youtube_dl/extractor/kickstarter.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import smuggle_url
   6
   7
   8 class KickStarterIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?kickstarter\.com/projects/(?P<id>[^/]*)/.*'
  10     _TESTS = [{
  11         'url': 'https://www.kickstarter.com/projects/1404461844/intersection-the-story-of-josh-grant/description',
  12         'md5': 'c81addca81327ffa66c642b5d8b08cab',
  13         'info_dict': {
  14             'id': '1404461844',
  15             'ext': 'mp4',
  16             'title': 'Intersection: The Story of Josh Grant by Kyle Cowling',
  17             'description': (
  18                 'A unique motocross documentary that examines the '
  19                 'life and mind of one of sports most elite athletes: Josh Grant.'
  20             ),
  21         },
  22     }, {
  23         'note': 'Embedded video (not using the native kickstarter video service)',
  24         'url': 'https://www.kickstarter.com/projects/597507018/pebble-e-paper-watch-for-iphone-and-android/posts/659178',
  25         'info_dict': {
  26             'id': '78704821',
  27             'ext': 'mp4',
  28             'uploader_id': 'pebble',
  29             'uploader': 'Pebble Technology',
  30             'title': 'Pebble iOS Notifications',
  31         },
  32         'add_ie': ['Vimeo'],
  33     }, {
  34         'url': 'https://www.kickstarter.com/projects/1420158244/power-drive-2000/widget/video.html',
  35         'info_dict': {
  36             'id': '1420158244',
  37             'ext': 'mp4',
  38             'title': 'Power Drive 2000',
  39         },
  40     }]
  41
  42     def _real_extract(self, url):
  43         video_id = self._match_id(url)
  44         webpage = self._download_webpage(url, video_id)
  45
  46         title = self._html_search_regex(
  47             r'<title>\s*(.*?)(?:\s*&mdash;\s*Kickstarter)?\s*</title>',
  48             webpage, 'title')
  49         video_url = self._search_regex(
  50             r'data-video-url="(.*?)"',
  51             webpage, 'video URL', default=None)
  52         if video_url is None:  # No native kickstarter, look for embedded videos
  53             return {
  54                 '_type': 'url_transparent',
  55                 'ie_key': 'Generic',
  56                 'url': smuggle_url(url, {'to_generic': True}),
  57                 'title': title,
  58             }
  59
  60         thumbnail = self._og_search_thumbnail(webpage, default=None)
  61         if thumbnail is None:
  62             thumbnail = self._html_search_regex(
  63                 r'<img[^>]+class="[^"]+\s*poster\s*[^"]+"[^>]+src="([^"]+)"',
  64                 webpage, 'thumbnail image', fatal=False)
  65         return {
  66             'id': video_id,
  67             'url': video_url,
  68             'title': title,
  69             'description': self._og_search_description(webpage, default=None),
  70             'thumbnail': thumbnail,
  71         }