From 6398079fb74416997445cd6b6d4b92a9688ef864 Mon Sep 17 00:00:00 2001 From: Jon Williams Date: Wed, 16 Mar 2016 14:04:43 -0400 Subject: [PATCH] Fix protocol relative urls --- parsers/baseparser.py | 1 + 1 file changed, 1 insertion(+) diff --git a/parsers/baseparser.py b/parsers/baseparser.py index ef2a9eb5..ba09ca70 100644 --- a/parsers/baseparser.py +++ b/parsers/baseparser.py @@ -149,6 +149,7 @@ def feed_urls(cls): # If no http://, prepend domain name domain = '/'.join(feeder_url.split('/')[:3]) + urls = [url if not re.search('^//', url) else concat('http:', url) for url in urls] urls = [url if '://' in url else concat(domain, url) for url in urls] all_urls = all_urls + [url for url in urls if