]>
Commit | Line | Data |
---|---|---|
0eb799ba JMF |
1 | from __future__ import unicode_literals |
2 | ||
bfd5c93a | 3 | import json |
78af8eb1 PH |
4 | import re |
5 | ||
6 | from .common import InfoExtractor | |
74ac9bdd JMF |
7 | from ..utils import ( |
8 | compat_urlparse, | |
a921f407 | 9 | get_meta_content, |
74ac9bdd | 10 | ) |
78af8eb1 PH |
11 | |
12 | ||
13 | class UstreamIE(InfoExtractor): | |
b702eceb | 14 | _VALID_URL = r'https?://www\.ustream\.tv/(?P<type>recorded|embed|embed/recorded)/(?P<videoID>\d+)' |
0eb799ba | 15 | IE_NAME = 'ustream' |
ca6aada4 | 16 | _TEST = { |
0eb799ba | 17 | 'url': 'http://www.ustream.tv/recorded/20274954', |
0eb799ba JMF |
18 | 'md5': '088f151799e8f572f84eb62f17d73e5c', |
19 | 'info_dict': { | |
9e875391 S |
20 | 'id': '20274954', |
21 | 'ext': 'flv', | |
22 | 'uploader': 'Young Americans for Liberty', | |
23 | 'title': 'Young Americans for Liberty February 7, 2012 2:28 AM', | |
0eb799ba | 24 | }, |
6f5ac90c | 25 | } |
78af8eb1 PH |
26 | |
27 | def _real_extract(self, url): | |
28 | m = re.match(self._VALID_URL, url) | |
b702eceb | 29 | video_id = m.group('videoID') |
30 | ||
9e875391 S |
31 | # some sites use this embed format (see: http://github.com/rg3/youtube-dl/issues/2990) |
32 | if m.group('type') == 'embed/recorded': | |
b702eceb | 33 | video_id = m.group('videoID') |
b702eceb | 34 | desktop_url = 'http://www.ustream.tv/recorded/' + video_id |
35 | return self.url_result(desktop_url, 'Ustream') | |
5c386252 | 36 | if m.group('type') == 'embed': |
37 | video_id = m.group('videoID') | |
38 | webpage = self._download_webpage(url, video_id) | |
9e875391 S |
39 | desktop_video_id = self._html_search_regex( |
40 | r'ContentVideoIds=\["([^"]*?)"\]', webpage, 'desktop_video_id') | |
5c386252 | 41 | desktop_url = 'http://www.ustream.tv/recorded/' + desktop_video_id |
42 | return self.url_result(desktop_url, 'Ustream') | |
43 | ||
0eb799ba | 44 | video_url = 'http://tcdn.ustream.tv/video/%s' % video_id |
78af8eb1 PH |
45 | webpage = self._download_webpage(url, video_id) |
46 | ||
47 | self.report_extraction(video_id) | |
48 | ||
49 | video_title = self._html_search_regex(r'data-title="(?P<title>.+)"', | |
0eb799ba | 50 | webpage, 'title') |
78af8eb1 PH |
51 | |
52 | uploader = self._html_search_regex(r'data-content-type="channel".*?>(?P<uploader>.*?)</a>', | |
0eb799ba | 53 | webpage, 'uploader', fatal=False, flags=re.DOTALL) |
78af8eb1 PH |
54 | |
55 | thumbnail = self._html_search_regex(r'<link rel="image_src" href="(?P<thumb>.*?)"', | |
0eb799ba JMF |
56 | webpage, 'thumbnail', fatal=False) |
57 | ||
58 | return { | |
59 | 'id': video_id, | |
60 | 'url': video_url, | |
61 | 'ext': 'flv', | |
62 | 'title': video_title, | |
63 | 'uploader': uploader, | |
64 | 'thumbnail': thumbnail, | |
65 | } | |
66 | ||
bfd5c93a | 67 | |
bfd5c93a | 68 | class UstreamChannelIE(InfoExtractor): |
69 | _VALID_URL = r'https?://www\.ustream\.tv/channel/(?P<slug>.+)' | |
0eb799ba | 70 | IE_NAME = 'ustream:channel' |
bfd5c93a | 71 | |
72 | def _real_extract(self, url): | |
73 | m = re.match(self._VALID_URL, url) | |
74 | slug = m.group('slug') | |
a921f407 JMF |
75 | webpage = self._download_webpage(url, slug) |
76 | channel_id = get_meta_content('ustream:channel_id', webpage) | |
bfd5c93a | 77 | |
bfd5c93a | 78 | BASE = 'http://www.ustream.tv' |
79 | next_url = '/ajax/socialstream/videos/%s/1.json' % channel_id | |
a921f407 | 80 | video_ids = [] |
bfd5c93a | 81 | while next_url: |
74ac9bdd | 82 | reply = json.loads(self._download_webpage(compat_urlparse.urljoin(BASE, next_url), channel_id)) |
a921f407 | 83 | video_ids.extend(re.findall(r'data-content-id="(\d.*)"', reply['data'])) |
bfd5c93a | 84 | next_url = reply['nextUrl'] |
bfd5c93a | 85 | |
bfd5c93a | 86 | urls = ['http://www.ustream.tv/recorded/' + vid for vid in video_ids] |
87 | url_entries = [self.url_result(eurl, 'Ustream') for eurl in urls] | |
74ac9bdd | 88 | return self.playlist_result(url_entries, channel_id) |