]>
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 ( |
6b19647d | 8 | compat_urllib_parse_unquote, |
8e4e89f1 | 9 | compat_urlparse, |
1cc79574 PH |
10 | ) |
11 | from ..utils import ( | |
d5bb814d | 12 | ExtractorError, |
8e4e89f1 | 13 | clean_html, |
1cc79574 | 14 | get_element_by_id, |
8e4e89f1 JMF |
15 | ) |
16 | ||
c11a0611 | 17 | |
8e4e89f1 | 18 | class VeeHDIE(InfoExtractor): |
c0ade33e | 19 | _VALID_URL = r'https?://veehd\.com/video/(?P<id>\d+)' |
8e4e89f1 | 20 | |
f7f1df1d YCH |
21 | # Seems VeeHD videos have multiple copies on several servers, all of |
22 | # whom have different MD5 checksums, so omit md5 field in all tests | |
23 | _TESTS = [{ | |
aa24de39 | 24 | 'url': 'http://veehd.com/video/4639434_Solar-Sinter', |
c11a0611 | 25 | 'info_dict': { |
aa24de39 | 26 | 'id': '4639434', |
df53b635 | 27 | 'ext': 'mp4', |
aa24de39 S |
28 | 'title': 'Solar Sinter', |
29 | 'uploader_id': 'VideoEyes', | |
30 | 'description': 'md5:46a840e8692ddbaffb5f81d9885cb457', | |
8e4e89f1 | 31 | }, |
f7f1df1d YCH |
32 | 'skip': 'Video deleted', |
33 | }, { | |
34 | 'url': 'http://veehd.com/video/4905758_Elysian-Fields-Channeling', | |
35 | 'info_dict': { | |
36 | 'id': '4905758', | |
37 | 'ext': 'mp4', | |
38 | 'title': 'Elysian Fields - Channeling', | |
39 | 'description': 'md5:360e4e95fdab58aefbea0f2a19e5604b', | |
40 | 'uploader_id': 'spotted', | |
41 | } | |
42 | }, { | |
482a1258 | 43 | 'url': 'http://veehd.com/video/2046729_2012-2009-DivX-Trailer', |
f7f1df1d | 44 | 'info_dict': { |
482a1258 | 45 | 'id': '2046729', |
f7f1df1d | 46 | 'ext': 'avi', |
482a1258 YCH |
47 | 'title': '2012 (2009) DivX Trailer', |
48 | 'description': 'md5:75435ee95255e6a9838ac6f6f3a2396b', | |
49 | 'uploader_id': 'Movie_Trailers', | |
f7f1df1d YCH |
50 | } |
51 | }] | |
8e4e89f1 JMF |
52 | |
53 | def _real_extract(self, url): | |
1cc79574 | 54 | video_id = self._match_id(url) |
8e4e89f1 | 55 | |
c11a0611 | 56 | # VeeHD seems to send garbage on the first request. |
067aa17e | 57 | # See https://github.com/ytdl-org/youtube-dl/issues/2102 |
c11a0611 | 58 | self._download_webpage(url, video_id, 'Requesting webpage') |
8e4e89f1 | 59 | webpage = self._download_webpage(url, video_id) |
d5bb814d S |
60 | |
61 | if 'This video has been removed<' in webpage: | |
62 | raise ExtractorError('Video %s has been removed' % video_id, expected=True) | |
63 | ||
c11a0611 PH |
64 | player_path = self._search_regex( |
65 | r'\$\("#playeriframe"\).attr\({src : "(.+?)"', | |
66 | webpage, 'player path') | |
8e4e89f1 | 67 | player_url = compat_urlparse.urljoin(url, player_path) |
c11a0611 PH |
68 | |
69 | self._download_webpage(player_url, video_id, 'Requesting player page') | |
70 | player_page = self._download_webpage( | |
71 | player_url, video_id, 'Downloading player page') | |
a798e64c | 72 | |
f7f1df1d YCH |
73 | video_url = None |
74 | ||
c11a0611 | 75 | config_json = self._search_regex( |
a798e64c S |
76 | r'value=\'config=({.+?})\'', player_page, 'config json', default=None) |
77 | ||
78 | if config_json: | |
79 | config = json.loads(config_json) | |
6b19647d | 80 | video_url = compat_urllib_parse_unquote(config['clip']['url']) |
f7f1df1d YCH |
81 | |
82 | if not video_url: | |
83 | video_url = self._html_search_regex( | |
84 | r'<embed[^>]+type="video/divx"[^>]+src="([^"]+)"', | |
85 | player_page, 'video url', default=None) | |
86 | ||
87 | if not video_url: | |
a798e64c S |
88 | iframe_src = self._search_regex( |
89 | r'<iframe[^>]+src="/?([^"]+)"', player_page, 'iframe url') | |
90 | iframe_url = 'http://veehd.com/%s' % iframe_src | |
91 | ||
92 | self._download_webpage(iframe_url, video_id, 'Requesting iframe page') | |
93 | iframe_page = self._download_webpage( | |
94 | iframe_url, video_id, 'Downloading iframe page') | |
95 | ||
96 | video_url = self._search_regex( | |
97 | r"file\s*:\s*'([^']+)'", iframe_page, 'video url') | |
8e4e89f1 | 98 | |
8e4e89f1 | 99 | title = clean_html(get_element_by_id('videoName', webpage).rpartition('|')[0]) |
a798e64c S |
100 | uploader_id = self._html_search_regex( |
101 | r'<a href="/profile/\d+">(.+?)</a>', | |
102 | webpage, 'uploader') | |
103 | thumbnail = self._search_regex( | |
104 | r'<img id="veehdpreview" src="(.+?)"', | |
105 | webpage, 'thumbnail') | |
106 | description = self._html_search_regex( | |
107 | r'<td class="infodropdown".*?<div>(.*?)<ul', | |
108 | webpage, 'description', flags=re.DOTALL) | |
8e4e89f1 JMF |
109 | |
110 | return { | |
111 | '_type': 'video', | |
112 | 'id': video_id, | |
113 | 'title': title, | |
114 | 'url': video_url, | |
8e4e89f1 JMF |
115 | 'uploader_id': uploader_id, |
116 | 'thumbnail': thumbnail, | |
117 | 'description': description, | |
118 | } |