1
0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2025-08-15 17:08:29 +00:00

NBCSportsIE, NBCNewsIE

This commit is contained in:
doe1080 2025-07-16 23:24:51 +09:00
parent ff08c27cc8
commit faa4f15b71
2 changed files with 202 additions and 222 deletions

View File

@ -981,26 +981,6 @@ class GenericIE(InfoExtractor):
'thumbnail': r're:https?://cdn\.vox-cdn\.com/thumbor/.+\.jpg',
},
'playlist_count': 3,
}, {
# 🔍85
# NBCNewsIE
'url': 'http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html',
'info_dict': {
'id': 'x_dtl_oa_LettermanliftPR_160608',
'ext': 'mp4',
'title': 'David Letterman: A Preview',
},
'skip': 'Site changed',
}, {
# 🔍86
# NBCSportsIE
'url': 'http://www.riderfans.com/forum/showthread.php?121827-Freeman&s=e98fa1ea6dc08e886b1678d35212494a',
'info_dict': {
'id': 'ln7x1qSThw4k',
'ext': 'flv',
'title': "PFT Live: New leader in the 'new-look' defense",
},
'skip': 'Invalid URL',
}, {
# ✅124
# YoutubeIE

View File

@ -138,95 +138,88 @@ def _extract_nbcu_video(self, url, display_id, old_ie_key=None):
class NBCIE(NBCUniversalBaseIE):
_VALID_URL = r'https?(?P<permalink>://(?:www\.)?nbc\.com/(?:classic-tv/)?[^/?#]+/video/[^/?#]+/(?P<id>\w+))'
_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',
'episode': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s',
'episode_number': 86,
'season': 'Season 2',
'season_number': 2,
'series': 'Tonight',
'duration': 236.504,
'tags': 'count:2',
'thumbnail': r're:https?://.+\.jpg',
'categories': ['Series/The Tonight Show Starring Jimmy Fallon'],
'media_type': 'Full Episode',
'age_limit': 14,
'_old_archive_ids': ['theplatform 2848237'],
},
'params': {
'skip_download': 'm3u8',
},
_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',
'episode': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s',
'episode_number': 86,
'season': 'Season 2',
'season_number': 2,
'series': 'Tonight',
'duration': 236.504,
'tags': 'count:2',
'thumbnail': r're:https?://.+\.jpg',
'categories': ['Series/The Tonight Show Starring Jimmy Fallon'],
'media_type': 'Full Episode',
'age_limit': 14,
'_old_archive_ids': ['theplatform 2848237'],
},
{
'url': 'https://www.nbc.com/the-golden-globe-awards/video/oprah-winfrey-receives-cecil-b-de-mille-award-at-the-2018-golden-globes/3646439',
'info_dict': {
'id': '3646439',
'ext': 'mp4',
'title': 'Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes',
'episode': 'Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes',
'episode_number': 1,
'season': 'Season 75',
'season_number': 75,
'series': 'Golden Globes',
'description': 'Oprah Winfrey receives the Cecil B. de Mille Award at the 75th Annual Golden Globe Awards.',
'uploader': 'NBCU-COM',
'upload_date': '20180107',
'timestamp': 1515312000,
'duration': 569.703,
'tags': 'count:8',
'thumbnail': r're:https?://.+\.jpg',
'media_type': 'Highlight',
'age_limit': 0,
'categories': ['Series/The Golden Globe Awards'],
'_old_archive_ids': ['theplatform 3646439'],
},
'params': {
'skip_download': 'm3u8',
},
'params': {
'skip_download': 'm3u8',
},
{
# Needs to be extracted from webpage instead of GraphQL
'url': 'https://www.nbc.com/paris2024/video/ali-truwit-found-purpose-pool-after-her-life-changed/para24_sww_alitruwittodayshow_240823',
'info_dict': {
'id': 'para24_sww_alitruwittodayshow_240823',
'ext': 'mp4',
'title': 'Ali Truwit found purpose in the pool after her life changed',
'description': 'md5:c16d7489e1516593de1cc5d3f39b9bdb',
'uploader': 'NBCU-SPORTS',
'duration': 311.077,
'thumbnail': r're:https?://.+\.jpg',
'episode': 'Ali Truwit found purpose in the pool after her life changed',
'timestamp': 1724435902.0,
'upload_date': '20240823',
'_old_archive_ids': ['theplatform para24_sww_alitruwittodayshow_240823'],
},
'params': {
'skip_download': 'm3u8',
},
}, {
'url': 'https://www.nbc.com/the-golden-globe-awards/video/oprah-winfrey-receives-cecil-b-de-mille-award-at-the-2018-golden-globes/3646439',
'info_dict': {
'id': '3646439',
'ext': 'mp4',
'title': 'Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes',
'episode': 'Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes',
'episode_number': 1,
'season': 'Season 75',
'season_number': 75,
'series': 'Golden Globes',
'description': 'Oprah Winfrey receives the Cecil B. de Mille Award at the 75th Annual Golden Globe Awards.',
'uploader': 'NBCU-COM',
'upload_date': '20180107',
'timestamp': 1515312000,
'duration': 569.703,
'tags': 'count:8',
'thumbnail': r're:https?://.+\.jpg',
'media_type': 'Highlight',
'age_limit': 0,
'categories': ['Series/The Golden Globe Awards'],
'_old_archive_ids': ['theplatform 3646439'],
},
{
'url': 'https://www.nbc.com/quantum-leap/video/bens-first-leap-nbcs-quantum-leap/NBCE125189978',
'only_matching': True,
'params': {
'skip_download': 'm3u8',
},
{
'url': 'https://www.nbc.com/classic-tv/charles-in-charge/video/charles-in-charge-pilot/n3310',
'only_matching': True,
}, {
# Needs to be extracted from webpage instead of GraphQL
'url': 'https://www.nbc.com/paris2024/video/ali-truwit-found-purpose-pool-after-her-life-changed/para24_sww_alitruwittodayshow_240823',
'info_dict': {
'id': 'para24_sww_alitruwittodayshow_240823',
'ext': 'mp4',
'title': 'Ali Truwit found purpose in the pool after her life changed',
'description': 'md5:c16d7489e1516593de1cc5d3f39b9bdb',
'uploader': 'NBCU-SPORTS',
'duration': 311.077,
'thumbnail': r're:https?://.+\.jpg',
'episode': 'Ali Truwit found purpose in the pool after her life changed',
'timestamp': 1724435902.0,
'upload_date': '20240823',
'_old_archive_ids': ['theplatform para24_sww_alitruwittodayshow_240823'],
},
{
# Percent escaped url
'url': 'https://www.nbc.com/up-all-night/video/day-after-valentine%27s-day/n2189',
'only_matching': True,
'params': {
'skip_download': 'm3u8',
},
]
}, {
'url': 'https://www.nbc.com/quantum-leap/video/bens-first-leap-nbcs-quantum-leap/NBCE125189978',
'only_matching': True,
}, {
'url': 'https://www.nbc.com/classic-tv/charles-in-charge/video/charles-in-charge-pilot/n3310',
'only_matching': True,
}, {
# Percent escaped url
'url': 'https://www.nbc.com/up-all-night/video/day-after-valentine%27s-day/n2189',
'only_matching': True,
}]
_SOFTWARE_STATEMENT = 'eyJhbGciOiJSUzI1NiJ9.eyJzdWIiOiI1Yzg2YjdkYy04NDI3LTRjNDUtOGQwZi1iNDkzYmE3MmQwYjQiLCJuYmYiOjE1Nzg3MDM2MzEsImlzcyI6ImF1dGguYWRvYmUuY29tIiwiaWF0IjoxNTc4NzAzNjMxfQ.QQKIsBhAjGQTMdAqRTqhcz2Cddr4Y2hEjnSiOeKKki4nLrkDOsjQMmqeTR0hSRarraxH54wBgLvsxI7LHwKMvr7G8QpynNAxylHlQD3yhN9tFhxt4KR5wW3as02B-W2TznK9bhNWPKIyHND95Uo2Mi6rEQoq8tM9O09WPWaanE5BX_-r6Llr6dPq5F0Lpx2QOn2xYRb1T4nFxdFTNoss8GBds8OvChTiKpXMLHegLTc1OS4H_1a8tO_37jDwSdJuZ8iTyRLV4kZ2cpL6OL5JPMObD4-HQiec_dfcYgMKPiIfP9ZqdXpec2SVaCLsWEk86ZYvD97hLIQrK5rrKd1y-A'
def _real_extract(self, url):
@ -378,6 +371,15 @@ class NBCSportsIE(InfoExtractor):
'url': 'https://www.nbcsports.com/boston/video/report-card-pats-secondary-no-match-josh-allen',
'only_matching': True,
}]
_WEBPAGE_TESTS = [{
'url': 'http://www.riderfans.com/forum/showthread.php?121827-Freeman&s=e98fa1ea6dc08e886b1678d35212494a',
'info_dict': {
'id': 'ln7x1qSThw4k',
'ext': 'flv',
'title': "PFT Live: New leader in the 'new-look' defense",
},
'skip': 'Invalid URL',
}]
def _real_extract(self, url):
video_id = self._match_id(url)
@ -389,7 +391,7 @@ def _real_extract(self, url):
class NBCSportsStreamIE(AdobePassIE):
_WORKING = False
_VALID_URL = r'https?://stream\.nbcsports\.com/.+?\bpid=(?P<id>\d+)'
_TEST = {
_TESTS = [{
'url': 'http://stream.nbcsports.com/nbcsn/generic?pid=206559',
'info_dict': {
'id': '206559',
@ -402,7 +404,7 @@ class NBCSportsStreamIE(AdobePassIE):
'skip_download': True,
},
'skip': 'Requires Adobe Pass Authentication',
}
}]
def _real_extract(self, url):
video_id = self._match_id(url)
@ -449,98 +451,100 @@ class NBCNewsIE(ThePlatformIE): # XXX: Do not subclass from concrete IE
_VALID_URL = r'(?x)https?://(?:www\.)?(?:nbcnews|today|msnbc)\.com/([^/]+/)*(?:.*-)?(?P<id>[^/?]+)'
_EMBED_REGEX = [r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//www\.nbcnews\.com/widget/video-embed/[^"\']+)\1']
_TESTS = [
{
'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
'md5': 'fb3dcd2d7b1dd9804305fa2fc95ab610', # md5 tends to fluctuate
'info_dict': {
'id': '269389891880',
'ext': 'mp4',
'title': 'How Twitter Reacted To The Snowden Interview',
'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
'timestamp': 1401363060,
'upload_date': '20140529',
'duration': 46.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/140529/p_tweet_snow_140529.jpg',
},
_TESTS = [{
'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
'md5': 'fb3dcd2d7b1dd9804305fa2fc95ab610', # md5 tends to fluctuate
'info_dict': {
'id': '269389891880',
'ext': 'mp4',
'title': 'How Twitter Reacted To The Snowden Interview',
'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
'timestamp': 1401363060,
'upload_date': '20140529',
'duration': 46.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/140529/p_tweet_snow_140529.jpg',
},
{
'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/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',
},
{
'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
'md5': '40d0e48c68896359c80372306ece0fc3',
'info_dict': {
'id': '394064451844',
'ext': 'mp4',
'title': 'Nightly News with Brian Williams Full Broadcast (February 4)',
'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
'timestamp': 1423104900,
'upload_date': '20150205',
'duration': 1236.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/nn_netcast_150204.jpg',
},
'skip': 'This page is unavailable.',
}, {
'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
'md5': '40d0e48c68896359c80372306ece0fc3',
'info_dict': {
'id': '394064451844',
'ext': 'mp4',
'title': 'Nightly News with Brian Williams Full Broadcast (February 4)',
'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
'timestamp': 1423104900,
'upload_date': '20150205',
'duration': 1236.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/nn_netcast_150204.jpg',
},
{
'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
'md5': 'ffb59bcf0733dc3c7f0ace907f5e3939',
'info_dict': {
'id': 'n431456',
'ext': 'mp4',
'title': "Volkswagen U.S. Chief: We 'Totally Screwed Up'",
'description': 'md5:d22d1281a24f22ea0880741bb4dd6301',
'upload_date': '20150922',
'timestamp': 1442917800,
'duration': 37.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/x_lon_vwhorn_150922.jpg',
},
}, {
'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
'md5': 'ffb59bcf0733dc3c7f0ace907f5e3939',
'info_dict': {
'id': 'n431456',
'ext': 'mp4',
'title': "Volkswagen U.S. Chief: We 'Totally Screwed Up'",
'description': 'md5:d22d1281a24f22ea0880741bb4dd6301',
'upload_date': '20150922',
'timestamp': 1442917800,
'duration': 37.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/x_lon_vwhorn_150922.jpg',
},
{
'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
'md5': '693d1fa21d23afcc9b04c66b227ed9ff',
'info_dict': {
'id': '669831235788',
'ext': 'mp4',
'title': 'See the aurora borealis from space in stunning new NASA video',
'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
'upload_date': '20160420',
'timestamp': 1461152093,
'duration': 69.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/201604/2016-04-20T11-35-09-133Z--1280x720.jpg',
},
}, {
'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
'md5': '693d1fa21d23afcc9b04c66b227ed9ff',
'info_dict': {
'id': '669831235788',
'ext': 'mp4',
'title': 'See the aurora borealis from space in stunning new NASA video',
'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
'upload_date': '20160420',
'timestamp': 1461152093,
'duration': 69.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/201604/2016-04-20T11-35-09-133Z--1280x720.jpg',
},
{
'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924',
'md5': '6d236bf4f3dddc226633ce6e2c3f814d',
'info_dict': {
'id': '314487875924',
'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',
'duration': 940.0,
},
'skip': 'Invalid URL',
}, {
'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924',
'md5': '6d236bf4f3dddc226633ce6e2c3f814d',
'info_dict': {
'id': '314487875924',
'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',
'duration': 940.0,
},
{
'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952',
'only_matching': True,
'skip': 'Invalid URL',
}, {
'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,
}]
_WEBPAGE_TESTS = [{
'url': 'http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html',
'info_dict': {
'id': 'x_dtl_oa_LettermanliftPR_160608',
'ext': 'mp4',
'title': 'David Letterman: A Preview',
},
{
# 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,
},
]
'skip': 'Site changed',
}]
def _real_extract(self, url):
video_id = self._match_id(url)
@ -602,7 +606,6 @@ class NBCOlympicsIE(InfoExtractor):
_VALID_URL = r'https?://www\.nbcolympics\.com/videos?/(?P<id>[0-9a-z-]+)'
_TESTS = [{
# Geo-restricted to US
'url': 'https://www.nbcolympics.com/videos/watch-final-minutes-team-usas-mens-basketball-gold',
'info_dict': {
'id': 'SAwGfPlQ1q01',
@ -610,13 +613,14 @@ class NBCOlympicsIE(InfoExtractor):
'display_id': 'watch-final-minutes-team-usas-mens-basketball-gold',
'title': 'Watch the final minutes of Team USA\'s men\'s basketball gold',
'description': 'md5:f704f591217305c9559b23b877aa8d31',
'episode': 'Watch the final minutes of Team USA\'s men\'s basketball gold',
'uploader': 'NBCU-SPORTS',
'duration': 387.053,
'thumbnail': r're:https://.+/.+\.jpg',
'chapters': [],
'thumbnail': r're:https?://.+\.jpg',
'timestamp': 1723346984,
'upload_date': '20240811',
},
'skip': 'Geo-restricted to US',
}, {
'url': 'http://www.nbcolympics.com/video/justin-roses-son-leo-was-tears-after-his-dad-won-gold',
'only_matching': True,
@ -652,33 +656,31 @@ class NBCOlympicsStreamIE(AdobePassIE):
_WORKING = False
IE_NAME = 'nbcolympics:stream'
_VALID_URL = r'https?://stream\.nbcolympics\.com/(?P<id>[0-9a-z-]+)'
_TESTS = [
{
'note': 'Tokenized m3u8 source URL',
'url': 'https://stream.nbcolympics.com/womens-soccer-group-round-11',
'info_dict': {
'id': '2019740',
'ext': 'mp4',
'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}$",
},
'params': {
'skip_download': 'm3u8',
},
'skip': 'Livestream',
}, {
'note': 'Plain m3u8 source URL',
'url': 'https://stream.nbcolympics.com/gymnastics-event-finals-mens-floor-pommel-horse-womens-vault-bars',
'info_dict': {
'id': '2021729',
'ext': 'mp4',
'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}$',
},
'params': {
'skip_download': 'm3u8',
},
'skip': 'Livestream',
_TESTS = [{
'note': 'Tokenized m3u8 source URL',
'url': 'https://stream.nbcolympics.com/womens-soccer-group-round-11',
'info_dict': {
'id': '2019740',
'ext': 'mp4',
'title': r"re:Women's Group Stage - Netherlands vs\. Brazil [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$",
},
]
'params': {
'skip_download': 'm3u8',
},
'skip': 'Livestream',
}, {
'note': 'Plain m3u8 source URL',
'url': 'https://stream.nbcolympics.com/gymnastics-event-finals-mens-floor-pommel-horse-womens-vault-bars',
'info_dict': {
'id': '2021729',
'ext': 'mp4',
'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}$',
},
'params': {
'skip_download': 'm3u8',
},
'skip': 'Livestream',
}]
def _real_extract(self, url):
display_id = self._match_id(url)
@ -758,9 +760,7 @@ class NBCStationsIE(InfoExtractor):
'channel_id': 'KNBC',
'channel': 'nbclosangeles',
},
'params': {
'skip_download': 'm3u8',
},
'skip': 'Site changed',
}, {
'url': 'https://www.telemundoarizona.com/responde/huracan-complica-reembolso-para-televidente-de-tucson/2247002/',
'info_dict': {