SAX performence problem

Question

Hi all
I am suposed to replace a paser already made wich is very slow by a parser made like SAX, DOM or JDOM.
My problem is that my SAx program takes the same time to anylise than the otehr rogram. My question is . Is there another way to get all the values (faster) of an XML document than this way:
public void startElement(String uri,String name,String qname,Attributes atts)throws SAXException{
//      System.out.print(" uri:"+uri+'\n');
//      System.out.print(" name:"+name+'\n');
//      System.out.print(" qname:"+qname+'\n');
//      System.out.print("\n att:"+atts+'\n');
          if (name.trim().equals("MEMBRE")){
               a++;
     if(objetMembre.nomMembre!=null){
               objetVPN.membres.addElement(objetMembre);
               objetMembre=new ObjetMembre();
/* if (name.trim().equals("SITE")){
               a++;
//     System.out.print(a+"un membre");
     if(objetMembre.nomMembre!=null){
     //          System.out.println(a+"ajout dans objetVPN");
               objetVPN.membres.addElement(objetMembre);
     //          System.out.println("new Membre");
               objetMembre=new ObjetMembre();
if (name.trim().equals("UTILISATEUR")){
               a++;
//     System.out.print(a+"un membre");
     if(objetMembre.nomMembre!=null){
     //          System.out.println(a+"ajout dans objetVPN");
               objetVPN.membres.addElement(objetMembre);
     //          System.out.println("new Membre");
               objetMembre=new ObjetMembre();
          else if (name.trim().equals("VERSION_OUTIL"))
     balise=VERSION_OUTIL;
     public void characters(char[] ch,int start,int length){
     StringBuffer temp=new StringBuffer();
     switch(balise){
     case VERSION_OUTIL:
          temp.append(ch,start,length);
          if (!temp.toString().trim().equals("")){
          versionOutil=temp.toString();
          temp=new StringBuffer();
     break;

Answer 1

Hi Cecile,
The performance problem may be in the code you have hooked into the method
public void characters(char[] ch,int start,int length)
Depending on the parser, this method may be called once for every character, or once for every value, or any arbitrary number of times, particularly if the node contains the likes of &
It also gets called for whitespace between nodes.
Anyhow, in your implementation of this method, you're creating new objects, which can get performance-costly.
For best performance, try the following approach:
Create a CharArrayWriter object as a member of your class. Inside the characters method, simply write the characters to the chararraywriter instance.
Inside the startElement method, call reset() on the writer
Inside the endElement method, call toString() on the writer to extract the value of the current node. This is very convenient because you have the name and the complete value of the node available to do your switch.
Using this in combination with the Xerces (or other SAX parser) should give you pretty good results.