[brightcove] Allow whitespace around attribute names in embedded code
[youtube-dl] / youtube_dl / extractor / kamcord.py
1 from __future__ import unicode_literals
2
3 from .common import InfoExtractor
4 from ..compat import compat_str
5 from ..utils import (
6     int_or_none,
7     qualities,
8 )
9
10
11 class KamcordIE(InfoExtractor):
12     _VALID_URL = r'https?://(?:www\.)?kamcord\.com/v/(?P<id>[^/?#&]+)'
13     _TEST = {
14         'url': 'https://www.kamcord.com/v/hNYRduDgWb4',
15         'md5': 'c3180e8a9cfac2e86e1b88cb8751b54c',
16         'info_dict': {
17             'id': 'hNYRduDgWb4',
18             'ext': 'mp4',
19             'title': 'Drinking Madness',
20             'uploader': 'jacksfilms',
21             'uploader_id': '3044562',
22             'view_count': int,
23             'like_count': int,
24             'comment_count': int,
25         },
26     }
27
28     def _real_extract(self, url):
29         video_id = self._match_id(url)
30
31         webpage = self._download_webpage(url, video_id)
32
33         video = self._parse_json(
34             self._search_regex(
35                 r'window\.__props\s*=\s*({.+?});?(?:\n|\s*</script)',
36                 webpage, 'video'),
37             video_id)['video']
38
39         title = video['title']
40
41         formats = self._extract_m3u8_formats(
42             video['play']['hls'], video_id, 'mp4', entry_protocol='m3u8_native')
43         self._sort_formats(formats)
44
45         uploader = video.get('user', {}).get('username')
46         uploader_id = video.get('user', {}).get('id')
47
48         view_count = int_or_none(video.get('viewCount'))
49         like_count = int_or_none(video.get('heartCount'))
50         comment_count = int_or_none(video.get('messageCount'))
51
52         preference_key = qualities(('small', 'medium', 'large'))
53
54         thumbnails = [{
55             'url': thumbnail_url,
56             'id': thumbnail_id,
57             'preference': preference_key(thumbnail_id),
58         } for thumbnail_id, thumbnail_url in (video.get('thumbnail') or {}).items()
59             if isinstance(thumbnail_id, compat_str) and isinstance(thumbnail_url, compat_str)]
60
61         return {
62             'id': video_id,
63             'title': title,
64             'uploader': uploader,
65             'uploader_id': uploader_id,
66             'view_count': view_count,
67             'like_count': like_count,
68             'comment_count': comment_count,
69             'thumbnails': thumbnails,
70             'formats': formats,
71         }