1 from __future__ import unicode_literals
7 from .common import InfoExtractor
8 from .theplatform import ThePlatformIE
9 from .adobepass import AdobePassIE
10 from ..compat import compat_urllib_parse_unquote
22 class NBCIE(AdobePassIE):
23 _VALID_URL = r'https?(?P<permalink>://(?:www\.)?nbc\.com/(?:classic-tv/)?[^/]+/video/[^/]+/(?P<id>n?\d+))'
27 'url': 'http://www.nbc.com/the-tonight-show/video/jimmy-fallon-surprises-fans-at-ben-jerrys/2848237',
31 'title': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s',
32 'description': 'Jimmy gives out free scoops of his new "Tonight Dough" ice cream flavor by surprising customers at the Ben & Jerry\'s scoop shop.',
33 'timestamp': 1424246400,
34 'upload_date': '20150218',
35 'uploader': 'NBCU-COM',
39 'skip_download': True,
43 'url': 'http://www.nbc.com/saturday-night-live/video/star-wars-teaser/2832821',
47 'title': 'Star Wars Teaser',
48 'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442',
49 'timestamp': 1417852800,
50 'upload_date': '20141206',
51 'uploader': 'NBCU-COM',
55 'skip_download': True,
57 'skip': 'Only works from US',
60 # HLS streams requires the 'hdnea3' cookie
61 'url': 'http://www.nbc.com/Kings/video/goliath/n1806',
63 'id': '101528f5a9e8127b107e98c5e6ce4638',
66 '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.',
67 'timestamp': 1237100400,
68 'upload_date': '20090315',
69 'uploader': 'NBCU-COM',
72 'skip_download': True,
74 'skip': 'Only works from US',
77 'url': 'https://www.nbc.com/classic-tv/charles-in-charge/video/charles-in-charge-pilot/n3310',
78 'only_matching': True,
82 'url': 'https://www.nbc.com/up-all-night/video/day-after-valentine%27s-day/n2189',
83 'only_matching': True,
87 def _real_extract(self, url):
88 permalink, video_id = re.match(self._VALID_URL, url).groups()
89 permalink = 'http' + compat_urllib_parse_unquote(permalink)
90 response = self._download_json(
91 'https://friendship.nbc.co/v2/graphql', video_id, query={
93 page(name: "%s", platform: web, type: VIDEO, userId: "0") {
95 ... on VideoPageData {
111 video_data = response['data']['page']['data']
116 video_id = video_data['mpxGuid']
117 title = video_data['secondaryTitle']
118 if video_data.get('locked'):
119 resource = self._get_mvpd_resource(
120 'nbcentertainment', title, video_id,
121 video_data.get('rating'))
122 query['auth'] = self._extract_mvpd_auth(
123 url, video_id, 'nbcentertainment', resource)
124 theplatform_url = smuggle_url(update_url_query(
125 'http://link.theplatform.com/s/NnzsPC/media/guid/%s/%s' % (video_data.get('mpxAccountId') or '2410887629', video_id),
126 query), {'force_smil_url': True})
128 '_type': 'url_transparent',
131 'url': theplatform_url,
132 'description': video_data.get('description'),
133 'tags': video_data.get('keywords'),
134 'season_number': int_or_none(video_data.get('seasonNumber')),
135 'episode_number': int_or_none(video_data.get('episodeNumber')),
137 'series': video_data.get('seriesShortTitle'),
138 'ie_key': 'ThePlatform',
142 class NBCSportsVPlayerIE(InfoExtractor):
143 _VALID_URL = r'https?://vplayer\.nbcsports\.com/(?:[^/]+/)+(?P<id>[0-9a-zA-Z_]+)'
146 'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_embed/select/9CsDKds0kvHI',
148 'id': '9CsDKds0kvHI',
150 'description': 'md5:df390f70a9ba7c95ff1daace988f0d8d',
151 'title': 'Tyler Kalinoski hits buzzer-beater to lift Davidson',
152 'timestamp': 1426270238,
153 'upload_date': '20150313',
154 'uploader': 'NBCU-SPORTS',
157 'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_embed/select/media/_hqLjQ95yx8Z',
158 'only_matching': True,
162 def _extract_url(webpage):
163 iframe_m = re.search(
164 r'<iframe[^>]+src="(?P<url>https?://vplayer\.nbcsports\.com/[^"]+)"', webpage)
166 return iframe_m.group('url')
168 def _real_extract(self, url):
169 video_id = self._match_id(url)
170 webpage = self._download_webpage(url, video_id)
171 theplatform_url = self._og_search_video_url(webpage).replace(
172 'vplayer.nbcsports.com', 'player.theplatform.com')
173 return self.url_result(theplatform_url, 'ThePlatform')
176 class NBCSportsIE(InfoExtractor):
177 # Does not include https because its certificate is invalid
178 _VALID_URL = r'https?://(?:www\.)?nbcsports\.com//?(?:[^/]+/)+(?P<id>[0-9a-z-]+)'
181 'url': 'http://www.nbcsports.com//college-basketball/ncaab/tom-izzo-michigan-st-has-so-much-respect-duke',
183 'id': 'PHJSaFWbrTY9',
185 'title': 'Tom Izzo, Michigan St. has \'so much respect\' for Duke',
186 'description': 'md5:ecb459c9d59e0766ac9c7d5d0eda8113',
187 'uploader': 'NBCU-SPORTS',
188 'upload_date': '20150330',
189 'timestamp': 1427726529,
193 def _real_extract(self, url):
194 video_id = self._match_id(url)
195 webpage = self._download_webpage(url, video_id)
196 return self.url_result(
197 NBCSportsVPlayerIE._extract_url(webpage), 'NBCSportsVPlayer')
200 class NBCSportsStreamIE(AdobePassIE):
201 _VALID_URL = r'https?://stream\.nbcsports\.com/.+?\bpid=(?P<id>\d+)'
203 'url': 'http://stream.nbcsports.com/nbcsn/generic?pid=206559',
207 'title': 'Amgen Tour of California Women\'s Recap',
208 'description': 'md5:66520066b3b5281ada7698d0ea2aa894',
212 'skip_download': True,
214 'skip': 'Requires Adobe Pass Authentication',
217 def _real_extract(self, url):
218 video_id = self._match_id(url)
219 live_source = self._download_json(
220 'http://stream.nbcsports.com/data/live_sources_%s.json' % video_id,
222 video_source = live_source['videoSources'][0]
223 title = video_source['title']
225 for k in ('source', 'msl4source', 'iossource', 'hlsv4'):
227 source_url = video_source.get(sk) or video_source.get(sk + 'Alt')
231 source_url = video_source['ottStreamUrl']
232 is_live = video_source.get('type') == 'live' or video_source.get('status') == 'Live'
233 resource = self._get_mvpd_resource('nbcsports', title, video_id, '')
234 token = self._extract_mvpd_auth(url, video_id, 'nbcsports', resource)
235 tokenized_url = self._download_json(
236 'https://token.playmakerservices.com/cdn',
237 video_id, data=json.dumps({
238 'requestorId': 'nbcsports',
240 'application': 'NBCSports',
242 'platform': 'desktop',
244 'url': video_source['sourceUrl'],
245 'token': base64.b64encode(token.encode()).decode(),
246 'resourceId': base64.b64encode(resource.encode()).decode(),
247 }).encode())['tokenizedUrl']
248 formats = self._extract_m3u8_formats(tokenized_url, video_id, 'mp4')
249 self._sort_formats(formats)
252 'title': self._live_title(title) if is_live else title,
253 'description': live_source.get('description'),
259 class CSNNEIE(InfoExtractor):
260 _VALID_URL = r'https?://(?:www\.)?csnne\.com/video/(?P<id>[0-9a-z-]+)'
263 'url': 'http://www.csnne.com/video/snc-evening-update-wright-named-red-sox-no-5-starter',
265 'id': 'yvBLLUgQ8WU0',
267 'title': 'SNC evening update: Wright named Red Sox\' No. 5 starter.',
268 'description': 'md5:1753cfee40d9352b19b4c9b3e589b9e3',
269 'timestamp': 1459369979,
270 'upload_date': '20160330',
271 'uploader': 'NBCU-SPORTS',
275 def _real_extract(self, url):
276 display_id = self._match_id(url)
277 webpage = self._download_webpage(url, display_id)
279 '_type': 'url_transparent',
280 'ie_key': 'ThePlatform',
281 'url': self._html_search_meta('twitter:player:stream', webpage),
282 'display_id': display_id,
286 class NBCNewsIE(ThePlatformIE):
287 _VALID_URL = r'(?x)https?://(?:www\.)?(?:nbcnews|today|msnbc)\.com/([^/]+/)*(?:.*-)?(?P<id>[^/?]+)'
291 'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
292 'md5': 'cf4bc9e6ce0130f00f545d80ecedd4bf',
294 'id': '269389891880',
296 'title': 'How Twitter Reacted To The Snowden Interview',
297 'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
298 'timestamp': 1401363060,
299 'upload_date': '20140529',
303 'url': 'http://www.nbcnews.com/feature/dateline-full-episodes/full-episode-family-business-n285156',
304 'md5': 'fdbf39ab73a72df5896b6234ff98518a',
306 'id': '529953347624',
308 'title': 'FULL EPISODE: Family Business',
309 'description': 'md5:757988edbaae9d7be1d585eb5d55cc04',
311 'skip': 'This page is unavailable.',
314 'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
315 'md5': '8eb831eca25bfa7d25ddd83e85946548',
317 'id': '394064451844',
319 'title': 'Nightly News with Brian Williams Full Broadcast (February 4)',
320 'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
321 'timestamp': 1423104900,
322 'upload_date': '20150205',
326 'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
327 'md5': '4a8c4cec9e1ded51060bdda36ff0a5c0',
331 'title': "Volkswagen U.S. Chief: We 'Totally Screwed Up'",
332 'description': 'md5:d22d1281a24f22ea0880741bb4dd6301',
333 'upload_date': '20150922',
334 'timestamp': 1442917800,
338 'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
339 'md5': '118d7ca3f0bea6534f119c68ef539f71',
341 'id': '669831235788',
343 'title': 'See the aurora borealis from space in stunning new NASA video',
344 'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
345 'upload_date': '20160420',
346 'timestamp': 1461152093,
350 'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924',
351 'md5': '6d236bf4f3dddc226633ce6e2c3f814d',
353 'id': '314487875924',
355 'title': 'The chaotic GOP immigration vote',
356 '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.',
357 'thumbnail': r're:^https?://.*\.jpg$',
358 'timestamp': 1406937606,
359 'upload_date': '20140802',
363 'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952',
364 'only_matching': True,
367 # From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html
368 'url': 'http://www.nbcnews.com/widget/video-embed/701714499682',
369 'only_matching': True,
373 def _real_extract(self, url):
374 video_id = self._match_id(url)
375 webpage = self._download_webpage(url, video_id)
377 data = self._parse_json(self._search_regex(
378 r'window\.__data\s*=\s*({.+});', webpage,
379 'bootstrap json'), video_id, js_to_json)
380 video_data = try_get(data, lambda x: x['video']['current'], dict)
382 video_data = data['article']['content'][0]['primaryMedia']['video']
383 title = video_data['headline']['primary']
386 for va in video_data.get('videoAssets', []):
387 public_url = va.get('publicUrl')
390 if '://link.theplatform.com/' in public_url:
391 public_url = update_url_query(public_url, {'format': 'redirect'})
392 format_id = va.get('format')
393 if format_id == 'M3U':
394 formats.extend(self._extract_m3u8_formats(
395 public_url, video_id, 'mp4', 'm3u8_native',
396 m3u8_id=format_id, fatal=False))
398 tbr = int_or_none(va.get('bitrate'), 1000)
400 format_id += '-%d' % tbr
402 'format_id': format_id,
404 'width': int_or_none(va.get('width')),
405 'height': int_or_none(va.get('height')),
409 self._sort_formats(formats)
412 closed_captioning = video_data.get('closedCaptioning')
413 if closed_captioning:
414 for cc_url in closed_captioning.values():
417 subtitles.setdefault('en', []).append({
424 'description': try_get(video_data, lambda x: x['description']['primary']),
425 'thumbnail': try_get(video_data, lambda x: x['primaryImage']['url']['primary']),
426 'duration': parse_duration(video_data.get('duration')),
427 'timestamp': unified_timestamp(video_data.get('datePublished')),
429 'subtitles': subtitles,
433 class NBCOlympicsIE(InfoExtractor):
434 IE_NAME = 'nbcolympics'
435 _VALID_URL = r'https?://www\.nbcolympics\.com/video/(?P<id>[a-z-]+)'
438 # Geo-restricted to US
439 'url': 'http://www.nbcolympics.com/video/justin-roses-son-leo-was-tears-after-his-dad-won-gold',
440 'md5': '54fecf846d05429fbaa18af557ee523a',
442 'id': 'WjTBzDXx5AUq',
443 'display_id': 'justin-roses-son-leo-was-tears-after-his-dad-won-gold',
445 'title': 'Rose\'s son Leo was in tears after his dad won gold',
446 'description': 'Olympic gold medalist Justin Rose gets emotional talking to the impact his win in men\'s golf has already had on his children.',
447 'timestamp': 1471274964,
448 'upload_date': '20160815',
449 'uploader': 'NBCU-SPORTS',
453 def _real_extract(self, url):
454 display_id = self._match_id(url)
456 webpage = self._download_webpage(url, display_id)
458 drupal_settings = self._parse_json(self._search_regex(
459 r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);',
460 webpage, 'drupal settings'), display_id)
462 iframe_url = drupal_settings['vod']['iframe_url']
463 theplatform_url = iframe_url.replace(
464 'vplayer.nbcolympics.com', 'player.theplatform.com')
467 '_type': 'url_transparent',
468 'url': theplatform_url,
469 'ie_key': ThePlatformIE.ie_key(),
470 'display_id': display_id,
474 class NBCOlympicsStreamIE(AdobePassIE):
475 IE_NAME = 'nbcolympics:stream'
476 _VALID_URL = r'https?://stream\.nbcolympics\.com/(?P<id>[0-9a-z-]+)'
478 'url': 'http://stream.nbcolympics.com/2018-winter-olympics-nbcsn-evening-feb-8',
482 'title': 're:Curling, Alpine, Luge [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
486 'skip_download': True,
489 _DATA_URL_TEMPLATE = 'http://stream.nbcolympics.com/data/%s_%s.json'
491 def _real_extract(self, url):
492 display_id = self._match_id(url)
493 webpage = self._download_webpage(url, display_id)
494 pid = self._search_regex(r'pid\s*=\s*(\d+);', webpage, 'pid')
495 resource = self._search_regex(
496 r"resource\s*=\s*'(.+)';", webpage,
497 'resource').replace("' + pid + '", pid)
498 event_config = self._download_json(
499 self._DATA_URL_TEMPLATE % ('event_config', pid),
501 title = self._live_title(event_config['eventTitle'])
502 source_url = self._download_json(
503 self._DATA_URL_TEMPLATE % ('live_sources', pid),
504 pid)['videoSources'][0]['sourceUrl']
505 media_token = self._extract_mvpd_auth(
506 url, pid, event_config.get('requestorId', 'NBCOlympics'), resource)
507 formats = self._extract_m3u8_formats(self._download_webpage(
508 'http://sp.auth.adobe.com/tvs/v1/sign', pid, query={
510 'mediaToken': base64.b64encode(media_token.encode()),
511 'resource': base64.b64encode(resource.encode()),
514 self._sort_formats(formats)
518 'display_id': display_id,