yt_dlp/extractor/tmz.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     ExtractorError,
   6     get_element_by_attribute,
   7 )
   8
   9
  10 class TMZIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?tmz\.com/.*'
  12     _TESTS = [
  13         {
  14             'url': 'http://www.tmz.com/videos/0-cegprt2p/',
  15             'info_dict': {
  16                 'id': 'http://www.tmz.com/videos/0-cegprt2p/',
  17                 'ext': 'mp4',
  18                 'title': 'No Charges Against Hillary Clinton? Harvey Says It Ain\'t Over Yet',
  19                 'description': 'Harvey talks about Director Comey’s decision not to prosecute Hillary Clinton.',
  20                 'timestamp': 1467831837,
  21                 'uploader': 'TMZ Staff',
  22                 'upload_date': '20160706',
  23                 'thumbnail': 'https://imagez.tmz.com/image/5e/4by3/2016/07/06/5eea7dc01baa5c2e83eb06930c170e46_xl.jpg',
  24                 'duration': 772.0,
  25             },
  26         },
  27         {
  28             'url': 'https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/',
  29             'info_dict': {
  30                 'id': 'https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/',
  31                 'ext': 'mp4',
  32                 'title': 'Angry Bagel Shop Guy Says He Doesn\'t Trust Women',
  33                 'description': 'The enraged man who went viral for ranting about women on dating sites before getting ragdolled in a bagel shop is defending his misogyny ... he says it\'s women\'s fault in the first place.',
  34                 'timestamp': 1562889485,
  35                 'uploader': 'TMZ Staff',
  36                 'upload_date': '20190711',
  37                 'thumbnail': 'https://imagez.tmz.com/image/a8/4by3/2019/07/12/a85480d27b2f50a7bfea2322151d67a5_xl.jpg',
  38                 'duration': 123.0,
  39             },
  40         },
  41         {
  42             'url': 'http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert',
  43             'md5': '5429c85db8bde39a473a56ca8c4c5602',
  44             'info_dict': {
  45                 'id': 'http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert',
  46                 'ext': 'mp4',
  47                 'title': 'Bobby Brown Tells Crowd ... Bobbi Kristina is Awake',
  48                 'description': 'Bobby Brown stunned his audience during a concert Saturday night, when he told the crowd, "Bobbi is awake.  She\'s watching me."',
  49                 'timestamp': 1429467813,
  50                 'uploader': 'TMZ Staff',
  51                 'upload_date': '20150419',
  52                 'duration': 29.0,
  53                 'thumbnail': 'https://imagez.tmz.com/image/15/4by3/2015/04/20/1539c7ae136359fc979236fa6a9449dd_xl.jpg',
  54             },
  55         },
  56         {
  57             'url': 'http://www.tmz.com/2015/09/19/patti-labelle-concert-fan-stripping-kicked-out-nicki-minaj/',
  58             'info_dict': {
  59                 'id': 'http://www.tmz.com/2015/09/19/patti-labelle-concert-fan-stripping-kicked-out-nicki-minaj/',
  60                 'ext': 'mp4',
  61                 'title': 'Patti LaBelle -- Goes Nuclear On Stripping Fan',
  62                 'description': 'Patti LaBelle made it known loud and clear last night ... NO '
  63                 'ONE gets on her stage and strips down.',
  64                 'timestamp': 1442683746,
  65                 'uploader': 'TMZ Staff',
  66                 'upload_date': '20150919',
  67                 'duration': 104.0,
  68                 'thumbnail': 'https://imagez.tmz.com/image/5e/4by3/2015/09/20/5e57d7575062528082994e18ac3f0f48_xl.jpg',
  69             },
  70         },
  71         {
  72             'url': 'http://www.tmz.com/2016/01/28/adam-silver-sting-drake-blake-griffin/',
  73             'info_dict': {
  74                 'id': 'http://www.tmz.com/2016/01/28/adam-silver-sting-drake-blake-griffin/',
  75                 'ext': 'mp4',
  76                 'title': 'NBA\'s Adam Silver -- Blake Griffin\'s a Great Guy ... He\'ll Learn from This',
  77                 'description': 'Two pretty parts of this video with NBA Commish Adam Silver.',
  78                 'timestamp': 1454010989,
  79                 'uploader': 'TMZ Staff',
  80                 'upload_date': '20160128',
  81                 'duration': 59.0,
  82                 'thumbnail': 'https://imagez.tmz.com/image/38/4by3/2016/01/29/3856e83e0beb57059ec412122b842fb1_xl.jpg',
  83             },
  84         },
  85         {
  86             'url': 'http://www.tmz.com/2016/10/27/donald-trump-star-vandal-arrested-james-otis/',
  87             'info_dict': {
  88                 'id': 'http://www.tmz.com/2016/10/27/donald-trump-star-vandal-arrested-james-otis/',
  89                 'ext': 'mp4',
  90                 'title': 'Trump Star Vandal -- I\'m Not Afraid of Donald or the Cops!',
  91                 'description': 'James Otis is the the guy who took a pickaxe to Donald Trump\'s star on the Walk of Fame, and he tells TMZ .. he\'s ready and willing to go to jail for the crime.',
  92                 'timestamp': 1477500095,
  93                 'uploader': 'TMZ Staff',
  94                 'upload_date': '20161026',
  95                 'thumbnail': 'https://imagez.tmz.com/image/0d/4by3/2016/10/27/0d904814d4a75dcf9cc3b8cfd1edc1a3_xl.jpg',
  96                 'duration': 128.0,
  97             },
  98         },
  99         {
 100             'url': 'https://www.tmz.com/videos/2020-10-31-103120-beverly-hills-protest-4878209/',
 101             'info_dict': {
 102                 'id': 'https://www.tmz.com/videos/2020-10-31-103120-beverly-hills-protest-4878209/',
 103                 'ext': 'mp4',
 104                 'title': 'Cops Use Billy Clubs Against Pro-Trump and Anti-Fascist '
 105                 'Demonstrators',
 106                 'description': 'Beverly Hills may be an omen of what\'s coming next week, '
 107                 'because things got crazy on the streets and cops started '
 108                 'swinging their billy clubs at both Anti-Fascist and Pro-Trump '
 109                 'demonstrators.',
 110                 'timestamp': 1604182772,
 111                 'uploader': 'TMZ Staff',
 112                 'upload_date': '20201031',
 113                 'duration': 96.0,
 114                 'thumbnail': 'https://imagez.tmz.com/image/f3/4by3/2020/10/31/f37bd5a8aef84497866f425130c58be3_xl.jpg',
 115             },
 116         },
 117         {
 118             'url': 'https://www.tmz.com/2020/11/05/gervonta-davis-car-crash-hit-and-run-police/',
 119             'info_dict': {
 120                 'id': 'Dddb6IGe-ws',
 121                 'ext': 'mp4',
 122                 'title': 'SICK LAMBO GERVONTA DAVIS IN HIS NEW RIDE RIGHT AFTER KO AFTER LEO  EsNews Boxing',
 123                 'uploader': 'ESNEWS',
 124                 'description': 'md5:49675bc58883ccf80474b8aa701e1064',
 125                 'upload_date': '20201102',
 126                 'uploader_id': '@ESNEWS',
 127                 'uploader_url': 'https://www.youtube.com/@ESNEWS',
 128                 'like_count': int,
 129                 'channel_id': 'UCI-Oq7oFGakzSzHFlTtsUsQ',
 130                 'channel': 'ESNEWS',
 131                 'view_count': int,
 132                 'duration': 225,
 133                 'live_status': 'not_live',
 134                 'thumbnail': 'https://i.ytimg.com/vi_webp/Dddb6IGe-ws/maxresdefault.webp',
 135                 'channel_url': 'https://www.youtube.com/channel/UCI-Oq7oFGakzSzHFlTtsUsQ',
 136                 'channel_follower_count': int,
 137                 'playable_in_embed': True,
 138                 'categories': ['Sports'],
 139                 'age_limit': 0,
 140                 'tags': 'count:10',
 141                 'availability': 'public',
 142                 'comment_count': int,
 143             },
 144         },
 145         {
 146             'url': 'https://www.tmz.com/2020/11/19/conor-mcgregor-dustin-poirier-contract-fight-ufc-257-fight-island/',
 147             'info_dict': {
 148                 'id': '1329448013937471491',
 149                 'ext': 'mp4',
 150                 'title': 'The Mac Life - BREAKING: Conor McGregor (@thenotoriousmma) has signed his bout agreement for his rematch with Dustin Poirier for January 23.',
 151                 'uploader': 'The Mac Life',
 152                 'description': 'md5:56e6009bbc3d12498e10d08a8e1f1c69',
 153                 'upload_date': '20201119',
 154                 'display_id': '1329450007125225473',
 155                 'uploader_id': 'TheMacLife',
 156                 'timestamp': 1605800556,
 157                 'thumbnail': 'https://pbs.twimg.com/media/EnMmfT8XYAExgxJ.jpg?name=small',
 158                 'like_count': int,
 159                 'duration': 11.812,
 160                 'uploader_url': 'https://twitter.com/TheMacLife',
 161                 'age_limit': 0,
 162                 'repost_count': int,
 163                 'tags': [],
 164                 'comment_count': int,
 165             },
 166         },
 167     ]
 168
 169     def _real_extract(self, url):
 170         webpage = self._download_webpage(url, url)
 171         jsonld = self._search_json_ld(webpage, url)
 172         if not jsonld or 'url' not in jsonld:
 173             # try to extract from YouTube Player API
 174             # see https://developers.google.com/youtube/iframe_api_reference#Video_Queueing_Functions
 175             match_obj = re.search(r'\.cueVideoById\(\s*(?P<quote>[\'"])(?P<id>.*?)(?P=quote)', webpage)
 176             if match_obj:
 177                 return self.url_result(match_obj.group('id'))
 178             # try to extract from twitter
 179             blockquote_el = get_element_by_attribute('class', 'twitter-tweet', webpage)
 180             if blockquote_el:
 181                 matches = re.findall(
 182                     r'<a[^>]+href=\s*(?P<quote>[\'"])(?P<link>.*?)(?P=quote)',
 183                     blockquote_el)
 184                 if matches:
 185                     for _, match in matches:
 186                         if '/status/' in match:
 187                             return self.url_result(match)
 188             raise ExtractorError('No video found!')
 189         if id not in jsonld:
 190             jsonld['id'] = url
 191         return jsonld