yt_dlp/extractor/movieclips.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     smuggle_url,
   4     float_or_none,
   5     parse_iso8601,
   6     update_url_query,
   7 )
   8
   9
  10 class MovieClipsIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?movieclips\.com/videos/.+-(?P<id>\d+)(?:\?|$)'
  12     _TEST = {
  13         'url': 'http://www.movieclips.com/videos/warcraft-trailer-1-561180739597',
  14         'md5': '42b5a0352d4933a7bd54f2104f481244',
  15         'info_dict': {
  16             'id': 'pKIGmG83AqD9',
  17             'ext': 'mp4',
  18             'title': 'Warcraft Trailer 1',
  19             'description': 'Watch Trailer 1 from Warcraft (2016). Legendary’s WARCRAFT is a 3D epic adventure of world-colliding conflict based.',
  20             'thumbnail': r're:^https?://.*\.jpg$',
  21             'timestamp': 1446843055,
  22             'upload_date': '20151106',
  23             'uploader': 'Movieclips',
  24         },
  25         'add_ie': ['ThePlatform'],
  26     }
  27
  28     def _real_extract(self, url):
  29         video_id = self._match_id(url)
  30         webpage = self._download_webpage(url, video_id)
  31         video = next(v for v in self._parse_json(self._search_regex(
  32             r'var\s+__REACT_ENGINE__\s*=\s*({.+});',
  33             webpage, 'react engine'), video_id)['playlist']['videos'] if v['id'] == video_id)
  34
  35         return {
  36             '_type': 'url_transparent',
  37             'ie_key': 'ThePlatform',
  38             'url': smuggle_url(update_url_query(
  39                 video['contentUrl'], {'mbr': 'true'}), {'force_smil_url': True}),
  40             'title': self._og_search_title(webpage),
  41             'description': self._html_search_meta('description', webpage),
  42             'duration': float_or_none(video.get('duration')),
  43             'timestamp': parse_iso8601(video.get('dateCreated')),
  44             'thumbnail': video.get('defaultImage'),
  45             'uploader': video.get('provider'),
  46         }