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

# coding: utf-8
from __future__ import unicode_literals

import re

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+)$|
                    https?://
                        (?:
                            videomore\.ru/
                            (?:
                                embed|
                                [^/]+/[^/]+
                            )/|
                            (?:
                                (?:player\.)?videomore\.ru|
                                siren\.more\.tv/player
                            )/[^/]*\?.*?\btrack_id=|
                            odysseus\.more.tv/player/(?P<partner_id>\d+)/
                        )
                        (?P<id>\d+)
                        (?:[/?#&]|\.(?:xml|json)|$)
                    '''
    _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

    @staticmethod
    def _extract_url(webpage):
        mobj = re.search(
            r'<object[^>]+data=(["\'])https?://videomore\.ru/player\.swf\?.*config=(?P<url>https?://videomore\.ru/(?:[^/]+/)+\d+\.xml).*\1',
            webpage)
        if not mobj:
            mobj = re.search(
                r'<iframe[^>]+src=([\'"])(?P<url>https?://videomore\.ru/embed/\d+)',
                webpage)

        if mobj:
            return mobj.group('url')

    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)
        self._sort_formats(formats)

        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 S	1	# coding: utf-8
	2	from __future__ import unicode_literals
	3
	4	import re
	5
	6	from .common import InfoExtractor
29f7c58a	7	from ..compat import (
29f7c58a	8	compat_str,
29f7c58a	9	)
030dfb04 S	10	from ..utils import (
030dfb04 S	11	int_or_none,
4dfbf869	12	parse_qs,
030dfb04 S	13	)
	14
	15
29f7c58a	16	class VideomoreBaseIE(InfoExtractor):
	17	_API_BASE_URL = 'https://more.tv/api/v3/web/'
	18	_VALID_URL_BASE = r'https?://(?:videomore\.ru\|more\.tv)/'
	19
	20	def _download_page_data(self, display_id):
	21	return self._download_json(
	22	self._API_BASE_URL + 'PageData', display_id, query={
	23	'url': '/' + display_id,
	24	})['attributes']['response']['data']
	25
	26	def _track_url_result(self, track):
	27	track_vod = track['trackVod']
	28	video_url = track_vod.get('playerLink') or track_vod['link']
	29	return self.url_result(
	30	video_url, VideomoreIE.ie_key(), track_vod.get('hubId'))
	31
	32
030dfb04	33	class VideomoreIE(InfoExtractor):
69759a59	34	IE_NAME = 'videomore'
19d69913 S	35	_VALID_URL = r'''(?x)
19d69913 S	36	videomore:(?P<sid>\d+)$\|
29f7c58a	37	https?://
19d69913	38	(?:
29f7c58a	39	videomore\.ru/
19d69913 S	40	(?:
	41	embed\|
	42	[^/]+/[^/]+
	43	)/\|
29f7c58a	44	(?:
	45	(?:player\.)?videomore\.ru\|
	46	siren\.more\.tv/player
	47	)/[^/]\?.?\btrack_id=\|
	48	odysseus\.more.tv/player/(?P<partner_id>\d+)/
19d69913 S	49	)
	50	(?P<id>\d+)
	51	(?:[/?#&]\|\.(?:xml\|json)\|$)
	52	'''
030dfb04 S	53	_TESTS = [{
030dfb04 S	54	'url': 'http://videomore.ru/kino_v_detalayah/5_sezon/367617',
010d034f	55	'md5': '44455a346edc0d509ac5b5a5b531dc35',
030dfb04 S	56	'info_dict': {
	57	'id': '367617',
	58	'ext': 'flv',
010d034f	59	'title': 'Кино в деталях 5 сезон В гостях Алексей Чумаков и Юлия Ковальчук',
7a0b07c7 S	60	'series': 'Кино в деталях',
7a0b07c7 S	61	'episode': 'В гостях Алексей Чумаков и Юлия Ковальчук',
ec85ded8	62	'thumbnail': r're:^https?://.*\.jpg',
030dfb04	63	'duration': 2910,
030dfb04	64	'view_count': int,
010d034f S	65	'comment_count': int,
010d034f S	66	'age_limit': 16,
030dfb04	67	},
29f7c58a	68	'skip': 'The video is not available for viewing.',
7a0b07c7 S	69	}, {
	70	'url': 'http://videomore.ru/embed/259974',
	71	'info_dict': {
	72	'id': '259974',
29f7c58a	73	'ext': 'mp4',
010d034f	74	'title': 'Молодежка 2 сезон 40 серия',
7a0b07c7	75	'series': 'Молодежка',
29f7c58a	76	'season': '2 сезон',
010d034f	77	'episode': '40 серия',
ec85ded8	78	'thumbnail': r're:^https?://.*\.jpg',
29f7c58a	79	'duration': 2789,
7a0b07c7	80	'view_count': int,
010d034f	81	'age_limit': 16,
7a0b07c7 S	82	},
	83	'params': {
	84	'skip_download': True,
	85	},
	86	}, {
	87	'url': 'http://videomore.ru/molodezhka/sezon_promo/341073',
	88	'info_dict': {
	89	'id': '341073',
	90	'ext': 'flv',
010d034f	91	'title': 'Промо Команда проиграла из-за Бакина?',
7a0b07c7	92	'episode': 'Команда проиграла из-за Бакина?',
ec85ded8	93	'thumbnail': r're:^https?://.*\.jpg',
7a0b07c7 S	94	'duration': 29,
	95	'age_limit': 16,
	96	'view_count': int,
	97	},
	98	'params': {
	99	'skip_download': True,
	100	},
29f7c58a	101	'skip': 'The video is not available for viewing.',
030dfb04 S	102	}, {
	103	'url': 'http://videomore.ru/elki_3?track_id=364623',
	104	'only_matching': True,
	105	}, {
	106	'url': 'http://videomore.ru/embed/364623',
	107	'only_matching': True,
	108	}, {
	109	'url': 'http://videomore.ru/video/tracks/364623.xml',
	110	'only_matching': True,
	111	}, {
	112	'url': 'http://videomore.ru/video/tracks/364623.json',
	113	'only_matching': True,
	114	}, {
	115	'url': 'http://videomore.ru/video/tracks/158031/quotes/33248',
	116	'only_matching': True,
	117	}, {
	118	'url': 'videomore:367617',
	119	'only_matching': True,
19d69913 S	120	}, {
	121	'url': 'https://player.videomore.ru/?partner_id=97&track_id=736234&autoplay=0&userToken=',
	122	'only_matching': True,
29f7c58a	123	}, {
	124	'url': 'https://odysseus.more.tv/player/1788/352317',
	125	'only_matching': True,
	126	}, {
	127	'url': 'https://siren.more.tv/player/config?track_id=352317&partner_id=1788&user_token=',
	128	'only_matching': True,
030dfb04	129	}]
29f7c58a	130	_GEO_BYPASS = False
030dfb04 S	131
	132	@staticmethod
	133	def _extract_url(webpage):
	134	mobj = re.search(
12f211d0	135	r'<object[^>]+data=(["\'])https?://videomore\.ru/player\.swf\?.config=(?P<url>https?://videomore\.ru/(?:[^/]+/)+\d+\.xml).\1',
030dfb04	136	webpage)
146969e0 YCH	137	if not mobj:
	138	mobj = re.search(
	139	r'<iframe[^>]+src=([\'"])(?P<url>https?://videomore\.ru/embed/\d+)',
	140	webpage)
	141
030dfb04 S	142	if mobj:
	143	return mobj.group('url')
	144
	145	def _real_extract(self, url):
5ad28e7f	146	mobj = self._match_valid_url(url)
030dfb04	147	video_id = mobj.group('sid') or mobj.group('id')
4dfbf869	148	partner_id = mobj.group('partner_id') or parse_qs(url).get('partner_id', [None])[0] or '97'
29f7c58a	149
	150	item = self._download_json(
	151	'https://siren.more.tv/player/config', video_id, query={
	152	'partner_id': partner_id,
	153	'track_id': video_id,
	154	})['data']['playlist']['items'][0]
	155
	156	title = item.get('title')
	157	series = item.get('project_name')
	158	season = item.get('season_name')
	159	episode = item.get('episode_name')
	160	if not title:
	161	title = []
	162	for v in (series, season, episode):
	163	if v:
	164	title.append(v)
	165	title = ' '.join(title)
	166
	167	streams = item.get('streams') or []
	168	for protocol in ('DASH', 'HLS'):
	169	stream_url = item.get(protocol.lower() + '_url')
	170	if stream_url:
	171	streams.append({'protocol': protocol, 'url': stream_url})
	172
	173	formats = []
	174	for stream in streams:
	175	stream_url = stream.get('url')
	176	if not stream_url:
	177	continue
	178	protocol = stream.get('protocol')
	179	if protocol == 'DASH':
	180	formats.extend(self._extract_mpd_formats(
	181	stream_url, video_id, mpd_id='dash', fatal=False))
	182	elif protocol == 'HLS':
	183	formats.extend(self._extract_m3u8_formats(
	184	stream_url, video_id, 'mp4', 'm3u8_native',
	185	m3u8_id='hls', fatal=False))
	186	elif protocol == 'MSS':
	187	formats.extend(self._extract_ism_formats(
	188	stream_url, video_id, ism_id='mss', fatal=False))
	189
	190	if not formats:
	191	error = item.get('error')
	192	if error:
	193	if error in ('Данное видео недоступно для просмотра на территории этой страны', 'Данное видео доступно для просмотра только на территории России'):
b7da73eb	194	self.raise_geo_restricted(countries=['RU'], metadata_available=True)
b7da73eb	195	self.raise_no_formats(error, expected=True)
19dbaeec	196	self._sort_formats(formats)
030dfb04	197
030dfb04 S	198	return {
	199	'id': video_id,
	200	'title': title,
7a0b07c7	201	'series': series,
29f7c58a	202	'season': season,
7a0b07c7	203	'episode': episode,
29f7c58a	204	'thumbnail': item.get('thumbnail_url'),
	205	'duration': int_or_none(item.get('duration')),
	206	'view_count': int_or_none(item.get('views')),
	207	'age_limit': int_or_none(item.get('min_age')),
030dfb04 S	208	'formats': formats,
	209	}
	210
	211
29f7c58a	212	class VideomoreVideoIE(VideomoreBaseIE):
030dfb04	213	IE_NAME = 'videomore:video'
29f7c58a	214	_VALID_URL = VideomoreBaseIE._VALID_URL_BASE + r'(?P<id>(?:(?:[^/]+/){2})?[^/?#&]+)(?:/\|[?#&].?)$'
030dfb04 S	215	_TESTS = [{
	216	# single video with og:video:iframe
	217	'url': 'http://videomore.ru/elki_3',
	218	'info_dict': {
	219	'id': '364623',
	220	'ext': 'flv',
	221	'title': 'Ёлки 3',
	222	'description': '',
ec85ded8	223	'thumbnail': r're:^https?://.*\.jpg',
030dfb04 S	224	'duration': 5579,
	225	'age_limit': 6,
	226	'view_count': int,
	227	},
	228	'params': {
	229	'skip_download': True,
	230	},
29f7c58a	231	'skip': 'Requires logging in',
030dfb04	232	}, {
dfb1b146	233	# season single series with og:video:iframe
030dfb04	234	'url': 'http://videomore.ru/poslednii_ment/1_sezon/14_seriya',
29f7c58a	235	'info_dict': {
	236	'id': '352317',
	237	'ext': 'mp4',
	238	'title': 'Последний мент 1 сезон 14 серия',
	239	'series': 'Последний мент',
	240	'season': '1 сезон',
	241	'episode': '14 серия',
	242	'thumbnail': r're:^https?://.*\.jpg',
	243	'duration': 2464,
	244	'age_limit': 16,
	245	'view_count': int,
	246	},
	247	'params': {
	248	'skip_download': True,
	249	},
030dfb04 S	250	}, {
	251	'url': 'http://videomore.ru/sejchas_v_seti/serii_221-240/226_vypusk',
	252	'only_matching': True,
	253	}, {
	254	# single video without og:video:iframe
	255	'url': 'http://videomore.ru/marin_i_ego_druzya',
	256	'info_dict': {
	257	'id': '359073',
	258	'ext': 'flv',
	259	'title': '1 серия. Здравствуй, Аквавилль!',
	260	'description': 'md5:c6003179538b5d353e7bcd5b1372b2d7',
ec85ded8	261	'thumbnail': r're:^https?://.*\.jpg',
030dfb04 S	262	'duration': 754,
	263	'age_limit': 6,
	264	'view_count': int,
	265	},
	266	'params': {
	267	'skip_download': True,
	268	},
29f7c58a	269	'skip': 'redirects to https://more.tv/'
19d69913 S	270	}, {
	271	'url': 'https://videomore.ru/molodezhka/6_sezon/29_seriya?utm_so',
	272	'only_matching': True,
29f7c58a	273	}, {
	274	'url': 'https://more.tv/poslednii_ment/1_sezon/14_seriya',
	275	'only_matching': True,
030dfb04 S	276	}]
	277
	278	@classmethod
	279	def suitable(cls, url):
	280	return False if VideomoreIE.suitable(url) else super(VideomoreVideoIE, cls).suitable(url)
	281
	282	def _real_extract(self, url):
	283	display_id = self._match_id(url)
29f7c58a	284	return self._track_url_result(self._download_page_data(display_id))
030dfb04 S	285
030dfb04 S	286
29f7c58a	287	class VideomoreSeasonIE(VideomoreBaseIE):
030dfb04	288	IE_NAME = 'videomore:season'
29f7c58a	289	_VALID_URL = VideomoreBaseIE._VALID_URL_BASE + r'(?!embed)(?P<id>[^/]+/[^/?#&]+)(?:/\|[?#&].?)$'
030dfb04	290	_TESTS = [{
29f7c58a	291	'url': 'http://videomore.ru/molodezhka/film_o_filme',
030dfb04	292	'info_dict': {
29f7c58a	293	'id': 'molodezhka/film_o_filme',
29f7c58a	294	'title': 'Фильм о фильме',
030dfb04	295	},
29f7c58a	296	'playlist_mincount': 3,
19d69913 S	297	}, {
	298	'url': 'http://videomore.ru/molodezhka/sezon_promo?utm_so',
	299	'only_matching': True,
29f7c58a	300	}, {
	301	'url': 'https://more.tv/molodezhka/film_o_filme',
	302	'only_matching': True,
030dfb04 S	303	}]
030dfb04 S	304
19d69913 S	305	@classmethod
	306	def suitable(cls, url):
	307	return (False if (VideomoreIE.suitable(url) or VideomoreVideoIE.suitable(url))
	308	else super(VideomoreSeasonIE, cls).suitable(url))
	309
030dfb04 S	310	def _real_extract(self, url):
030dfb04 S	311	display_id = self._match_id(url)
29f7c58a	312	season = self._download_page_data(display_id)
	313	season_id = compat_str(season['id'])
	314	tracks = self._download_json(
	315	self._API_BASE_URL + 'seasons/%s/tracks' % season_id,
	316	season_id)['data']
19d69913	317	entries = []
29f7c58a	318	for track in tracks:
	319	entries.append(self._track_url_result(track))
	320	return self.playlist_result(entries, display_id, season.get('title'))