1
0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-09 12:42:28 +01:00
yt-dlp/youtube_dl/extractor/spankbang.py
2015-04-06 21:24:17 +06:00

61 lines
1.9 KiB
Python

from __future__ import unicode_literals
import re
from .common import InfoExtractor
class SpankBangIE(InfoExtractor):
_VALID_URL = r'https?://(?:(?:www|[a-z]{2})\.)?spankbang\.com/(?P<id>[\da-z]+)/video'
_TEST = {
'url': 'http://spankbang.com/3vvn/video/fantasy+solo',
'md5': '1cc433e1d6aa14bc376535b8679302f7',
'info_dict': {
'id': '3vvn',
'ext': 'mp4',
'title': 'fantasy solo',
'description': 'dillion harper masturbates on a bed',
'thumbnail': 're:^https?://.*\.jpg$',
'uploader': 'silly2587',
'age_limit': 18,
}
}
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
stream_key = self._html_search_regex(
r'''var\s+stream_key\s*=\s*['"](.+?)['"]''',
webpage, 'stream key')
formats = [{
'url': 'http://spankbang.com/_%s/%s/title/%sp__mp4' % (video_id, stream_key, height),
'ext': 'mp4',
'format_id': '%sp' % height,
'height': int(height),
} for height in re.findall(r'<span[^>]+q_(\d+)p', webpage)]
self._sort_formats(formats)
title = self._html_search_regex(
r'(?s)<h1>(.+?)</h1>', webpage, 'title')
description = self._search_regex(
r'class="desc"[^>]*>([^<]+)',
webpage, 'description', default=None)
thumbnail = self._og_search_thumbnail(webpage)
uploader = self._search_regex(
r'class="user"[^>]*>([^<]+)',
webpage, 'uploader', fatal=False)
age_limit = self._rta_search(webpage)
return {
'id': video_id,
'title': title,
'description': description,
'thumbnail': thumbnail,
'uploader': uploader,
'formats': formats,
'age_limit': age_limit,
}