XML Filtering using XPath

[Eurig Jones]

Hi,

I have a problem to solve. I need to filter very large XML documents using XPath. I need to provide a list of xpaths to generate a smaller versions of the documents. for example...

Input:

<aaa>
    <bbb>Hello</bbb>
    <ccc>123</ccc>
</aaa>
<zzz>
    <yyy>
        <xxx>David</xxx>
        <vvv>true</vvv>
    </yyy>
</zzz>

XPath

aaa/bbb
zzz/yyy

Output:

<aaa>
    <bbb>Hello</bbb>
</aaa>
<zzz>
    <yyy>
        <xxx>David</xxx>
        <vvv>true</vvv>
    </yyy>
</zzz>

What is a good way to do this using Java APIs?

Thanks!
Eurig

[William Brogden]

If your input XML document is truly "very large" - too large to get in memory as a DOM - the standard library XPath will not help you because it requires a DOM to work with.

For documents too big for DOM, you are stuck with some form of SAX or StAX parsing.

I found Harold's chapter on SAX filters to be very helpful when thinking about big SAX problems.

You might find the servingXML toolkit to be helpful.

Bill

[Shane Lee]

How large an xml file william are we talking before looking into streaming xml instead of parsing as a DOM and using xpath and xquery to query the data?