[SpankBang] Remove regexp type prefix from _TEST url.
[youtube-dl] / youtube_dl / extractor / spankbang.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 import re
6
7 class SpankBangIE(InfoExtractor):
8     """Extractor for http://spankbang.com"""
9     
10     _VALID_URL = r"https?://(?:www\.)?spankbang\.com/(?P<id>\w+)/video/.*"
11     
12     _TEST = {
13         "url": "http://spankbang.com/3vvn/video/fantasy+solo",
14         "md5": "1cc433e1d6aa14bc376535b8679302f7",
15         "info_dict": {
16             "id": "3vvn",
17             "title": "fantasy solo",
18             "description": "Watch fantasy solo free HD porn video - 05 minutes - dillion harper masturbates on a bed free adult movies.",
19             "format": "720p",
20             "format_id": "720p",
21             "ext": "mp4",
22             "url": "http://spankbang.com/_3vvn/IjE0MjgyNjY5MTcuMzUi.IaGrcF-vDrvktMhjd-1fWixiCzU/title/720p__mp4"
23         }
24     }
25
26     def _real_extract(self, url):
27         video_id = self._match_id(url)
28         webpage = self._download_webpage(url, video_id)
29
30         title = self._html_search_regex(r"<h1>(?:<img.+?>)?(.*?)</h1>", webpage, "title")
31         
32         stream_key = self._html_search_regex(r"""var\s+stream_key\s*[=]\s*['"](.+?)['"]\s*;""", webpage, "stream_key")
33         
34         qualities = re.findall(r"<span.+?>([0-9]+p).*?</span>", webpage)
35         
36         formats = []
37         for q in sorted(qualities):
38             formats.append({
39                 "format_id": q,
40                 "format": q,
41                 "ext": "mp4",
42                 "url": "http://spankbang.com/_{0}/{1}/title/{2}__mp4".format(video_id, stream_key, q)
43             })
44
45         return {
46             "id": video_id,
47             "title": title,
48             "description": self._og_search_description(webpage),
49             "formats": formats
50         }
51
52 # vim: tabstop=4 expandtab