mirror of
https://codeberg.org/polarisfm/youtube-dl
synced 2024-11-13 21:34:31 +01:00
57 lines
1.8 KiB
Python
57 lines
1.8 KiB
Python
#coding: utf-8
|
|
|
|
import re
|
|
|
|
from .common import InfoExtractor
|
|
from ..utils import (
|
|
ExtractorError,
|
|
HEADRequest,
|
|
)
|
|
|
|
|
|
class AparatIE(InfoExtractor):
|
|
_VALID_URL = r'^https?://(?:www\.)?aparat\.com/(?:v/|video/video/embed/videohash/)(?P<id>[a-zA-Z0-9]+)'
|
|
|
|
_TEST = {
|
|
u'url': u'http://www.aparat.com/v/wP8On',
|
|
u'file': u'wP8On.mp4',
|
|
u'md5': u'6714e0af7e0d875c5a39c4dc4ab46ad1',
|
|
u'info_dict': {
|
|
u"title": u"تیم گلکسی 11 - زومیت",
|
|
},
|
|
#u'skip': u'Extremely unreliable',
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
m = re.match(self._VALID_URL, url)
|
|
video_id = m.group('id')
|
|
|
|
# Note: There is an easier-to-parse configuration at
|
|
# http://www.aparat.com/video/video/config/videohash/%video_id
|
|
# but the URL in there does not work
|
|
embed_url = (u'http://www.aparat.com/video/video/embed/videohash/' +
|
|
video_id + u'/vt/frame')
|
|
webpage = self._download_webpage(embed_url, video_id)
|
|
|
|
video_urls = re.findall(r'fileList\[[0-9]+\]\s*=\s*"([^"]+)"', webpage)
|
|
for i, video_url in enumerate(video_urls):
|
|
req = HEADRequest(video_url)
|
|
res = self._request_webpage(
|
|
req, video_id, note=u'Testing video URL %d' % i, errnote=False)
|
|
if res:
|
|
break
|
|
else:
|
|
raise ExtractorError(u'No working video URLs found')
|
|
|
|
title = self._search_regex(r'\s+title:\s*"([^"]+)"', webpage, u'title')
|
|
thumbnail = self._search_regex(
|
|
r'\s+image:\s*"([^"]+)"', webpage, u'thumbnail', fatal=False)
|
|
|
|
return {
|
|
'id': video_id,
|
|
'title': title,
|
|
'url': video_url,
|
|
'ext': 'mp4',
|
|
'thumbnail': thumbnail,
|
|
}
|