summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorRobert Speicher <rspeicher@gmail.com>2016-03-14 16:07:51 -0400
committerRobert Speicher <rspeicher@gmail.com>2016-03-14 18:05:01 -0400
commit41de7b345b0abdaba2f0d7614ebdb1cc7310a5fb (patch)
tree2d6f9353aeab777469fee6fbcf66c0b8d1ab2cf2
parentb782e7c9732275175d79b485fcddc85e575d48dd (diff)
downloadgitlab-ce-rs-issue-14220.tar.gz
Be more intelligent about sanitizing links with unsafe protocolsrs-issue-14220
This prevents false matches on relative links like `[database](database.md)`. Closes https://gitlab.com/gitlab-org/gitlab-ce/issues/14220
-rw-r--r--lib/banzai/filter/sanitization_filter.rb9
-rw-r--r--spec/lib/banzai/filter/sanitization_filter_spec.rb20
2 files changed, 27 insertions, 2 deletions
diff --git a/lib/banzai/filter/sanitization_filter.rb b/lib/banzai/filter/sanitization_filter.rb
index abd79b329ae..e8011519608 100644
--- a/lib/banzai/filter/sanitization_filter.rb
+++ b/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
diff --git a/spec/lib/banzai/filter/sanitization_filter_spec.rb b/spec/lib/banzai/filter/sanitization_filter_spec.rb
index 4a7b00c7660..27ce312b11c 100644
--- a/spec/lib/banzai/filter/sanitization_filter_spec.rb
+++ b/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)