Support for parsing CDATA tags.

lib/oga/parser.y

@@ -1,8 +1,9 @@
 class Oga::Parser
 
 token T_NEWLINE T_SPACE
-token T_STRING
+token T_STRING T_TEXT
 token T_DOCTYPE_START T_DOCTYPE_END T_DOCTYPE_TYPE
+token T_CDATA_START T_CDATA_END
 
 options no_result_var
 
@@ -19,6 +20,7 @@ rule
 
   expression
     : doctype
+    | cdata
     ;
 
   # Doctypes
@@ -46,6 +48,16 @@ rule
       }
     ;
 
+  # CDATA tags
+
+  cdata
+    # <![CDATA[]]>
+    : T_CDATA_START T_CDATA_END { s(:cdata) }
+
+    # <![CDATA[foo]]>
+    | T_CDATA_START T_TEXT T_CDATA_END { s(:cdata, val[1]) }
+    ;
+
   whitespaces
     : whitespaces whitespace
     | whitespace

spec/oga/parser/cdata_spec.rb

@@ -0,0 +1,20 @@
+require 'spec_helper'
+
+describe Oga::Parser do
+  context 'cdata tags' do
+    example 'parse a cdata tag' do
+      parse_html('<![CDATA[foo]]>').should == s(:document, s(:cdata, 'foo'))
+    end
+
+    example 'parse an element inside a cdata tag' do
+      parse_html('<![CDATA[<p>foo</p>]]>').should == s(
+        :document,
+        s(:cdata, '<p>foo</p>')
+      )
+    end
+
+    example 'parse double brackets inside a cdata tag' do
+      parse_html('<![CDATA[]]]]>').should == s(:document, s(:cdata, ']]'))
+    end
+  end
+end