[nbc] fix extraction for msnbc videos(fixes #11466)

This commit is contained in:
Remita Amine 2016-12-17 18:11:13 +01:00
parent 04bf59ff64
commit 6e416b210c

View file

@ -9,6 +9,7 @@ from ..utils import (
lowercase_escape, lowercase_escape,
smuggle_url, smuggle_url,
unescapeHTML, unescapeHTML,
update_url_query,
) )
@ -208,7 +209,7 @@ class NBCNewsIE(ThePlatformIE):
'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880', 'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
'md5': 'af1adfa51312291a017720403826bb64', 'md5': 'af1adfa51312291a017720403826bb64',
'info_dict': { 'info_dict': {
'id': '269389891880', 'id': 'p_tweet_snow_140529',
'ext': 'mp4', 'ext': 'mp4',
'title': 'How Twitter Reacted To The Snowden Interview', 'title': 'How Twitter Reacted To The Snowden Interview',
'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64', 'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
@ -232,7 +233,7 @@ class NBCNewsIE(ThePlatformIE):
'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844', 'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
'md5': '73135a2e0ef819107bbb55a5a9b2a802', 'md5': '73135a2e0ef819107bbb55a5a9b2a802',
'info_dict': { 'info_dict': {
'id': '394064451844', 'id': 'nn_netcast_150204',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Nightly News with Brian Williams Full Broadcast (February 4)', 'title': 'Nightly News with Brian Williams Full Broadcast (February 4)',
'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5', 'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
@ -245,7 +246,7 @@ class NBCNewsIE(ThePlatformIE):
'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456', 'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
'md5': 'a49e173825e5fcd15c13fc297fced39d', 'md5': 'a49e173825e5fcd15c13fc297fced39d',
'info_dict': { 'info_dict': {
'id': '529953347624', 'id': 'x_lon_vwhorn_150922',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Volkswagen U.S. Chief:\xa0 We Have Totally Screwed Up', 'title': 'Volkswagen U.S. Chief:\xa0 We Have Totally Screwed Up',
'description': 'md5:c8be487b2d80ff0594c005add88d8351', 'description': 'md5:c8be487b2d80ff0594c005add88d8351',
@ -258,7 +259,7 @@ class NBCNewsIE(ThePlatformIE):
'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788', 'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
'md5': '118d7ca3f0bea6534f119c68ef539f71', 'md5': '118d7ca3f0bea6534f119c68ef539f71',
'info_dict': { 'info_dict': {
'id': '669831235788', 'id': 'tdy_al_space_160420',
'ext': 'mp4', 'ext': 'mp4',
'title': 'See the aurora borealis from space in stunning new NASA video', 'title': 'See the aurora borealis from space in stunning new NASA video',
'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1', 'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
@ -271,7 +272,7 @@ class NBCNewsIE(ThePlatformIE):
'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924', 'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924',
'md5': '6d236bf4f3dddc226633ce6e2c3f814d', 'md5': '6d236bf4f3dddc226633ce6e2c3f814d',
'info_dict': { 'info_dict': {
'id': '314487875924', 'id': 'n_hayes_Aimm_140801_272214',
'ext': 'mp4', 'ext': 'mp4',
'title': 'The chaotic GOP immigration vote', '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.', '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.',
@ -279,7 +280,6 @@ class NBCNewsIE(ThePlatformIE):
'timestamp': 1406937606, 'timestamp': 1406937606,
'upload_date': '20140802', 'upload_date': '20140802',
'uploader': 'NBCU-NEWS', 'uploader': 'NBCU-NEWS',
'categories': ['MSNBC/Topics/Franchise/Best of last night', 'MSNBC/Topics/General/Congress'],
}, },
}, },
{ {
@ -311,28 +311,41 @@ class NBCNewsIE(ThePlatformIE):
else: else:
# "feature" and "nightly-news" pages use theplatform.com # "feature" and "nightly-news" pages use theplatform.com
video_id = mobj.group('mpx_id') video_id = mobj.group('mpx_id')
if not video_id.isdigit(): webpage = self._download_webpage(url, video_id)
webpage = self._download_webpage(url, video_id)
info = None filter_param = 'byId'
bootstrap_json = self._search_regex( bootstrap_json = self._search_regex(
[r'(?m)(?:var\s+(?:bootstrapJson|playlistData)|NEWS\.videoObj)\s*=\s*({.+});?\s*$', [r'(?m)(?:var\s+(?:bootstrapJson|playlistData)|NEWS\.videoObj)\s*=\s*({.+});?\s*$',
r'videoObj\s*:\s*({.+})', r'data-video="([^"]+)"'], r'videoObj\s*:\s*({.+})', r'data-video="([^"]+)"',
webpage, 'bootstrap json', default=None) r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);'],
webpage, 'bootstrap json', default=None)
if bootstrap_json:
bootstrap = self._parse_json( bootstrap = self._parse_json(
bootstrap_json, video_id, transform_source=unescapeHTML) bootstrap_json, video_id, transform_source=unescapeHTML)
info = None
if 'results' in bootstrap: if 'results' in bootstrap:
info = bootstrap['results'][0]['video'] info = bootstrap['results'][0]['video']
elif 'video' in bootstrap: elif 'video' in bootstrap:
info = bootstrap['video'] info = bootstrap['video']
elif 'msnbcVideoInfo' in bootstrap:
info = bootstrap['msnbcVideoInfo']['meta']
elif 'msnbcThePlatform' in bootstrap:
info = bootstrap['msnbcThePlatform']['videoPlayer']['video']
else: else:
info = bootstrap info = bootstrap
video_id = info['mpxId']
if 'guid' in info:
video_id = info['guid']
filter_param = 'byGuid'
elif 'mpxId' in info:
video_id = info['mpxId']
return { return {
'_type': 'url_transparent', '_type': 'url_transparent',
'id': video_id, 'id': video_id,
# http://feed.theplatform.com/f/2E2eJC/nbcnews also works # http://feed.theplatform.com/f/2E2eJC/nbcnews also works
'url': 'http://feed.theplatform.com/f/2E2eJC/nnd_NBCNews?byId=%s' % video_id, 'url': update_url_query('http://feed.theplatform.com/f/2E2eJC/nnd_NBCNews', {filter_param: video_id}),
'ie_key': 'ThePlatformFeed', 'ie_key': 'ThePlatformFeed',
} }