Parse the following document into a DOM tree, and use
org.apache.xml.serialize.XMLSerializer to serialize it.  

Document:

    <?xml version="1.0" encoding="UTF-8"?>
    <top><![CDATA[]]]]><![CDATA[>]]>
    </top> 

Result:
    <?xml version="1.0" encoding="UTF-8"?>
    <top><![CDATA[]]>]]>
    </top>
  
In the second line of the result:

    <![CDATA[  is the CDATA header

    ]] is the content of the first CDATA section

    > is the content of the second CDATA section

    ]]> is the CDATA terminator

The problem is XMLSerializer concatentating CDATA sections without checking if
that produces "]]>".