Be more intelligent about sanitizing links with unsafe protocols

lib/banzai/filter/sanitization_filter.rb

@@ -7,7 +7,7 @@ module Banzai
     #
     # Extends HTML::Pipeline::SanitizationFilter with a custom whitelist.
     class SanitizationFilter < HTML::Pipeline::SanitizationFilter
-      UNSAFE_PROTOCOLS = %w(javascript :javascript data vbscript).freeze
+      UNSAFE_PROTOCOLS = %w(data javascript vbscript).freeze
 
       def whitelist
         whitelist = super
@@ -64,7 +64,12 @@ module Banzai
           return unless node.name == 'a'
           return unless node.has_attribute?('href')
 
-          if node['href'].start_with?(*UNSAFE_PROTOCOLS)
+          begin
+            uri = Addressable::URI.parse(node['href'])
+            uri.scheme.strip! if uri.scheme
+
+            node.remove_attribute('href') if UNSAFE_PROTOCOLS.include?(uri.scheme)
+          rescue Addressable::URI::InvalidURIError
             node.remove_attribute('href')
           end
         end

spec/lib/banzai/filter/sanitization_filter_spec.rb

@@ -149,10 +149,20 @@ describe Banzai::Filter::SanitizationFilter, lib: true do
         output: '<a href="java"></a>'
       },
 
+      'protocol-based JS injection: invalid URL char' => {
+        input: '<img src=java\script:alert("XSS")>',
+        output: '<img>'
+      },
+
       'protocol-based JS injection: spaces and entities' => {
         input:  '<a href=" &#14;  javascript:alert(\'XSS\');">foo</a>',
         output: '<a href="">foo</a>'
       },
+
+      'protocol whitespace' => {
+        input: '<a href=" http://example.com/"></a>',
+        output: '<a href="http://example.com/"></a>'
+      }
     }
 
     protocols.each do |name, data|
@@ -177,6 +187,16 @@ describe Banzai::Filter::SanitizationFilter, lib: true do
       expect(output.to_html).to eq '<a>XSS</a>'
     end
 
+    it 'disallows invalid URIs' do
+      expect(Addressable::URI).to receive(:parse).with('foo://example.com').
+        and_raise(Addressable::URI::InvalidURIError)
+
+      input = '<a href="foo://example.com">Foo</a>'
+      output = filter(input)
+
+      expect(output.to_html).to eq '<a>Foo</a>'
+    end
+
     it 'allows non-standard anchor schemes' do
       exp = %q{<a href="irc://irc.freenode.net/git">IRC</a>}
       act = filter(exp)