_ 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': 'A unique motocross documentary that examines the '
  17                 'life and mind of one of sports most elite athletes: Josh Grant.',
  18         },
  19     }, {
  20         'note': 'Embedded video (not using the native kickstarter video service)',
  21         'url': 'https://www.kickstarter.com/projects/597507018/pebble-e-paper-watch-for-iphone-and-android/posts/659178',
  22         'info_dict': {
  23             'id': '78704821',
  24             'ext': 'mp4',
  25             'uploader_id': 'pebble',
  26             'uploader': 'Pebble Technology',
  27             'title': 'Pebble iOS Notifications',
  28         }
  29     }]
  30
  31     def _real_extract(self, url):
  32         video_id = self._match_id(url)
  33         webpage = self._download_webpage(url, video_id)
  34
  35         title = self._html_search_regex(
  36             r'<title>\s*(.*?)(?:\s*&mdash; Kickstarter)?\s*</title>',
  37             webpage, 'title')
  38         video_url = self._search_regex(
  39             r'data-video-url="(.*?)"',
  40             webpage, 'video URL', default=None)
  41         if video_url is None:  # No native kickstarter, look for embedded videos
  42             return {
  43                 '_type': 'url_transparent',
  44                 'ie_key': 'Generic',
  45                 'url': url,
  46                 'title': title,
  47             }
  48
  49         return {
  50             'id': video_id,
  51             'url': video_url,
  52             'title': title,
  53             'description': self._og_search_description(webpage),
  54             'thumbnail': self._og_search_thumbnail(webpage),
  55         }