Trouble selecting nodes between two siblings

[Dave Alvarado]

Hi,

I'm using Java 6. Let's say I have a document with these nodes ...

    <h2>Description</h2>
    <p>...</p>
    <p>...</p>
    <p>...</p>
    <h2>Cost</h2>

I want to write an Xpath expression to select all the nodes (they may not necessarily be "p"'s) between the <h2> nodes with text descriptions "Description" and "Cost", as appear above. How do I do this?

I tried

//*[@id='content-primary']/h2[text()='Description'] and preceding-sibling::h2[text()='Cost']

but get the Java exception, "com.sun.org.apache.xpath.internal.XPathException: Can not convert #BOOLEAN to a NodeList!". - Dave

[Paul Clapham]

Yes, that expression is of the form "a and b" and so its value is boolean. But presumably your code tells the XPath object to treat it as a node-list (because that's what you expect it to produce).

The "and" needs to be inside the [brackets] which qualify the nodes you're looking for. Something like this?

*[following-sibling::h2[text()='Cost'] and preceding-sibling::h2[text()='Description']]

[Dave Alvarado]

Hi,

I changed the a little (The first h2 marker has text, "Description" and the second h2 marker has text, "Cost") ...

	public static final String EVENT_DETAILS_XPATH = "//*[following-sibling::h2[text()='Description'] and preceding-sibling::h2[text()='Cost']]";

but I'm not getting any results. I also tried removing the "//" from the beginning of the xpath. The URL I'm running this on is here -- http://www.foresightdesign.org/events/detail.php?id=1871 . What else am I missing from the above expression? - Dave

[Paul Clapham]

Aha. You're missing the namespace. Your "h2" in the XPath expression refers to an h2 element which is in no namespace, but the h2 elements in the document you linked to are actually in the XHTML namespace.

So provide a namespace context for your XPath object -- let's say it maps the "http://www.w3.org/1999/xhtml" namespace URI to the prefix "x" -- and then use "x:h2" instead of "h2" for the element names.

[Dave Alvarado]

Thanks, but I don't think this is a namespace issue. For example, this expression accurately returns an element ...

    public static final String EVENT_DETAILS_XPATH = "//*[@id='content-primary']/h2[text()='Description']";	

Notice the use of "h2" without a preceding namespace. The above returns the "<h2>Description</h2>" element from the page. Here is the Java code I'm using to find the element ...

	public static List<Node> findElements(final Element rootElement, String xpathStr) throws XPathExpressionException { 
		final List<Node> results = new ArrayList<Node>();
		final XPathFactory factory = XPathFactory.newInstance();
		final XPath xpath = factory.newXPath();
		final NodeList result = (NodeList) xpath.evaluate(xpathStr, rootElement, XPathConstants.NODESET);
		for (int i = 0; i < result.getLength(); i++) {
		    final Node node = result.item(i);
		    results.add(node); 
		}	// for
		return results;
	}	// findElements

[Paul Clapham]

Okay. So simplify: do you get anything from this XPath expression?

//*[following-sibling::h2[text()='Description']]

[Paul Clapham]

Dave Alvarado wrote:The first h2 marker has text, "Description" and the second h2 marker has text, "Cost"...

Then shouldn't you be looking for nodes which have a preceding h2 sibling with text 'Description' and a following h2 sibling with text 'Cost'?

[Dave Alvarado]

Ah, you're right. I misunderstood following and preceding. Problem solved. Thanks, -