Traversing DOM very very slow

Question

I've an XML Document quite big (240K) with a root, 800 first level nodes and an average of 6-10 second level nodes for each first level.
I read the document from a file (very fast) and the I loop 2 or 3 times using data to transfer them in the legacy system.
I din't find a single operation very slow but the sum of all traversing time is very large (more than 10 minutes).
May be I'm wrong, but I didn't find any loop or programming error.
Any suggestion or idea will be appreciated.
TIA
Tullio

Answer 1

As a simple example, consider the file Tullio.xml at the end of this posting. Using the sample code:import oracle.xml.parser.v2.*;
import java.io.*;
import org.w3c.dom.*;
public class Tullio {
public static void main(String[] a) throws Throwable {
DOMParser d = new DOMParser();
FileReader f = new FileReader("Tullio.xml");
d.parse(f);
XMLDocument x = d.getDocument();
long start = 0;
long end = 0;
long running = 0;
long times = 0;
for (; times < 20; times++) {
start = System.currentTimeMillis();
NodeList nl = x.selectNodes("//oggetto[@template='foo']");
end = System.currentTimeMillis();
running = running + (end - start);
int nodes = nl.getLength();
for (int z = 0; z < nodes; z++) {
processOggetto((Element)nl.item(z));
System.out.println("Avg select took " + (running / 20L) + "ms.");
times = 0;
running = 0;
for (; times < 20; times++) {
start = System.currentTimeMillis();
NodeList nl = x.selectNodes("/qualcosa/oggetto[@template='foo']");
end = System.currentTimeMillis();
running = running + (end - start);
int nodes = nl.getLength();
for (int z = 0; z < nodes; z++) {
processOggetto((Element)nl.item(z));
System.out.println("Avg select took " + (running / 20L) + "ms.");
static void processOggetto(Element e) throws Throwable {
// System.out.println(((XMLElement)e).valueOf("nome"));
}I get the results:
==================
Avg select took 40ms.
Avg select took 7ms.
So using the // is quite a bit slower. On your file with "6-10 nested levels of elements", it may make an even bigger difference.
Tullio.xml
<qualcosa>
<oggetto template="foo">
<nome>FooName_Start</nome>
</oggetto>
<oggetto template="foo2">
<nome>FooName</nome>
</oggetto>
<!-- Ripete "foo2" 300 volte -->
<oggetto template="foo">
<nome>FooName_Start</nome>
</oggetto>
</qualcosa> null