	231	'age_limit': 16,
	232	'view_count': int,
	233	},
	234	'params': {
	235	'skip_download': True,
	236	},
030dfb04 S	237	}, {
	238	'url': 'http://videomore.ru/sejchas_v_seti/serii_221-240/226_vypusk',
	239	'only_matching': True,
	240	}, {
	241	# single video without og:video:iframe
	242	'url': 'http://videomore.ru/marin_i_ego_druzya',
	243	'info_dict': {
	244	'id': '359073',
	245	'ext': 'flv',
	246	'title': '1 серия. Здравствуй, Аквавилль!',
	247	'description': 'md5:c6003179538b5d353e7bcd5b1372b2d7',
ec85ded8	248	'thumbnail': r're:^https?://.*\.jpg',
030dfb04 S	249	'duration': 754,
	250	'age_limit': 6,
	251	'view_count': int,
	252	},
	253	'params': {
	254	'skip_download': True,
	255	},
29f7c58a	256	'skip': 'redirects to https://more.tv/'
19d69913 S	257	}, {
	258	'url': 'https://videomore.ru/molodezhka/6_sezon/29_seriya?utm_so',
	259	'only_matching': True,
29f7c58a	260	}, {
	261	'url': 'https://more.tv/poslednii_ment/1_sezon/14_seriya',
	262	'only_matching': True,
030dfb04 S	263	}]
	264
	265	@classmethod
	266	def suitable(cls, url):
	267	return False if VideomoreIE.suitable(url) else super(VideomoreVideoIE, cls).suitable(url)
	268
	269	def _real_extract(self, url):
	270	display_id = self._match_id(url)
29f7c58a	271	return self._track_url_result(self._download_page_data(display_id))
030dfb04 S	272
030dfb04 S	273
29f7c58a	274	class VideomoreSeasonIE(VideomoreBaseIE):
030dfb04	275	IE_NAME = 'videomore:season'
29f7c58a	276	_VALID_URL = VideomoreBaseIE._VALID_URL_BASE + r'(?!embed)(?P<id>[^/]+/[^/?#&]+)(?:/\|[?#&].?)$'
030dfb04	277	_TESTS = [{
29f7c58a	278	'url': 'http://videomore.ru/molodezhka/film_o_filme',
030dfb04	279	'info_dict': {
29f7c58a	280	'id': 'molodezhka/film_o_filme',
29f7c58a	281	'title': 'Фильм о фильме',
030dfb04	282	},
29f7c58a	283	'playlist_mincount': 3,
19d69913 S	284	}, {
	285	'url': 'http://videomore.ru/molodezhka/sezon_promo?utm_so',
	286	'only_matching': True,
29f7c58a	287	}, {
	288	'url': 'https://more.tv/molodezhka/film_o_filme',
	289	'only_matching': True,
030dfb04 S	290	}]
030dfb04 S	291
19d69913 S	292	@classmethod
	293	def suitable(cls, url):
	294	return (False if (VideomoreIE.suitable(url) or VideomoreVideoIE.suitable(url))
	295	else super(VideomoreSeasonIE, cls).suitable(url))
	296
030dfb04 S	297	def _real_extract(self, url):
030dfb04 S	298	display_id = self._match_id(url)
29f7c58a	299	season = self._download_page_data(display_id)
	300	season_id = compat_str(season['id'])
	301	tracks = self._download_json(
	302	self._API_BASE_URL + 'seasons/%s/tracks' % season_id,
	303	season_id)['data']
19d69913	304	entries = []
29f7c58a	305	for track in tracks:
	306	entries.append(self._track_url_result(track))
	307	return self.playlist_result(entries, display_id, season.get('title'))