mirror of
https://github.com/ytdl-org/youtube-dl.git
synced 2024-11-14 05:17:25 +00:00
98 lines
3.1 KiB
Python
98 lines
3.1 KiB
Python
# -*- coding: utf-8 -*-
|
|
from __future__ import unicode_literals
|
|
|
|
import re
|
|
|
|
from .common import InfoExtractor
|
|
from ..utils import (
|
|
ExtractorError,
|
|
determine_ext,
|
|
compat_urllib_parse,
|
|
compat_urllib_request,
|
|
)
|
|
|
|
|
|
class GorillaVidIE(InfoExtractor):
|
|
IE_DESC = 'GorillaVid.in, daclips.in and movpod.in'
|
|
_VALID_URL = r'''(?x)
|
|
https?://(?P<host>(?:www\.)?
|
|
(?:daclips\.in|gorillavid\.in|movpod\.in))/
|
|
(?:embed-)?(?P<id>[0-9a-zA-Z]+)(?:-[0-9]+x[0-9]+\.html)?
|
|
'''
|
|
|
|
_FILE_NOT_FOUND_REGEX = r'>(?:404 - )?File Not Found<'
|
|
|
|
_TESTS = [{
|
|
'url': 'http://gorillavid.in/06y9juieqpmi',
|
|
'md5': '5ae4a3580620380619678ee4875893ba',
|
|
'info_dict': {
|
|
'id': '06y9juieqpmi',
|
|
'ext': 'flv',
|
|
'title': 'Rebecca Black My Moment Official Music Video Reaction-6GK87Rc8bzQ',
|
|
'thumbnail': 're:http://.*\.jpg',
|
|
},
|
|
}, {
|
|
'url': 'http://gorillavid.in/embed-z08zf8le23c6-960x480.html',
|
|
'md5': 'c9e293ca74d46cad638e199c3f3fe604',
|
|
'info_dict': {
|
|
'id': 'z08zf8le23c6',
|
|
'ext': 'mp4',
|
|
'title': 'Say something nice',
|
|
'thumbnail': 're:http://.*\.jpg',
|
|
},
|
|
}, {
|
|
'url': 'http://daclips.in/3rso4kdn6f9m',
|
|
'md5': '1ad8fd39bb976eeb66004d3a4895f106',
|
|
'info_dict': {
|
|
'id': '3rso4kdn6f9m',
|
|
'ext': 'mp4',
|
|
'title': 'Micro Pig piglets ready on 16th July 2009-bG0PdrCdxUc',
|
|
'thumbnail': 're:http://.*\.jpg',
|
|
}
|
|
}, {
|
|
'url': 'http://movpod.in/0wguyyxi1yca',
|
|
'only_matching': True,
|
|
}]
|
|
|
|
def _real_extract(self, url):
|
|
mobj = re.match(self._VALID_URL, url)
|
|
video_id = mobj.group('id')
|
|
|
|
webpage = self._download_webpage('http://%s/%s' % (mobj.group('host'), video_id), video_id)
|
|
|
|
if re.search(self._FILE_NOT_FOUND_REGEX, webpage) is not None:
|
|
raise ExtractorError('Video %s does not exist' % video_id, expected=True)
|
|
|
|
fields = dict(re.findall(r'''(?x)<input\s+
|
|
type="hidden"\s+
|
|
name="([^"]+)"\s+
|
|
(?:id="[^"]+"\s+)?
|
|
value="([^"]*)"
|
|
''', webpage))
|
|
|
|
if fields['op'] == 'download1':
|
|
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')
|
|
|
|
title = self._search_regex(r'style="z-index: [0-9]+;">([^<]+)</span>', webpage, 'title')
|
|
video_url = self._search_regex(r'file\s*:\s*\'(http[^\']+)\',', webpage, 'file url')
|
|
thumbnail = self._search_regex(r'image\s*:\s*\'(http[^\']+)\',', webpage, 'thumbnail', fatal=False)
|
|
|
|
formats = [{
|
|
'format_id': 'sd',
|
|
'url': video_url,
|
|
'ext': determine_ext(video_url),
|
|
'quality': 1,
|
|
}]
|
|
|
|
return {
|
|
'id': video_id,
|
|
'title': title,
|
|
'thumbnail': thumbnail,
|
|
'formats': formats,
|
|
}
|