git.bitcoin.ninja Git - youtube-dl/blob - youtube_dl/extractor/kickstarter.py

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5
   6
   7 class KickStarterIE(InfoExtractor):
   8     _VALID_URL = r'https?://www\.kickstarter\.com/projects/(?P<id>[^/]*)/.*'
   9     _TESTS = [{
  10         'url': 'https://www.kickstarter.com/projects/1404461844/intersection-the-story-of-josh-grant?ref=home_location',
  11         'md5': 'c81addca81327ffa66c642b5d8b08cab',
  12         'info_dict': {
  13             'id': '1404461844',
  14             'ext': 'mp4',
  15             'title': 'Intersection: The Story of Josh Grant by Kyle Cowling',
  16             'description': (
  17                 'A unique motocross documentary that examines the '
  18                 'life and mind of one of sports most elite athletes: Josh Grant.'
  19             ),
  20         },
  21     }, {
  22         'note': 'Embedded video (not using the native kickstarter video service)',
  23         'url': 'https://www.kickstarter.com/projects/597507018/pebble-e-paper-watch-for-iphone-and-android/posts/659178',
  24         'info_dict': {
  25             'id': '78704821',
  26             'ext': 'mp4',
  27             'uploader_id': 'pebble',
  28             'uploader': 'Pebble Technology',
  29             'title': 'Pebble iOS Notifications',
  30         }
  31     }]
  32
  33     def _real_extract(self, url):
  34         video_id = self._match_id(url)
  35         webpage = self._download_webpage(url, video_id)
  36
  37         title = self._html_search_regex(
  38             r'<title>\s*(.*?)(?:\s*&mdash; Kickstarter)?\s*</title>',
  39             webpage, 'title')
  40         video_url = self._search_regex(
  41             r'data-video-url="(.*?)"',
  42             webpage, 'video URL', default=None)
  43         if video_url is None:  # No native kickstarter, look for embedded videos
  44             return {
  45                 '_type': 'url_transparent',
  46                 'ie_key': 'Generic',
  47                 'url': url,
  48                 'title': title,
  49             }
  50
  51         return {
  52             'id': video_id,
  53             'url': video_url,
  54             'title': title,
  55             'description': self._og_search_description(webpage),
  56             'thumbnail': self._og_search_thumbnail(webpage),
  57         }