[TMZ] Fix TMZ.com extractor

This commit is contained in:
Diego Fernando Rodríguez Varón 2020-11-01 21:25:34 -05:00
parent 764876a01f
commit 60351178a5
1 changed files with 58 additions and 15 deletions

View File

@ -5,7 +5,7 @@ from .common import InfoExtractor
class TMZIE(InfoExtractor): class TMZIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?tmz\.com/videos/(?P<id>[^/?#]+)' _VALID_URL = r'https?://(?:www\.)?tmz\.com/videos/.*(?P<id>[^/?#]{10,10})'
_TESTS = [{ _TESTS = [{
'url': 'http://www.tmz.com/videos/0_okj015ty/', 'url': 'http://www.tmz.com/videos/0_okj015ty/',
'md5': '4d22a51ef205b6c06395d8394f72d560', 'md5': '4d22a51ef205b6c06395d8394f72d560',
@ -13,14 +13,30 @@ class TMZIE(InfoExtractor):
'id': '0_okj015ty', 'id': '0_okj015ty',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Kim Kardashian\'s Boobs Unlock a Mystery!', 'title': 'Kim Kardashian\'s Boobs Unlock a Mystery!',
'description': 'Did Kim Kardasain try to one-up Khloe by one-upping Kylie??? Or is she just showing off her amazing boobs?',
'timestamp': 1394747163, 'timestamp': 1394747163,
'uploader_id': 'batchUser', 'uploader_id': 'batchUser',
'upload_date': '20140313', 'upload_date': '20140313',
} }
}, { }, {
'url': 'http://www.tmz.com/videos/0-cegprt2p/', 'url': 'http://www.tmz.com/videos/0-cegprt2p/',
'only_matching': True, 'info_dict': {
'id': '0_cegprt2p',
'ext': 'mp4',
'title': "No Charges Against Hillary Clinton? Harvey Says It Ain't Over Yet",
'timestamp': 1467831837,
'uploader_id': 'batchUser',
'upload_date': '20160706',
}
}, {
'url': 'https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/',
'info_dict': {
'id': '0_zcsejvcr',
'ext': 'mxf',
'title': "Angry Bagel Shop Guy Says He Doesn't Trust Women",
'timestamp': 1562889485,
'uploader_id': 'batchUser',
'upload_date': '20190711',
}
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -30,27 +46,54 @@ class TMZIE(InfoExtractor):
class TMZArticleIE(InfoExtractor): class TMZArticleIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?tmz\.com/\d{4}/\d{2}/\d{2}/(?P<id>[^/]+)/?' _VALID_URL = r'https?://(?:www\.)?tmz\.com/\d{4}/\d{2}/\d{2}/(?P<id>[^/]+)/?'
_TEST = { _TESTS = [{
'url': 'http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert', 'url': 'http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert',
'md5': '3316ff838ae5bb7f642537825e1e90d2', 'md5': '5429c85db8bde39a473a56ca8c4c5602',
'info_dict': { 'info_dict': {
'id': '0_6snoelag', 'id': '0_6snoelag',
'ext': 'mov', 'ext': 'mp4',
'title': 'Bobby Brown Tells Crowd ... Bobbi Kristina is Awake', 'title': 'Bobby Brown Tells Crowd ... Bobbi Kristina is Awake',
'description': 'Bobby Brown stunned his audience during a concert Saturday night, when he told the crowd, "Bobbi is awake. She\'s watching me."',
'timestamp': 1429467813, 'timestamp': 1429467813,
'upload_date': '20150419', 'upload_date': '20150419',
'uploader_id': 'batchUser', 'uploader_id': 'batchUser',
} }
} }, {
'url': 'http://www.tmz.com/2015/09/19/patti-labelle-concert-fan-stripping-kicked-out-nicki-minaj/',
'info_dict': {
'id': '0_jerz7s3l',
'ext': 'mp4',
'title': 'Patti LaBelle -- Goes Nuclear On Stripping Fan',
'timestamp': 1442683746,
'upload_date': '20150919',
'uploader_id': 'batchUser',
}
}, {
'url': 'http://www.tmz.com/2016/01/28/adam-silver-sting-drake-blake-griffin/',
'info_dict': {
'id': '0_ytz87kk7',
'ext': 'mp4',
'title': "NBA's Adam Silver -- Blake Griffin's a Great Guy ... He'll Learn from This",
'timestamp': 1454010989,
'upload_date': '20160128',
'uploader_id': 'batchUser',
}
}, {
'url': 'http://www.tmz.com/2016/10/27/donald-trump-star-vandal-arrested-james-otis/',
'info_dict': {
'id': '0_isigfatu',
'ext': 'mp4',
'title': "Trump Star Vandal -- I'm Not Afraid of Donald or the Cops!",
'timestamp': 1477500095,
'upload_date': '20161026',
'uploader_id': 'batchUser',
}
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id) webpage = self._download_webpage(url, video_id)
embedded_video_info = self._parse_json(self._html_search_regex( params = self._html_search_regex(r'TMZ.actions.clickLink\(([\s\S]+?)\)',
r'tmzVideoEmbed\(({.+?})\);', webpage, 'embedded video info'), webpage, 'embedded video info').split(',')
video_id) new_url = params[0].strip("'\"")
if new_url != url:
return self.url_result( return self.url_result(new_url)
'http://www.tmz.com/videos/%s/' % embedded_video_info['id'])