1
1
mirror of https://github.com/ytdl-org/youtube-dl synced 2024-11-27 23:56:51 +01:00

[streamcloud] Add IE (Fixes #1801)

This commit is contained in:
Philipp Hagemeister 2013-11-22 17:19:22 +01:00
parent ab009f59ef
commit 02e4ebbbad
2 changed files with 66 additions and 0 deletions

View File

@ -128,6 +128,7 @@ from .spiegel import SpiegelIE
from .stanfordoc import StanfordOpenClassroomIE from .stanfordoc import StanfordOpenClassroomIE
from .statigram import StatigramIE from .statigram import StatigramIE
from .steam import SteamIE from .steam import SteamIE
from .streamcloud import StreamcloudIE
from .sztvhu import SztvHuIE from .sztvhu import SztvHuIE
from .teamcoco import TeamcocoIE from .teamcoco import TeamcocoIE
from .techtalks import TechTalksIE from .techtalks import TechTalksIE

View File

@ -0,0 +1,65 @@
# coding: utf-8
import re
import time
from .common import InfoExtractor
from ..utils import (
compat_urllib_parse,
compat_urllib_request,
)
class StreamcloudIE(InfoExtractor):
IE_NAME = u'streamcloud.eu'
_VALID_URL = r'https?://streamcloud\.eu/(?P<id>[a-zA-Z0-9_-]+)/(?P<fname>[^#?]*)\.html'
_TEST = {
u'url': u'http://streamcloud.eu/skp9j99s4bpz/youtube-dl_test_video_____________-BaW_jenozKc.mp4.html',
u'file': u'skp9j99s4bpz.mp4',
u'md5': u'6bea4c7fa5daaacc2a946b7146286686',
u'info_dict': {
u'title': u'youtube-dl test video \'/\\ ä ↭',
u'duration': 9,
},
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
orig_webpage = self._download_webpage(url, video_id)
fields = re.findall(r'''(?x)<input\s+
type="(?:hidden|submit)"\s+
name="([^"]+)"\s+
(?:id="[^"]+"\s+)?
value="([^"]*)"
''', orig_webpage)
post = compat_urllib_parse.urlencode(fields)
self.to_screen('%s: Waiting for timeout' % video_id)
time.sleep(12)
headers = {
b'Content-Type': b'application/x-www-form-urlencoded',
}
req = compat_urllib_request.Request(url, post, headers)
webpage = self._download_webpage(
req, video_id, note=u'Downloading video page ...')
title = self._html_search_regex(
r'<h1[^>]*>([^<]+)<', webpage, u'title')
video_url = self._search_regex(
r'file:\s*"([^"]+)"', webpage, u'video URL')
duration_str = self._search_regex(
r'duration:\s*"?([0-9]+)"?', webpage, u'duration', fatal=False)
duration = None if duration_str is None else int(duration_str)
thumbnail = self._search_regex(
r'image:\s*"([^"]+)"', webpage, u'thumbnail URL', fatal=False)
return {
'id': video_id,
'title': title,
'url': video_url,
'duration': duration,
'thumbnail': thumbnail,
}