_ Git - youtube-dl/blob - youtube_dl/extractor/wwe.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_str
   5 from ..utils import urljoin
   6
   7
   8 class WWEIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:\w+\.)?wwe.com/(?:.*/)?videos/(?P<id>[\w-]+)'
  10     _TESTS = [{
  11         'url': 'https://www.wwe.com/videos/daniel-bryan-vs-andrade-cien-almas-smackdown-live-sept-4-2018',
  12         'md5': '30cbc824b51f4010ea885bfcaec76972',
  13         'info_dict': {
  14             'id': '40048199',
  15             'ext': 'mp4',
  16             'title': 'Daniel Bryan vs. Andrade "Cien" Almas: SmackDown LIVE, Sept. 4, 2018',
  17             'description': 'Still fuming after he and his wife Brie Bella were attacked by The Miz and Maryse last week, Daniel Bryan takes care of some unfinished business with Andrade "Cien" Almas.',
  18             'thumbnail': r're:^https?://.*\.jpg$',
  19         }
  20     }, {
  21         'url': 'https://de.wwe.com/videos/gran-metalik-vs-tony-nese-wwe-205-live-sept-4-2018',
  22         'only_matching': True,
  23     }]
  24
  25     def _real_extract(self, url):
  26         display_id = self._match_id(url)
  27         webpage = self._download_webpage(url, display_id)
  28
  29         drupal_settings = self._parse_json(
  30             self._html_search_regex(
  31                 r'(?s)Drupal\.settings\s*,\s*({.+?})\);',
  32                 webpage, 'drupal settings'),
  33             display_id)
  34
  35         player = drupal_settings['WWEVideoLanding']['initialVideo']
  36         metadata = player['playlist'][0]
  37
  38         id = compat_str(metadata['nid'])
  39         title = metadata.get('title') or self._og_search_title(webpage)
  40         video_url = 'https:' + metadata['file']
  41         thumbnail = None
  42         if metadata.get('image') is not None:
  43             thumbnail = urljoin(url, metadata.get('image'))
  44         description = metadata.get('description')
  45
  46         formats = self._extract_m3u8_formats(video_url, id, 'mp4')
  47
  48         return {
  49             'id': id,
  50             'title': title,
  51             'formats': formats,
  52             'url': video_url,
  53             'display_id': display_id,
  54             'thumbnail': thumbnail,
  55             'description': description,
  56         }