]>
Commit | Line | Data |
---|---|---|
58284890 | 1 | import base64 |
1139935d RA |
2 | import json |
3 | import re | |
0bc56fa6 JMF |
4 | |
5 | from .common import InfoExtractor | |
574b2a73 | 6 | from .theplatform import ThePlatformIE |
fdf9b959 | 7 | from .adobepass import AdobePassIE |
d0c5fabc | 8 | from ..compat import compat_urllib_parse_unquote |
1cc79574 | 9 | from ..utils import ( |
11398b92 | 10 | ExtractorError, |
895e5c03 | 11 | int_or_none, |
9160a0c6 | 12 | parse_age_limit, |
895e5c03 | 13 | parse_duration, |
3e376d18 | 14 | RegexNotFoundError, |
b46b65ed | 15 | smuggle_url, |
11398b92 | 16 | str_or_none, |
17 | traverse_obj, | |
895e5c03 | 18 | try_get, |
11398b92 | 19 | unified_strdate, |
895e5c03 | 20 | unified_timestamp, |
6e416b21 | 21 | update_url_query, |
11398b92 | 22 | url_basename, |
23 | variadic, | |
37e64add | 24 | ) |
0bc56fa6 JMF |
25 | |
26 | ||
6368e2e6 | 27 | class NBCIE(ThePlatformIE): # XXX: Do not subclass from concrete IE |
d673ab65 | 28 | _VALID_URL = r'https?(?P<permalink>://(?:www\.)?nbc\.com/(?:classic-tv/)?[^/]+/video/[^/]+/(?P<id>n?\d+))' |
58c1f6f0 S |
29 | |
30 | _TESTS = [ | |
31 | { | |
fdf9b959 | 32 | 'url': 'http://www.nbc.com/the-tonight-show/video/jimmy-fallon-surprises-fans-at-ben-jerrys/2848237', |
58c1f6f0 | 33 | 'info_dict': { |
fdf9b959 | 34 | 'id': '2848237', |
e881c4bc | 35 | 'ext': 'mp4', |
5c8a3f86 JMF |
36 | 'title': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s', |
37 | '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 | 38 | 'timestamp': 1424246400, |
39 | 'upload_date': '20150218', | |
40 | 'uploader': 'NBCU-COM', | |
58c1f6f0 | 41 | }, |
e881c4bc YCH |
42 | 'params': { |
43 | # m3u8 download | |
44 | 'skip_download': True, | |
45 | }, | |
020cf5eb | 46 | }, |
b9b3ab45 YCH |
47 | { |
48 | 'url': 'http://www.nbc.com/saturday-night-live/video/star-wars-teaser/2832821', | |
49 | 'info_dict': { | |
e881c4bc YCH |
50 | 'id': '2832821', |
51 | 'ext': 'mp4', | |
b9b3ab45 YCH |
52 | 'title': 'Star Wars Teaser', |
53 | 'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442', | |
79ba9140 | 54 | 'timestamp': 1417852800, |
55 | 'upload_date': '20141206', | |
56 | 'uploader': 'NBCU-COM', | |
b9b3ab45 | 57 | }, |
e881c4bc YCH |
58 | 'params': { |
59 | # m3u8 download | |
60 | 'skip_download': True, | |
61 | }, | |
b9b3ab45 | 62 | 'skip': 'Only works from US', |
0fe2ff78 | 63 | }, |
e6e90515 YCH |
64 | { |
65 | # HLS streams requires the 'hdnea3' cookie | |
66 | 'url': 'http://www.nbc.com/Kings/video/goliath/n1806', | |
67 | 'info_dict': { | |
fdf9b959 | 68 | 'id': '101528f5a9e8127b107e98c5e6ce4638', |
e6e90515 YCH |
69 | 'ext': 'mp4', |
70 | 'title': 'Goliath', | |
71 | '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.', | |
72 | 'timestamp': 1237100400, | |
73 | 'upload_date': '20090315', | |
74 | 'uploader': 'NBCU-COM', | |
75 | }, | |
76 | 'params': { | |
77 | 'skip_download': True, | |
78 | }, | |
79 | 'skip': 'Only works from US', | |
d673ab65 L |
80 | }, |
81 | { | |
82 | 'url': 'https://www.nbc.com/classic-tv/charles-in-charge/video/charles-in-charge-pilot/n3310', | |
83 | 'only_matching': True, | |
84 | }, | |
d0c5fabc T |
85 | { |
86 | # Percent escaped url | |
87 | 'url': 'https://www.nbc.com/up-all-night/video/day-after-valentine%27s-day/n2189', | |
88 | 'only_matching': True, | |
89 | } | |
58c1f6f0 | 90 | ] |
020cf5eb JMF |
91 | |
92 | def _real_extract(self, url): | |
5ad28e7f | 93 | permalink, video_id = self._match_valid_url(url).groups() |
d0c5fabc | 94 | permalink = 'http' + compat_urllib_parse_unquote(permalink) |
48ff5590 | 95 | video_data = self._download_json( |
311ee457 | 96 | 'https://friendship.nbc.co/v2/graphql', video_id, query={ |
48ff5590 RA |
97 | 'query': '''query bonanzaPage( |
98 | $app: NBCUBrands! = nbc | |
99 | $name: String! | |
100 | $oneApp: Boolean | |
101 | $platform: SupportedPlatforms! = web | |
102 | $type: EntityPageType! = VIDEO | |
103 | $userId: String! | |
104 | ) { | |
105 | bonanzaPage( | |
106 | app: $app | |
107 | name: $name | |
108 | oneApp: $oneApp | |
109 | platform: $platform | |
110 | type: $type | |
111 | userId: $userId | |
112 | ) { | |
113 | metadata { | |
311ee457 RA |
114 | ... on VideoPageData { |
115 | description | |
116 | episodeNumber | |
117 | keywords | |
118 | locked | |
119 | mpxAccountId | |
120 | mpxGuid | |
121 | rating | |
48ff5590 | 122 | resourceId |
311ee457 RA |
123 | seasonNumber |
124 | secondaryTitle | |
125 | seriesShortTitle | |
126 | } | |
127 | } | |
128 | } | |
48ff5590 RA |
129 | }''', |
130 | 'variables': json.dumps({ | |
131 | 'name': permalink, | |
132 | 'oneApp': True, | |
133 | 'userId': '0', | |
134 | }), | |
135 | })['data']['bonanzaPage']['metadata'] | |
2eeb588e RA |
136 | query = { |
137 | 'mbr': 'true', | |
138 | 'manifest': 'm3u', | |
139 | } | |
311ee457 | 140 | video_id = video_data['mpxGuid'] |
9160a0c6 TSJ |
141 | tp_path = 'NnzsPC/media/guid/%s/%s' % (video_data.get('mpxAccountId') or '2410887629', video_id) |
142 | tpm = self._download_theplatform_metadata(tp_path, video_id) | |
143 | title = tpm.get('title') or video_data.get('secondaryTitle') | |
311ee457 | 144 | if video_data.get('locked'): |
2eeb588e | 145 | resource = self._get_mvpd_resource( |
48ff5590 RA |
146 | video_data.get('resourceId') or 'nbcentertainment', |
147 | title, video_id, video_data.get('rating')) | |
2eeb588e RA |
148 | query['auth'] = self._extract_mvpd_auth( |
149 | url, video_id, 'nbcentertainment', resource) | |
150 | theplatform_url = smuggle_url(update_url_query( | |
311ee457 | 151 | 'http://link.theplatform.com/s/NnzsPC/media/guid/%s/%s' % (video_data.get('mpxAccountId') or '2410887629', video_id), |
2eeb588e | 152 | query), {'force_smil_url': True}) |
9160a0c6 TSJ |
153 | |
154 | # Empty string or 0 can be valid values for these. So the check must be `is None` | |
155 | description = video_data.get('description') | |
156 | if description is None: | |
157 | description = tpm.get('description') | |
158 | episode_number = int_or_none(video_data.get('episodeNumber')) | |
159 | if episode_number is None: | |
160 | episode_number = int_or_none(tpm.get('nbcu$airOrder')) | |
161 | rating = video_data.get('rating') | |
162 | if rating is None: | |
163 | try_get(tpm, lambda x: x['ratings'][0]['rating']) | |
164 | season_number = int_or_none(video_data.get('seasonNumber')) | |
165 | if season_number is None: | |
166 | season_number = int_or_none(tpm.get('nbcu$seasonNumber')) | |
167 | series = video_data.get('seriesShortTitle') | |
168 | if series is None: | |
169 | series = tpm.get('nbcu$seriesShortTitle') | |
170 | tags = video_data.get('keywords') | |
171 | if tags is None or len(tags) == 0: | |
172 | tags = tpm.get('keywords') | |
173 | ||
2eeb588e | 174 | return { |
e881c4bc | 175 | '_type': 'url_transparent', |
9160a0c6 TSJ |
176 | 'age_limit': parse_age_limit(rating), |
177 | 'description': description, | |
178 | 'episode': title, | |
179 | 'episode_number': episode_number, | |
e881c4bc | 180 | 'id': video_id, |
9160a0c6 TSJ |
181 | 'ie_key': 'ThePlatform', |
182 | 'season_number': season_number, | |
183 | 'series': series, | |
184 | 'tags': tags, | |
2eeb588e RA |
185 | 'title': title, |
186 | 'url': theplatform_url, | |
e881c4bc | 187 | } |
020cf5eb JMF |
188 | |
189 | ||
a2a4d5fa | 190 | class NBCSportsVPlayerIE(InfoExtractor): |
29f7c58a | 191 | _VALID_URL_BASE = r'https?://(?:vplayer\.nbcsports\.com|(?:www\.)?nbcsports\.com/vplayer)/' |
192 | _VALID_URL = _VALID_URL_BASE + r'(?:[^/]+/)+(?P<id>[0-9a-zA-Z_]+)' | |
bfd973ec | 193 | _EMBED_REGEX = [r'(?:iframe[^>]+|var video|div[^>]+data-(?:mpx-)?)[sS]rc\s?=\s?"(?P<url>%s[^\"]+)' % _VALID_URL_BASE] |
a28ccbab | 194 | |
5cbb2699 | 195 | _TESTS = [{ |
12ea5c79 | 196 | 'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_embed/select/9CsDKds0kvHI', |
a28ccbab YCH |
197 | 'info_dict': { |
198 | 'id': '9CsDKds0kvHI', | |
12ea5c79 | 199 | 'ext': 'mp4', |
a28ccbab YCH |
200 | 'description': 'md5:df390f70a9ba7c95ff1daace988f0d8d', |
201 | 'title': 'Tyler Kalinoski hits buzzer-beater to lift Davidson', | |
79ba9140 | 202 | 'timestamp': 1426270238, |
203 | 'upload_date': '20150313', | |
204 | 'uploader': 'NBCU-SPORTS', | |
bfa0e270 | 205 | 'duration': 72.818, |
206 | 'chapters': [], | |
207 | 'thumbnail': r're:^https?://.*\.jpg$' | |
a28ccbab | 208 | } |
5cbb2699 | 209 | }, { |
bfa0e270 | 210 | 'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_embed/select/media/PEgOtlNcC_y2', |
5cbb2699 | 211 | 'only_matching': True, |
29f7c58a | 212 | }, { |
213 | 'url': 'https://www.nbcsports.com/vplayer/p/BxmELC/nbcsports/select/PHJSaFWbrTY9?form=html&autoPlay=true', | |
214 | 'only_matching': True, | |
5cbb2699 | 215 | }] |
a28ccbab YCH |
216 | |
217 | def _real_extract(self, url): | |
218 | video_id = self._match_id(url) | |
219 | webpage = self._download_webpage(url, video_id) | |
bfa0e270 | 220 | theplatform_url = self._html_search_regex(r'tp:releaseUrl="(.+?)"', webpage, 'url') |
a28ccbab YCH |
221 | return self.url_result(theplatform_url, 'ThePlatform') |
222 | ||
223 | ||
a2a4d5fa | 224 | class NBCSportsIE(InfoExtractor): |
29f7c58a | 225 | _VALID_URL = r'https?://(?:www\.)?nbcsports\.com//?(?!vplayer/)(?:[^/]+/)+(?P<id>[0-9a-z-]+)' |
a2a4d5fa | 226 | |
29f7c58a | 227 | _TESTS = [{ |
228 | # iframe src | |
a2a4d5fa | 229 | 'url': 'http://www.nbcsports.com//college-basketball/ncaab/tom-izzo-michigan-st-has-so-much-respect-duke', |
a2a4d5fa YCH |
230 | 'info_dict': { |
231 | 'id': 'PHJSaFWbrTY9', | |
29f7c58a | 232 | 'ext': 'mp4', |
a2a4d5fa YCH |
233 | 'title': 'Tom Izzo, Michigan St. has \'so much respect\' for Duke', |
234 | 'description': 'md5:ecb459c9d59e0766ac9c7d5d0eda8113', | |
0738187f YCH |
235 | 'uploader': 'NBCU-SPORTS', |
236 | 'upload_date': '20150330', | |
237 | 'timestamp': 1427726529, | |
bfa0e270 | 238 | 'chapters': [], |
239 | 'thumbnail': 'https://hdliveextra-a.akamaihd.net/HD/image_sports/NBCU_Sports_Group_-_nbcsports/253/303/izzodps.jpg', | |
240 | 'duration': 528.395, | |
a2a4d5fa | 241 | } |
29f7c58a | 242 | }, { |
243 | # data-mpx-src | |
244 | 'url': 'https://www.nbcsports.com/philadelphia/philadelphia-phillies/bruce-bochy-hector-neris-hes-idiot', | |
245 | 'only_matching': True, | |
246 | }, { | |
247 | # data-src | |
248 | 'url': 'https://www.nbcsports.com/boston/video/report-card-pats-secondary-no-match-josh-allen', | |
249 | 'only_matching': True, | |
250 | }] | |
a2a4d5fa YCH |
251 | |
252 | def _real_extract(self, url): | |
253 | video_id = self._match_id(url) | |
254 | webpage = self._download_webpage(url, video_id) | |
255 | return self.url_result( | |
256 | NBCSportsVPlayerIE._extract_url(webpage), 'NBCSportsVPlayer') | |
257 | ||
258 | ||
1139935d RA |
259 | class NBCSportsStreamIE(AdobePassIE): |
260 | _VALID_URL = r'https?://stream\.nbcsports\.com/.+?\bpid=(?P<id>\d+)' | |
261 | _TEST = { | |
262 | 'url': 'http://stream.nbcsports.com/nbcsn/generic?pid=206559', | |
263 | 'info_dict': { | |
264 | 'id': '206559', | |
265 | 'ext': 'mp4', | |
266 | 'title': 'Amgen Tour of California Women\'s Recap', | |
267 | 'description': 'md5:66520066b3b5281ada7698d0ea2aa894', | |
268 | }, | |
269 | 'params': { | |
270 | # m3u8 download | |
271 | 'skip_download': True, | |
272 | }, | |
273 | 'skip': 'Requires Adobe Pass Authentication', | |
274 | } | |
275 | ||
276 | def _real_extract(self, url): | |
277 | video_id = self._match_id(url) | |
278 | live_source = self._download_json( | |
279 | 'http://stream.nbcsports.com/data/live_sources_%s.json' % video_id, | |
280 | video_id) | |
281 | video_source = live_source['videoSources'][0] | |
282 | title = video_source['title'] | |
283 | source_url = None | |
284 | for k in ('source', 'msl4source', 'iossource', 'hlsv4'): | |
285 | sk = k + 'Url' | |
286 | source_url = video_source.get(sk) or video_source.get(sk + 'Alt') | |
287 | if source_url: | |
288 | break | |
289 | else: | |
290 | source_url = video_source['ottStreamUrl'] | |
291 | is_live = video_source.get('type') == 'live' or video_source.get('status') == 'Live' | |
292 | resource = self._get_mvpd_resource('nbcsports', title, video_id, '') | |
293 | token = self._extract_mvpd_auth(url, video_id, 'nbcsports', resource) | |
294 | tokenized_url = self._download_json( | |
295 | 'https://token.playmakerservices.com/cdn', | |
296 | video_id, data=json.dumps({ | |
297 | 'requestorId': 'nbcsports', | |
298 | 'pid': video_id, | |
299 | 'application': 'NBCSports', | |
300 | 'version': 'v1', | |
301 | 'platform': 'desktop', | |
302 | 'cdn': 'akamai', | |
303 | 'url': video_source['sourceUrl'], | |
304 | 'token': base64.b64encode(token.encode()).decode(), | |
305 | 'resourceId': base64.b64encode(resource.encode()).decode(), | |
306 | }).encode())['tokenizedUrl'] | |
307 | formats = self._extract_m3u8_formats(tokenized_url, video_id, 'mp4') | |
1139935d RA |
308 | return { |
309 | 'id': video_id, | |
39ca3b5c | 310 | 'title': title, |
1139935d RA |
311 | 'description': live_source.get('description'), |
312 | 'formats': formats, | |
313 | 'is_live': is_live, | |
314 | } | |
315 | ||
316 | ||
6368e2e6 | 317 | class NBCNewsIE(ThePlatformIE): # XXX: Do not subclass from concrete IE |
a843464a | 318 | _VALID_URL = r'(?x)https?://(?:www\.)?(?:nbcnews|today|msnbc)\.com/([^/]+/)*(?:.*-)?(?P<id>[^/?]+)' |
bfd973ec | 319 | _EMBED_REGEX = [r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//www\.nbcnews\.com/widget/video-embed/[^"\']+)\1'] |
0bc56fa6 | 320 | |
87fe568c | 321 | _TESTS = [ |
87fe568c | 322 | { |
574b2a73 | 323 | 'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880', |
895e5c03 | 324 | 'md5': 'cf4bc9e6ce0130f00f545d80ecedd4bf', |
87fe568c | 325 | 'info_dict': { |
a843464a | 326 | 'id': '269389891880', |
10e3d734 | 327 | 'ext': 'mp4', |
87fe568c JMF |
328 | 'title': 'How Twitter Reacted To The Snowden Interview', |
329 | 'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64', | |
0437307a RA |
330 | 'timestamp': 1401363060, |
331 | 'upload_date': '20140529', | |
87fe568c | 332 | }, |
87fe568c | 333 | }, |
2df54b4b S |
334 | { |
335 | 'url': 'http://www.nbcnews.com/feature/dateline-full-episodes/full-episode-family-business-n285156', | |
336 | 'md5': 'fdbf39ab73a72df5896b6234ff98518a', | |
337 | 'info_dict': { | |
0437307a | 338 | 'id': '529953347624', |
2df54b4b S |
339 | 'ext': 'mp4', |
340 | 'title': 'FULL EPISODE: Family Business', | |
341 | 'description': 'md5:757988edbaae9d7be1d585eb5d55cc04', | |
342 | }, | |
574b2a73 | 343 | 'skip': 'This page is unavailable.', |
2df54b4b | 344 | }, |
d9aa2b78 RS |
345 | { |
346 | 'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844', | |
895e5c03 | 347 | 'md5': '8eb831eca25bfa7d25ddd83e85946548', |
d9aa2b78 | 348 | 'info_dict': { |
a843464a | 349 | 'id': '394064451844', |
d9aa2b78 RS |
350 | 'ext': 'mp4', |
351 | 'title': 'Nightly News with Brian Williams Full Broadcast (February 4)', | |
352 | 'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5', | |
0437307a | 353 | 'timestamp': 1423104900, |
0437307a | 354 | 'upload_date': '20150205', |
d9aa2b78 RS |
355 | }, |
356 | }, | |
574b2a73 | 357 | { |
358 | 'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456', | |
895e5c03 | 359 | 'md5': '4a8c4cec9e1ded51060bdda36ff0a5c0', |
574b2a73 | 360 | 'info_dict': { |
895e5c03 | 361 | 'id': 'n431456', |
574b2a73 | 362 | 'ext': 'mp4', |
895e5c03 RA |
363 | 'title': "Volkswagen U.S. Chief: We 'Totally Screwed Up'", |
364 | 'description': 'md5:d22d1281a24f22ea0880741bb4dd6301', | |
0437307a RA |
365 | 'upload_date': '20150922', |
366 | 'timestamp': 1442917800, | |
574b2a73 | 367 | }, |
574b2a73 | 368 | }, |
cb7d4d0e | 369 | { |
370 | 'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788', | |
371 | 'md5': '118d7ca3f0bea6534f119c68ef539f71', | |
372 | 'info_dict': { | |
a843464a | 373 | 'id': '669831235788', |
cb7d4d0e | 374 | 'ext': 'mp4', |
375 | 'title': 'See the aurora borealis from space in stunning new NASA video', | |
376 | 'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1', | |
377 | 'upload_date': '20160420', | |
378 | 'timestamp': 1461152093, | |
0437307a RA |
379 | }, |
380 | }, | |
381 | { | |
382 | 'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924', | |
383 | 'md5': '6d236bf4f3dddc226633ce6e2c3f814d', | |
384 | 'info_dict': { | |
a843464a | 385 | 'id': '314487875924', |
0437307a RA |
386 | 'ext': 'mp4', |
387 | 'title': 'The chaotic GOP immigration vote', | |
388 | '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 | 389 | 'thumbnail': r're:^https?://.*\.jpg$', |
0437307a RA |
390 | 'timestamp': 1406937606, |
391 | 'upload_date': '20140802', | |
cb7d4d0e | 392 | }, |
393 | }, | |
3f125c8c S |
394 | { |
395 | 'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952', | |
396 | 'only_matching': True, | |
397 | }, | |
5de008e8 YCH |
398 | { |
399 | # From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html | |
400 | 'url': 'http://www.nbcnews.com/widget/video-embed/701714499682', | |
401 | 'only_matching': True, | |
402 | }, | |
87fe568c | 403 | ] |
0bc56fa6 JMF |
404 | |
405 | def _real_extract(self, url): | |
a843464a | 406 | video_id = self._match_id(url) |
895e5c03 RA |
407 | webpage = self._download_webpage(url, video_id) |
408 | ||
135dfa2c | 409 | data = self._search_nextjs_data(webpage, video_id)['props']['initialState'] |
895e5c03 RA |
410 | video_data = try_get(data, lambda x: x['video']['current'], dict) |
411 | if not video_data: | |
412 | video_data = data['article']['content'][0]['primaryMedia']['video'] | |
413 | title = video_data['headline']['primary'] | |
414 | ||
415 | formats = [] | |
416 | for va in video_data.get('videoAssets', []): | |
417 | public_url = va.get('publicUrl') | |
418 | if not public_url: | |
419 | continue | |
420 | if '://link.theplatform.com/' in public_url: | |
421 | public_url = update_url_query(public_url, {'format': 'redirect'}) | |
422 | format_id = va.get('format') | |
423 | if format_id == 'M3U': | |
424 | formats.extend(self._extract_m3u8_formats( | |
425 | public_url, video_id, 'mp4', 'm3u8_native', | |
426 | m3u8_id=format_id, fatal=False)) | |
427 | continue | |
428 | tbr = int_or_none(va.get('bitrate'), 1000) | |
429 | if tbr: | |
430 | format_id += '-%d' % tbr | |
431 | formats.append({ | |
432 | 'format_id': format_id, | |
433 | 'url': public_url, | |
434 | 'width': int_or_none(va.get('width')), | |
435 | 'height': int_or_none(va.get('height')), | |
436 | 'tbr': tbr, | |
437 | 'ext': 'mp4', | |
438 | }) | |
6e416b21 | 439 | |
895e5c03 RA |
440 | subtitles = {} |
441 | closed_captioning = video_data.get('closedCaptioning') | |
442 | if closed_captioning: | |
443 | for cc_url in closed_captioning.values(): | |
444 | if not cc_url: | |
445 | continue | |
446 | subtitles.setdefault('en', []).append({ | |
447 | 'url': cc_url, | |
448 | }) | |
6e416b21 | 449 | |
a843464a | 450 | return { |
a843464a | 451 | 'id': video_id, |
895e5c03 RA |
452 | 'title': title, |
453 | 'description': try_get(video_data, lambda x: x['description']['primary']), | |
454 | 'thumbnail': try_get(video_data, lambda x: x['primaryImage']['url']['primary']), | |
455 | 'duration': parse_duration(video_data.get('duration')), | |
456 | 'timestamp': unified_timestamp(video_data.get('datePublished')), | |
457 | 'formats': formats, | |
458 | 'subtitles': subtitles, | |
a843464a | 459 | } |
be457302 YCH |
460 | |
461 | ||
462 | class NBCOlympicsIE(InfoExtractor): | |
58284890 | 463 | IE_NAME = 'nbcolympics' |
3e376d18 | 464 | _VALID_URL = r'https?://www\.nbcolympics\.com/videos?/(?P<id>[0-9a-z-]+)' |
be457302 YCH |
465 | |
466 | _TEST = { | |
467 | # Geo-restricted to US | |
468 | 'url': 'http://www.nbcolympics.com/video/justin-roses-son-leo-was-tears-after-his-dad-won-gold', | |
469 | 'md5': '54fecf846d05429fbaa18af557ee523a', | |
470 | 'info_dict': { | |
471 | 'id': 'WjTBzDXx5AUq', | |
472 | 'display_id': 'justin-roses-son-leo-was-tears-after-his-dad-won-gold', | |
473 | 'ext': 'mp4', | |
474 | 'title': 'Rose\'s son Leo was in tears after his dad won gold', | |
475 | 'description': 'Olympic gold medalist Justin Rose gets emotional talking to the impact his win in men\'s golf has already had on his children.', | |
476 | 'timestamp': 1471274964, | |
477 | 'upload_date': '20160815', | |
478 | 'uploader': 'NBCU-SPORTS', | |
479 | }, | |
480 | } | |
481 | ||
482 | def _real_extract(self, url): | |
483 | display_id = self._match_id(url) | |
484 | ||
485 | webpage = self._download_webpage(url, display_id) | |
486 | ||
3e376d18 W |
487 | try: |
488 | drupal_settings = self._parse_json(self._search_regex( | |
489 | r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);', | |
490 | webpage, 'drupal settings'), display_id) | |
491 | ||
492 | iframe_url = drupal_settings['vod']['iframe_url'] | |
493 | theplatform_url = iframe_url.replace( | |
494 | 'vplayer.nbcolympics.com', 'player.theplatform.com') | |
495 | except RegexNotFoundError: | |
496 | theplatform_url = self._search_regex( | |
497 | r"([\"'])embedUrl\1: *([\"'])(?P<embedUrl>.+)\2", | |
498 | webpage, 'embedding URL', group="embedUrl") | |
be457302 YCH |
499 | |
500 | return { | |
501 | '_type': 'url_transparent', | |
502 | 'url': theplatform_url, | |
503 | 'ie_key': ThePlatformIE.ie_key(), | |
504 | 'display_id': display_id, | |
505 | } | |
58284890 RA |
506 | |
507 | ||
508 | class NBCOlympicsStreamIE(AdobePassIE): | |
509 | IE_NAME = 'nbcolympics:stream' | |
510 | _VALID_URL = r'https?://stream\.nbcolympics\.com/(?P<id>[0-9a-z-]+)' | |
bb36a55c | 511 | _TESTS = [ |
512 | { | |
513 | 'note': 'Tokenized m3u8 source URL', | |
514 | 'url': 'https://stream.nbcolympics.com/womens-soccer-group-round-11', | |
515 | 'info_dict': { | |
516 | 'id': '2019740', | |
517 | 'ext': 'mp4', | |
518 | 'title': r"re:Women's Group Stage - Netherlands vs\. Brazil [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$", | |
519 | }, | |
520 | 'params': { | |
521 | 'skip_download': 'm3u8', | |
522 | }, | |
523 | }, { | |
524 | 'note': 'Plain m3u8 source URL', | |
525 | 'url': 'https://stream.nbcolympics.com/gymnastics-event-finals-mens-floor-pommel-horse-womens-vault-bars', | |
526 | 'info_dict': { | |
527 | 'id': '2021729', | |
528 | 'ext': 'mp4', | |
529 | 'title': r're:Event Finals: M Floor, W Vault, M Pommel, W Uneven Bars [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', | |
530 | }, | |
531 | 'params': { | |
532 | 'skip_download': 'm3u8', | |
533 | }, | |
58284890 | 534 | }, |
bb36a55c | 535 | ] |
58284890 RA |
536 | |
537 | def _real_extract(self, url): | |
538 | display_id = self._match_id(url) | |
539 | webpage = self._download_webpage(url, display_id) | |
540 | pid = self._search_regex(r'pid\s*=\s*(\d+);', webpage, 'pid') | |
bb36a55c | 541 | |
58284890 | 542 | event_config = self._download_json( |
bb36a55c | 543 | f'http://stream.nbcolympics.com/data/event_config_{pid}.json', |
544 | pid, 'Downloading event config')['eventConfig'] | |
545 | ||
546 | title = event_config['eventTitle'] | |
547 | is_live = {'live': True, 'replay': False}.get(event_config.get('eventStatus')) | |
bb36a55c | 548 | |
58284890 | 549 | source_url = self._download_json( |
bb36a55c | 550 | f'https://api-leap.nbcsports.com/feeds/assets/{pid}?application=NBCOlympics&platform=desktop&format=nbc-player&env=staging', |
551 | pid, 'Downloading leap config' | |
552 | )['videoSources'][0]['cdnSources']['primary'][0]['sourceUrl'] | |
553 | ||
554 | if event_config.get('cdnToken'): | |
555 | ap_resource = self._get_mvpd_resource( | |
556 | event_config.get('resourceId', 'NBCOlympics'), | |
557 | re.sub(r'[^\w\d ]+', '', event_config['eventTitle']), pid, | |
558 | event_config.get('ratingId', 'NO VALUE')) | |
559 | media_token = self._extract_mvpd_auth(url, pid, event_config.get('requestorId', 'NBCOlympics'), ap_resource) | |
560 | ||
561 | source_url = self._download_json( | |
562 | 'https://tokens.playmakerservices.com/', pid, 'Retrieving tokenized URL', | |
563 | data=json.dumps({ | |
564 | 'application': 'NBCSports', | |
565 | 'authentication-type': 'adobe-pass', | |
566 | 'cdn': 'akamai', | |
567 | 'pid': pid, | |
568 | 'platform': 'desktop', | |
569 | 'requestorId': 'NBCOlympics', | |
570 | 'resourceId': base64.b64encode(ap_resource.encode()).decode(), | |
571 | 'token': base64.b64encode(media_token.encode()).decode(), | |
572 | 'url': source_url, | |
573 | 'version': 'v1', | |
574 | }).encode(), | |
575 | )['akamai'][0]['tokenizedUrl'] | |
576 | ||
577 | formats = self._extract_m3u8_formats(source_url, pid, 'mp4', live=is_live) | |
578 | for f in formats: | |
579 | # -http_seekable requires ffmpeg 4.3+ but it doesnt seem possible to | |
580 | # download with ffmpeg without this option | |
0a5a191a | 581 | f['downloader_options'] = {'ffmpeg_args': ['-seekable', '0', '-http_seekable', '0', '-icy', '0']} |
58284890 RA |
582 | |
583 | return { | |
584 | 'id': pid, | |
585 | 'display_id': display_id, | |
586 | 'title': title, | |
587 | 'formats': formats, | |
bb36a55c | 588 | 'is_live': is_live, |
58284890 | 589 | } |
11398b92 | 590 | |
591 | ||
592 | class NBCStationsIE(InfoExtractor): | |
593 | _DOMAIN_RE = '|'.join(map(re.escape, ( | |
594 | 'nbcbayarea', 'nbcboston', 'nbcchicago', 'nbcconnecticut', 'nbcdfw', 'nbclosangeles', | |
595 | 'nbcmiami', 'nbcnewyork', 'nbcphiladelphia', 'nbcsandiego', 'nbcwashington', | |
596 | 'necn', 'telemundo52', 'telemundoarizona', 'telemundochicago', 'telemundonuevainglaterra', | |
597 | ))) | |
598 | _VALID_URL = rf'https?://(?:www\.)?(?P<site>{_DOMAIN_RE})\.com/(?:[^/?#]+/)*(?P<id>[^/?#]+)/?(?:$|[#?])' | |
599 | ||
600 | _TESTS = [{ | |
601 | 'url': 'https://www.nbclosangeles.com/news/local/large-structure-fire-in-downtown-la-prompts-smoke-odor-advisory/2968618/', | |
602 | 'md5': '462041d91bd762ef5a38b7d85d6dc18f', | |
603 | 'info_dict': { | |
604 | 'id': '2968618', | |
605 | 'ext': 'mp4', | |
606 | 'title': 'Large Structure Fire in Downtown LA Prompts Smoke Odor Advisory', | |
607 | 'description': None, | |
608 | 'timestamp': 1661135892, | |
609 | 'upload_date': '20220821', | |
610 | 'uploader': 'NBC 4', | |
611 | 'uploader_id': 'KNBC', | |
612 | 'channel': 'nbclosangeles', | |
613 | }, | |
614 | }, { | |
615 | 'url': 'https://www.telemundoarizona.com/responde/huracan-complica-reembolso-para-televidente-de-tucson/2247002/', | |
616 | 'md5': '0917dcf7885be1023a9220630d415f67', | |
617 | 'info_dict': { | |
618 | 'id': '2247002', | |
619 | 'ext': 'mp4', | |
620 | 'title': 'Huracán complica que televidente de Tucson reciba reembolso', | |
621 | 'description': 'md5:af298dc73aab74d4fca6abfb12acb6cf', | |
622 | 'timestamp': 1660886507, | |
623 | 'upload_date': '20220819', | |
624 | 'uploader': 'Telemundo Arizona', | |
625 | 'uploader_id': 'KTAZ', | |
626 | 'channel': 'telemundoarizona', | |
627 | }, | |
628 | }] | |
629 | ||
630 | _RESOLUTIONS = { | |
631 | '1080': '1920', | |
632 | '720': '1280', | |
633 | '540': '960', | |
634 | '360': '640', | |
635 | '234': '416', | |
636 | } | |
637 | ||
638 | def _real_extract(self, url): | |
639 | channel, video_id = self._match_valid_url(url).group('site', 'id') | |
640 | webpage = self._download_webpage(url, video_id) | |
641 | ||
642 | nbc_data = self._search_json( | |
304ad45a | 643 | r'<script>var\s*nbc\s*=', webpage, 'NBC JSON data', video_id) |
11398b92 | 644 | pdk_acct = nbc_data.get('pdkAcct') or 'Yh1nAC' |
645 | fw_ssid = traverse_obj(nbc_data, ('video', 'fwSSID')) | |
646 | fw_network_id = traverse_obj(nbc_data, ('video', 'fwNetworkID'), default='382114') | |
647 | ||
648 | video_data = self._parse_json(self._html_search_regex( | |
649 | r'data-videos="([^"]*)"', webpage, 'video data', default='{}'), video_id) | |
650 | video_data = variadic(video_data)[0] | |
651 | video_data.update(self._parse_json(self._html_search_regex( | |
652 | r'data-meta="([^"]*)"', webpage, 'metadata', default='{}'), video_id)) | |
653 | ||
654 | formats = [] | |
655 | ||
656 | if video_data.get('mpx_is_livestream') == '1': | |
657 | live = True | |
658 | player_id = traverse_obj( | |
659 | video_data, 'mpx_m3upid', ('video', 'meta', 'mpx_m3upid'), 'mpx_pid', | |
660 | ('video', 'meta', 'mpx_pid'), 'pid_streaming_web_medium') | |
661 | query = { | |
662 | 'mbr': 'true', | |
663 | 'assetTypes': 'LegacyRelease', | |
664 | 'fwsitesection': fw_ssid, | |
665 | 'fwNetworkID': fw_network_id, | |
666 | 'pprofile': 'ots_desktop_html', | |
667 | 'sensitive': 'false', | |
668 | 'w': '1920', | |
669 | 'h': '1080', | |
670 | 'rnd': '1660303', | |
671 | 'mode': 'LIVE', | |
672 | 'format': 'SMIL', | |
673 | 'tracking': 'true', | |
674 | 'formats': 'M3U+none,MPEG-DASH+none,MPEG4,MP3', | |
675 | 'vpaid': 'script', | |
676 | 'schema': '2.0', | |
677 | 'SDK': 'PDK+6.1.3', | |
678 | } | |
679 | info = { | |
680 | 'title': f'{channel} livestream', | |
681 | } | |
682 | ||
683 | else: | |
684 | live = False | |
685 | player_id = traverse_obj( | |
686 | video_data, ('video', 'meta', 'pid_streaming_web_high'), 'pid_streaming_web_high', | |
687 | ('video', 'meta', 'mpx_pid'), 'mpx_pid') | |
688 | ||
689 | date_string = traverse_obj(video_data, 'date_string', 'date_gmt') | |
690 | if date_string: | |
691 | date_string = self._search_regex( | |
692 | r'datetime="([^"]+)"', date_string, 'date string', fatal=False) | |
693 | else: | |
694 | date_string = traverse_obj( | |
695 | nbc_data, ('dataLayer', 'adobe', 'prop70'), ('dataLayer', 'adobe', 'eVar70'), | |
696 | ('dataLayer', 'adobe', 'eVar59')) | |
697 | ||
698 | video_url = traverse_obj(video_data, ('video', 'meta', 'mp4_url'), 'mp4_url') | |
699 | if video_url: | |
700 | height = url_basename(video_url).split('-')[1].split('p')[0] | |
701 | formats.append({ | |
702 | 'url': video_url, | |
703 | 'ext': 'mp4', | |
704 | 'width': int_or_none(self._RESOLUTIONS.get(height)), | |
705 | 'height': int_or_none(height), | |
706 | 'format_id': f'http-{height}', | |
707 | }) | |
708 | ||
709 | query = { | |
710 | 'mbr': 'true', | |
711 | 'assetTypes': 'LegacyRelease', | |
712 | 'fwsitesection': fw_ssid, | |
713 | 'fwNetworkID': fw_network_id, | |
714 | 'format': 'redirect', | |
715 | 'manifest': 'm3u', | |
716 | 'Tracking': 'true', | |
717 | 'Embedded': 'true', | |
718 | 'formats': 'MPEG4', | |
719 | } | |
720 | info = { | |
721 | 'title': video_data.get('title') or traverse_obj( | |
722 | nbc_data, ('dataLayer', 'contenttitle'), ('dataLayer', 'title'), | |
723 | ('dataLayer', 'adobe', 'prop22'), ('dataLayer', 'id')), | |
724 | 'description': traverse_obj(video_data, 'summary', 'excerpt', 'video_hero_text'), | |
725 | 'upload_date': str_or_none(unified_strdate(date_string)), | |
726 | 'timestamp': int_or_none(unified_timestamp(date_string)), | |
727 | } | |
728 | ||
729 | if not player_id: | |
730 | raise ExtractorError( | |
731 | 'No video player ID or livestream player ID found in webpage', expected=True) | |
732 | ||
733 | headers = {'Origin': f'https://www.{channel}.com'} | |
734 | manifest, urlh = self._download_webpage_handle( | |
735 | f'https://link.theplatform.com/s/{pdk_acct}/{player_id}', video_id, | |
736 | headers=headers, query=query, note='Downloading manifest') | |
737 | if live: | |
738 | manifest_url = self._search_regex(r'<video src="([^"]*)', manifest, 'manifest URL') | |
739 | else: | |
740 | manifest_url = urlh.geturl() | |
741 | ||
742 | formats.extend(self._extract_m3u8_formats( | |
743 | manifest_url, video_id, 'mp4', headers=headers, m3u8_id='hls', | |
744 | fatal=live, live=live, errnote='No HLS formats found')) | |
11398b92 | 745 | |
746 | return { | |
747 | 'id': str_or_none(video_id), | |
748 | 'channel': channel, | |
749 | 'uploader': str_or_none(nbc_data.get('on_air_name')), | |
750 | 'uploader_id': str_or_none(nbc_data.get('callLetters')), | |
751 | 'formats': formats, | |
752 | 'is_live': live, | |
753 | **info, | |
754 | } |