Merge remote-tracking branch 'origin/master'
[youtube-dl] / 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     _TESTS = [{
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         '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         'playlist': [
25             {
26                 'info_dict': {
27                     'id': '78704821',
28                     'ext': 'mp4',
29                     'uploader_id': 'pebble',
30                     'uploader': 'Pebble Technology',
31                     'title': 'Pebble iOS Notifications',
32                 }
33             }
34         ],
35     }]
36
37     def _real_extract(self, url):
38         m = re.match(self._VALID_URL, url)
39         video_id = m.group('id')
40         webpage = self._download_webpage(url, video_id)
41
42         title = self._html_search_regex(
43             r'<title>\s*(.*?)(?:\s*&mdash; Kickstarter)?\s*</title>',
44             webpage, 'title')
45         video_url = self._search_regex(
46             r'data-video-url="(.*?)"',
47             webpage, 'video URL', default=None)
48         if video_url is None:  # No native kickstarter, look for embedded videos
49             return {
50                 '_type': 'url_transparent',
51                 'ie_key': 'Generic',
52                 'url': url,
53                 'title': title,
54             }
55
56         return {
57             'id': video_id,
58             'url': video_url,
59             'title': title,
60             'description': self._og_search_description(webpage),
61             'thumbnail': self._og_search_thumbnail(webpage),
62         }