yt_dlp/extractor/cnn.py

   1 from .common import InfoExtractor
   2 from .turner import TurnerBaseIE
   3 from ..utils import merge_dicts, try_call, url_basename
   4
   5
   6 class CNNIE(TurnerBaseIE):
   7     _VALID_URL = r'''(?x)https?://(?:(?P<sub_domain>edition|www|money)\.)?cnn\.com/(?:video/(?:data/.+?|\?)/)?videos?/
   8         (?P<path>.+?/(?P<title>[^/]+?)(?:\.(?:[a-z\-]+)|(?=&)))'''
   9
  10     _TESTS = [{
  11         'url': 'http://edition.cnn.com/video/?/video/sports/2013/06/09/nadal-1-on-1.cnn',
  12         'md5': '3e6121ea48df7e2259fe73a0628605c4',
  13         'info_dict': {
  14             'id': 'sports/2013/06/09/nadal-1-on-1.cnn',
  15             'ext': 'mp4',
  16             'title': 'Nadal wins 8th French Open title',
  17             'description': 'World Sport\'s Amanda Davies chats with 2013 French Open champion Rafael Nadal.',
  18             'duration': 135,
  19             'upload_date': '20130609',
  20         },
  21         'expected_warnings': ['Failed to download m3u8 information'],
  22     }, {
  23         'url': 'http://edition.cnn.com/video/?/video/us/2013/08/21/sot-student-gives-epic-speech.georgia-institute-of-technology&utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+rss%2Fcnn_topstories+%28RSS%3A+Top+Stories%29',
  24         'md5': 'b5cc60c60a3477d185af8f19a2a26f4e',
  25         'info_dict': {
  26             'id': 'us/2013/08/21/sot-student-gives-epic-speech.georgia-institute-of-technology',
  27             'ext': 'mp4',
  28             'title': "Student's epic speech stuns new freshmen",
  29             'description': 'A Georgia Tech student welcomes the incoming freshmen with an epic speech backed by music from "2001: A Space Odyssey."',
  30             'upload_date': '20130821',
  31         },
  32         'expected_warnings': ['Failed to download m3u8 information'],
  33     }, {
  34         'url': 'http://www.cnn.com/video/data/2.0/video/living/2014/12/22/growing-america-nashville-salemtown-board-episode-1.hln.html',
  35         'md5': 'f14d02ebd264df951feb2400e2c25a1b',
  36         'info_dict': {
  37             'id': 'living/2014/12/22/growing-america-nashville-salemtown-board-episode-1.hln',
  38             'ext': 'mp4',
  39             'title': 'Nashville Ep. 1: Hand crafted skateboards',
  40             'description': 'md5:e7223a503315c9f150acac52e76de086',
  41             'upload_date': '20141222',
  42         },
  43         'expected_warnings': ['Failed to download m3u8 information'],
  44     }, {
  45         'url': 'http://money.cnn.com/video/news/2016/08/19/netflix-stunning-stats.cnnmoney/index.html',
  46         'md5': '52a515dc1b0f001cd82e4ceda32be9d1',
  47         'info_dict': {
  48             'id': '/video/news/2016/08/19/netflix-stunning-stats.cnnmoney',
  49             'ext': 'mp4',
  50             'title': '5 stunning stats about Netflix',
  51             'description': 'Did you know that Netflix has more than 80 million members? Here are five facts about the online video distributor that you probably didn\'t know.',
  52             'upload_date': '20160819',
  53         },
  54         'params': {
  55             # m3u8 download
  56             'skip_download': True,
  57         },
  58     }, {
  59         'url': 'http://cnn.com/video/?/video/politics/2015/03/27/pkg-arizona-senator-church-attendance-mandatory.ktvk',
  60         'only_matching': True,
  61     }, {
  62         'url': 'http://cnn.com/video/?/video/us/2015/04/06/dnt-baker-refuses-anti-gay-order.wkmg',
  63         'only_matching': True,
  64     }, {
  65         'url': 'http://edition.cnn.com/videos/arts/2016/04/21/olympic-games-cultural-a-z-brazil.cnn',
  66         'only_matching': True,
  67     }]
  68
  69     _CONFIG = {
  70         # http://edition.cnn.com/.element/apps/cvp/3.0/cfg/spider/cnn/expansion/config.xml
  71         'edition': {
  72             'data_src': 'http://edition.cnn.com/video/data/3.0/video/%s/index.xml',
  73             'media_src': 'http://pmd.cdn.turner.com/cnn/big',
  74         },
  75         # http://money.cnn.com/.element/apps/cvp2/cfg/config.xml
  76         'money': {
  77             'data_src': 'http://money.cnn.com/video/data/4.0/video/%s.xml',
  78             'media_src': 'http://ht3.cdn.turner.com/money/big',
  79         },
  80     }
  81
  82     def _extract_timestamp(self, video_data):
  83         # TODO: fix timestamp extraction
  84         return None
  85
  86     def _real_extract(self, url):
  87         sub_domain, path, page_title = self._match_valid_url(url).groups()
  88         if sub_domain not in ('money', 'edition'):
  89             sub_domain = 'edition'
  90         config = self._CONFIG[sub_domain]
  91         return self._extract_cvp_info(
  92             config['data_src'] % path, page_title, {
  93                 'default': {
  94                     'media_src': config['media_src'],
  95                 },
  96                 'f4m': {
  97                     'host': 'cnn-vh.akamaihd.net',
  98                 },
  99             })
 100
 101
 102 class CNNBlogsIE(InfoExtractor):
 103     _VALID_URL = r'https?://[^\.]+\.blogs\.cnn\.com/.+'
 104     _TEST = {
 105         'url': 'http://reliablesources.blogs.cnn.com/2014/02/09/criminalizing-journalism/',
 106         'md5': '3e56f97b0b6ffb4b79f4ea0749551084',
 107         'info_dict': {
 108             'id': 'bestoftv/2014/02/09/criminalizing-journalism.cnn',
 109             'ext': 'mp4',
 110             'title': 'Criminalizing journalism?',
 111             'description': 'Glenn Greenwald responds to comments made this week on Capitol Hill that journalists could be criminal accessories.',
 112             'upload_date': '20140209',
 113         },
 114         'expected_warnings': ['Failed to download m3u8 information'],
 115         'add_ie': ['CNN'],
 116     }
 117
 118     def _real_extract(self, url):
 119         webpage = self._download_webpage(url, url_basename(url))
 120         cnn_url = self._html_search_regex(r'data-url="(.+?)"', webpage, 'cnn url')
 121         return self.url_result(cnn_url, CNNIE.ie_key())
 122
 123
 124 class CNNArticleIE(InfoExtractor):
 125     _VALID_URL = r'https?://(?:(?:edition|www)\.)?cnn\.com/(?!videos?/)'
 126     _TEST = {
 127         'url': 'http://www.cnn.com/2014/12/21/politics/obama-north-koreas-hack-not-war-but-cyber-vandalism/',
 128         'md5': '689034c2a3d9c6dc4aa72d65a81efd01',
 129         'info_dict': {
 130             'id': 'bestoftv/2014/12/21/ip-north-korea-obama.cnn',
 131             'ext': 'mp4',
 132             'title': 'Obama: Cyberattack not an act of war',
 133             'description': 'md5:0a802a40d2376f60e6b04c8d5bcebc4b',
 134             'upload_date': '20141221',
 135         },
 136         'expected_warnings': ['Failed to download m3u8 information'],
 137         'add_ie': ['CNN'],
 138     }
 139
 140     def _real_extract(self, url):
 141         webpage = self._download_webpage(url, url_basename(url))
 142         cnn_url = self._html_search_regex(r"video:\s*'([^']+)'", webpage, 'cnn url')
 143         return self.url_result('http://cnn.com/video/?/video/' + cnn_url, CNNIE.ie_key())
 144
 145
 146 class CNNIndonesiaIE(InfoExtractor):
 147     _VALID_URL = r'https?://www\.cnnindonesia\.com/[\w-]+/(?P<upload_date>\d{8})\d+-\d+-(?P<id>\d+)/(?P<display_id>[\w-]+)'
 148     _TESTS = [{
 149         'url': 'https://www.cnnindonesia.com/ekonomi/20220909212635-89-845885/alasan-harga-bbm-di-indonesia-masih-disubsidi',
 150         'info_dict': {
 151             'id': '845885',
 152             'ext': 'mp4',
 153             'description': 'md5:e7954bfa6f1749bc9ef0c079a719c347',
 154             'upload_date': '20220909',
 155             'title': 'Alasan Harga BBM di Indonesia Masih Disubsidi',
 156             'timestamp': 1662859088,
 157             'duration': 120.0,
 158             'thumbnail': r're:https://akcdn\.detik\.net\.id/visual/2022/09/09/thumbnail-ekopedia-alasan-harga-bbm-disubsidi_169\.jpeg',
 159             'tags': ['ekopedia', 'subsidi bbm', 'subsidi', 'bbm', 'bbm subsidi', 'harga pertalite naik'],
 160             'age_limit': 0,
 161             'release_timestamp': 1662859088,
 162             'release_date': '20220911',
 163             'uploader': 'Asfahan Yahsyi',
 164         },
 165     }, {
 166         'url': 'https://www.cnnindonesia.com/internasional/20220911104341-139-846189/video-momen-charles-disambut-meriah-usai-dilantik-jadi-raja-inggris',
 167         'info_dict': {
 168             'id': '846189',
 169             'ext': 'mp4',
 170             'upload_date': '20220911',
 171             'duration': 76.0,
 172             'timestamp': 1662869995,
 173             'description': 'md5:ece7b003b3ee7d81c6a5cfede7d5397d',
 174             'thumbnail': r're:https://akcdn\.detik\.net\.id/visual/2022/09/11/thumbnail-video-1_169\.jpeg',
 175             'title': 'VIDEO: Momen Charles Disambut Meriah usai Dilantik jadi Raja Inggris',
 176             'tags': ['raja charles', 'raja charles iii', 'ratu elizabeth', 'ratu elizabeth meninggal dunia', 'raja inggris', 'inggris'],
 177             'age_limit': 0,
 178             'release_date': '20220911',
 179             'uploader': 'REUTERS',
 180             'release_timestamp': 1662869995,
 181         },
 182     }]
 183
 184     def _real_extract(self, url):
 185         upload_date, video_id, display_id = self._match_valid_url(url).group('upload_date', 'id', 'display_id')
 186         webpage = self._download_webpage(url, display_id)
 187
 188         json_ld_list = list(self._yield_json_ld(webpage, display_id))
 189         json_ld_data = self._json_ld(json_ld_list, display_id)
 190         embed_url = next(
 191             json_ld.get('embedUrl') for json_ld in json_ld_list if json_ld.get('@type') == 'VideoObject')
 192
 193         return merge_dicts(json_ld_data, {
 194             '_type': 'url_transparent',
 195             'url': embed_url,
 196             'upload_date': upload_date,
 197             'tags': try_call(lambda: self._html_search_meta('keywords', webpage).split(', ')),
 198         })