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

from __future__ import unicode_literals

import re

from .common import InfoExtractor
from ..compat import (
    compat_parse_qs,
    compat_urllib_parse_urlparse,
)
from ..utils import (
    HEADRequest,
    determine_ext,
    int_or_none,
    parse_iso8601,
    strip_or_none,
    try_get,
)


class IGNBaseIE(InfoExtractor):
    def _call_api(self, slug):
        return self._download_json(
            'http://apis.ign.com/{0}/v3/{0}s/slug/{1}'.format(self._PAGE_TYPE, slug), slug)


class IGNIE(IGNBaseIE):
    """
    Extractor for some of the IGN sites, like www.ign.com, es.ign.com de.ign.com.
    Some videos of it.ign.com are also supported
    """

    _VALID_URL = r'https?://(?:.+?\.ign|www\.pcmag)\.com/videos/(?:\d{4}/\d{2}/\d{2}/)?(?P<id>[^/?&#]+)'
    IE_NAME = 'ign.com'
    _PAGE_TYPE = 'video'

    _TESTS = [{
        'url': 'http://www.ign.com/videos/2013/06/05/the-last-of-us-review',
        'md5': 'd2e1586d9987d40fad7867bf96a018ea',
        'info_dict': {
            'id': '8f862beef863986b2785559b9e1aa599',
            'ext': 'mp4',
            'title': 'The Last of Us Review',
            'description': 'md5:c8946d4260a4d43a00d5ae8ed998870c',
            'timestamp': 1370440800,
            'upload_date': '20130605',
            'tags': 'count:9',
        }
    }, {
        'url': 'http://www.pcmag.com/videos/2015/01/06/010615-whats-new-now-is-gogo-snooping-on-your-data',
        'md5': 'f1581a6fe8c5121be5b807684aeac3f6',
        'info_dict': {
            'id': 'ee10d774b508c9b8ec07e763b9125b91',
            'ext': 'mp4',
            'title': 'What\'s New Now: Is GoGo Snooping on Your Data?',
            'description': 'md5:817a20299de610bd56f13175386da6fa',
            'timestamp': 1420571160,
            'upload_date': '20150106',
            'tags': 'count:4',
        }
    }, {
        'url': 'https://www.ign.com/videos/is-a-resident-evil-4-remake-on-the-way-ign-daily-fix',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        display_id = self._match_id(url)
        video = self._call_api(display_id)
        video_id = video['videoId']
        metadata = video['metadata']
        title = metadata.get('longTitle') or metadata.get('title') or metadata['name']

        formats = []
        refs = video.get('refs') or {}

        m3u8_url = refs.get('m3uUrl')
        if m3u8_url:
            formats.extend(self._extract_m3u8_formats(
                m3u8_url, video_id, 'mp4', 'm3u8_native',
                m3u8_id='hls', fatal=False))

        f4m_url = refs.get('f4mUrl')
        if f4m_url:
            formats.extend(self._extract_f4m_formats(
                f4m_url, video_id, f4m_id='hds', fatal=False))

        for asset in (video.get('assets') or []):
            asset_url = asset.get('url')
            if not asset_url:
                continue
            formats.append({
                'url': asset_url,
                'tbr': int_or_none(asset.get('bitrate'), 1000),
                'fps': int_or_none(asset.get('frame_rate')),
                'height': int_or_none(asset.get('height')),
                'width': int_or_none(asset.get('width')),
            })

        mezzanine_url = try_get(video, lambda x: x['system']['mezzanineUrl'])
        if mezzanine_url:
            formats.append({
                'ext': determine_ext(mezzanine_url, 'mp4'),
                'format_id': 'mezzanine',
                'quality': 1,
                'url': mezzanine_url,
            })

        self._sort_formats(formats)

        thumbnails = []
        for thumbnail in (video.get('thumbnails') or []):
            thumbnail_url = thumbnail.get('url')
            if not thumbnail_url:
                continue
            thumbnails.append({
                'url': thumbnail_url,
            })

        tags = []
        for tag in (video.get('tags') or []):
            display_name = tag.get('displayName')
            if not display_name:
                continue
            tags.append(display_name)

        return {
            'id': video_id,
            'title': title,
            'description': strip_or_none(metadata.get('description')),
            'timestamp': parse_iso8601(metadata.get('publishDate')),
            'duration': int_or_none(metadata.get('duration')),
            'display_id': display_id,
            'thumbnails': thumbnails,
            'formats': formats,
            'tags': tags,
        }


class IGNVideoIE(InfoExtractor):
    _VALID_URL = r'https?://.+?\.ign\.com/(?:[a-z]{2}/)?[^/]+/(?P<id>\d+)/(?:video|trailer)/'
    _TESTS = [{
        'url': 'http://me.ign.com/en/videos/112203/video/how-hitman-aims-to-be-different-than-every-other-s',
        'md5': 'dd9aca7ed2657c4e118d8b261e5e9de1',
        'info_dict': {
            'id': 'e9be7ea899a9bbfc0674accc22a36cc8',
            'ext': 'mp4',
            'title': 'How Hitman Aims to Be Different Than Every Other Stealth Game - NYCC 2015',
            'description': 'Taking out assassination targets in Hitman has never been more stylish.',
            'timestamp': 1444665600,
            'upload_date': '20151012',
        }
    }, {
        'url': 'http://me.ign.com/ar/angry-birds-2/106533/video/lrd-ldyy-lwl-lfylm-angry-birds',
        'only_matching': True,
    }, {
        # Youtube embed
        'url': 'https://me.ign.com/ar/ratchet-clank-rift-apart/144327/trailer/embed',
        'only_matching': True,
    }, {
        # Twitter embed
        'url': 'http://adria.ign.com/sherlock-season-4/9687/trailer/embed',
        'only_matching': True,
    }, {
        # Vimeo embed
        'url': 'https://kr.ign.com/bic-2018/3307/trailer/embed',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        video_id = self._match_id(url)
        req = HEADRequest(url.rsplit('/', 1)[0] + '/embed')
        url = self._request_webpage(req, video_id).geturl()
        ign_url = compat_parse_qs(
            compat_urllib_parse_urlparse(url).query).get('url', [None])[0]
        if ign_url:
            return self.url_result(ign_url, IGNIE.ie_key())
        return self.url_result(url)


class IGNArticleIE(IGNBaseIE):
    _VALID_URL = r'https?://.+?\.ign\.com/(?:articles(?:/\d{4}/\d{2}/\d{2})?|(?:[a-z]{2}/)?feature/\d+)/(?P<id>[^/?&#]+)'
    _PAGE_TYPE = 'article'
    _TESTS = [{
        'url': 'http://me.ign.com/en/feature/15775/100-little-things-in-gta-5-that-will-blow-your-mind',
        'info_dict': {
            'id': '524497489e4e8ff5848ece34',
            'title': '100 Little Things in GTA 5 That Will Blow Your Mind',
        },
        'playlist': [
            {
                'info_dict': {
                    'id': '5ebbd138523268b93c9141af17bec937',
                    'ext': 'mp4',
                    'title': 'GTA 5 Video Review',
                    'description': 'Rockstar drops the mic on this generation of games. Watch our review of the masterly Grand Theft Auto V.',
                    'timestamp': 1379339880,
                    'upload_date': '20130916',
                },
            },
            {
                'info_dict': {
                    'id': '638672ee848ae4ff108df2a296418ee2',
                    'ext': 'mp4',
                    'title': '26 Twisted Moments from GTA 5 in Slow Motion',
                    'description': 'The twisted beauty of GTA 5 in stunning slow motion.',
                    'timestamp': 1386878820,
                    'upload_date': '20131212',
                },
            },
        ],
        'params': {
            'playlist_items': '2-3',
            'skip_download': True,
        },
    }, {
        'url': 'http://www.ign.com/articles/2014/08/15/rewind-theater-wild-trailer-gamescom-2014?watch',
        'info_dict': {
            'id': '53ee806780a81ec46e0790f8',
            'title': 'Rewind Theater - Wild Trailer Gamescom 2014',
        },
        'playlist_count': 2,
    }, {
        # videoId pattern
        'url': 'http://www.ign.com/articles/2017/06/08/new-ducktales-short-donalds-birthday-doesnt-go-as-planned',
        'only_matching': True,
    }, {
        # Youtube embed
        'url': 'https://www.ign.com/articles/2021-mvp-named-in-puppy-bowl-xvii',
        'only_matching': True,
    }, {
        # IMDB embed
        'url': 'https://www.ign.com/articles/2014/08/07/sons-of-anarchy-final-season-trailer',
        'only_matching': True,
    }, {
        # Facebook embed
        'url': 'https://www.ign.com/articles/2017/09/20/marvels-the-punisher-watch-the-new-trailer-for-the-netflix-series',
        'only_matching': True,
    }, {
        # Brightcove embed
        'url': 'https://www.ign.com/articles/2016/01/16/supergirl-goes-flying-with-martian-manhunter-in-new-clip',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        display_id = self._match_id(url)
        article = self._call_api(display_id)

        def entries():
            media_url = try_get(article, lambda x: x['mediaRelations'][0]['media']['metadata']['url'])
            if media_url:
                yield self.url_result(media_url, IGNIE.ie_key())
            for content in (article.get('content') or []):
                for video_url in re.findall(r'(?:\[(?:ignvideo\s+url|youtube\s+clip_id)|<iframe[^>]+src)="([^"]+)"', content):
                    yield self.url_result(video_url)

        return self.playlist_result(
            entries(), article.get('articleId'),
            strip_or_none(try_get(article, lambda x: x['metadata']['headline'])))
Commit	Line	Data
40c716d2 JMF	1	from __future__ import unicode_literals
40c716d2 JMF	2
2ef648d3	3	import re
2ef648d3 JMF	4
2ef648d3 JMF	5	from .common import InfoExtractor
cc2db878	6	from ..compat import (
	7	compat_parse_qs,
	8	compat_urllib_parse_urlparse,
	9	)
adccf336	10	from ..utils import (
cc2db878	11	HEADRequest,
cc2db878	12	determine_ext,
adccf336	13	int_or_none,
adccf336	14	parse_iso8601,
cc2db878	15	strip_or_none,
cc2db878	16	try_get,
adccf336	17	)
2ef648d3	18
a95967f8	19
cc2db878	20	class IGNBaseIE(InfoExtractor):
	21	def _call_api(self, slug):
	22	return self._download_json(
	23	'http://apis.ign.com/{0}/v3/{0}s/slug/{1}'.format(self._PAGE_TYPE, slug), slug)
	24
	25
	26	class IGNIE(IGNBaseIE):
a95967f8 JMF	27	"""
	28	Extractor for some of the IGN sites, like www.ign.com, es.ign.com de.ign.com.
	29	Some videos of it.ign.com are also supported
	30	"""
	31
cc2db878	32	_VALID_URL = r'https?://(?:.+?\.ign\|www\.pcmag)\.com/videos/(?:\d{4}/\d{2}/\d{2}/)?(?P<id>[^/?&#]+)'
40c716d2	33	IE_NAME = 'ign.com'
cc2db878	34	_PAGE_TYPE = 'video'
2ef648d3	35
cc2db878	36	_TESTS = [{
	37	'url': 'http://www.ign.com/videos/2013/06/05/the-last-of-us-review',
	38	'md5': 'd2e1586d9987d40fad7867bf96a018ea',
	39	'info_dict': {
	40	'id': '8f862beef863986b2785559b9e1aa599',
	41	'ext': 'mp4',
	42	'title': 'The Last of Us Review',
	43	'description': 'md5:c8946d4260a4d43a00d5ae8ed998870c',
	44	'timestamp': 1370440800,
	45	'upload_date': '20130605',
	46	'tags': 'count:9',
	47	}
	48	}, {
	49	'url': 'http://www.pcmag.com/videos/2015/01/06/010615-whats-new-now-is-gogo-snooping-on-your-data',
	50	'md5': 'f1581a6fe8c5121be5b807684aeac3f6',
	51	'info_dict': {
	52	'id': 'ee10d774b508c9b8ec07e763b9125b91',
	53	'ext': 'mp4',
	54	'title': 'What\'s New Now: Is GoGo Snooping on Your Data?',
	55	'description': 'md5:817a20299de610bd56f13175386da6fa',
	56	'timestamp': 1420571160,
	57	'upload_date': '20150106',
	58	'tags': 'count:4',
	59	}
	60	}, {
	61	'url': 'https://www.ign.com/videos/is-a-resident-evil-4-remake-on-the-way-ign-daily-fix',
	62	'only_matching': True,
	63	}]
a95967f8	64
2ef648d3	65	def _real_extract(self, url):
cc2db878	66	display_id = self._match_id(url)
	67	video = self._call_api(display_id)
	68	video_id = video['videoId']
	69	metadata = video['metadata']
	70	title = metadata.get('longTitle') or metadata.get('title') or metadata['name']
adccf336	71
adccf336	72	formats = []
cc2db878	73	refs = video.get('refs') or {}
	74
	75	m3u8_url = refs.get('m3uUrl')
adccf336	76	if m3u8_url:
f889ac45	77	formats.extend(self._extract_m3u8_formats(
	78	m3u8_url, video_id, 'mp4', 'm3u8_native',
	79	m3u8_id='hls', fatal=False))
cc2db878	80
cc2db878	81	f4m_url = refs.get('f4mUrl')
adccf336	82	if f4m_url:
f889ac45	83	formats.extend(self._extract_f4m_formats(
f889ac45	84	f4m_url, video_id, f4m_id='hds', fatal=False))
cc2db878	85
	86	for asset in (video.get('assets') or []):
	87	asset_url = asset.get('url')
	88	if not asset_url:
	89	continue
adccf336	90	formats.append({
cc2db878	91	'url': asset_url,
	92	'tbr': int_or_none(asset.get('bitrate'), 1000),
	93	'fps': int_or_none(asset.get('frame_rate')),
adccf336	94	'height': int_or_none(asset.get('height')),
	95	'width': int_or_none(asset.get('width')),
	96	})
cc2db878	97
	98	mezzanine_url = try_get(video, lambda x: x['system']['mezzanineUrl'])
	99	if mezzanine_url:
	100	formats.append({
	101	'ext': determine_ext(mezzanine_url, 'mp4'),
	102	'format_id': 'mezzanine',
f983b875	103	'quality': 1,
cc2db878	104	'url': mezzanine_url,
	105	})
	106
adccf336	107	self._sort_formats(formats)
adccf336	108
cc2db878	109	thumbnails = []
	110	for thumbnail in (video.get('thumbnails') or []):
	111	thumbnail_url = thumbnail.get('url')
	112	if not thumbnail_url:
	113	continue
	114	thumbnails.append({
	115	'url': thumbnail_url,
	116	})
adccf336	117
cc2db878	118	tags = []
	119	for tag in (video.get('tags') or []):
	120	display_name = tag.get('displayName')
	121	if not display_name:
	122	continue
	123	tags.append(display_name)
2ef648d3	124
40c716d2	125	return {
cc2db878	126	'id': video_id,
	127	'title': title,
	128	'description': strip_or_none(metadata.get('description')),
adccf336	129	'timestamp': parse_iso8601(metadata.get('publishDate')),
adccf336	130	'duration': int_or_none(metadata.get('duration')),
cc2db878	131	'display_id': display_id,
adccf336	132	'thumbnails': thumbnails,
adccf336	133	'formats': formats,
cc2db878	134	'tags': tags,
40c716d2	135	}
2ef648d3 JMF	136
2ef648d3 JMF	137
cc2db878	138	class IGNVideoIE(InfoExtractor):
cc2db878	139	_VALID_URL = r'https?://.+?\.ign\.com/(?:[a-z]{2}/)?[^/]+/(?P<id>\d+)/(?:video\|trailer)/'
52fadd5f	140	_TESTS = [{
cc2db878	141	'url': 'http://me.ign.com/en/videos/112203/video/how-hitman-aims-to-be-different-than-every-other-s',
cc2db878	142	'md5': 'dd9aca7ed2657c4e118d8b261e5e9de1',
40c716d2	143	'info_dict': {
cc2db878	144	'id': 'e9be7ea899a9bbfc0674accc22a36cc8',
40c716d2	145	'ext': 'mp4',
cc2db878	146	'title': 'How Hitman Aims to Be Different Than Every Other Stealth Game - NYCC 2015',
	147	'description': 'Taking out assassination targets in Hitman has never been more stylish.',
	148	'timestamp': 1444665600,
	149	'upload_date': '20151012',
a95967f8	150	}
cc2db878	151	}, {
	152	'url': 'http://me.ign.com/ar/angry-birds-2/106533/video/lrd-ldyy-lwl-lfylm-angry-birds',
	153	'only_matching': True,
	154	}, {
	155	# Youtube embed
	156	'url': 'https://me.ign.com/ar/ratchet-clank-rift-apart/144327/trailer/embed',
	157	'only_matching': True,
	158	}, {
	159	# Twitter embed
	160	'url': 'http://adria.ign.com/sherlock-season-4/9687/trailer/embed',
	161	'only_matching': True,
	162	}, {
	163	# Vimeo embed
	164	'url': 'https://kr.ign.com/bic-2018/3307/trailer/embed',
	165	'only_matching': True,
52fadd5f	166	}]
ee6adb16	167
a95967f8	168	def _real_extract(self, url):
cc2db878	169	video_id = self._match_id(url)
	170	req = HEADRequest(url.rsplit('/', 1)[0] + '/embed')
	171	url = self._request_webpage(req, video_id).geturl()
	172	ign_url = compat_parse_qs(
	173	compat_urllib_parse_urlparse(url).query).get('url', [None])[0]
	174	if ign_url:
	175	return self.url_result(ign_url, IGNIE.ie_key())
	176	return self.url_result(url)
adccf336	177
adccf336	178
cc2db878	179	class IGNArticleIE(IGNBaseIE):
	180	_VALID_URL = r'https?://.+?\.ign\.com/(?:articles(?:/\d{4}/\d{2}/\d{2})?\|(?:[a-z]{2}/)?feature/\d+)/(?P<id>[^/?&#]+)'
	181	_PAGE_TYPE = 'article'
adccf336	182	_TESTS = [{
cc2db878	183	'url': 'http://me.ign.com/en/feature/15775/100-little-things-in-gta-5-that-will-blow-your-mind',
adccf336	184	'info_dict': {
cc2db878	185	'id': '524497489e4e8ff5848ece34',
	186	'title': '100 Little Things in GTA 5 That Will Blow Your Mind',
	187	},
	188	'playlist': [
	189	{
	190	'info_dict': {
	191	'id': '5ebbd138523268b93c9141af17bec937',
	192	'ext': 'mp4',
	193	'title': 'GTA 5 Video Review',
	194	'description': 'Rockstar drops the mic on this generation of games. Watch our review of the masterly Grand Theft Auto V.',
	195	'timestamp': 1379339880,
	196	'upload_date': '20130916',
	197	},
	198	},
	199	{
	200	'info_dict': {
	201	'id': '638672ee848ae4ff108df2a296418ee2',
	202	'ext': 'mp4',
	203	'title': '26 Twisted Moments from GTA 5 in Slow Motion',
	204	'description': 'The twisted beauty of GTA 5 in stunning slow motion.',
	205	'timestamp': 1386878820,
	206	'upload_date': '20131212',
	207	},
	208	},
	209	],
	210	'params': {
	211	'playlist_items': '2-3',
	212	'skip_download': True,
	213	},
607d65fb	214	}, {
cc2db878	215	'url': 'http://www.ign.com/articles/2014/08/15/rewind-theater-wild-trailer-gamescom-2014?watch',
adccf336	216	'info_dict': {
cc2db878	217	'id': '53ee806780a81ec46e0790f8',
	218	'title': 'Rewind Theater - Wild Trailer Gamescom 2014',
	219	},
	220	'playlist_count': 2,
	221	}, {
	222	# videoId pattern
	223	'url': 'http://www.ign.com/articles/2017/06/08/new-ducktales-short-donalds-birthday-doesnt-go-as-planned',
	224	'only_matching': True,
	225	}, {
	226	# Youtube embed
	227	'url': 'https://www.ign.com/articles/2021-mvp-named-in-puppy-bowl-xvii',
	228	'only_matching': True,
	229	}, {
	230	# IMDB embed
	231	'url': 'https://www.ign.com/articles/2014/08/07/sons-of-anarchy-final-season-trailer',
	232	'only_matching': True,
	233	}, {
	234	# Facebook embed
	235	'url': 'https://www.ign.com/articles/2017/09/20/marvels-the-punisher-watch-the-new-trailer-for-the-netflix-series',
	236	'only_matching': True,
	237	}, {
	238	# Brightcove embed
	239	'url': 'https://www.ign.com/articles/2016/01/16/supergirl-goes-flying-with-martian-manhunter-in-new-clip',
	240	'only_matching': True,
adccf336	241	}]
cc2db878	242
	243	def _real_extract(self, url):
	244	display_id = self._match_id(url)
	245	article = self._call_api(display_id)
	246
	247	def entries():
	248	media_url = try_get(article, lambda x: x['mediaRelations'][0]['media']['metadata']['url'])
	249	if media_url:
	250	yield self.url_result(media_url, IGNIE.ie_key())
	251	for content in (article.get('content') or []):
	252	for video_url in re.findall(r'(?:\[(?:ignvideo\s+url\|youtube\s+clip_id)\|<iframe[^>]+src)="([^"]+)"', content):
	253	yield self.url_result(video_url)
	254
	255	return self.playlist_result(
	256	entries(), article.get('articleId'),
	257	strip_or_none(try_get(article, lambda x: x['metadata']['headline'])))