webpage = self._download_webpage(webpage_url, video_id)
self.report_extraction(video_id)
- mobj = re.search(r'<link rel=\'image_src\' href=\'(http://is[0-9].myvideo\.de/de/movie[0-9]+/[a-f0-9]+)/thumbs/.*?\.jpg\' />',
+ mobj = re.search(r'<link rel=\'image_src\' href=\'(http://is[0-9].myvideo\.de/de/movie[0-9]+/[a-f0-9]+)/thumbs/.*?\.jpg\'',
webpage)
if mobj is None:
self._downloader.report_error(u'unable to extract media URL')
self._downloader.report_error(u'unable to find video information')
video_url = unescapeHTML(m.group('url'))
- m = re.search(r"class='player_page_h1'>\s+<a.*?>(?P<title>.*?)</a>", webpage)
+ m = re.search(r"<h1 class='player_page_h1'.*?>(?P<title>.*?)</h1>", webpage, flags=re.DOTALL)
if not m:
self._downloader.trouble(u'Cannot find video title')
- title = unescapeHTML(m.group('title'))
+ title = clean_html(m.group('title'))
m = re.search(r'<meta property="og:description" content="(?P<desc>.*?)"', webpage)
if m:
videoName=m.group('name')
webpage=self._download_webpage(url, video_id, 'Downloading \"%s\" page' % videoName)
# If the url includes the language we get the title translated
- title_RE=r'<h1><span id="altHeadline" >(?P<title>.*)</span></h1>'
+ title_RE=r'<span id="altHeadline" >(?P<title>.*)</span>'
title=re.search(title_RE, webpage).group('title')
info_RE=r'''<script\ type="text/javascript">var\ talkDetails\ =(.*?)
"id":(?P<videoID>[\d]+).*?
return [info]
class SpiegelIE(InfoExtractor):
- _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P<videoID>[0-9]+)(?:\.html)?(?:#.*)$'
+ _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P<videoID>[0-9]+)(?:\.html)?(?:#.*)?$'
def _real_extract(self, url):
m = re.match(self._VALID_URL, url)
}
return [info]
-class liveleakIE(InfoExtractor):
+class LiveLeakIE(InfoExtractor):
- _VALID_URL = r'^(?:http?://)?(?:\w+\.)?liveleak\.com/view\?(?:.*?)i=(?P<video_id>\d+)(?:.*)'
+ _VALID_URL = r'^(?:http?://)?(?:\w+\.)?liveleak\.com/view\?(?:.*?)i=(?P<video_id>[\w_]+)(?:.*)'
IE_NAME = u'liveleak'
def _real_extract(self, url):
self._downloader.trouble(u'ERROR: invalid URL: %s' % url)
return
- video_id = mobj.group(1)
- if video_id.endswith('/index.html'):
- video_id = video_id[:-len('/index.html')]
+ video_id = mobj.group('video_id')
webpage = self._download_webpage(url, video_id)
- video_url = u'http://edge.liveleak.com/80281E/u/u/ll2_player_files/mp55/player.swf?config=http://www.liveleak.com/player?a=config%26item_token=' + video_id
+ m = re.search(r'file: "(.*?)",', webpage)
+ if not m:
+ self._downloader.report_error(u'unable to find video url')
+ return
+ video_url = m.group(1)
+
m = re.search(r'<meta property="og:title" content="(?P<title>.*?)"', webpage)
if not m:
self._downloader.trouble(u'Cannot find video title')
- title = unescapeHTML(m.group('title'))
+ title = unescapeHTML(m.group('title')).replace('LiveLeak.com -', '').strip()
m = re.search(r'<meta property="og:description" content="(?P<desc>.*?)"', webpage)
if m:
else:
desc = None
+ m = re.search(r'By:.*?(\w+)</a>', webpage)
+ if m:
+ uploader = clean_html(m.group(1))
+ else:
+ uploader = None
info = {
'id': video_id,
'url': video_url,
'ext': 'mp4',
'title': title,
- 'description': desc
+ 'description': desc,
+ 'uploader': uploader
}
return [info]
TEDIE(),
MySpassIE(),
SpiegelIE(),
- liveleakIE(),
+ LiveLeakIE(),
GenericIE()
]