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

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     parse_duration,
   8     int_or_none,
   9 )
  10
  11
  12 class XXXYMoviesIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?xxxymovies\.com/videos/(?P<id>\d+)/(?P<display_id>[^/]+)'
  14     _TEST = {
  15         'url': 'http://xxxymovies.com/videos/138669/ecstatic-orgasm-sofcore/',
  16         'md5': '810b1bdbbffff89dd13bdb369fe7be4b',
  17         'info_dict': {
  18             'id': '138669',
  19             'display_id': 'ecstatic-orgasm-sofcore',
  20             'ext': 'mp4',
  21             'title': 'Ecstatic Orgasm Sofcore',
  22             'duration': 931,
  23             'age_limit': 18,
  24         }
  25     }
  26
  27     def _real_extract(self, url):
  28         mobj = re.match(self._VALID_URL, url)
  29
  30         video_id = mobj.group('id')
  31         display_id = mobj.group('display_id')
  32
  33         webpage = self._download_webpage(url, video_id)
  34
  35         video_url = self._html_search_regex(
  36             r"video_url\s*:\s*'([^']+)'", webpage, 'video URL')
  37
  38         title = self._html_search_regex(
  39             r'<title>(.*?)\s*-\s*XXXYMovies.com</title>', webpage, 'title')
  40
  41         thumbnail = self._html_search_regex(
  42             r'preview_url\s*:\s*\'(.*?)\'', webpage, 'thumbnail', fatal=False)
  43
  44         categories = self._html_search_meta(
  45             'keywords', webpage, 'categories', default='').split(',')
  46
  47         duration = parse_duration(self._search_regex(
  48             r'<span>Duration:</span>\s*(\d+:\d+)', webpage, 'duration', fatal=False))
  49
  50         view_count = int_or_none(self._html_search_regex(
  51             r'<div class="video_views">\s*(\d+)', webpage, 'view count', fatal=False))
  52
  53         return {
  54             'id': video_id,
  55             'display_id': display_id,
  56             'url': video_url,
  57             'title': title,
  58             'thumbnail': thumbnail,
  59             'categories': categories,
  60             'duration': duration,
  61             'view_count': view_count,
  62             'age_limit': 18,
  63         }