mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-21 20:46:36 -05:00
Merge pull request #878 from yasoob/master
Added Vbox7.com InfoExtractor and tests.
This commit is contained in:
commit
cd453d38bb
2 changed files with 44 additions and 0 deletions
|
@ -500,5 +500,14 @@
|
|||
"info_dict":{
|
||||
"title":"TAME"
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Vbox7",
|
||||
"url": "http://vbox7.com/play:249bb972c2",
|
||||
"file": "249bb972c2.flv",
|
||||
"md5": "9c70d6d956f888bdc08c124acc120cfe",
|
||||
"info_dict":{
|
||||
"title":"Смях! Чудо - чист за секунди - Скрита камера"
|
||||
}
|
||||
}
|
||||
]
|
||||
|
|
|
@ -4534,6 +4534,40 @@ def _real_extract(self, url):
|
|||
'artist': artist,
|
||||
}]
|
||||
|
||||
class Vbox7IE(InfoExtractor):
|
||||
"""Information Extractor for Vbox7"""
|
||||
_VALID_URL = r'(?:http://)?(?:www\.)?vbox7\.com/play:([^/]+)'
|
||||
|
||||
def _real_extract(self,url):
|
||||
mobj = re.match(self._VALID_URL, url)
|
||||
if mobj is None:
|
||||
raise ExtractorError(u'Invalid URL: %s' % url)
|
||||
video_id = mobj.group(1)
|
||||
|
||||
redirect_page, urlh = self._download_webpage_handle(url, video_id)
|
||||
redirect_url = urlh.geturl() + re.search(r'window\.location = \'(.*)\';', redirect_page).group(1)
|
||||
webpage = self._download_webpage(redirect_url, video_id, u'Downloading redirect page')
|
||||
|
||||
title = re.search(r'<title>(.*)</title>', webpage)
|
||||
title = (title.group(1)).split('/')[0].strip()
|
||||
|
||||
ext = "flv"
|
||||
info_url = "http://vbox7.com/play/magare.do"
|
||||
data = compat_urllib_parse.urlencode({'as3':'1','vid':video_id})
|
||||
info_request = compat_urllib_request.Request(info_url, data)
|
||||
info_request.add_header('Content-Type', 'application/x-www-form-urlencoded')
|
||||
info_response = self._download_webpage(info_request, video_id, u'Downloading info webpage')
|
||||
if info_response is None:
|
||||
raise ExtractorError(u'Unable to extract the media url')
|
||||
(final_url, thumbnail_url) = map(lambda x: x.split('=')[1], info_response.split('&'))
|
||||
|
||||
return [{
|
||||
'id': video_id,
|
||||
'url': final_url,
|
||||
'ext': ext,
|
||||
'title': title,
|
||||
'thumbnail': thumbnail_url,
|
||||
}]
|
||||
|
||||
def gen_extractors():
|
||||
""" Return a list of an instance of every supported extractor.
|
||||
|
@ -4598,6 +4632,7 @@ def gen_extractors():
|
|||
TeamcocoIE(),
|
||||
XHamsterIE(),
|
||||
HypemIE(),
|
||||
Vbox7IE(),
|
||||
GenericIE()
|
||||
]
|
||||
|
||||
|
|
Loading…
Reference in a new issue