youtube-dl/youtube_dl/extractor/eporner.py

59 lines
2.0 KiB
Python
Raw Normal View History

2014-08-31 05:28:53 +09:00
# coding: utf-8
from __future__ import unicode_literals
import re
from .common import InfoExtractor
2014-09-01 06:51:36 +09:00
from ..utils import (
parse_duration,
str_to_int,
)
2014-08-31 05:28:53 +09:00
class EpornerIE(InfoExtractor):
2014-09-01 06:51:36 +09:00
_VALID_URL = r'https?://(?:www\.)?eporner\.com/hd-porn/(?P<id>\d+)/(?P<title_dash>[\w-]+)/?'
2014-08-31 05:28:53 +09:00
_TEST = {
'url': 'http://www.eporner.com/hd-porn/95008/Infamous-Tiffany-Teen-Strip-Tease-Video/',
'md5': '3b427ae4b9d60619106de3185c2987cd',
'info_dict': {
'id': '95008',
'ext': 'flv',
'title': 'Infamous Tiffany Teen Strip Tease Video',
2014-09-01 06:51:36 +09:00
'duration': 194,
'view_count': int,
2014-09-02 06:07:48 +09:00
'age_limit': 18,
2014-08-31 05:28:53 +09:00
}
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
webpage = self._download_webpage(url, video_id)
2014-09-01 06:51:36 +09:00
title = self._html_search_regex(
r'<title>(.*?) - EPORNER', webpage, 'title')
2014-08-31 05:28:53 +09:00
2014-09-01 06:51:36 +09:00
redirect_code = self._html_search_regex(
r'<script type="text/javascript" src="/config5/%s/([a-f\d]+)/">' % video_id,
webpage, 'redirect_code')
redirect_url = 'http://www.eporner.com/config5/%s/%s' % (video_id, redirect_code)
2014-09-02 06:07:48 +09:00
player_code = self._download_webpage(
redirect_url, video_id, note='Downloading player config')
2014-09-01 06:51:36 +09:00
video_url = self._html_search_regex(
2014-09-02 06:07:48 +09:00
r'file: "(.*?)",', player_code, 'video_url')
2014-08-31 05:28:53 +09:00
2014-09-01 06:51:36 +09:00
duration = parse_duration(self._search_regex(
r'class="mbtim">([0-9:]+)</div>', webpage, 'duration',
fatal=False))
view_count = str_to_int(self._search_regex(
r'id="cinemaviews">\s*([0-9,]+)\s*<small>views',
webpage, 'view count', fatal=False))
2014-08-31 05:28:53 +09:00
return {
'id': video_id,
'url': video_url,
'title': title,
2014-09-01 06:51:36 +09:00
'duration': duration,
'view_count': view_count,
2014-09-02 06:07:48 +09:00
'age_limit': self._rta_search(webpage),
2014-08-31 05:28:53 +09:00
}