Urgent - Need an answer - XML Java API

Question

This is the 3rd time I've posted this question in last week - Oracle Support: How can I get an answer to this?
This question is about how to free up the memory of XMLDocuments, Parsers, etc using the Java APIs - not the PLSQL wrappers.
I'm writing an app using the DocumentSplitter class (from Steve M.'s book) that allows for the SAX parsing of large XML documents. But this question is really about the Oracle Java XML API.
Is there a way to free the memory of each of the "temporary" XMLDocuments that get created? I know there's a XMLNodeCover.freeDocument(int id)
method in the xmlplsql.jar- where do you get the id from? I'm trying XMLNodeCover.getDocumentId(curDoc) - but that doesn't seem to work.
And can I use that same method for each Node and Element that gets created also?
Also, would like to free the memory for the XSL processor, SAXParser, and XSLStylesheet - but this is even less obvious (although not as critical).
I'm actually getting out-of-memory errors even sooner when using the SAX parser then when using the DOM parser, and I think its due to this issue.
I know that there are PL/SQL wrappers for all these, but I'm trying to use the DocumentSplitter and associated custom classes
in only Java stored procs, and then writing a single XSLT.transform wrapper. This way its largely invisible to the PLSQL layer.
Thanks

Answer 1

Hi - thanks for your response.
Actually, yes, I did change the split method to return a CLOB. The big picture is I'm trying to use the DocumentSplitter
class to assist in the transformation of large XML documents to html or tab-delimited output. Using the standard DOM parser,
I eventually get an out-of-memory situation. So now I'm trying the DocumentSplitter to do a "mini" XSL transform on each <ROW>.
The good news is it works for small to medium size reports. But its slower and runs out of memory quicker.
Below is some of the relevant code. I also wrote associated classes to wrap the DocumentSplitter and to be the 'handler' that transfoms the XML when a <ROW> is found, and returns a string (see 'endElement' method below).
I'm freeing the CLOBS in PLSQL after reading them - do I need to free them from JVM memory seprately? If so, how can I do that if I need to pass them to PLSQL?
Thanks much,
Eric
// Split a large XML document into N subdocuments, each one identified
// by the an opening 'splitElement' element. Invoke an XMLDocumentHandler
// to process each subdocument encountered.
public CLOB split(String splitElement, CLOB inputCLOB, CLOB outputCLOB)
throws Exception, SAXParseException,SAXException,IOException, SQLException {
this.inputCLOB = inputCLOB;
this.outputCLOB = outputCLOB;
this.splitOnElement = splitElement;
// Create a new SAXParser
SAXParser parser = new SAXParser();
// Register the current instance of this class as the Document Handler
parser.setDocumentHandler(this);
// Create ascii stream from CLOB source
InputStream inputStream = inputCLOB.getAsciiStream();
// Start parsing the stream of XML
parser.parse(inputStream);
Writer out = this.outputCLOB.getCharacterOutputStream();
// Set buffer to be the optimal size
int chunk = outputCLOB.getChunkSize();
char[] buffer = new char[chunk];
// Create a reader from the output string
StringReader reader = new StringReader(this.outputString);
int length = -1;
// Now write from output string to output CLOB
while ((length = reader.read(buffer)) != -1) {
out.write(buffer, 0, length);
// Close streams
reader.close();
out.close();
inputStream.close();
return outputCLOB;
portion of endElement method:
// Handle the </Element> "end-element" parsing event
public void endElement(String name) throws SAXException {
// If we're NOT building a subdocument, we don't care. Just return.
if (curDoc == null) return;
// If this is the endElement event for the subdocument splitElement
// then we're done with the subdocument and are ready to call the
// handler to handle it.
if (name.equals(splitOnElement)) {
if (curDoc != null) {
try {
// Call the XMLDocumentHandler.handle() method for current subdoc
//Append the output to the growing output string
if (this.outputString == null) {
this.outputString = handler.handleDocument(curDoc);
} else {
this.outputString = this.outputString + handler.handleDocument(curDoc);
catch (Exception e) {
System.out.println(e.getMessage());
e.printStackTrace();
}