youtube_dl/extractor/empflix.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import fix_xml_ampersands
   7
   8
   9 class EmpflixIE(InfoExtractor):
  10     _VALID_URL = r'^https?://www\.empflix\.com/videos/.*?-(?P<id>[0-9]+)\.html'
  11     _TEST = {
  12         'url': 'http://www.empflix.com/videos/Amateur-Finger-Fuck-33051.html',
  13         'md5': 'b1bc15b6412d33902d6e5952035fcabc',
  14         'info_dict': {
  15             'id': '33051',
  16             'ext': 'mp4',
  17             'title': 'Amateur Finger Fuck',
  18             'description': 'Amateur solo finger fucking.',
  19             'age_limit': 18,
  20         }
  21     }
  22
  23     def _real_extract(self, url):
  24         mobj = re.match(self._VALID_URL, url)
  25         video_id = mobj.group('id')
  26
  27         webpage = self._download_webpage(url, video_id)
  28         age_limit = self._rta_search(webpage)
  29
  30         video_title = self._html_search_regex(
  31             r'name="title" value="(?P<title>[^"]*)"', webpage, 'title')
  32         video_description = self._html_search_regex(
  33             r'name="description" value="([^"]*)"', webpage, 'description', fatal=False)
  34
  35         cfg_url = self._html_search_regex(
  36             r'flashvars\.config = escape\("([^"]+)"',
  37             webpage, 'flashvars.config')
  38
  39         cfg_xml = self._download_xml(
  40             cfg_url, video_id, note='Downloading metadata',
  41             transform_source=fix_xml_ampersands)
  42
  43         formats = [
  44             {
  45                 'url': item.find('videoLink').text,
  46                 'format_id': item.find('res').text,
  47             } for item in cfg_xml.findall('./quality/item')
  48         ]
  49         thumbnail = cfg_xml.find('./startThumb').text
  50
  51         return {
  52             'id': video_id,
  53             'title': video_title,
  54             'description': video_description,
  55             'thumbnail': thumbnail,
  56             'formats': formats,
  57             'age_limit': age_limit,
  58         }