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

   1 import re
   2 import json
   3 from .common import InfoExtractor
   4
   5
   6 class DotsubIE(InfoExtractor):
   7     _VALID_URL = r'(?:http://)?(?:www\.)?dotsub\.com/view/([^/]+)'
   8     _TEST = {
   9         u'url': u'http://dotsub.com/view/aed3b8b2-1889-4df5-ae63-ad85f5572f27',
  10         u'file': u'aed3b8b2-1889-4df5-ae63-ad85f5572f27.flv',
  11         u'md5': u'0914d4d69605090f623b7ac329fea66e',
  12         u'info_dict': {
  13             u"title": u"Pyramids of Waste (2010), AKA The Lightbulb Conspiracy - Planned obsolescence documentary",
  14             u"uploader": u"4v4l0n42",
  15             u'description': u'Pyramids of Waste (2010) also known as "The lightbulb conspiracy" is a documentary about how our economic system based on consumerism  and planned obsolescence is breaking our planet down.\r\n\r\nSolutions to this can be found at:\r\nhttp://robotswillstealyourjob.com\r\nhttp://www.federicopistono.org\r\n\r\nhttp://opensourceecology.org\r\nhttp://thezeitgeistmovement.com',
  16             u'thumbnail': u'http://dotsub.com/media/aed3b8b2-1889-4df5-ae63-ad85f5572f27/p'
  17         }
  18     }
  19
  20     def _real_extract(self, url):
  21         mobj = re.match(self._VALID_URL, url)
  22         video_id = mobj.group(1)
  23         info_url = "https://dotsub.com/api/media/%s/metadata" %(video_id)
  24         webpage = self._download_webpage(info_url, video_id)
  25         info = json.loads(webpage)
  26         video_url = info['mediaURI']
  27         uploader = info['user']
  28         description = info['description']
  29         view_count = info['numberOfViews']
  30         title = info['title']
  31         thumbnail_url = info['screenshotURI']
  32         ext = 'flv'
  33         return [{
  34             'id':          video_id,
  35             'url':         video_url,
  36             'ext':         ext,
  37             'title':       title,
  38             'thumbnail':   thumbnail_url,
  39             'description': description,
  40             'uploader':    uploader,
  41             'view_count':  view_count,
  42         }]