Getting double underscores in xml when using StaxEventItemWriter

[GON]

Hi all,

I'm new to spring batch and trying to use the StaxEventItemReader/Writer using XstreamMarshaller. The job runs ok but some of the xml tags/variables contains underscore, i.e job_ident, and they all come out with double underscores in the output file(job__ident). I think this is caused by Xstream using _ as escape character by default, thus resulting in the double underscores.

Is there a way to configure StaxEventItemWriter or the XstreamMarshaller so that Xstream will use another escape character?

Thanks,
GON

[Dave Syer]

That's not really anything to do with Batch - we just delegate to the OXM marshaller. Try the Spring OXM (http://static.springsource.org/sprin.../html/oxm.html) and XStream documentation for more help.

[GON]

Hi

In case others will experience the same problem some time in the future:

I solved this by extending the XstreamMarshaller and overriding the method

protected void marshalSaxHandlers(Object graph, ContentHandler contentHandler, LexicalHandler lexicalHandler) throws XmlMappingException
  {
    SaxWriter saxWriter = new SaxWriter();
    saxWriter.setContentHandler(contentHandler);
    marshal(graph, saxWriter);
  }

with

@Override
	protected void marshalSaxHandlers(Object graph, ContentHandler contentHandler, LexicalHandler lexicalHandler) throws XmlMappingException {
		SaxWriter saxWriter = new SaxWriter(new XmlFriendlyReplacer("ddd","_"));
	    saxWriter.setContentHandler(contentHandler);
	    marshal(graph, saxWriter);
	}

Don't know if this is the best solution, but it seems to work ok for me so far.

[gonzalad]

Hello,

I've just had the same problem today with Spring Batch.
I know XStreamMashaller class is defined by Spring Framework, but since this problem happens mostly with Spring Batch usage (I mostly use XStreamMashaller in my batch applications), I wanted to have a Spring Batch user or dev point of view.

I find XStreamMashaller configuration a bit tricky whenever I need to handle '_' escaping.

Is it possible to simplify this and to have the same behaviour for every xml source / result type (stream / sax / dom / stax) ?

For stream source / result type, we just need to configure streamDriver in XStreamMashaller.
For Sax, Stax, DOM, we need to override one of the following methods :

• marshalSaxHandlers
• marshalDomNode
• ou equivalent unmarshal

Marshalling Stream Results
For instance, when using the following code to unmarshal an object :

StringResult lResult = new StringResult();
DocFileNet lDocument = new DocFileNet();
lDocument.setCanalOperateur("canal1");
marshaller.marshal(lDocument, lResult);

Customizing XStreamMarshaller is sufficient to handle _ escaping.
So I have :

<bean id="indexFileNetMarshaller" class="org.springframework.oxm.xstream.XStreamMarshaller">
    <property name="streamDriver">
          <bean class="com.thoughtworks.xstream.io.xml.XppDriver">
              <constructor-arg>
                  <bean class="com.thoughtworks.xstream.io.xml.XmlFriendlyReplacer">
                      <constructor-arg index="0" value="_-"/>
                      <constructor-arg index="1" value="_"/>
                  </bean>
              </constructor-arg>
          </bean>
      </property>
      <property name="annotatedClasses">
          <list>
              <value>com.xxx.demat.filenet.entite.DocFileNet</value>
          </list>
      </property>
</bean>

Marshalling With Spring Batch StaxEventItemWriter

I need to override marshalSaxHandlers method in XStreamMarshaller as noted in this thread by GONP.

Thanks very much for your input.