Help: to get XML child data in faster way?

Question

Dear all,
With the following xml file, how can I use xslt by Java to get any <data> child element's string content. For example, to get "ee ff gg hh" in <Data name="inputs2">ee ff gg hh</Data> [this is in <Info Date="03-05-2003"> branch]?
I want to get it in a quick way, so I do not want to get it by DOM. The question is what is the best way to get it in Java? Is it possible by SAX combined with XSLT (XPATH) - with Java Code?
<?xml version="1.0" encoding="UTF-8"?>
<Customer>
    <Infos>
        <Info Date="03-05-2003" >
            <Inputs name="John">
                <Data name="inputs1">a b c d</Data>
                <Data name="inputs2">aa bb cc dd</Data>
                <Data name="inputsn">an bn cn dn</Data>
            </Inputs>
        </Info>
        <Info Date="03-05-2003" >
            <Inputs name="Mike">
                <Data name="inputs1">e f g h</Data>
                <Data name="inputs2">ee ff gg hh</Data>
                <Data name="inputsn">en fn gn hn</Data>
            </Inputs>
        </Info>
    </Infos>
</Customer>Thanks
- John

Answer 1

The problem is my file could be very large in size,
and if "load the whole thing into memory", it will
throw out of memory exception. So the best thing would
be dynamically load it if possible.
SAX and XSLT actually can do some help, since whenever
it finds my requested node, it can stop processing -
but the problem here is: it has to reparse the XML
file from the beginning every time to find the
requested node - a very time-consuming job...
Any better solution?how about using sax to parse the file, and temporarily insert the records into a DB? You could then use SQL to do your queries.
parsing/inserting might take awhile; but a properly tuned DB (and the java.sql.statement.addBatch()) will minimize this - probably down to the order of a DOM parse.
Other solutions:
- regex and perl/python on the XML file
- splitting the big XML file into little XML files (one file per <Inputs> tag maybe?) and dealing with an individual node at a time
- parsing the file as SAX and caching the info you need (i.e. parse it once and cache the values); then look them up afterwards