[generic] Follow redirects specified by Refresh HTTP header

This commit is contained in:
Sergey M․ 2015-03-17 23:51:40 +06:00
parent b1337948eb
commit 84f8101606

View File

@ -1270,8 +1270,14 @@ def filter_video(urls):
if not found: if not found:
found = re.search( found = re.search(
r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")' r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")'
r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};url=\'?([^\'"]+)', r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};\s*(?:URL|url)=\'?([^\'"]+)',
webpage) webpage)
if not found:
# Look also in Refresh HTTP header
refresh_header = head_response.headers.get('Refresh')
if refresh_header:
found = re.search(
r'[0-9]{,2};\s*(?:URL|url)=(.+)', refresh_header)
if found: if found:
new_url = found.group(1) new_url = found.group(1)
self.report_following_redirect(new_url) self.report_following_redirect(new_url)