import os
import re
-import xml.etree.ElementTree
from .common import InfoExtractor
from .youtube import YoutubeIE
ExtractorError,
HEADRequest,
+ parse_xml,
smuggle_url,
unescapeHTML,
unified_strdate,
'params': {
'skip_download': False,
}
- }
+ },
+ # embed.ly video
+ {
+ 'url': 'http://www.tested.com/science/weird/460206-tested-grinding-coffee-2000-frames-second/',
+ 'info_dict': {
+ 'id': '9ODmcdjQcHQ',
+ 'ext': 'mp4',
+ 'title': 'Tested: Grinding Coffee at 2000 Frames Per Second',
+ 'upload_date': '20140225',
+ 'description': 'md5:06a40fbf30b220468f1e0957c0f558ff',
+ 'uploader': 'Tested',
+ 'uploader_id': 'testedcom',
+ },
+ # No need to test YoutubeIE here
+ 'params': {
+ 'skip_download': True,
+ },
+ },
]
def report_download_webpage(self, video_id):
else:
assert ':' in default_search
return self.url_result(default_search + url)
- video_id = os.path.splitext(url.split('/')[-1])[0]
+ video_id = os.path.splitext(url.rstrip('/').split('/')[-1])[0]
self.to_screen('%s: Requesting header' % video_id)
# Is it an RSS feed?
try:
- doc = xml.etree.ElementTree.fromstring(webpage.encode('utf-8'))
+ doc = parse_xml(webpage)
if doc.tag == 'rss':
return self._extract_rss(url, video_id, doc)
except compat_xml_parse_error:
if mobj is not None:
return self.url_result(mobj.group('url'), 'Facebook')
+ # Look for embedded VK player
+ mobj = re.search(r'<iframe[^>]+?src=(["\'])(?P<url>https?://vk\.com/video_ext\.php.+?)\1', webpage)
+ if mobj is not None:
+ return self.url_result(mobj.group('url'), 'VK')
+
# 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')
+ # Look for embed.ly
+ mobj = re.search(r'class=["\']embedly-card["\'][^>]href=["\'](?P<url>[^"\']+)', webpage)
+ if mobj is not None:
+ return self.url_result(mobj.group('url'))
+ mobj = re.search(r'class=["\']embedly-embed["\'][^>]src=["\'][^"\']*url=(?P<url>[^&]+)', webpage)
+ if mobj is not None:
+ return self.url_result(compat_urllib_parse.unquote(mobj.group('url')))
+
# Start with something easy: JW Player in SWFObject
mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
if mobj is None: