Using XPATH to Marshall xml

[MattMacchia]

Hey Guys,
Im a newbie trying to setup spring batch and I have a question about the best way to handle my data import.
So, currently I need to import some xml data, but my data is deeply nested and I need to only extract two values out of about 50 in each record.

I see the XMLStreamMarshaller but it doesn't seem to support namespaces very effectively. Is there a way I could use an XPATH query to bind my values to my object?

thanks guys.
-matt

[Dave Syer]

I'm not really sure where you are getting an XMLStreamMarshaller from (there is an XStreamMarshaller in Spring OXM, I believe - maybe that?). Anyway, you could ask this question in the Web Services forum because that us the home of the OXM framework that we use in Batch. If you don't get anything there you might need to write your own EventReaderDeserializer.

[MattMacchia]

Ill give that a try.
Thanks!

[MattMacchia]

So, we tackled the xml issue by implementing our own EventReaderDesierializer.
It takes a list of namespaces that we are looking for and finds those values and puts them in a map. It's no xpath, but it at least will handle deep structures. Note: The namespaces need to be unique, you will get collisions if you don't.

public class NamespaceBasedEventReaderDeserializer implements EventReaderDeserializer
{
	private Log logger = LogFactory.getLog(NamespaceBasedEventReaderDeserializer.class);
	private List<String> namespaces = new ArrayList<String>();

	public Object deserializeFragment(XMLEventReader eventReader)
	{
		Map<String, String> aliases = new HashMap<String, String>();
		if(namespaces.isEmpty())
		{
			logger.warn("We are not looking for any data in our pathBasedEventReader - ");
			return aliases;
		}
		List<String> localPaths = new ArrayList<String>(namespaces);

		while(eventReader.hasNext())
		{
			XMLEvent event = (XMLEvent) eventReader.next();
			for(Iterator<String> iterator = localPaths.iterator(); iterator.hasNext();)
			{
				String key = iterator.next();
				if(StringUtils.contains(event.getLocation().toString(), key)
						&& event.isStartElement())
				{
					XMLEvent found = (XMLEvent) eventReader.next();
					if(found.getEventType() == XMLStreamConstants.CHARACTERS)
					{
						String value = found.toString().trim();
						aliases.put(key, value);
						iterator.remove();
					}
				}
			}
		}
		return aliases;
	}

	/**
	 * @return the namespaces
	 */
	public List<String> getNamespaces()
	{
		return namespaces;
	}

	/**
	 * @param namespaces the namespaces to set
	 */
	public void setNamespaces(List<String> namespaces)
	{
		this.namespaces = namespaces;
	}


}

Here is how we call it:

<property name="itemReader">
	<bean class="org.springframework.batch.item.xml.StaxEventItemReader">
		<property name="fragmentRootElementName"
			value="order" />
		<property name="resource"
			value="com/bc/batch/mlregister/mldata.xml" />
		<property name="fragmentDeserializer">
			<bean
				class="com.bc.batch.NamespaceBasedEventReaderDeserializer">
				<property name="namespaces">
					<list>
						<value>customerCode</value>
						<value>email</value>
					</list>
				</property>
			</bean>
		</property>
	</bean>
</property>

[lucasward]

Do you mind creating an issue against 1.1 with the code you just posted? That way we can look at it for the enhancements release.

[MattMacchia]

No problem.
Consider it done.