1
0
mirror of https://github.com/l1ving/youtube-dl synced 2024-11-23 02:42:52 +08:00
youtube-dl/youtube_dl/extractor/shared.py

64 lines
1.9 KiB
Python
Raw Normal View History

2014-07-24 22:12:45 +08:00
from __future__ import unicode_literals
import re
import base64
from .common import InfoExtractor
from ..compat import (
compat_urllib_parse,
compat_urllib_request,
)
2014-07-24 22:12:45 +08:00
from ..utils import (
ExtractorError,
int_or_none,
)
class SharedIE(InfoExtractor):
_VALID_URL = r'http://shared\.sx/(?P<id>[\da-z]{10})'
_TEST = {
'url': 'http://shared.sx/0060718775',
2014-08-13 19:24:46 +08:00
'md5': '106fefed92a8a2adb8c98e6a0652f49b',
2014-07-24 22:12:45 +08:00
'info_dict': {
'id': '0060718775',
'ext': 'mp4',
2014-08-13 19:24:46 +08:00
'title': 'Bmp4',
2014-07-24 22:12:45 +08:00
},
}
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
2014-07-24 22:12:45 +08:00
if '>File does not exist<' in webpage:
raise ExtractorError(
'Video %s does not exist' % video_id, expected=True)
2014-07-24 22:12:45 +08:00
download_form = dict(re.findall(
r'<input type="hidden" name="([^"]+)" value="([^"]*)"', webpage))
request = compat_urllib_request.Request(
url, compat_urllib_parse.urlencode(download_form))
2014-07-24 22:12:45 +08:00
request.add_header('Content-Type', 'application/x-www-form-urlencoded')
video_page = self._download_webpage(
request, video_id, 'Downloading video page')
2014-07-24 22:12:45 +08:00
video_url = self._html_search_regex(
r'data-url="([^"]+)"', video_page, 'video URL')
title = base64.b64decode(self._html_search_meta(
'full:title', webpage, 'title')).decode('utf-8')
filesize = int_or_none(self._html_search_meta(
'full:size', webpage, 'file size', fatal=False))
2014-07-24 22:12:45 +08:00
thumbnail = self._html_search_regex(
r'data-poster="([^"]+)"', video_page, 'thumbnail', default=None)
2014-07-24 22:12:45 +08:00
return {
'id': video_id,
'url': video_url,
'ext': 'mp4',
'filesize': filesize,
'title': title,
'thumbnail': thumbnail,
2014-11-24 03:41:03 +08:00
}