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

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7
   8
   9 class KickStarterIE(InfoExtractor):
  10     _VALID_URL = r'https?://www\.kickstarter\.com/projects/(?P<id>[^/]*)/.*'
  11     _TEST = {
  12         'url': 'https://www.kickstarter.com/projects/1404461844/intersection-the-story-of-josh-grant?ref=home_location',
  13         'md5': 'c81addca81327ffa66c642b5d8b08cab',
  14         'info_dict': {
  15             'id': '1404461844',
  16             'ext': 'mp4',
  17             'title': 'Intersection: The Story of Josh Grant by Kyle Cowling',
  18             'description': 'A unique motocross documentary that examines the '
  19                 'life and mind of one of sports most elite athletes: Josh Grant.',
  20         },
  21     }
  22
  23     def _real_extract(self, url):
  24         m = re.match(self._VALID_URL, url)
  25         video_id = m.group('id')
  26         webpage = self._download_webpage(url, video_id)
  27
  28         video_url = self._search_regex(r'data-video-url="(.*?)"',
  29             webpage, 'video URL')
  30         video_title = self._html_search_regex(r'<title>(.*?)</title>',
  31             webpage, 'title').rpartition('— Kickstarter')[0].strip()
  32
  33         return {
  34             'id': video_id,
  35             'url': video_url,
  36             'title': video_title,
  37             'description': self._og_search_description(webpage),
  38             'thumbnail': self._og_search_thumbnail(webpage),
  39         }