yt_dlp/extractor/bleacherreport.py

   1 from .common import InfoExtractor
   2 from .amp import AMPIE
   3 from ..utils import (
   4     ExtractorError,
   5     int_or_none,
   6     parse_iso8601,
   7     str_or_none,
   8 )
   9
  10
  11 class BleacherReportIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:www\.)?bleacherreport\.com/articles/(?P<id>\d+)'
  13     _TESTS = [{
  14         'url': 'http://bleacherreport.com/articles/2496438-fsu-stat-projections-is-jalen-ramsey-best-defensive-player-in-college-football',
  15         'md5': 'a3ffc3dc73afdbc2010f02d98f990f20',
  16         'info_dict': {
  17             'id': '2496438',
  18             'ext': 'mp4',
  19             'title': 'FSU Stat Projections: Is Jalen Ramsey Best Defensive Player in College Football?',
  20             'uploader_id': '3992341',
  21             'description': 'CFB, ACC, Florida State',
  22             'timestamp': 1434380212,
  23             'upload_date': '20150615',
  24             'uploader': 'Team Stream Now ',
  25         },
  26         'skip': 'Video removed',
  27     }, {
  28         'url': 'http://bleacherreport.com/articles/2586817-aussie-golfers-get-fright-of-their-lives-after-being-chased-by-angry-kangaroo',
  29         'md5': '6a5cd403418c7b01719248ca97fb0692',
  30         'info_dict': {
  31             'id': '2586817',
  32             'ext': 'webm',
  33             'title': 'Aussie Golfers Get Fright of Their Lives After Being Chased by Angry Kangaroo',
  34             'timestamp': 1446839961,
  35             'uploader': 'Sean Fay',
  36             'description': 'md5:b1601e2314c4d8eec23b6eafe086a757',
  37             'uploader_id': '6466954',
  38             'upload_date': '20151011',
  39         },
  40         'add_ie': ['Youtube'],
  41     }]
  42
  43     def _real_extract(self, url):
  44         article_id = self._match_id(url)
  45
  46         article_data = self._download_json('http://api.bleacherreport.com/api/v1/articles/%s' % article_id, article_id)['article']
  47
  48         thumbnails = []
  49         primary_photo = article_data.get('primaryPhoto')
  50         if primary_photo:
  51             thumbnails = [{
  52                 'url': primary_photo['url'],
  53                 'width': primary_photo.get('width'),
  54                 'height': primary_photo.get('height'),
  55             }]
  56
  57         info = {
  58             '_type': 'url_transparent',
  59             'id': article_id,
  60             'title': article_data['title'],
  61             'uploader': article_data.get('author', {}).get('name'),
  62             'uploader_id': str_or_none(article_data.get('authorId')),
  63             'timestamp': parse_iso8601(article_data.get('createdAt')),
  64             'thumbnails': thumbnails,
  65             'comment_count': int_or_none(article_data.get('commentsCount')),
  66             'view_count': int_or_none(article_data.get('hitCount')),
  67         }
  68
  69         video = article_data.get('video')
  70         if video:
  71             video_type = video['type']
  72             if video_type in ('cms.bleacherreport.com', 'vid.bleacherreport.com'):
  73                 info['url'] = 'http://bleacherreport.com/video_embed?id=%s' % video['id']
  74             elif video_type == 'youtube.com':
  75                 info['url'] = video['id']
  76             elif video_type == 'vine.co':
  77                 info['url'] = 'https://vine.co/v/%s' % video['id']
  78             else:
  79                 info['url'] = video_type + video['id']
  80             return info
  81         else:
  82             raise ExtractorError('no video in the article', expected=True)
  83
  84
  85 class BleacherReportCMSIE(AMPIE):
  86     _VALID_URL = r'https?://(?:www\.)?bleacherreport\.com/video_embed\?id=(?P<id>[0-9a-f-]{36}|\d{5})'
  87     _TESTS = [{
  88         'url': 'http://bleacherreport.com/video_embed?id=8fd44c2f-3dc5-4821-9118-2c825a98c0e1&library=video-cms',
  89         'md5': '670b2d73f48549da032861130488c681',
  90         'info_dict': {
  91             'id': '8fd44c2f-3dc5-4821-9118-2c825a98c0e1',
  92             'ext': 'mp4',
  93             'title': 'Cena vs. Rollins Would Expose the Heavyweight Division',
  94             'description': 'md5:984afb4ade2f9c0db35f3267ed88b36e',
  95             'upload_date': '20150723',
  96             'timestamp': 1437679032,
  97
  98         },
  99         'expected_warnings': [
 100             'Unable to download f4m manifest'
 101         ]
 102     }]
 103
 104     def _real_extract(self, url):
 105         video_id = self._match_id(url)
 106         info = self._extract_feed_info('http://vid.bleacherreport.com/videos/%s.akamai' % video_id)
 107         info['id'] = video_id
 108         return info