1
1
mirror of https://github.com/ytdl-org/youtube-dl synced 2024-12-21 13:37:46 +01:00
youtube-dl/youtube_dl/extractor/slutload.py
2016-02-14 15:37:17 +06:00

45 lines
1.4 KiB
Python

from __future__ import unicode_literals
import re
from .common import InfoExtractor
class SlutloadIE(InfoExtractor):
_VALID_URL = r'^https?://(?:\w+\.)?slutload\.com/video/[^/]+/(?P<id>[^/]+)/?$'
_TEST = {
'url': 'http://www.slutload.com/video/virginie-baisee-en-cam/TD73btpBqSxc/',
'md5': '0cf531ae8006b530bd9df947a6a0df77',
'info_dict': {
'id': 'TD73btpBqSxc',
'ext': 'mp4',
'title': 'virginie baisee en cam',
'age_limit': 18,
'thumbnail': 're:https?://.*?\.jpg'
}
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
webpage = self._download_webpage(url, video_id)
video_title = self._html_search_regex(r'<h1><strong>([^<]+)</strong>',
webpage, 'title').strip()
video_url = self._html_search_regex(
r'(?s)<div id="vidPlayer"\s+data-url="([^"]+)"',
webpage, 'video URL')
thumbnail = self._html_search_regex(
r'(?s)<div id="vidPlayer"\s+.*?previewer-file="([^"]+)"',
webpage, 'thumbnail', fatal=False)
return {
'id': video_id,
'url': video_url,
'title': video_title,
'thumbnail': thumbnail,
'age_limit': 18
}