]> git.bitcoin.ninja Git - youtube-dl/blob - youtube_dl/extractor/nbc.py
[JWPlatform] Use non-capturing group in RE
[youtube-dl] / youtube_dl / extractor / nbc.py
1 from __future__ import unicode_literals
2
3 import re
4
5 from .common import InfoExtractor
6 from .theplatform import ThePlatformIE
7 from .adobepass import AdobePassIE
8 from ..utils import (
9     find_xpath_attr,
10     smuggle_url,
11     unescapeHTML,
12     update_url_query,
13     int_or_none,
14 )
15
16
17 class NBCIE(AdobePassIE):
18     _VALID_URL = r'https?(?P<permalink>://(?:www\.)?nbc\.com/(?:classic-tv/)?[^/]+/video/[^/]+/(?P<id>n?\d+))'
19
20     _TESTS = [
21         {
22             'url': 'http://www.nbc.com/the-tonight-show/video/jimmy-fallon-surprises-fans-at-ben-jerrys/2848237',
23             'info_dict': {
24                 'id': '2848237',
25                 'ext': 'mp4',
26                 'title': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s',
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.',
28                 'timestamp': 1424246400,
29                 'upload_date': '20150218',
30                 'uploader': 'NBCU-COM',
31             },
32             'params': {
33                 # m3u8 download
34                 'skip_download': True,
35             },
36         },
37         {
38             'url': 'http://www.nbc.com/saturday-night-live/video/star-wars-teaser/2832821',
39             'info_dict': {
40                 'id': '2832821',
41                 'ext': 'mp4',
42                 'title': 'Star Wars Teaser',
43                 'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442',
44                 'timestamp': 1417852800,
45                 'upload_date': '20141206',
46                 'uploader': 'NBCU-COM',
47             },
48             'params': {
49                 # m3u8 download
50                 'skip_download': True,
51             },
52             'skip': 'Only works from US',
53         },
54         {
55             # HLS streams requires the 'hdnea3' cookie
56             'url': 'http://www.nbc.com/Kings/video/goliath/n1806',
57             'info_dict': {
58                 'id': '101528f5a9e8127b107e98c5e6ce4638',
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',
70         },
71         {
72             'url': 'https://www.nbc.com/classic-tv/charles-in-charge/video/charles-in-charge-pilot/n3310',
73             'only_matching': True,
74         },
75     ]
76
77     def _real_extract(self, url):
78         permalink, video_id = re.match(self._VALID_URL, url).groups()
79         permalink = 'http' + permalink
80         video_data = self._download_json(
81             'https://api.nbc.com/v3/videos', video_id, query={
82                 'filter[permalink]': permalink,
83             })['data'][0]['attributes']
84         query = {
85             'mbr': 'true',
86             'manifest': 'm3u',
87         }
88         video_id = video_data['guid']
89         title = video_data['title']
90         if video_data.get('entitlement') == 'auth':
91             resource = self._get_mvpd_resource(
92                 'nbcentertainment', title, video_id,
93                 video_data.get('vChipRating'))
94             query['auth'] = self._extract_mvpd_auth(
95                 url, video_id, 'nbcentertainment', resource)
96         theplatform_url = smuggle_url(update_url_query(
97             'http://link.theplatform.com/s/NnzsPC/media/guid/2410887629/' + video_id,
98             query), {'force_smil_url': True})
99         return {
100             '_type': 'url_transparent',
101             'id': video_id,
102             'title': title,
103             'url': theplatform_url,
104             'description': video_data.get('description'),
105             'keywords': video_data.get('keywords'),
106             'season_number': int_or_none(video_data.get('seasonNumber')),
107             'episode_number': int_or_none(video_data.get('episodeNumber')),
108             'series': video_data.get('showName'),
109             'ie_key': 'ThePlatform',
110         }
111
112
113 class NBCSportsVPlayerIE(InfoExtractor):
114     _VALID_URL = r'https?://vplayer\.nbcsports\.com/(?:[^/]+/)+(?P<id>[0-9a-zA-Z_]+)'
115
116     _TESTS = [{
117         'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_embed/select/9CsDKds0kvHI',
118         'info_dict': {
119             'id': '9CsDKds0kvHI',
120             'ext': 'mp4',
121             'description': 'md5:df390f70a9ba7c95ff1daace988f0d8d',
122             'title': 'Tyler Kalinoski hits buzzer-beater to lift Davidson',
123             'timestamp': 1426270238,
124             'upload_date': '20150313',
125             'uploader': 'NBCU-SPORTS',
126         }
127     }, {
128         'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_embed/select/media/_hqLjQ95yx8Z',
129         'only_matching': True,
130     }]
131
132     @staticmethod
133     def _extract_url(webpage):
134         iframe_m = re.search(
135             r'<iframe[^>]+src="(?P<url>https?://vplayer\.nbcsports\.com/[^"]+)"', webpage)
136         if iframe_m:
137             return iframe_m.group('url')
138
139     def _real_extract(self, url):
140         video_id = self._match_id(url)
141         webpage = self._download_webpage(url, video_id)
142         theplatform_url = self._og_search_video_url(webpage).replace(
143             'vplayer.nbcsports.com', 'player.theplatform.com')
144         return self.url_result(theplatform_url, 'ThePlatform')
145
146
147 class NBCSportsIE(InfoExtractor):
148     # Does not include https because its certificate is invalid
149     _VALID_URL = r'https?://(?:www\.)?nbcsports\.com//?(?:[^/]+/)+(?P<id>[0-9a-z-]+)'
150
151     _TEST = {
152         'url': 'http://www.nbcsports.com//college-basketball/ncaab/tom-izzo-michigan-st-has-so-much-respect-duke',
153         'info_dict': {
154             'id': 'PHJSaFWbrTY9',
155             'ext': 'flv',
156             'title': 'Tom Izzo, Michigan St. has \'so much respect\' for Duke',
157             'description': 'md5:ecb459c9d59e0766ac9c7d5d0eda8113',
158             'uploader': 'NBCU-SPORTS',
159             'upload_date': '20150330',
160             'timestamp': 1427726529,
161         }
162     }
163
164     def _real_extract(self, url):
165         video_id = self._match_id(url)
166         webpage = self._download_webpage(url, video_id)
167         return self.url_result(
168             NBCSportsVPlayerIE._extract_url(webpage), 'NBCSportsVPlayer')
169
170
171 class CSNNEIE(InfoExtractor):
172     _VALID_URL = r'https?://(?:www\.)?csnne\.com/video/(?P<id>[0-9a-z-]+)'
173
174     _TEST = {
175         'url': 'http://www.csnne.com/video/snc-evening-update-wright-named-red-sox-no-5-starter',
176         'info_dict': {
177             'id': 'yvBLLUgQ8WU0',
178             'ext': 'mp4',
179             'title': 'SNC evening update: Wright named Red Sox\' No. 5 starter.',
180             'description': 'md5:1753cfee40d9352b19b4c9b3e589b9e3',
181             'timestamp': 1459369979,
182             'upload_date': '20160330',
183             'uploader': 'NBCU-SPORTS',
184         }
185     }
186
187     def _real_extract(self, url):
188         display_id = self._match_id(url)
189         webpage = self._download_webpage(url, display_id)
190         return {
191             '_type': 'url_transparent',
192             'ie_key': 'ThePlatform',
193             'url': self._html_search_meta('twitter:player:stream', webpage),
194             'display_id': display_id,
195         }
196
197
198 class NBCNewsIE(ThePlatformIE):
199     _VALID_URL = r'''(?x)https?://(?:www\.)?(?:nbcnews|today|msnbc)\.com/
200         (?:video/.+?/(?P<id>\d+)|
201         ([^/]+/)*(?:.*-)?(?P<mpx_id>[^/?]+))
202         '''
203
204     _TESTS = [
205         {
206             'url': 'http://www.nbcnews.com/video/nbc-news/52753292',
207             'md5': '47abaac93c6eaf9ad37ee6c4463a5179',
208             'info_dict': {
209                 'id': '52753292',
210                 'ext': 'flv',
211                 'title': 'Crew emerges after four-month Mars food study',
212                 'description': 'md5:24e632ffac72b35f8b67a12d1b6ddfc1',
213             },
214         },
215         {
216             'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
217             'md5': 'af1adfa51312291a017720403826bb64',
218             'info_dict': {
219                 'id': 'p_tweet_snow_140529',
220                 'ext': 'mp4',
221                 'title': 'How Twitter Reacted To The Snowden Interview',
222                 'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
223                 'uploader': 'NBCU-NEWS',
224                 'timestamp': 1401363060,
225                 'upload_date': '20140529',
226             },
227         },
228         {
229             'url': 'http://www.nbcnews.com/feature/dateline-full-episodes/full-episode-family-business-n285156',
230             'md5': 'fdbf39ab73a72df5896b6234ff98518a',
231             'info_dict': {
232                 'id': '529953347624',
233                 'ext': 'mp4',
234                 'title': 'FULL EPISODE: Family Business',
235                 'description': 'md5:757988edbaae9d7be1d585eb5d55cc04',
236             },
237             'skip': 'This page is unavailable.',
238         },
239         {
240             'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
241             'md5': '73135a2e0ef819107bbb55a5a9b2a802',
242             'info_dict': {
243                 'id': 'nn_netcast_150204',
244                 'ext': 'mp4',
245                 'title': 'Nightly News with Brian Williams Full Broadcast (February 4)',
246                 'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
247                 'timestamp': 1423104900,
248                 'uploader': 'NBCU-NEWS',
249                 'upload_date': '20150205',
250             },
251         },
252         {
253             'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
254             'md5': 'a49e173825e5fcd15c13fc297fced39d',
255             'info_dict': {
256                 'id': 'x_lon_vwhorn_150922',
257                 'ext': 'mp4',
258                 'title': 'Volkswagen U.S. Chief:\xa0 We Have Totally Screwed Up',
259                 'description': 'md5:c8be487b2d80ff0594c005add88d8351',
260                 'upload_date': '20150922',
261                 'timestamp': 1442917800,
262                 'uploader': 'NBCU-NEWS',
263             },
264         },
265         {
266             'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
267             'md5': '118d7ca3f0bea6534f119c68ef539f71',
268             'info_dict': {
269                 'id': 'tdy_al_space_160420',
270                 'ext': 'mp4',
271                 'title': 'See the aurora borealis from space in stunning new NASA video',
272                 'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
273                 'upload_date': '20160420',
274                 'timestamp': 1461152093,
275                 'uploader': 'NBCU-NEWS',
276             },
277         },
278         {
279             'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924',
280             'md5': '6d236bf4f3dddc226633ce6e2c3f814d',
281             'info_dict': {
282                 'id': 'n_hayes_Aimm_140801_272214',
283                 'ext': 'mp4',
284                 'title': 'The chaotic GOP immigration vote',
285                 '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.',
286                 'thumbnail': r're:^https?://.*\.jpg$',
287                 'timestamp': 1406937606,
288                 'upload_date': '20140802',
289                 'uploader': 'NBCU-NEWS',
290             },
291         },
292         {
293             'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952',
294             'only_matching': True,
295         },
296         {
297             # From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html
298             'url': 'http://www.nbcnews.com/widget/video-embed/701714499682',
299             'only_matching': True,
300         },
301     ]
302
303     def _real_extract(self, url):
304         mobj = re.match(self._VALID_URL, url)
305         video_id = mobj.group('id')
306         if video_id is not None:
307             all_info = self._download_xml('http://www.nbcnews.com/id/%s/displaymode/1219' % video_id, video_id)
308             info = all_info.find('video')
309
310             return {
311                 'id': video_id,
312                 'title': info.find('headline').text,
313                 'ext': 'flv',
314                 'url': find_xpath_attr(info, 'media', 'type', 'flashVideo').text,
315                 'description': info.find('caption').text,
316                 'thumbnail': find_xpath_attr(info, 'media', 'type', 'thumbnail').text,
317             }
318         else:
319             # "feature" and "nightly-news" pages use theplatform.com
320             video_id = mobj.group('mpx_id')
321             webpage = self._download_webpage(url, video_id)
322
323             filter_param = 'byId'
324             bootstrap_json = self._search_regex(
325                 [r'(?m)(?:var\s+(?:bootstrapJson|playlistData)|NEWS\.videoObj)\s*=\s*({.+});?\s*$',
326                  r'videoObj\s*:\s*({.+})', r'data-video="([^"]+)"',
327                  r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);'],
328                 webpage, 'bootstrap json', default=None)
329             if bootstrap_json:
330                 bootstrap = self._parse_json(
331                     bootstrap_json, video_id, transform_source=unescapeHTML)
332
333                 info = None
334                 if 'results' in bootstrap:
335                     info = bootstrap['results'][0]['video']
336                 elif 'video' in bootstrap:
337                     info = bootstrap['video']
338                 elif 'msnbcVideoInfo' in bootstrap:
339                     info = bootstrap['msnbcVideoInfo']['meta']
340                 elif 'msnbcThePlatform' in bootstrap:
341                     info = bootstrap['msnbcThePlatform']['videoPlayer']['video']
342                 else:
343                     info = bootstrap
344
345                 if 'guid' in info:
346                     video_id = info['guid']
347                     filter_param = 'byGuid'
348                 elif 'mpxId' in info:
349                     video_id = info['mpxId']
350
351             return {
352                 '_type': 'url_transparent',
353                 'id': video_id,
354                 # http://feed.theplatform.com/f/2E2eJC/nbcnews also works
355                 'url': update_url_query('http://feed.theplatform.com/f/2E2eJC/nnd_NBCNews', {filter_param: video_id}),
356                 'ie_key': 'ThePlatformFeed',
357             }
358
359
360 class NBCOlympicsIE(InfoExtractor):
361     _VALID_URL = r'https?://www\.nbcolympics\.com/video/(?P<id>[a-z-]+)'
362
363     _TEST = {
364         # Geo-restricted to US
365         'url': 'http://www.nbcolympics.com/video/justin-roses-son-leo-was-tears-after-his-dad-won-gold',
366         'md5': '54fecf846d05429fbaa18af557ee523a',
367         'info_dict': {
368             'id': 'WjTBzDXx5AUq',
369             'display_id': 'justin-roses-son-leo-was-tears-after-his-dad-won-gold',
370             'ext': 'mp4',
371             'title': 'Rose\'s son Leo was in tears after his dad won gold',
372             'description': 'Olympic gold medalist Justin Rose gets emotional talking to the impact his win in men\'s golf has already had on his children.',
373             'timestamp': 1471274964,
374             'upload_date': '20160815',
375             'uploader': 'NBCU-SPORTS',
376         },
377     }
378
379     def _real_extract(self, url):
380         display_id = self._match_id(url)
381
382         webpage = self._download_webpage(url, display_id)
383
384         drupal_settings = self._parse_json(self._search_regex(
385             r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);',
386             webpage, 'drupal settings'), display_id)
387
388         iframe_url = drupal_settings['vod']['iframe_url']
389         theplatform_url = iframe_url.replace(
390             'vplayer.nbcolympics.com', 'player.theplatform.com')
391
392         return {
393             '_type': 'url_transparent',
394             'url': theplatform_url,
395             'ie_key': ThePlatformIE.ie_key(),
396             'display_id': display_id,
397         }