XOM API for Processing XML with Java

[Wolfgang Hoschek <wolfgang.hoschek AT mac.com>]

On Mar 15, 2006, at 2:00 AM, Verachten Bruno wrote:

> Hi,
>
> I'm fairly new to NUX & XOM, I only downloaded the jars yesterday.
> I'm trying to validate a BIG xml file thanks to XOM.
> I found a snippet in the Javadoc, so I tried this :
>
> Map schemaLocations = new HashMap();
> schemaLocations.put(new File("/tmp/p2pio.xsd"),
>                         "http://dsd.lbl.gov/p2pio-1.0";);
> Builder builder = new
> BuilderFactory().createW3CBuilder(schemaLocations);
>
> That sounds good to me, except that I need a StaxBuilder.
>
> StaxBuilder staxBuilder = new StaxBuilder();
> Document document = staxBuilder.build(reader);
>
> How can I link the two builders?

StAX isn't required for that. Both SAX and STAX are fully streaming  
APIs, hence can operate on arbitrary large input documents, and so  
can a Builder based on them, provided it has an appropriate NodeFactory.

If you only want to validate against a W3C XML Schema, configure the  
builder to have a NodeFactory that simply throws all events away,  
e.g. a XOMUtil.getNullNodeFactory(). If you'd like to build all or  
parts of the document, configure it with whatever is the  
corresponding NodeFactory. All this can be done by overriding  
BuilderFactory.newBuilder(). Example:

        protected Builder newBuilder(XMLReader parser, boolean validate) {
                return new Builder(parser, validate, 
XOMUtil.getNullNodeFactory());             
        }

If a STAX impl becomes available that can validate against W3C  
schema, such STAX impl could be plugged in instead by feeding its  
XMLInputFactory into the following nux-1.5 STAX method:

        /**
	 * Constructs and returns a Builder implementation that uses a STAX  
parser instead
	 * of a SAX parser. Can be used for polymorphic pluggability of SAX  
vs. STAX.
         *
         * @param inputFactory
	 *            a factory constructing STAX {@link XMLStreamReader}  
instances.
	 *            May be <code>null</code> in which case a default  
factory is
         *            used, producing a parser that will be namespace aware,
	 *            non-validating, support DTDs, and in text coalescing  
mode.
	 *            In this case the preferred implementation is Woodstox,  
if available.
         * @param factory
	 *            the node factory to stream into. May be <code>null</ 
code> in
	 *            which case the default XOM NodeFactory is used,  
building the
         *            full XML tree.
         * @return a Builder implementation using STAX instead of SAX
         */
	public static Builder createBuilder(XMLInputFactory inputFactory,  
NodeFactory factory)



I think Tatu is getting close to releasing a woodstox STAX impl using  
generic Sun-MSV validation, so with some luck that might cover not  
only RelaxNG but also W3C XML Schema. Tatu?

This week I'm swamped and at a family funeral. I hope I might get  
around to upload 1.5 some time around end of next week.

Thanks to Tatu's incredible and tireless efforts around the STAX spec  
and woodstox impl, it will soon be a conformant, reliable, complete  
and efficient alternative to SAX.

Wolfgang.