yt_dlp/extractor/foxnews.py

   1 import re
   2
   3 from .amp import AMPIE
   4 from .common import InfoExtractor
   5
   6
   7 class FoxNewsIE(AMPIE):
   8     IE_NAME = 'foxnews'
   9     IE_DESC = 'Fox News and Fox Business Video'
  10     _VALID_URL = r'https?://(?P<host>video\.(?:insider\.)?fox(?:news|business)\.com)/v/(?:video-embed\.html\?video_id=)?(?P<id>\d+)'
  11     _TESTS = [
  12         {
  13             'url': 'http://video.foxnews.com/v/3937480/frozen-in-time/#sp=show-clips',
  14             'md5': '32aaded6ba3ef0d1c04e238d01031e5e',
  15             'info_dict': {
  16                 'id': '3937480',
  17                 'ext': 'flv',
  18                 'title': 'Frozen in Time',
  19                 'description': '16-year-old girl is size of toddler',
  20                 'duration': 265,
  21                 'timestamp': 1304411491,
  22                 'upload_date': '20110503',
  23                 'thumbnail': r're:^https?://.*\.jpg$',
  24             },
  25         },
  26         {
  27             'url': 'http://video.foxnews.com/v/3922535568001/rep-luis-gutierrez-on-if-obamas-immigration-plan-is-legal/#sp=show-clips',
  28             'md5': '5846c64a1ea05ec78175421b8323e2df',
  29             'info_dict': {
  30                 'id': '3922535568001',
  31                 'ext': 'mp4',
  32                 'title': "Rep. Luis Gutierrez on if Obama's immigration plan is legal",
  33                 'description': "Congressman discusses president's plan",
  34                 'duration': 292,
  35                 'timestamp': 1417662047,
  36                 'upload_date': '20141204',
  37                 'thumbnail': r're:^https?://.*\.jpg$',
  38             },
  39             'params': {
  40                 # m3u8 download
  41                 'skip_download': True,
  42             },
  43         },
  44         {
  45             'url': 'http://video.foxnews.com/v/video-embed.html?video_id=3937480&d=video.foxnews.com',
  46             'only_matching': True,
  47         },
  48         {
  49             'url': 'http://video.foxbusiness.com/v/4442309889001',
  50             'only_matching': True,
  51         },
  52         {
  53             # From http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words
  54             'url': 'http://video.insider.foxnews.com/v/video-embed.html?video_id=5099377331001&autoplay=true&share_url=http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words&share_title=Student%20Group:%20Saying%20%27Politically%20Correct,%27%20%27Trash%27%20and%20%27Lame%27%20Is%20Offensive&share=true',
  55             'only_matching': True,
  56         },
  57     ]
  58
  59     @classmethod
  60     def _extract_embed_urls(cls, url, webpage):
  61         for mobj in re.finditer(
  62                 r'''(?x)
  63                     <(?:script|(?:amp-)?iframe)[^>]+\bsrc=["\']
  64                     (?:https?:)?//video\.foxnews\.com/v/(?:video-embed\.html|embed\.js)\?
  65                     (?:[^>"\']+&)?(?:video_)?id=(?P<video_id>\d+)
  66                 ''', webpage):
  67             yield f'https://video.foxnews.com/v/video-embed.html?video_id={mobj.group("video_id")}'
  68
  69     def _real_extract(self, url):
  70         host, video_id = self._match_valid_url(url).groups()
  71
  72         info = self._extract_feed_info(
  73             'http://%s/v/feed/video/%s.js?template=fox' % (host, video_id))
  74         info['id'] = video_id
  75         return info
  76
  77
  78 class FoxNewsArticleIE(InfoExtractor):
  79     _VALID_URL = r'https?://(?:www\.)?(?:insider\.)?foxnews\.com/(?!v)([^/]+/)+(?P<id>[a-z-]+)'
  80     IE_NAME = 'foxnews:article'
  81
  82     _TESTS = [{
  83         # data-video-id
  84         'url': 'http://www.foxnews.com/politics/2016/09/08/buzz-about-bud-clinton-camp-denies-claims-wore-earpiece-at-forum.html',
  85         'md5': '83d44e1aff1433e7a29a7b537d1700b5',
  86         'info_dict': {
  87             'id': '5116295019001',
  88             'ext': 'mp4',
  89             'title': 'Trump and Clinton asked to defend positions on Iraq War',
  90             'description': 'Veterans react on \'The Kelly File\'',
  91             'timestamp': 1473301045,
  92             'upload_date': '20160908',
  93         },
  94     }, {
  95         # iframe embed
  96         'url': 'http://www.foxnews.com/us/2018/03/09/parkland-survivor-kyle-kashuv-on-meeting-trump-his-app-to-prevent-another-school-shooting.amp.html?__twitter_impression=true',
  97         'info_dict': {
  98             'id': '5748266721001',
  99             'ext': 'flv',
 100             'title': 'Kyle Kashuv has a positive message for the Trump White House',
 101             'description': 'Marjory Stoneman Douglas student disagrees with classmates.',
 102             'thumbnail': r're:^https?://.*\.jpg$',
 103             'duration': 229,
 104             'timestamp': 1520594670,
 105             'upload_date': '20180309',
 106         },
 107         'params': {
 108             'skip_download': True,
 109         },
 110     }, {
 111         'url': 'http://insider.foxnews.com/2016/08/25/univ-wisconsin-student-group-pushing-silence-certain-words',
 112         'only_matching': True,
 113     }]
 114
 115     def _real_extract(self, url):
 116         display_id = self._match_id(url)
 117         webpage = self._download_webpage(url, display_id)
 118
 119         video_id = self._html_search_regex(
 120             r'data-video-id=([\'"])(?P<id>[^\'"]+)\1',
 121             webpage, 'video ID', group='id', default=None)
 122         if video_id:
 123             return self.url_result(
 124                 'http://video.foxnews.com/v/' + video_id, FoxNewsIE.ie_key())
 125
 126         return self.url_result(
 127             next(FoxNewsIE._extract_embed_urls(url, webpage)), FoxNewsIE.ie_key())