_ Git - youtube-dl/blob - youtube_dl/extractor/eporner.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import int_or_none
   8
   9 class EpornerIE(InfoExtractor):
  10     _VALID_URL = r'http?://(?:www\.)?eporner\.com/hd-porn/(?P<id>\d+)/(?P<title_dash>[\w-]+)/?'
  11     _TEST = {
  12         'url': 'http://www.eporner.com/hd-porn/95008/Infamous-Tiffany-Teen-Strip-Tease-Video/',
  13         'md5': '3b427ae4b9d60619106de3185c2987cd',
  14         'info_dict': {
  15             'id': '95008',
  16             'ext': 'flv',
  17             'title': 'Infamous Tiffany Teen Strip Tease Video',
  18             'duration': 194
  19         }
  20     }
  21
  22     def _real_extract(self, url):
  23         mobj = re.match(self._VALID_URL, url)
  24         video_id = mobj.group('id')
  25         webpage = self._download_webpage(url, video_id)
  26         title = self._html_search_regex(r'<title>(.*?) - EPORNER', webpage, 'title')
  27
  28         redirect_code = self._html_search_regex(r'<script type="text/javascript" src="/config5/'+str(video_id)+'/([a-f\d]+)/">', webpage, 'redirect_code')
  29         redirect_url = 'http://www.eporner.com/config5/' + str(video_id) +'/'+ redirect_code
  30         webpage2 = self._download_webpage(redirect_url, video_id)
  31         video_url = self._html_search_regex(r'file: "(.*?)",', webpage2, 'video_url')
  32
  33         mobj = re.search(r'class="mbtim">(?P<minutes>\d+):(?P<seconds>\d+)</div>', webpage)
  34         duration = int(mobj.group('minutes')) * 60 + int(mobj.group('seconds')) if mobj else None
  35
  36         mobj = re.search(r'id="cinemaviews">((?P<thousands>\d+),)?(?P<units>\d+)<small>views', webpage)
  37         view_count = int(mobj.group('thousands')) * 1000 + int(mobj.group('units')) if mobj else None
  38
  39         return {
  40             'id': video_id,
  41             'url': video_url,
  42             'title': title,
  43             'duration': int_or_none(duration),
  44             'view_count': int_or_none(view_count),
  45         }