lib/feedzirra/parser/atom.rb in feedzirra-0.5.0 vs lib/feedzirra/parser/atom.rb in feedzirra-0.6.0

@@ -1,7 +1,6 @@
 module Feedzirra
-
   module Parser
     # Parser for dealing with Atom feeds.
     class Atom
       include SAXMachine
       include FeedUtilities
@@ -22,9 +21,41 @@
       end
 
       def feed_url
         @feed_url ||= links.first
       end
+
+      def self.preprocess(xml)
+        Preprocessor.new(xml).to_xml
+      end
+
+      class Preprocessor
+        def initialize(xml)
+          @xml = xml
+        end
+
+        def to_xml
+          process_nodes
+          doc.to_xml
+        end
+
+        private
+
+        def process_nodes
+          nodes.each { |node| node.content = raw_html(node) unless node.cdata? }
+        end
+
+        def nodes
+          doc.search 'entry > content[type="xhtml"]'
+        end
+
+        def raw_html(node)
+          CGI.unescape_html node.inner_html
+        end
+
+        def doc
+          @doc ||= Nokogiri::XML @xml
+        end
+      end
     end
   end
-
 end