yt_dlp/extractor/kickstarter.py

   1 from .common import InfoExtractor
   2 from ..utils import smuggle_url
   3
   4
   5 class KickStarterIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:www\.)?kickstarter\.com/projects/(?P<id>[^/]*)/.*'
   7     _TESTS = [{
   8         'url': 'https://www.kickstarter.com/projects/1404461844/intersection-the-story-of-josh-grant/description',
   9         'md5': 'c81addca81327ffa66c642b5d8b08cab',
  10         'info_dict': {
  11             'id': '1404461844',
  12             'ext': 'mp4',
  13             'title': 'Intersection: The Story of Josh Grant by Kyle Cowling',
  14             'description': (
  15                 'A unique motocross documentary that examines the '
  16                 'life and mind of one of sports most elite athletes: Josh Grant.'
  17             ),
  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         'add_ie': ['Vimeo'],
  30     }, {
  31         'url': 'https://www.kickstarter.com/projects/1420158244/power-drive-2000/widget/video.html',
  32         'info_dict': {
  33             'id': '1420158244',
  34             'ext': 'mp4',
  35             'title': 'Power Drive 2000',
  36         },
  37     }]
  38
  39     def _real_extract(self, url):
  40         video_id = self._match_id(url)
  41         webpage = self._download_webpage(url, video_id)
  42
  43         title = self._html_search_regex(
  44             r'<title>\s*(.*?)(?:\s*&mdash;\s*Kickstarter)?\s*</title>',
  45             webpage, 'title')
  46         video_url = self._search_regex(
  47             r'data-video-url="(.*?)"',
  48             webpage, 'video URL', default=None)
  49         if video_url is None:  # No native kickstarter, look for embedded videos
  50             return {
  51                 '_type': 'url_transparent',
  52                 'ie_key': 'Generic',
  53                 'url': smuggle_url(url, {'to_generic': True}),
  54                 'title': title,
  55             }
  56
  57         thumbnail = self._og_search_thumbnail(webpage, default=None)
  58         if thumbnail is None:
  59             thumbnail = self._html_search_regex(
  60                 r'<img[^>]+class="[^"]+\s*poster\s*[^"]+"[^>]+src="([^"]+)"',
  61                 webpage, 'thumbnail image', fatal=False)
  62         return {
  63             'id': video_id,
  64             'url': video_url,
  65             'title': title,
  66             'description': self._og_search_description(webpage, default=None),
  67             'thumbnail': thumbnail,
  68         }