mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-10 15:14:57 +00:00
40 lines
1.3 KiB
Python
40 lines
1.3 KiB
Python
|
import re
|
||
|
|
||
|
from .common import InfoExtractor
|
||
|
|
||
|
class NuvidIE(InfoExtractor):
|
||
|
_VALID_URL = r'^https?://(?:www|m)\.nuvid\.com/video/(?P<videoid>\d+)'
|
||
|
_TEST = {
|
||
|
u'url': u'http://m.nuvid.com/video/1310741/',
|
||
|
u'file': u'1310741.mp4',
|
||
|
u'md5': u'eab207b7ac4fccfb4e23c86201f11277',
|
||
|
u'info_dict': {
|
||
|
u"title": u"Horny babes show their awesome bodeis and",
|
||
|
u"age_limit": 18,
|
||
|
}
|
||
|
}
|
||
|
|
||
|
def _real_extract(self, url):
|
||
|
mobj = re.match(self._VALID_URL, url)
|
||
|
|
||
|
video_id = mobj.group('videoid')
|
||
|
|
||
|
# Get webpage content
|
||
|
murl = url.replace('//www.', '//m.')
|
||
|
webpage = self._download_webpage(murl, video_id)
|
||
|
|
||
|
video_title = self._html_search_regex(r'<div class="title">\s+<h2[^>]*>([^<]+)</h2>', webpage, 'video_title').strip()
|
||
|
|
||
|
video_url = 'http://m.nuvid.com'+self._html_search_regex(r'href="(/mp4/[^"]+)"[^>]*data-link_type="mp4"', webpage, 'video_url')
|
||
|
|
||
|
video_thumb = self._html_search_regex(r'href="(/thumbs/[^"]+)"[^>]*data-link_type="thumbs"', webpage, 'video_thumb')
|
||
|
|
||
|
info = {'id': video_id,
|
||
|
'url': video_url,
|
||
|
'title': video_title,
|
||
|
'thumbnail': video_thumb,
|
||
|
'ext': 'mp4',
|
||
|
'age_limit': 18}
|
||
|
|
||
|
return [info]
|