Merge remote-tracking branch 'origin/master'

This commit is contained in:
Philipp Hagemeister 2014-02-04 16:33:06 +01:00
commit acf1555d76
3 changed files with 116 additions and 1 deletions

View File

@ -127,6 +127,7 @@ from .mit import TechTVMITIE, MITIE
from .mixcloud import MixcloudIE from .mixcloud import MixcloudIE
from .mpora import MporaIE from .mpora import MporaIE
from .mofosex import MofosexIE from .mofosex import MofosexIE
from .mooshare import MooshareIE
from .mtv import ( from .mtv import (
MTVIE, MTVIE,
MTVIggyIE, MTVIggyIE,

View File

@ -31,7 +31,7 @@ class LifeNewsIE(InfoExtractor):
webpage = self._download_webpage('http://lifenews.ru/mobile/news/%s' % video_id, video_id, 'Downloading page') webpage = self._download_webpage('http://lifenews.ru/mobile/news/%s' % video_id, video_id, 'Downloading page')
video_url = self._html_search_regex( video_url = self._html_search_regex(
r'<video.*?src="([^"]+)"></video>', webpage, 'video URL') r'<video.*?src="([^"]+)".*?></video>', webpage, 'video URL')
thumbnail = self._html_search_regex( thumbnail = self._html_search_regex(
r'<video.*?poster="([^"]+)".*?"></video>', webpage, 'video thumbnail') r'<video.*?poster="([^"]+)".*?"></video>', webpage, 'video thumbnail')

View File

@ -0,0 +1,114 @@
from __future__ import unicode_literals
import re
import time
from .common import InfoExtractor
from ..utils import (
ExtractorError,
compat_urllib_request,
compat_urllib_parse,
)
class MooshareIE(InfoExtractor):
IE_NAME = 'mooshare'
IE_DESC = 'Mooshare.biz'
_VALID_URL = r'http://mooshare\.biz/(?P<id>[\da-z]{12})'
_TESTS = [
{
'url': 'http://mooshare.biz/8dqtk4bjbp8g',
'md5': '4e14f9562928aecd2e42c6f341c8feba',
'info_dict': {
'id': '8dqtk4bjbp8g',
'ext': 'mp4',
'title': 'Comedy Football 2011 - (part 1-2)',
'duration': 893,
},
},
{
'url': 'http://mooshare.biz/aipjtoc4g95j',
'info_dict': {
'id': 'aipjtoc4g95j',
'ext': 'mp4',
'title': 'Orange Caramel Dashing Through the Snow',
'duration': 212,
},
'params': {
# rtmp download
'skip_download': True,
}
}
]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
page = self._download_webpage(url, video_id, 'Downloading page')
if re.search(r'>Video Not Found or Deleted<', page) is not None:
raise ExtractorError(u'Video %s does not exist' % video_id, expected=True)
hash_key = self._html_search_regex(r'<input type="hidden" name="hash" value="([^"]+)">', page, 'hash')
title = self._html_search_regex(r'(?m)<div class="blockTitle">\s*<h2>Watch ([^<]+)</h2>', page, 'title')
download_form = {
'op': 'download1',
'id': video_id,
'hash': hash_key,
}
request = compat_urllib_request.Request(
'http://mooshare.biz/8dqtk4bjbp8g', compat_urllib_parse.urlencode(download_form))
request.add_header('Content-Type', 'application/x-www-form-urlencoded')
self.to_screen('%s: Waiting for timeout' % video_id)
time.sleep(5)
video_page = self._download_webpage(request, video_id, 'Downloading video page')
thumbnail = self._html_search_regex(r'image:\s*"([^"]+)",', video_page, 'thumbnail', fatal=False)
duration_str = self._html_search_regex(r'duration:\s*"(\d+)",', video_page, 'duration', fatal=False)
duration = int(duration_str) if duration_str is not None else None
formats = []
# SD video
mobj = re.search(r'(?m)file:\s*"(?P<url>[^"]+)",\s*provider:', video_page)
if mobj is not None:
formats.append({
'url': mobj.group('url'),
'format_id': 'sd',
'format': 'SD',
})
# HD video
mobj = re.search(r'\'hd-2\': { file: \'(?P<url>[^\']+)\' },', video_page)
if mobj is not None:
formats.append({
'url': mobj.group('url'),
'format_id': 'hd',
'format': 'HD',
})
# rtmp video
mobj = re.search(r'(?m)file: "(?P<playpath>[^"]+)",\s*streamer: "(?P<rtmpurl>rtmp://[^"]+)",', video_page)
if mobj is not None:
formats.append({
'url': mobj.group('rtmpurl'),
'play_path': mobj.group('playpath'),
'rtmp_live': False,
'ext': 'mp4',
'format_id': 'rtmp',
'format': 'HD',
})
return {
'id': video_id,
'title': title,
'thumbnail': thumbnail,
'duration': duration,
'formats': formats,
}