]>
Commit | Line | Data |
---|---|---|
c11a0611 PH |
1 | from __future__ import unicode_literals |
2 | ||
8e4e89f1 JMF |
3 | import re |
4 | import json | |
5 | ||
6 | from .common import InfoExtractor | |
1cc79574 | 7 | from ..compat import ( |
8e4e89f1 | 8 | compat_urlparse, |
1cc79574 PH |
9 | ) |
10 | from ..utils import ( | |
d5bb814d | 11 | ExtractorError, |
8e4e89f1 | 12 | clean_html, |
1cc79574 | 13 | get_element_by_id, |
8e4e89f1 JMF |
14 | ) |
15 | ||
c11a0611 | 16 | |
8e4e89f1 | 17 | class VeeHDIE(InfoExtractor): |
c0ade33e | 18 | _VALID_URL = r'https?://veehd\.com/video/(?P<id>\d+)' |
8e4e89f1 JMF |
19 | |
20 | _TEST = { | |
aa24de39 | 21 | 'url': 'http://veehd.com/video/4639434_Solar-Sinter', |
c11a0611 | 22 | 'info_dict': { |
aa24de39 | 23 | 'id': '4639434', |
df53b635 | 24 | 'ext': 'mp4', |
aa24de39 S |
25 | 'title': 'Solar Sinter', |
26 | 'uploader_id': 'VideoEyes', | |
27 | 'description': 'md5:46a840e8692ddbaffb5f81d9885cb457', | |
8e4e89f1 JMF |
28 | }, |
29 | } | |
30 | ||
31 | def _real_extract(self, url): | |
1cc79574 | 32 | video_id = self._match_id(url) |
8e4e89f1 | 33 | |
c11a0611 PH |
34 | # VeeHD seems to send garbage on the first request. |
35 | # See https://github.com/rg3/youtube-dl/issues/2102 | |
36 | self._download_webpage(url, video_id, 'Requesting webpage') | |
8e4e89f1 | 37 | webpage = self._download_webpage(url, video_id) |
d5bb814d S |
38 | |
39 | if 'This video has been removed<' in webpage: | |
40 | raise ExtractorError('Video %s has been removed' % video_id, expected=True) | |
41 | ||
c11a0611 PH |
42 | player_path = self._search_regex( |
43 | r'\$\("#playeriframe"\).attr\({src : "(.+?)"', | |
44 | webpage, 'player path') | |
8e4e89f1 | 45 | player_url = compat_urlparse.urljoin(url, player_path) |
c11a0611 PH |
46 | |
47 | self._download_webpage(player_url, video_id, 'Requesting player page') | |
48 | player_page = self._download_webpage( | |
49 | player_url, video_id, 'Downloading player page') | |
a798e64c | 50 | |
c11a0611 | 51 | config_json = self._search_regex( |
a798e64c S |
52 | r'value=\'config=({.+?})\'', player_page, 'config json', default=None) |
53 | ||
54 | if config_json: | |
55 | config = json.loads(config_json) | |
56 | video_url = compat_urlparse.unquote(config['clip']['url']) | |
57 | else: | |
58 | iframe_src = self._search_regex( | |
59 | r'<iframe[^>]+src="/?([^"]+)"', player_page, 'iframe url') | |
60 | iframe_url = 'http://veehd.com/%s' % iframe_src | |
61 | ||
62 | self._download_webpage(iframe_url, video_id, 'Requesting iframe page') | |
63 | iframe_page = self._download_webpage( | |
64 | iframe_url, video_id, 'Downloading iframe page') | |
65 | ||
66 | video_url = self._search_regex( | |
67 | r"file\s*:\s*'([^']+)'", iframe_page, 'video url') | |
8e4e89f1 | 68 | |
8e4e89f1 | 69 | title = clean_html(get_element_by_id('videoName', webpage).rpartition('|')[0]) |
a798e64c S |
70 | uploader_id = self._html_search_regex( |
71 | r'<a href="/profile/\d+">(.+?)</a>', | |
72 | webpage, 'uploader') | |
73 | thumbnail = self._search_regex( | |
74 | r'<img id="veehdpreview" src="(.+?)"', | |
75 | webpage, 'thumbnail') | |
76 | description = self._html_search_regex( | |
77 | r'<td class="infodropdown".*?<div>(.*?)<ul', | |
78 | webpage, 'description', flags=re.DOTALL) | |
8e4e89f1 JMF |
79 | |
80 | return { | |
81 | '_type': 'video', | |
82 | 'id': video_id, | |
83 | 'title': title, | |
84 | 'url': video_url, | |
85 | 'ext': 'mp4', | |
86 | 'uploader_id': uploader_id, | |
87 | 'thumbnail': thumbnail, | |
88 | 'description': description, | |
89 | } |