2 from __future__ import unicode_literals
4 from .common import InfoExtractor
5 from .jwplatform import JWPlatformIE
6 from .kaltura import KalturaIE
13 class TMZIE(InfoExtractor):
14 _VALID_URL = r'https?://(?:www\.)?tmz\.com/videos/(?P<id>[^/?#&]+)'
16 'url': 'http://www.tmz.com/videos/0-cegprt2p/',
17 'md5': '31f9223e20eef55954973359afa61a20',
21 'title': "No Charges Against Hillary Clinton? Harvey Says It Ain't Over Yet",
22 'description': 'md5:b714359fc18607715ebccbd2da8ff488',
23 'timestamp': 1467831837,
24 'upload_date': '20160706',
26 'add_ie': [JWPlatformIE.ie_key()],
28 'url': 'http://www.tmz.com/videos/0_okj015ty/',
29 'only_matching': True,
31 'url': 'https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/',
32 'only_matching': True,
34 'url': 'https://www.tmz.com/videos/2021-02-19-021921-floyd-mayweather-1043872/',
35 'only_matching': True,
38 def _real_extract(self, url):
39 video_id = self._match_id(url).replace('-', '_')
41 webpage = self._download_webpage(url, video_id, fatal=False)
43 tmz_video_id = self._search_regex(
44 r'nodeRef\s*:\s*["\']tmz:video:([\da-fA-F]{8}-[\da-fA-F]{4}-[\da-fA-F]{4}-[\da-fA-F]{4}-[\da-fA-F]{12})',
45 webpage, 'video id', default=None)
46 video = self._download_json(
47 'https://www.tmz.com/_/video/%s' % tmz_video_id, video_id,
50 message = video['message']
52 '_type': 'url_transparent',
53 'title': message.get('title'),
54 'description': message.get('description'),
55 'timestamp': unified_timestamp(message.get('published_at')),
56 'duration': int_or_none(message.get('duration')),
58 jwplatform_id = message.get('jwplayer_media_id')
61 'url': 'jwplatform:%s' % jwplatform_id,
62 'ie_key': JWPlatformIE.ie_key(),
65 kaltura_entry_id = message.get('kaltura_entry_id') or video_id
66 kaltura_partner_id = message.get('kaltura_partner_id') or '591531'
68 'url': 'kaltura:%s:%s' % (kaltura_partner_id, kaltura_entry_id),
69 'ie_key': KalturaIE.ie_key(),
73 return self.url_result(
74 'kaltura:591531:%s' % video_id, KalturaIE.ie_key(), video_id)
77 class TMZArticleIE(InfoExtractor):
78 _VALID_URL = r'https?://(?:www\.)?tmz\.com/\d{4}/\d{2}/\d{2}/(?P<id>[^/?#&]+)'
80 'url': 'http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert',
84 'title': 'Bobby Brown Tells Crowd ... Bobbi Kristina is Awake',
85 'description': 'Bobby Brown stunned his audience during a concert Saturday night, when he told the crowd, "Bobbi is awake. She\'s watching me."',
86 'timestamp': 1429466400,
87 'upload_date': '20150419',
90 'skip_download': True,
92 'add_ie': [JWPlatformIE.ie_key()],
95 def _real_extract(self, url):
96 video_id = self._match_id(url)
98 webpage = self._download_webpage(url, video_id)
100 tmz_url = self._search_regex(
101 r'clickLink\s*\(\s*["\'](?P<url>%s)' % TMZIE._VALID_URL, webpage,
102 'video id', default=None, group='url')
104 return self.url_result(tmz_url, ie=TMZIE.ie_key())
106 embedded_video_info = self._parse_json(self._html_search_regex(
107 r'tmzVideoEmbed\(({.+?})\);', webpage, 'embedded video info'),
109 return self.url_result(
110 'http://www.tmz.com/videos/%s/' % embedded_video_info['id'],