Tagging with JDOM helped me find this.
Long answer coming.... XPath does not have the native ability to parse the 'standard' way of adding 'attributes' to ProcessingInstructions. If you want to do the concatenation of the values as part of a single XPath expression I think you are out of luck.... actually, Martin's answer looks promising, but it will return a number of String values, not ProcessingInsructions. JDOM 2.x will need a Filters.string() on the XPath.compile(...) and you will get a List<String> result to path.evaluate(doc).... I think it's simpler to do it outside of the XPath. Especially given that there's only limited support for XPath2.0 by using the Saxon library with JDOM 2.x
As for doing it programmatically, JDOM 2.x helps a fair amount. Taking your example XML I did it two ways, the first way uses a custom Filter on the XPath resultset. The second way does effectively the same thing but restricting the PI's further in the loop.
public static void main(String[] args) throws Exception {
SAXBuilder saxb = new SAXBuilder();
Document doc = saxb.build(new File("data.xml"));
// This custom filter will return PI's that have the NAME="CONTENTTYPE" 'pseudo' attribute...
@SuppressWarnings("serial")
Filter<ProcessingInstruction> contenttypefilter = new AbstractFilter<ProcessingInstruction>() {
@Override
public ProcessingInstruction filter(Object obj) {
// because we know the XPath expression selects Processing Instructions
// we can safely cast here:
ProcessingInstruction pi = (ProcessingInstruction)obj;
if ("CONTENTTYPE".equals(pi.getPseudoAttributeValue("NAME"))) {
return pi;
}
return null;
}
};
XPathExpression<ProcessingInstruction> xp = XPathFactory.instance().compile(
// search for all METADATA PI's.
"//processing-instruction('METADATA')",
// The XPath will return ProcessingInstruction content, which we
// refine with our custom filter.
contenttypefilter);
StringBuilder sb = new StringBuilder();
for (ProcessingInstruction pi : xp.evaluate(doc)) {
sb.append(pi.getPseudoAttributeValue("VALUE")).append("\n");
}
System.out.println(sb);
}
This second way uses the simpler and pre-defined Filters.processingInstruction() but then does the additional filtering manually....
public static void main(String[] args) throws Exception {
SAXBuilder saxb = new SAXBuilder();
Document doc = saxb.build(new File("data.xml"));
XPathExpression<ProcessingInstruction> xp = XPathFactory.instance().compile(
// search for all METADATA PI's.
"//processing-instruction('METADATA')",
// Use the pre-defined filter to set the generic type
Filters.processinginstruction());
StringBuilder sb = new StringBuilder();
for (ProcessingInstruction pi : xp.evaluate(doc)) {
if (!"CONTENTTYPE".equals(pi.getPseudoAttributeValue("NAME"))) {
continue;
}
sb.append(pi.getPseudoAttributeValue("VALUE")).append("\n");
}
System.out.println(sb);
}
