HTML escape YAML after parsing to prevent invalidating YAML string

lib/gollum-lib/filter/yaml.rb

@@ -10,7 +10,7 @@ def extract(data)
     data.gsub!(YAML_FRONT_MATTER_REGEXP) do
       @markup.metadata ||= {}
       begin
-        frontmatter = ::YAML.safe_load(sanitize(Regexp.last_match[1]))
+        frontmatter = sanitize_frontmatter(::YAML.safe_load(Regexp.last_match[1]))
         @markup.metadata.merge!(frontmatter) if frontmatter.respond_to?(:keys) && frontmatter.respond_to?(:values)
       rescue ::Psych::SyntaxError, ::Psych::DisallowedClass, ::Psych::BadAlias => error
         @markup.metadata['errors'] ||= []
@@ -24,4 +24,23 @@ def extract(data)
   def process(data)
     data
   end
+
+  private
+
+  def sanitize_frontmatter(obj)
+    case obj
+    when Hash
+      obj.map do |k, v|
+        [sanitize(k.to_s), sanitize_frontmatter(v)]
+      end.to_h
+    when Array
+      obj.map! do |v|
+        sanitize_frontmatter(v)
+      end
+    when String
+      sanitize(obj)
+    else
+      obj
+    end
+  end
 end

test/filter/test_yaml.rb

@@ -0,0 +1,75 @@
+path = File.join(File.dirname(__FILE__), "..", "helper")
+require File.expand_path(path)
+
+context "Gollum::Filter::YAML" do
+  setup do
+    @page = mock_page
+    @markup = Gollum::Markup.new(@page)
+    @filter = Gollum::Filter::YAML.new(@markup)
+  end
+
+  def filter(content)
+    @filter.process(@filter.extract(content))
+  end
+
+  test 'process yaml' do
+    markup = <<~EOF
+    ---
+    Literal Scalar: |
+      abc
+
+      123
+    Folded Scalar: >
+      abc
+
+      123
+    Escaped: "abc\n\n123"
+    ---
+
+    # Markdown content here
+    EOF
+
+    result = {"Escaped"=> "abc\n" + "123",
+      "Folded Scalar" => "abc\n" + "123\n",
+      "Literal Scalar" => "abc\n" + "\n" + "123\n"
+    }
+
+    assert_equal "# Markdown content here\n", filter(markup)
+    assert_nil @markup.metadata['errors']
+    assert_equal result, @markup.metadata
+  end
+
+  test 'escape yaml' do
+    markup = <<~EOF
+    ---
+    BadStuffInKey<script>bad()</script>: foo
+    Literal Scalar: |
+      <script>foo</script>
+
+      123
+    Folded Scalar: >
+      >abc
+
+      <123
+    Escaped: "abc<script>123</script>"
+    NestedBadStuff:
+      Baz:
+        - [1, "<script>bad()</script>"]
+        - [1, 2, 3]
+    ---
+
+    # Markdown content here
+    EOF
+
+    result = {"Escaped"=> "abc",
+      "Folded Scalar" => "&gt;abc\n" + "&lt;123\n",
+      "Literal Scalar" => "\n" + "\n" + "123\n",
+      "BadStuffInKey" => "foo",
+      "NestedBadStuff"=> {"Baz"=>[[1, ""], [1, 2, 3]]}
+    }
+    filter(markup)   
+    assert_nil @markup.metadata['errors']
+    assert_equal result, @markup.metadata
+  end
+
+end

test/helper.rb

@@ -88,6 +88,10 @@ def metadata(val={})
     val
   end
 
+  def sanitizer
+    Gollum::Sanitization.new(Gollum::Markup.to_xml_opts)
+  end
+
   attr_reader :repo_is_bare
   attr_reader :base_path
 end