Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/escapist.py

   1 import json
   2 import re
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     compat_str,
   7     compat_urllib_parse,
   8
   9     ExtractorError,
  10 )
  11
  12
  13 class EscapistIE(InfoExtractor):
  14     _VALID_URL = r'^(https?://)?(www\.)?escapistmagazine\.com/videos/view/(?P<showname>[^/]+)/(?P<episode>[^/?]+)[/?]?.*$'
  15
  16     def _real_extract(self, url):
  17         mobj = re.match(self._VALID_URL, url)
  18         if mobj is None:
  19             raise ExtractorError(u'Invalid URL: %s' % url)
  20         showName = mobj.group('showname')
  21         videoId = mobj.group('episode')
  22
  23         self.report_extraction(videoId)
  24         webpage = self._download_webpage(url, videoId)
  25
  26         videoDesc = self._html_search_regex('<meta name="description" content="([^"]*)"',
  27             webpage, u'description', fatal=False)
  28
  29         imgUrl = self._html_search_regex('<meta property="og:image" content="([^"]*)"',
  30             webpage, u'thumbnail', fatal=False)
  31
  32         playerUrl = self._html_search_regex('<meta property="og:video" content="([^"]*)"',
  33             webpage, u'player url')
  34
  35         title = self._html_search_regex('<meta name="title" content="([^"]*)"',
  36             webpage, u'player url').split(' : ')[-1]
  37
  38         configUrl = self._search_regex('config=(.*)$', playerUrl, u'config url')
  39         configUrl = compat_urllib_parse.unquote(configUrl)
  40
  41         configJSON = self._download_webpage(configUrl, videoId,
  42                                             u'Downloading configuration',
  43                                             u'unable to download configuration')
  44
  45         # Technically, it's JavaScript, not JSON
  46         configJSON = configJSON.replace("'", '"')
  47
  48         try:
  49             config = json.loads(configJSON)
  50         except (ValueError,) as err:
  51             raise ExtractorError(u'Invalid JSON in configuration file: ' + compat_str(err))
  52
  53         playlist = config['playlist']
  54         videoUrl = playlist[1]['url']
  55
  56         info = {
  57             'id': videoId,
  58             'url': videoUrl,
  59             'uploader': showName,
  60             'upload_date': None,
  61             'title': title,
  62             'ext': 'mp4',
  63             'thumbnail': imgUrl,
  64             'description': videoDesc,
  65             'player_url': playerUrl,
  66         }
  67
  68         return [info]