How to generate CDATA block using JAXB?_问答_开发者

I am using JAXB to serialize my data to XML. The class code is simple as given below. I want to produce XML that contains CDATA blocks for the value of some Args. For example, current code produces this XML:

<command>
   <args>
      <arg name="test_id">1234</arg>
      <arg name="source">&lt;html>EMAIL&lt;/html></arg>
   </args>
</command>

I want to wrap the "source" arg in CDAT开发者_StackOverflow中文版A such that it looks like below:

<command>
   <args>
      <arg name="test_id">1234</arg>
      <arg name="source"><[![CDATA[<html>EMAIL</html>]]></arg>
   </args>
</command>

How can I achieve this in the below code?

@XmlRootElement(name="command")
public class Command {

        @XmlElementWrapper(name="args")
        protected List<Arg>  arg;
    }
@XmlRootElement(name="arg")
public class Arg {

        @XmlAttribute
        public String name;
        @XmlValue
        public String value;

        public Arg() {};

        static Arg make(final String name, final String value) {
            Arg a = new Arg();
            a.name=name; a.value=value;
            return a; }
    }

Note: I'm the EclipseLink JAXB (MOXy) lead and a member of the JAXB (JSR-222) expert group.

If you are using MOXy as your JAXB provider then you can leverage the @XmlCDATA extension:

package blog.cdata;

import javax.xml.bind.annotation.XmlRootElement;
import org.eclipse.persistence.oxm.annotations.XmlCDATA;

@XmlRootElement(name="c")
public class Customer {

   private String bio;

   @XmlCDATA
   public void setBio(String bio) {
      this.bio = bio;
   }

   public String getBio() {
      return bio;
   }

}

For More Information

http://bdoughan.blogspot.com/2010/07/cdata-cdata-run-run-data-run.html
http://blog.bdoughan.com/2011/05/specifying-eclipselink-moxy-as-your.html

Use JAXB's Marshaller#marshal(ContentHandler) to marshal into a ContentHandler object. Simply override the characters method on the ContentHandler implementation you are using (e.g. JDOM's SAXHandler, Apache's XMLSerializer, etc):

public class CDataContentHandler extends (SAXHandler|XMLSerializer|Other...) {
    // see http://www.w3.org/TR/xml/#syntax
    private static final Pattern XML_CHARS = Pattern.compile("[<>&]");

    public void characters(char[] ch, int start, int length) throws SAXException {
        boolean useCData = XML_CHARS.matcher(new String(ch,start,length)).find();
        if (useCData) super.startCDATA();
        super.characters(ch, start, length);
        if (useCData) super.endCDATA();
    }
}

This is much better than using the XMLSerializer.setCDataElements(...) method because you don't have to hardcode any list of elements. It automatically outputs CDATA blocks only when one is required.

Solution Review:

The answer of fred is just a workaround which will fail while validating the content when the Marshaller is linked to a Schema because you modify only the string literal and do not create CDATA sections. So if you only rewrite the String from foo to <![CDATA[foo]]> the length of the string is recognized by Xerces with 15 instead of 3.
The MOXy solution is implementation specific and does not work only with the classes of the JDK.
The solution with the getSerializer references to the deprecated XMLSerializer class.
The solution LSSerializer is just a pain.

I modified the solution of a2ndrade by using a XMLStreamWriter implementation. This solution works very well.

XMLOutputFactory xof = XMLOutputFactory.newInstance();
XMLStreamWriter streamWriter = xof.createXMLStreamWriter( System.out );
CDataXMLStreamWriter cdataStreamWriter = new CDataXMLStreamWriter( streamWriter );
marshaller.marshal( jaxbElement, cdataStreamWriter );
cdataStreamWriter.flush();
cdataStreamWriter.close();

Thats the CDataXMLStreamWriter implementation. The delegate class simply delegates all method calls to the given XMLStreamWriter implementation.

import java.util.regex.Pattern;
import javax.xml.stream.XMLStreamException;
import javax.xml.stream.XMLStreamWriter;

/**
 * Implementation which is able to decide to use a CDATA section for a string.
 */
public class CDataXMLStreamWriter extends DelegatingXMLStreamWriter
{
   private static final Pattern XML_CHARS = Pattern.compile( "[&<>]" );

   public CDataXMLStreamWriter( XMLStreamWriter del )
   {
      super( del );
   }

   @Override
   public void writeCharacters( String text ) throws XMLStreamException
   {
      boolean useCData = XML_CHARS.matcher( text ).find();
      if( useCData )
      {
         super.writeCData( text );
      }
      else
      {
         super.writeCharacters( text );
      }
   }
}

Here is the code sample referenced by the site mentioned above:

import java.io.File;
import java.io.StringWriter;

import javax.xml.bind.JAXBContext;
import javax.xml.bind.Marshaller;
import javax.xml.bind.Unmarshaller;
import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;

import org.apache.xml.serialize.OutputFormat;
import org.apache.xml.serialize.XMLSerializer;
import org.w3c.dom.Document;

public class JaxbCDATASample {

    public static void main(String[] args) throws Exception {
        // unmarshal a doc
        JAXBContext jc = JAXBContext.newInstance("...");
        Unmarshaller u = jc.createUnmarshaller();
        Object o = u.unmarshal(...);

        // create a JAXB marshaller
        Marshaller m = jc.createMarshaller();

        // get an Apache XMLSerializer configured to generate CDATA
        XMLSerializer serializer = getXMLSerializer();

        // marshal using the Apache XMLSerializer
        m.marshal(o, serializer.asContentHandler());
    }

    private static XMLSerializer getXMLSerializer() {
        // configure an OutputFormat to handle CDATA
        OutputFormat of = new OutputFormat();

        // specify which of your elements you want to be handled as CDATA.
        // The use of the '^' between the namespaceURI and the localname
        // seems to be an implementation detail of the xerces code.
        // When processing xml that doesn't use namespaces, simply omit the
        // namespace prefix as shown in the third CDataElement below.
        of.setCDataElements(
            new String[] { "ns1^foo",   // <ns1:foo>
                   "ns2^bar",   // <ns2:bar>
                   "^baz" });   // <baz>

        // set any other options you'd like
        of.setPreserveSpace(true);
        of.setIndenting(true);

        // create the serializer
        XMLSerializer serializer = new XMLSerializer(of);
        serializer.setOutputByteStream(System.out);

        return serializer;
    }
}

For the same reasons as Michael Ernst I wasn't that happy with most of the answers here. I could not use his solution as my requirement was to put CDATA tags in a defined set of fields - as in raiglstorfer's OutputFormat solution.

My solution is to marshal to a DOM document, and then do a null XSL transform to do the output. Transformers allow you to set which elements are wrapped in CDATA tags.

Document document = ...
jaxbMarshaller.marshal(jaxbObject, document);

Transformer nullTransformer = TransformerFactory.newInstance().newTransformer();
nullTransformer.setOutputProperty(OutputKeys.INDENT, "yes");
nullTransformer.setOutputProperty(OutputKeys.CDATA_SECTION_ELEMENTS, "myElement {myNamespace}myOtherElement");
nullTransformer.transform(new DOMSource(document), new StreamResult(writer/stream));

Further info here: http://javacoalface.blogspot.co.uk/2012/09/outputting-cdata-sections-with-jaxb.html

The following simple method adds CDATA support in JAX-B which does not support CDATA natively :

declare a custom simple type CDataString extending string to identify the fields that should be handled via CDATA
Create a custom CDataAdapter that parses and print content in CDataString
use JAXB bindings to link CDataString and you CDataAdapter. the CdataAdapter will add/remove to/from CdataStrings at Marshall/Unmarshall time
Declare a custom character escape handler that does not escape character when printing CDATA strings and set this as the Marshaller CharacterEscapeEncoder

Et voila, any CDataString element will be encapsulated with at Marshall time. At unmarshall time, the will automatically be removed.

Supplement of @a2ndrade's answer.

I find one class to extend in JDK 8. But noted that the class is in com.sun package. You can make one copy of the code in case this class may be removed in future JDK.

public class CDataContentHandler extends com.sun.xml.internal.txw2.output.XMLWriter {
  public CDataContentHandler(Writer writer, String encoding) throws IOException {
    super(writer, encoding);
  }

  // see http://www.w3.org/TR/xml/#syntax
  private static final Pattern XML_CHARS = Pattern.compile("[<>&]");

  public void characters(char[] ch, int start, int length) throws SAXException {
    boolean useCData = XML_CHARS.matcher(new String(ch, start, length)).find();
    if (useCData) {
      super.startCDATA();
    }
    super.characters(ch, start, length);
    if (useCData) {
      super.endCDATA();
    }
  }
}

How to use:

  JAXBContext jaxbContext = JAXBContext.newInstance(...class);
  Marshaller marshaller = jaxbContext.createMarshaller();
  StringWriter sw = new StringWriter();
  CDataContentHandler cdataHandler = new CDataContentHandler(sw,"utf-8");
  marshaller.marshal(gu, cdataHandler);
  System.out.println(sw.toString());

Result example:

<?xml version="1.0" encoding="utf-8"?>
<genericUser>
  <password><![CDATA[dskfj>><<]]></password>
  <username>UNKNOWN::UNKNOWN</username>
  <properties>
    <prop2>v2</prop2>
    <prop1><![CDATA[v1><]]></prop1>
  </properties>
  <timestamp/>
  <uuid>cb8cbc487ee542ec83e934e7702b9d26</uuid>
</genericUser>

As of Xerxes-J 2.9, XMLSerializer has been deprecated. The suggestion is to replace it with DOM Level 3 LSSerializer or JAXP's Transformation API for XML. Has anyone tried approach?

Just a word of warning: according to documentation of the javax.xml.transform.Transformer.setOutputProperty(...) you should use the syntax of qualified names, when indicating an element from another namespace. According to JavaDoc (Java 1.6 rt.jar):

"(...) For example, if a URI and local name were obtained from an element defined with , then the qualified name would be "{http://xyz.foo.com/yada/baz.html}foo. Note that no prefix is used."

Well this doesn't work - the implementing class from Java 1.6 rt.jar, meaning com.sun.org.apache.xalan.internal.xsltc.trax.TransformerImpl interprets elements belonging to a different namespace only then correctly, when they are declared as "http://xyz.foo.com/yada/baz.html:foo", because in the implementation someone is parsing it looking for the last colon. So instead of invoking:

transformer.setOutputProperty(OutputKeys.CDATA_SECTION_ELEMENTS, "{http://xyz.foo.com/yada/baz.html}foo")

which should work according to JavaDoc, but ends up being parsed as "http" and "//xyz.foo.com/yada/baz.html", you must invoke

transformer.setOutputProperty(OutputKeys.CDATA_SECTION_ELEMENTS, "http://xyz.foo.com/yada/baz.html:foo")

At least in Java 1.6.

The following code will prevent from encoding CDATA elements:

Marshaller marshaller = context.createMarshaller();
marshaller.setProperty(Marshaller.JAXB_ENCODING, "UTF-8");
marshaller.setProperty(Marshaller.JAXB_FORMATTED_OUTPUT, true);

StringWriter stringWriter = new StringWriter();
PrintWriter printWriter = new PrintWriter(stringWriter);
DataWriter dataWriter = new DataWriter(printWriter, "UTF-8", new CharacterEscapeHandler() {
    @Override
    public void escape(char[] buf, int start, int len, boolean b, Writer out) throws IOException {
        out.write(buf, start, len);
    }
});

marshaller.marshal(data, dataWriter);

System.out.println(stringWriter.toString());

It will also keep UTF-8 as your encoding.