[yt-dlp.git] / youtube_dl / extractor / nbc.py

from __future__ import unicode_literals

import re

from .common import InfoExtractor
from .theplatform import ThePlatformIE
from .adobepass import AdobePassIE
from ..utils import (
    find_xpath_attr,
    smuggle_url,
    unescapeHTML,
    update_url_query,
    int_or_none,
)


class NBCIE(AdobePassIE):
    _VALID_URL = r'(?P<permalink>https?://(?:www\.)?nbc\.com/[^/]+/video/[^/]+/(?P<id>n?\d+))'

    _TESTS = [
        {
            'url': 'http://www.nbc.com/the-tonight-show/video/jimmy-fallon-surprises-fans-at-ben-jerrys/2848237',
            'info_dict': {
                'id': '2848237',
                'ext': 'mp4',
                'title': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s',
                'description': 'Jimmy gives out free scoops of his new "Tonight Dough" ice cream flavor by surprising customers at the Ben & Jerry\'s scoop shop.',
                'timestamp': 1424246400,
                'upload_date': '20150218',
                'uploader': 'NBCU-COM',
            },
            'params': {
                # m3u8 download
                'skip_download': True,
            },
        },
        {
            'url': 'http://www.nbc.com/saturday-night-live/video/star-wars-teaser/2832821',
            'info_dict': {
                'id': '2832821',
                'ext': 'mp4',
                'title': 'Star Wars Teaser',
                'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442',
                'timestamp': 1417852800,
                'upload_date': '20141206',
                'uploader': 'NBCU-COM',
            },
            'params': {
                # m3u8 download
                'skip_download': True,
            },
            'skip': 'Only works from US',
        },
        {
            # HLS streams requires the 'hdnea3' cookie
            'url': 'http://www.nbc.com/Kings/video/goliath/n1806',
            'info_dict': {
                'id': '101528f5a9e8127b107e98c5e6ce4638',
                'ext': 'mp4',
                'title': 'Goliath',
                'description': 'When an unknown soldier saves the life of the King\'s son in battle, he\'s thrust into the limelight and politics of the kingdom.',
                'timestamp': 1237100400,
                'upload_date': '20090315',
                'uploader': 'NBCU-COM',
            },
            'params': {
                'skip_download': True,
            },
            'skip': 'Only works from US',
        }
    ]

    def _real_extract(self, url):
        permalink, video_id = re.match(self._VALID_URL, url).groups()
        video_data = self._download_json(
            'https://api.nbc.com/v3/videos', video_id, query={
                'filter[permalink]': permalink,
            })['data'][0]['attributes']
        query = {
            'mbr': 'true',
            'manifest': 'm3u',
        }
        video_id = video_data['guid']
        title = video_data['title']
        if video_data.get('entitlement') == 'auth':
            resource = self._get_mvpd_resource(
                'nbcentertainment', title, video_id,
                video_data.get('vChipRating'))
            query['auth'] = self._extract_mvpd_auth(
                url, video_id, 'nbcentertainment', resource)
        theplatform_url = smuggle_url(update_url_query(
            'http://link.theplatform.com/s/NnzsPC/media/guid/2410887629/' + video_id,
            query), {'force_smil_url': True})
        return {
            '_type': 'url_transparent',
            'id': video_id,
            'title': title,
            'url': theplatform_url,
            'description': video_data.get('description'),
            'keywords': video_data.get('keywords'),
            'season_number': int_or_none(video_data.get('seasonNumber')),
            'episode_number': int_or_none(video_data.get('episodeNumber')),
            'series': video_data.get('showName'),
            'ie_key': 'ThePlatform',
        }


class NBCSportsVPlayerIE(InfoExtractor):
    _VALID_URL = r'https?://vplayer\.nbcsports\.com/(?:[^/]+/)+(?P<id>[0-9a-zA-Z_]+)'

    _TESTS = [{
        'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_share/select/9CsDKds0kvHI',
        'info_dict': {
            'id': '9CsDKds0kvHI',
            'ext': 'flv',
            'description': 'md5:df390f70a9ba7c95ff1daace988f0d8d',
            'title': 'Tyler Kalinoski hits buzzer-beater to lift Davidson',
            'timestamp': 1426270238,
            'upload_date': '20150313',
            'uploader': 'NBCU-SPORTS',
        }
    }, {
        'url': 'http://vplayer.nbcsports.com/p/BxmELC/nbc_embedshare/select/_hqLjQ95yx8Z',
        'only_matching': True,
    }]

    @staticmethod
    def _extract_url(webpage):
        iframe_m = re.search(
            r'<iframe[^>]+src="(?P<url>https?://vplayer\.nbcsports\.com/[^"]+)"', webpage)
        if iframe_m:
            return iframe_m.group('url')

    def _real_extract(self, url):
        video_id = self._match_id(url)
        webpage = self._download_webpage(url, video_id)
        theplatform_url = self._og_search_video_url(webpage)
        return self.url_result(theplatform_url, 'ThePlatform')


class NBCSportsIE(InfoExtractor):
    # Does not include https because its certificate is invalid
    _VALID_URL = r'https?://(?:www\.)?nbcsports\.com//?(?:[^/]+/)+(?P<id>[0-9a-z-]+)'

    _TEST = {
        'url': 'http://www.nbcsports.com//college-basketball/ncaab/tom-izzo-michigan-st-has-so-much-respect-duke',
        'info_dict': {
            'id': 'PHJSaFWbrTY9',
            'ext': 'flv',
            'title': 'Tom Izzo, Michigan St. has \'so much respect\' for Duke',
            'description': 'md5:ecb459c9d59e0766ac9c7d5d0eda8113',
            'uploader': 'NBCU-SPORTS',
            'upload_date': '20150330',
            'timestamp': 1427726529,
        }
    }

    def _real_extract(self, url):
        video_id = self._match_id(url)
        webpage = self._download_webpage(url, video_id)
        return self.url_result(
            NBCSportsVPlayerIE._extract_url(webpage), 'NBCSportsVPlayer')


class CSNNEIE(InfoExtractor):
    _VALID_URL = r'https?://(?:www\.)?csnne\.com/video/(?P<id>[0-9a-z-]+)'

    _TEST = {
        'url': 'http://www.csnne.com/video/snc-evening-update-wright-named-red-sox-no-5-starter',
        'info_dict': {
            'id': 'yvBLLUgQ8WU0',
            'ext': 'mp4',
            'title': 'SNC evening update: Wright named Red Sox\' No. 5 starter.',
            'description': 'md5:1753cfee40d9352b19b4c9b3e589b9e3',
            'timestamp': 1459369979,
            'upload_date': '20160330',
            'uploader': 'NBCU-SPORTS',
        }
    }

    def _real_extract(self, url):
        display_id = self._match_id(url)
        webpage = self._download_webpage(url, display_id)
        return {
            '_type': 'url_transparent',
            'ie_key': 'ThePlatform',
            'url': self._html_search_meta('twitter:player:stream', webpage),
            'display_id': display_id,
        }


class NBCNewsIE(ThePlatformIE):
    _VALID_URL = r'''(?x)https?://(?:www\.)?(?:nbcnews|today|msnbc)\.com/
        (?:video/.+?/(?P<id>\d+)|
        ([^/]+/)*(?:.*-)?(?P<mpx_id>[^/?]+))
        '''

    _TESTS = [
        {
            'url': 'http://www.nbcnews.com/video/nbc-news/52753292',
            'md5': '47abaac93c6eaf9ad37ee6c4463a5179',
            'info_dict': {
                'id': '52753292',
                'ext': 'flv',
                'title': 'Crew emerges after four-month Mars food study',
                'description': 'md5:24e632ffac72b35f8b67a12d1b6ddfc1',
            },
        },
        {
            'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
            'md5': 'af1adfa51312291a017720403826bb64',
            'info_dict': {
                'id': 'p_tweet_snow_140529',
                'ext': 'mp4',
                'title': 'How Twitter Reacted To The Snowden Interview',
                'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
                'uploader': 'NBCU-NEWS',
                'timestamp': 1401363060,
                'upload_date': '20140529',
            },
        },
        {
            'url': 'http://www.nbcnews.com/feature/dateline-full-episodes/full-episode-family-business-n285156',
            'md5': 'fdbf39ab73a72df5896b6234ff98518a',
            'info_dict': {
                'id': '529953347624',
                'ext': 'mp4',
                'title': 'FULL EPISODE: Family Business',
                'description': 'md5:757988edbaae9d7be1d585eb5d55cc04',
            },
            'skip': 'This page is unavailable.',
        },
        {
            'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
            'md5': '73135a2e0ef819107bbb55a5a9b2a802',
            'info_dict': {
                'id': 'nn_netcast_150204',
                'ext': 'mp4',
                'title': 'Nightly News with Brian Williams Full Broadcast (February 4)',
                'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
                'timestamp': 1423104900,
                'uploader': 'NBCU-NEWS',
                'upload_date': '20150205',
            },
        },
        {
            'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
            'md5': 'a49e173825e5fcd15c13fc297fced39d',
            'info_dict': {
                'id': 'x_lon_vwhorn_150922',
                'ext': 'mp4',
                'title': 'Volkswagen U.S. Chief:\xa0 We Have Totally Screwed Up',
                'description': 'md5:c8be487b2d80ff0594c005add88d8351',
                'upload_date': '20150922',
                'timestamp': 1442917800,
                'uploader': 'NBCU-NEWS',
            },
        },
        {
            'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
            'md5': '118d7ca3f0bea6534f119c68ef539f71',
            'info_dict': {
                'id': 'tdy_al_space_160420',
                'ext': 'mp4',
                'title': 'See the aurora borealis from space in stunning new NASA video',
                'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
                'upload_date': '20160420',
                'timestamp': 1461152093,
                'uploader': 'NBCU-NEWS',
            },
        },
        {
            'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924',
            'md5': '6d236bf4f3dddc226633ce6e2c3f814d',
            'info_dict': {
                'id': 'n_hayes_Aimm_140801_272214',
                'ext': 'mp4',
                'title': 'The chaotic GOP immigration vote',
                'description': 'The Republican House votes on a border bill that has no chance of getting through the Senate or signed by the President and is drawing criticism from all sides.',
                'thumbnail': r're:^https?://.*\.jpg$',
                'timestamp': 1406937606,
                'upload_date': '20140802',
                'uploader': 'NBCU-NEWS',
            },
        },
        {
            'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952',
            'only_matching': True,
        },
        {
            # From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html
            'url': 'http://www.nbcnews.com/widget/video-embed/701714499682',
            'only_matching': True,
        },
    ]

    def _real_extract(self, url):
        mobj = re.match(self._VALID_URL, url)
        video_id = mobj.group('id')
        if video_id is not None:
            all_info = self._download_xml('http://www.nbcnews.com/id/%s/displaymode/1219' % video_id, video_id)
            info = all_info.find('video')

            return {
                'id': video_id,
                'title': info.find('headline').text,
                'ext': 'flv',
                'url': find_xpath_attr(info, 'media', 'type', 'flashVideo').text,
                'description': info.find('caption').text,
                'thumbnail': find_xpath_attr(info, 'media', 'type', 'thumbnail').text,
            }
        else:
            # "feature" and "nightly-news" pages use theplatform.com
            video_id = mobj.group('mpx_id')
            webpage = self._download_webpage(url, video_id)

            filter_param = 'byId'
            bootstrap_json = self._search_regex(
                [r'(?m)(?:var\s+(?:bootstrapJson|playlistData)|NEWS\.videoObj)\s*=\s*({.+});?\s*$',
                 r'videoObj\s*:\s*({.+})', r'data-video="([^"]+)"',
                 r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);'],
                webpage, 'bootstrap json', default=None)
            if bootstrap_json:
                bootstrap = self._parse_json(
                    bootstrap_json, video_id, transform_source=unescapeHTML)

                info = None
                if 'results' in bootstrap:
                    info = bootstrap['results'][0]['video']
                elif 'video' in bootstrap:
                    info = bootstrap['video']
                elif 'msnbcVideoInfo' in bootstrap:
                    info = bootstrap['msnbcVideoInfo']['meta']
                elif 'msnbcThePlatform' in bootstrap:
                    info = bootstrap['msnbcThePlatform']['videoPlayer']['video']
                else:
                    info = bootstrap

                if 'guid' in info:
                    video_id = info['guid']
                    filter_param = 'byGuid'
                elif 'mpxId' in info:
                    video_id = info['mpxId']

            return {
                '_type': 'url_transparent',
                'id': video_id,
                # http://feed.theplatform.com/f/2E2eJC/nbcnews also works
                'url': update_url_query('http://feed.theplatform.com/f/2E2eJC/nnd_NBCNews', {filter_param: video_id}),
                'ie_key': 'ThePlatformFeed',
            }


class NBCOlympicsIE(InfoExtractor):
    _VALID_URL = r'https?://www\.nbcolympics\.com/video/(?P<id>[a-z-]+)'

    _TEST = {
        # Geo-restricted to US
        'url': 'http://www.nbcolympics.com/video/justin-roses-son-leo-was-tears-after-his-dad-won-gold',
        'md5': '54fecf846d05429fbaa18af557ee523a',
        'info_dict': {
            'id': 'WjTBzDXx5AUq',
            'display_id': 'justin-roses-son-leo-was-tears-after-his-dad-won-gold',
            'ext': 'mp4',
            'title': 'Rose\'s son Leo was in tears after his dad won gold',
            'description': 'Olympic gold medalist Justin Rose gets emotional talking to the impact his win in men\'s golf has already had on his children.',
            'timestamp': 1471274964,
            'upload_date': '20160815',
            'uploader': 'NBCU-SPORTS',
        },
    }

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

        webpage = self._download_webpage(url, display_id)

        drupal_settings = self._parse_json(self._search_regex(
            r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);',
            webpage, 'drupal settings'), display_id)

        iframe_url = drupal_settings['vod']['iframe_url']
        theplatform_url = iframe_url.replace(
            'vplayer.nbcolympics.com', 'player.theplatform.com')

        return {
            '_type': 'url_transparent',
            'url': theplatform_url,
            'ie_key': ThePlatformIE.ie_key(),
            'display_id': display_id,
        }
Commit	Line	Data
cd7ee7aa JMF	1	from __future__ import unicode_literals
cd7ee7aa JMF	2
0bc56fa6	3	import re
0bc56fa6 JMF	4
0bc56fa6 JMF	5	from .common import InfoExtractor
574b2a73	6	from .theplatform import ThePlatformIE
fdf9b959	7	from .adobepass import AdobePassIE
1cc79574	8	from ..utils import (
37e64add	9	find_xpath_attr,
b46b65ed	10	smuggle_url,
0fe2ff78	11	unescapeHTML,
6e416b21	12	update_url_query,
fdf9b959	13	int_or_none,
37e64add	14	)
0bc56fa6 JMF	15
0bc56fa6 JMF	16
fdf9b959	17	class NBCIE(AdobePassIE):
52294cdd	18	_VALID_URL = r'(?P<permalink>https?://(?:www\.)?nbc\.com/[^/]+/video/[^/]+/(?P<id>n?\d+))'
58c1f6f0 S	19
	20	_TESTS = [
	21	{
fdf9b959	22	'url': 'http://www.nbc.com/the-tonight-show/video/jimmy-fallon-surprises-fans-at-ben-jerrys/2848237',
58c1f6f0	23	'info_dict': {
fdf9b959	24	'id': '2848237',
e881c4bc	25	'ext': 'mp4',
5c8a3f86 JMF	26	'title': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s',
5c8a3f86 JMF	27	'description': 'Jimmy gives out free scoops of his new "Tonight Dough" ice cream flavor by surprising customers at the Ben & Jerry\'s scoop shop.',
79ba9140	28	'timestamp': 1424246400,
	29	'upload_date': '20150218',
	30	'uploader': 'NBCU-COM',
58c1f6f0	31	},
e881c4bc YCH	32	'params': {
	33	# m3u8 download
	34	'skip_download': True,
	35	},
020cf5eb	36	},
b9b3ab45 YCH	37	{
	38	'url': 'http://www.nbc.com/saturday-night-live/video/star-wars-teaser/2832821',
	39	'info_dict': {
e881c4bc YCH	40	'id': '2832821',
e881c4bc YCH	41	'ext': 'mp4',
b9b3ab45 YCH	42	'title': 'Star Wars Teaser',
b9b3ab45 YCH	43	'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442',
79ba9140	44	'timestamp': 1417852800,
	45	'upload_date': '20141206',
	46	'uploader': 'NBCU-COM',
b9b3ab45	47	},
e881c4bc YCH	48	'params': {
	49	# m3u8 download
	50	'skip_download': True,
	51	},
b9b3ab45	52	'skip': 'Only works from US',
0fe2ff78	53	},
e6e90515 YCH	54	{
	55	# HLS streams requires the 'hdnea3' cookie
	56	'url': 'http://www.nbc.com/Kings/video/goliath/n1806',
	57	'info_dict': {
fdf9b959	58	'id': '101528f5a9e8127b107e98c5e6ce4638',
e6e90515 YCH	59	'ext': 'mp4',
	60	'title': 'Goliath',
	61	'description': 'When an unknown soldier saves the life of the King\'s son in battle, he\'s thrust into the limelight and politics of the kingdom.',
	62	'timestamp': 1237100400,
	63	'upload_date': '20090315',
	64	'uploader': 'NBCU-COM',
	65	},
	66	'params': {
	67	'skip_download': True,
	68	},
	69	'skip': 'Only works from US',
b9b3ab45	70	}
58c1f6f0	71	]
020cf5eb JMF	72
020cf5eb JMF	73	def _real_extract(self, url):
52294cdd	74	permalink, video_id = re.match(self._VALID_URL, url).groups()
2eeb588e RA	75	video_data = self._download_json(
2eeb588e RA	76	'https://api.nbc.com/v3/videos', video_id, query={
52294cdd	77	'filter[permalink]': permalink,
2eeb588e RA	78	})['data'][0]['attributes']
	79	query = {
	80	'mbr': 'true',
	81	'manifest': 'm3u',
	82	}
	83	video_id = video_data['guid']
	84	title = video_data['title']
	85	if video_data.get('entitlement') == 'auth':
	86	resource = self._get_mvpd_resource(
	87	'nbcentertainment', title, video_id,
	88	video_data.get('vChipRating'))
	89	query['auth'] = self._extract_mvpd_auth(
	90	url, video_id, 'nbcentertainment', resource)
	91	theplatform_url = smuggle_url(update_url_query(
	92	'http://link.theplatform.com/s/NnzsPC/media/guid/2410887629/' + video_id,
	93	query), {'force_smil_url': True})
	94	return {
e881c4bc	95	'_type': 'url_transparent',
e881c4bc	96	'id': video_id,
2eeb588e RA	97	'title': title,
	98	'url': theplatform_url,
	99	'description': video_data.get('description'),
	100	'keywords': video_data.get('keywords'),
	101	'season_number': int_or_none(video_data.get('seasonNumber')),
	102	'episode_number': int_or_none(video_data.get('episodeNumber')),
	103	'series': video_data.get('showName'),
	104	'ie_key': 'ThePlatform',
e881c4bc	105	}
020cf5eb JMF	106
020cf5eb JMF	107
a2a4d5fa	108	class NBCSportsVPlayerIE(InfoExtractor):
a2edf2e7	109	_VALID_URL = r'https?://vplayer\.nbcsports\.com/(?:[^/]+/)+(?P<id>[0-9a-zA-Z_]+)'
a28ccbab	110
5cbb2699	111	_TESTS = [{
a28ccbab	112	'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_share/select/9CsDKds0kvHI',
a28ccbab YCH	113	'info_dict': {
	114	'id': '9CsDKds0kvHI',
	115	'ext': 'flv',
	116	'description': 'md5:df390f70a9ba7c95ff1daace988f0d8d',
	117	'title': 'Tyler Kalinoski hits buzzer-beater to lift Davidson',
79ba9140	118	'timestamp': 1426270238,
	119	'upload_date': '20150313',
	120	'uploader': 'NBCU-SPORTS',
a28ccbab	121	}
5cbb2699	122	}, {
5cbb2699 YCH	123	'url': 'http://vplayer.nbcsports.com/p/BxmELC/nbc_embedshare/select/_hqLjQ95yx8Z',
	124	'only_matching': True,
	125	}]
a28ccbab	126
a2a4d5fa YCH	127	@staticmethod
	128	def _extract_url(webpage):
	129	iframe_m = re.search(
	130	r'<iframe[^>]+src="(?P<url>https?://vplayer\.nbcsports\.com/[^"]+)"', webpage)
	131	if iframe_m:
	132	return iframe_m.group('url')
	133
a28ccbab YCH	134	def _real_extract(self, url):
	135	video_id = self._match_id(url)
	136	webpage = self._download_webpage(url, video_id)
	137	theplatform_url = self._og_search_video_url(webpage)
	138	return self.url_result(theplatform_url, 'ThePlatform')
	139
	140
a2a4d5fa	141	class NBCSportsIE(InfoExtractor):
dfb1b146	142	# Does not include https because its certificate is invalid
92519402	143	_VALID_URL = r'https?://(?:www\.)?nbcsports\.com//?(?:[^/]+/)+(?P<id>[0-9a-z-]+)'
a2a4d5fa YCH	144
	145	_TEST = {
	146	'url': 'http://www.nbcsports.com//college-basketball/ncaab/tom-izzo-michigan-st-has-so-much-respect-duke',
a2a4d5fa YCH	147	'info_dict': {
	148	'id': 'PHJSaFWbrTY9',
	149	'ext': 'flv',
	150	'title': 'Tom Izzo, Michigan St. has \'so much respect\' for Duke',
	151	'description': 'md5:ecb459c9d59e0766ac9c7d5d0eda8113',
0738187f YCH	152	'uploader': 'NBCU-SPORTS',
	153	'upload_date': '20150330',
	154	'timestamp': 1427726529,
a2a4d5fa YCH	155	}
	156	}
	157
	158	def _real_extract(self, url):
	159	video_id = self._match_id(url)
	160	webpage = self._download_webpage(url, video_id)
	161	return self.url_result(
	162	NBCSportsVPlayerIE._extract_url(webpage), 'NBCSportsVPlayer')
	163
	164
9cf01f7f	165	class CSNNEIE(InfoExtractor):
92519402	166	_VALID_URL = r'https?://(?:www\.)?csnne\.com/video/(?P<id>[0-9a-z-]+)'
9cf01f7f	167
	168	_TEST = {
	169	'url': 'http://www.csnne.com/video/snc-evening-update-wright-named-red-sox-no-5-starter',
	170	'info_dict': {
	171	'id': 'yvBLLUgQ8WU0',
	172	'ext': 'mp4',
	173	'title': 'SNC evening update: Wright named Red Sox\' No. 5 starter.',
	174	'description': 'md5:1753cfee40d9352b19b4c9b3e589b9e3',
79ba9140	175	'timestamp': 1459369979,
	176	'upload_date': '20160330',
	177	'uploader': 'NBCU-SPORTS',
9cf01f7f	178	}
	179	}
	180
	181	def _real_extract(self, url):
	182	display_id = self._match_id(url)
	183	webpage = self._download_webpage(url, display_id)
	184	return {
	185	'_type': 'url_transparent',
	186	'ie_key': 'ThePlatform',
	187	'url': self._html_search_meta('twitter:player:stream', webpage),
	188	'display_id': display_id,
	189	}
	190
	191
574b2a73	192	class NBCNewsIE(ThePlatformIE):
0437307a	193	_VALID_URL = r'''(?x)https?://(?:www\.)?(?:nbcnews\|today\|msnbc)\.com/
a4f3d779	194	(?:video/.+?/(?P<id>\d+)\|
0437307a	195	([^/]+/)(?:.-)?(?P<mpx_id>[^/?]+))
87fe568c	196	'''
0bc56fa6	197
87fe568c JMF	198	_TESTS = [
	199	{
	200	'url': 'http://www.nbcnews.com/video/nbc-news/52753292',
	201	'md5': '47abaac93c6eaf9ad37ee6c4463a5179',
	202	'info_dict': {
	203	'id': '52753292',
	204	'ext': 'flv',
	205	'title': 'Crew emerges after four-month Mars food study',
	206	'description': 'md5:24e632ffac72b35f8b67a12d1b6ddfc1',
	207	},
0bc56fa6	208	},
87fe568c	209	{
574b2a73	210	'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
574b2a73	211	'md5': 'af1adfa51312291a017720403826bb64',
87fe568c	212	'info_dict': {
6e416b21	213	'id': 'p_tweet_snow_140529',
10e3d734	214	'ext': 'mp4',
87fe568c JMF	215	'title': 'How Twitter Reacted To The Snowden Interview',
87fe568c JMF	216	'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
0437307a RA	217	'uploader': 'NBCU-NEWS',
	218	'timestamp': 1401363060,
	219	'upload_date': '20140529',
87fe568c	220	},
87fe568c	221	},
2df54b4b S	222	{
	223	'url': 'http://www.nbcnews.com/feature/dateline-full-episodes/full-episode-family-business-n285156',
	224	'md5': 'fdbf39ab73a72df5896b6234ff98518a',
	225	'info_dict': {
0437307a	226	'id': '529953347624',
2df54b4b S	227	'ext': 'mp4',
	228	'title': 'FULL EPISODE: Family Business',
	229	'description': 'md5:757988edbaae9d7be1d585eb5d55cc04',
	230	},
574b2a73	231	'skip': 'This page is unavailable.',
2df54b4b	232	},
d9aa2b78 RS	233	{
d9aa2b78 RS	234	'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
574b2a73	235	'md5': '73135a2e0ef819107bbb55a5a9b2a802',
d9aa2b78	236	'info_dict': {
6e416b21	237	'id': 'nn_netcast_150204',
d9aa2b78 RS	238	'ext': 'mp4',
	239	'title': 'Nightly News with Brian Williams Full Broadcast (February 4)',
	240	'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
0437307a RA	241	'timestamp': 1423104900,
	242	'uploader': 'NBCU-NEWS',
	243	'upload_date': '20150205',
d9aa2b78 RS	244	},
d9aa2b78 RS	245	},
574b2a73	246	{
	247	'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
	248	'md5': 'a49e173825e5fcd15c13fc297fced39d',
	249	'info_dict': {
6e416b21	250	'id': 'x_lon_vwhorn_150922',
574b2a73	251	'ext': 'mp4',
0437307a RA	252	'title': 'Volkswagen U.S. Chief:\xa0 We Have Totally Screwed Up',
	253	'description': 'md5:c8be487b2d80ff0594c005add88d8351',
	254	'upload_date': '20150922',
	255	'timestamp': 1442917800,
	256	'uploader': 'NBCU-NEWS',
574b2a73	257	},
574b2a73	258	},
cb7d4d0e	259	{
	260	'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
	261	'md5': '118d7ca3f0bea6534f119c68ef539f71',
	262	'info_dict': {
6e416b21	263	'id': 'tdy_al_space_160420',
cb7d4d0e	264	'ext': 'mp4',
	265	'title': 'See the aurora borealis from space in stunning new NASA video',
	266	'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
	267	'upload_date': '20160420',
	268	'timestamp': 1461152093,
0437307a RA	269	'uploader': 'NBCU-NEWS',
	270	},
	271	},
	272	{
	273	'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924',
	274	'md5': '6d236bf4f3dddc226633ce6e2c3f814d',
	275	'info_dict': {
6e416b21	276	'id': 'n_hayes_Aimm_140801_272214',
0437307a RA	277	'ext': 'mp4',
	278	'title': 'The chaotic GOP immigration vote',
	279	'description': 'The Republican House votes on a border bill that has no chance of getting through the Senate or signed by the President and is drawing criticism from all sides.',
ec85ded8	280	'thumbnail': r're:^https?://.*\.jpg$',
0437307a RA	281	'timestamp': 1406937606,
	282	'upload_date': '20140802',
	283	'uploader': 'NBCU-NEWS',
cb7d4d0e	284	},
cb7d4d0e	285	},
3f125c8c S	286	{
	287	'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952',
	288	'only_matching': True,
	289	},
5de008e8 YCH	290	{
	291	# From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html
	292	'url': 'http://www.nbcnews.com/widget/video-embed/701714499682',
	293	'only_matching': True,
	294	},
87fe568c	295	]
0bc56fa6 JMF	296
	297	def _real_extract(self, url):
	298	mobj = re.match(self._VALID_URL, url)
	299	video_id = mobj.group('id')
87fe568c JMF	300	if video_id is not None:
	301	all_info = self._download_xml('http://www.nbcnews.com/id/%s/displaymode/1219' % video_id, video_id)
	302	info = all_info.find('video')
0bc56fa6	303
87fe568c JMF	304	return {
	305	'id': video_id,
	306	'title': info.find('headline').text,
	307	'ext': 'flv',
	308	'url': find_xpath_attr(info, 'media', 'type', 'flashVideo').text,
d890b4cc	309	'description': info.find('caption').text,
87fe568c JMF	310	'thumbnail': find_xpath_attr(info, 'media', 'type', 'thumbnail').text,
	311	}
	312	else:
d9aa2b78	313	# "feature" and "nightly-news" pages use theplatform.com
0437307a	314	video_id = mobj.group('mpx_id')
6e416b21 RA	315	webpage = self._download_webpage(url, video_id)
	316
	317	filter_param = 'byId'
	318	bootstrap_json = self._search_regex(
	319	[r'(?m)(?:var\s+(?:bootstrapJson\|playlistData)\|NEWS\.videoObj)\s=\s({.+});?\s*$',
	320	r'videoObj\s:\s({.+})', r'data-video="([^"]+)"',
	321	r'jQuery\.extend\(Drupal\.settings\s,\s({.+?})\);'],
	322	webpage, 'bootstrap json', default=None)
	323	if bootstrap_json:
0437307a RA	324	bootstrap = self._parse_json(
0437307a RA	325	bootstrap_json, video_id, transform_source=unescapeHTML)
6e416b21 RA	326
6e416b21 RA	327	info = None
0437307a RA	328	if 'results' in bootstrap:
	329	info = bootstrap['results'][0]['video']
	330	elif 'video' in bootstrap:
	331	info = bootstrap['video']
6e416b21 RA	332	elif 'msnbcVideoInfo' in bootstrap:
	333	info = bootstrap['msnbcVideoInfo']['meta']
	334	elif 'msnbcThePlatform' in bootstrap:
	335	info = bootstrap['msnbcThePlatform']['videoPlayer']['video']
574b2a73	336	else:
0437307a	337	info = bootstrap
6e416b21 RA	338
	339	if 'guid' in info:
	340	video_id = info['guid']
	341	filter_param = 'byGuid'
	342	elif 'mpxId' in info:
	343	video_id = info['mpxId']
87fe568c JMF	344
87fe568c JMF	345	return {
0437307a	346	'_type': 'url_transparent',
574b2a73	347	'id': video_id,
0437307a	348	# http://feed.theplatform.com/f/2E2eJC/nbcnews also works
6e416b21	349	'url': update_url_query('http://feed.theplatform.com/f/2E2eJC/nnd_NBCNews', {filter_param: video_id}),
0437307a	350	'ie_key': 'ThePlatformFeed',
87fe568c	351	}
be457302 YCH	352
	353
	354	class NBCOlympicsIE(InfoExtractor):
	355	_VALID_URL = r'https?://www\.nbcolympics\.com/video/(?P<id>[a-z-]+)'
	356
	357	_TEST = {
	358	# Geo-restricted to US
	359	'url': 'http://www.nbcolympics.com/video/justin-roses-son-leo-was-tears-after-his-dad-won-gold',
	360	'md5': '54fecf846d05429fbaa18af557ee523a',
	361	'info_dict': {
	362	'id': 'WjTBzDXx5AUq',
	363	'display_id': 'justin-roses-son-leo-was-tears-after-his-dad-won-gold',
	364	'ext': 'mp4',
	365	'title': 'Rose\'s son Leo was in tears after his dad won gold',
	366	'description': 'Olympic gold medalist Justin Rose gets emotional talking to the impact his win in men\'s golf has already had on his children.',
	367	'timestamp': 1471274964,
	368	'upload_date': '20160815',
	369	'uploader': 'NBCU-SPORTS',
	370	},
	371	}
	372
	373	def _real_extract(self, url):
	374	display_id = self._match_id(url)
	375
	376	webpage = self._download_webpage(url, display_id)
	377
	378	drupal_settings = self._parse_json(self._search_regex(
	379	r'jQuery\.extend\(Drupal\.settings\s,\s({.+?})\);',
	380	webpage, 'drupal settings'), display_id)
	381
	382	iframe_url = drupal_settings['vod']['iframe_url']
	383	theplatform_url = iframe_url.replace(
	384	'vplayer.nbcolympics.com', 'player.theplatform.com')
	385
	386	return {
	387	'_type': 'url_transparent',
	388	'url': theplatform_url,
	389	'ie_key': ThePlatformIE.ie_key(),
	390	'display_id': display_id,
	391	}