_ 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 (
   8     parse_duration,
   9     str_to_int,
  10 )
  11
  12
  13 class EpornerIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?eporner\.com/hd-porn/(?P<id>\d+)/(?P<title_dash>[\w-]+)/?'
  15     _TEST = {
  16         'url': 'http://www.eporner.com/hd-porn/95008/Infamous-Tiffany-Teen-Strip-Tease-Video/',
  17         'md5': '3b427ae4b9d60619106de3185c2987cd',
  18         'info_dict': {
  19             'id': '95008',
  20             'ext': 'flv',
  21             'title': 'Infamous Tiffany Teen Strip Tease Video',
  22             'duration': 194,
  23             'view_count': int,
  24             'age_limit': 18,
  25         }
  26     }
  27
  28     def _real_extract(self, url):
  29         mobj = re.match(self._VALID_URL, url)
  30         video_id = mobj.group('id')
  31         webpage = self._download_webpage(url, video_id)
  32         title = self._html_search_regex(
  33             r'<title>(.*?) - EPORNER', webpage, 'title')
  34
  35         redirect_code = self._html_search_regex(
  36             r'<script type="text/javascript" src="/config5/%s/([a-f\d]+)/">' % video_id,
  37             webpage, 'redirect_code')
  38         redirect_url = 'http://www.eporner.com/config5/%s/%s' % (video_id, redirect_code)
  39         player_code = self._download_webpage(
  40             redirect_url, video_id, note='Downloading player config')
  41         video_url = self._html_search_regex(
  42             r'file: "(.*?)",', player_code, 'video_url')
  43
  44         duration = parse_duration(self._search_regex(
  45             r'class="mbtim">([0-9:]+)</div>', webpage, 'duration',
  46             fatal=False))
  47         view_count = str_to_int(self._search_regex(
  48             r'id="cinemaviews">\s*([0-9,]+)\s*<small>views',
  49             webpage, 'view count', fatal=False))
  50
  51         return {
  52             'id': video_id,
  53             'url': video_url,
  54             'title': title,
  55             'duration': duration,
  56             'view_count': view_count,
  57             'age_limit': self._rta_search(webpage),
  58         }