# ooyala video
{
'url': 'http://www.rollingstone.com/music/videos/norwegian-dj-cashmere-cat-goes-spartan-on-with-me-premiere-20131219',
+ 'file': 'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ.mp4',
'md5': '5644c6ca5d5782c1d0d350dad9bd840c',
'info_dict': {
'id': 'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ',
'ext': 'mp4',
- 'title': '2cc213299525360.mov', #that's what we get
+ 'title': '2cc213299525360.mov', # that's what we get
},
},
]
def _real_extract(self, url):
parsed_url = compat_urlparse.urlparse(url)
if not parsed_url.scheme:
- self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
- return self.url_result('http://' + url)
+ default_search = self._downloader.params.get('default_search')
+ if default_search is None:
+ default_search = 'auto'
+
+ if default_search == 'auto':
+ if '/' in url:
+ self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
+ return self.url_result('http://' + url)
+ else:
+ return self.url_result('ytsearch:' + url)
+ else:
+ assert ':' in default_search
+ return self.url_result(default_search + url)
video_id = os.path.splitext(url.split('/')[-1])[0]
self.to_screen('%s: Requesting header' % video_id)
# Look for embedded (iframe) Vimeo player
mobj = re.search(
- r'<iframe[^>]+?src="(https?://player.vimeo.com/video/.+?)"', webpage)
+ r'<iframe[^>]+?src="((?:https?:)?//player.vimeo.com/video/.+?)"', webpage)
if mobj:
player_url = unescapeHTML(mobj.group(1))
surl = smuggle_url(player_url, {'Referer': url})
if mobj is not None:
return self.url_result(mobj.group('url'), 'Novamov')
+ # Look for embedded Facebook player
+ mobj = re.search(
+ r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
+ if mobj is not None:
+ return self.url_result(mobj.group('url'), 'Facebook')
+
+ # Look for embedded Huffington Post player
+ mobj = re.search(
+ r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live.huffingtonpost\.com/.+?)\1', webpage)
+ if mobj is not None:
+ return self.url_result(mobj.group('url'), 'HuffPost')
+
# Start with something easy: JW Player in SWFObject
mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
if mobj is None:
mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
if mobj is None:
# Broaden the search a little bit: JWPlayer JS loader
- mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http[^\'"]*)', webpage)
+ mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
if mobj is None:
# Try to find twitter cards info
mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)