Merge branch 'bvl-security-9-1-validate-urls-in-markdown-using-uri'

(security-9-1) Add correct `rel` attributes to external links when rendering markdown

See merge request !2097

1 files changed, 18 insertions, 18 deletions

diff --git a/lib/banzai/filter/external_link_filter.rb b/lib/banzai/filter/external_link_filter.rb
index d67d466bce8..7d15a0f6d44 100644
--- a/lib/banzai/filter/external_link_filter.rb
+++ b/lib/banzai/filter/external_link_filter.rb
@@ -2,16 +2,17 @@ module Banzai
   module Filter
     # HTML Filter to modify the attributes of external links
     class ExternalLinkFilter < HTML::Pipeline::Filter
+      SCHEMES = ['http', 'https', nil].freeze
+
       def call
         links.each do |node|
-          href = href_to_lowercase_scheme(node["href"].to_s)
+          uri = uri(node['href'].to_s)
+          next unless uri
 
-          unless node["href"].to_s == href
-            node.set_attribute('href', href)
-          end
+          node.set_attribute('href', uri.to_s)
 
-          if href =~ %r{\A(https?:)?//[^/]} && external_url?(href)
-            node.set_attribute('rel', 'nofollow noreferrer')
+          if SCHEMES.include?(uri.scheme) && external_url?(uri)
+            node.set_attribute('rel', 'nofollow noreferrer noopener')
             node.set_attribute('target', '_blank')
           end
         end
@@ -21,27 +22,26 @@ module Banzai
 
       private
 
+      def uri(href)
+        URI.parse(href)
+      rescue URI::InvalidURIError
+        nil
+      end
+
       def links
         query = 'descendant-or-self::a[@href and not(@href = "")]'
         doc.xpath(query)
       end
 
-      def href_to_lowercase_scheme(href)
-        scheme_match = href.match(/\A(\w+):\/\//)
-
-        if scheme_match
-          scheme_match.to_s.downcase + scheme_match.post_match
-        else
-          href
-        end
-      end
+      def external_url?(uri)
+        # Relative URLs miss a hostname
+        return false unless uri.hostname
 
-      def external_url?(url)
-        !url.start_with?(internal_url)
+        uri.hostname != internal_url.hostname
       end
 
       def internal_url
-        @internal_url ||= Gitlab.config.gitlab.url
+        @internal_url ||= URI.parse(Gitlab.config.gitlab.url)
       end
     end
   end