4 from .common
import InfoExtractor
6 compat_urllib_parse_unquote
,
16 class VeeHDIE(InfoExtractor
):
17 _VALID_URL
= r
'https?://veehd\.com/video/(?P<id>\d+)'
19 # Seems VeeHD videos have multiple copies on several servers, all of
20 # whom have different MD5 checksums, so omit md5 field in all tests
22 'url': 'http://veehd.com/video/4639434_Solar-Sinter',
26 'title': 'Solar Sinter',
27 'uploader_id': 'VideoEyes',
28 'description': 'md5:46a840e8692ddbaffb5f81d9885cb457',
30 'skip': 'Video deleted',
32 'url': 'http://veehd.com/video/4905758_Elysian-Fields-Channeling',
36 'title': 'Elysian Fields - Channeling',
37 'description': 'md5:360e4e95fdab58aefbea0f2a19e5604b',
38 'uploader_id': 'spotted',
41 'url': 'http://veehd.com/video/2046729_2012-2009-DivX-Trailer',
45 'title': '2012 (2009) DivX Trailer',
46 'description': 'md5:75435ee95255e6a9838ac6f6f3a2396b',
47 'uploader_id': 'Movie_Trailers',
51 def _real_extract(self
, url
):
52 video_id
= self
._match
_id
(url
)
54 # VeeHD seems to send garbage on the first request.
55 # See https://github.com/ytdl-org/youtube-dl/issues/2102
56 self
._download
_webpage
(url
, video_id
, 'Requesting webpage')
57 webpage
= self
._download
_webpage
(url
, video_id
)
59 if 'This video has been removed<' in webpage
:
60 raise ExtractorError('Video %s has been removed' % video_id
, expected
=True)
62 player_path
= self
._search
_regex
(
63 r
'\$\("#playeriframe"\).attr\({src : "(.+?)"',
64 webpage
, 'player path')
65 player_url
= compat_urlparse
.urljoin(url
, player_path
)
67 self
._download
_webpage
(player_url
, video_id
, 'Requesting player page')
68 player_page
= self
._download
_webpage
(
69 player_url
, video_id
, 'Downloading player page')
73 config_json
= self
._search
_regex
(
74 r
'value=\'config
=({.+?}
)\'', player_page, 'config json
', default=None)
77 config = json.loads(config_json)
78 video_url = compat_urllib_parse_unquote(config['clip
']['url
'])
81 video_url = self._html_search_regex(
82 r'<embed
[^
>]+type="video/divx"[^
>]+src
="([^"]+)"',
83 player_page, 'video url', default=None)
86 iframe_src = self._search_regex(
87 r'<iframe[^>]+src="/?
([^
"]+)"', player_page, 'iframe url
')
88 iframe_url = 'http
://veehd
.com
/%s' % iframe_src
90 self._download_webpage(iframe_url, video_id, 'Requesting iframe page
')
91 iframe_page = self._download_webpage(
92 iframe_url, video_id, 'Downloading iframe page
')
94 video_url = self._search_regex(
95 r"file\s*:\s*'([^
']+)'", iframe_page, 'video url')
97 title = clean_html(get_element_by_id('videoName', webpage).rpartition('|')[0])
98 uploader_id = self._html_search_regex(
99 r'<a href="/profile
/\d
+">(.+?)</a>',
101 thumbnail = self._search_regex(
102 r'<img id="veehdpreview
" src="(.+?
)"',
103 webpage, 'thumbnail')
104 description = self._html_search_regex(
105 r'<td class="infodropdown
".*?<div>(.*?)<ul',
106 webpage, 'description', flags=re.DOTALL)
113 'uploader_id': uploader_id,
114 'thumbnail': thumbnail,
115 'description': description,