1
0
mirror of https://codeberg.org/polarisfm/youtube-dl synced 2024-12-12 09:47:55 +01:00
youtube-dl/youtube_dl/extractor/vodlocker.py

64 lines
1.9 KiB
Python
Raw Normal View History

2014-07-09 18:21:46 +02:00
# -*- coding: utf-8 -*-
from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..utils import (
compat_urllib_parse,
compat_urllib_request,
)
class VodlockerIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?vodlocker.com/(?P<id>[0-9a-zA-Z]+)(?:\..*?)?'
_TESTS = [{
'url': 'http://vodlocker.com/e8wvyzz4sl42',
'md5': 'ce0c2d18fa0735f1bd91b69b0e54aacf',
'info_dict': {
'id': 'e8wvyzz4sl42',
'ext': 'mp4',
'title': 'Germany vs Brazil',
'thumbnail': 're:http://.*\.jpg',
},
}]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
webpage = self._download_webpage(url, video_id)
fields = dict(re.findall(r'''(?x)<input\s+
type="hidden"\s+
name="([^"]+)"\s+
(?:id="[^"]+"\s+)?
value="([^"]*)"
''', webpage))
if fields['op'] == 'download1':
self._sleep(3, video_id) # they do detect when requests happen too fast!
2014-07-09 18:21:46 +02:00
post = compat_urllib_parse.urlencode(fields)
req = compat_urllib_request.Request(url, post)
req.add_header('Content-type', 'application/x-www-form-urlencoded')
webpage = self._download_webpage(
req, video_id, 'Downloading video page')
2014-07-09 18:21:46 +02:00
title = self._search_regex(
r'id="file_title".*?>\s*(.*?)\s*<span', webpage, 'title')
thumbnail = self._search_regex(
r'image:\s*"(http[^\"]+)",', webpage, 'thumbnail')
url = self._search_regex(
r'file:\s*"(http[^\"]+)",', webpage, 'file url')
2014-07-09 18:21:46 +02:00
formats = [{
'format_id': 'sd',
'url': url,
}]
return {
'id': video_id,
'title': title,
'thumbnail': thumbnail,
'formats': formats,
}