XOM API for Processing XML with Java

[Wolfgang Hoschek <whoschek AT lbl.gov>]

When bnux deserializing I noticed Attributes.checkPrefixConflict and 
Attributes.get(String) as prominent hotspots.
This is with a WURFL document such as this one here: 
http://wurfl.sourceforge.net/wurfl.zip

The patch below eliminates some unnecessary iteration and uses indexes 
instead of iterators. Makes the hotspot completely go away. Otherwise, 
behaviour is unchanged.


    void checkPrefixConflict(Attribute attribute) {

        String prefix = attribute.getNamespacePrefix();
        String namespaceURI = attribute.getNamespaceURI();



        // Look for conflicts
//        Iterator iterator = attributes.iterator();
//        while (iterator.hasNext()) {
//            Attribute a = (Attribute) iterator.next();
        for (int i=attributes.size(); --i >= 0; ) {
                   Attribute a = (Attribute) attributes.get(i);
            if (a.getNamespacePrefix().equals(prefix)) {
            	 if (a.getNamespaceURI().equals(namespaceURI)) return; // 
no need to look further
               // there can't be any more attrs with the same prefix 
but different URI.
               // the original code would continue to iterate over the 
remaining attributes!
              throw new NamespaceConflictException(
                        "Prefix of " + attribute.getQualifiedName()
                        + " conflicts with " + a.getQualifiedName());
            }
        }
    }


    Attribute get(String localName, String namespaceURI) {

//        Iterator iterator = attributes.iterator();
//        while (iterator.hasNext()) {
//        Attribute a = (Attribute) iterator.next();
        for (int i=attributes.size(); --i >= 0; ) {
           Attribute a = (Attribute) attributes.get(i);
            if (a.getLocalName().equals(localName)
             && a.getNamespaceURI().equals(namespaceURI)) {
                return a;
            }
        }

        return null;

    }