Question

If I have an XPathNavigator positioned on a node, how can I get an XPath expression that represents the path to that node, from the root?

For example, if the XML is:

<data>
    <class name='dogs'>
        <item name='doberman />
        <item name='husky' />
    </class>
    <class name='cats'>
        <item name='persian' />
        <item name='tabby' />
    </class> </data>
</data>

…then the path to the persian cat could be expressed as /data/class[2]/item[1]

I can enumerate the ancestors of the node in question with SelectAncestors() (or I could iteratively climb up the parent relationship with SelectParent()), but that doesn’t get me the positional information.

Would I have to evaluate an XPath using position() for each ancestor, or is there a better way to do this?

Answer 1

Assuming you’re interested only in the xpath of xml elements, I implemented a brute force algorithm (i.e. traversing the XML structure) as extension methods on XmlElement. This is very similar to @Zenexer’s answer, although I had already started on my own version when he posted his.

Also, intrigued by Alexei’s tip about performance, I created a sort of test case using a somewhat complex XML file lyring around here. Then I implemented two versions of the same algorithm; one that depends on PreviousSibling, and other that iterates nodes sequentially. A third version relied on XPath’s position() function, but it didn’t work as expected and was discarded.

While you should check for yourself, in my machine the results showed a significant performance advantage for the iterative version — 1.7s against 21s scored by the siblings version.

Importart: these extension methods are declared inside a static class XmlElementExtension.

PreviousSibling version

    public static string GetXPath_UsingPreviousSiblings(this XmlElement element)
    {
        string path = "/" + element.Name;

        XmlElement parentElement = element.ParentNode as XmlElement;
        if (parentElement != null)
        {
            // Gets the position within the parent element, based on previous siblings of the same name.
            // However, this position is irrelevant if the element is unique under its parent:
            XPathNavigator navigator = parentElement.CreateNavigator();
            int count = Convert.ToInt32(navigator.Evaluate("count(" + element.Name + ")"));
            if (count > 1) // There's more than 1 element with the same name
            {
                int position = 1;
                XmlElement previousSibling = element.PreviousSibling as XmlElement;
                while (previousSibling != null)
                {
                    if (previousSibling.Name == element.Name)
                        position++;

                    previousSibling = previousSibling.PreviousSibling as XmlElement;
                }

                path = path + "[" + position + "]";
            }

            // Climbing up to the parent elements:
            path = parentElement.GetXPath_UsingPreviousSiblings() + path;
        }

        return path;
    }

Iterative version

    public static string GetXPath_SequentialIteration(this XmlElement element)
    {
        string path = "/" + element.Name;

        XmlElement parentElement = element.ParentNode as XmlElement;
        if (parentElement != null)
        {
            // Gets the position within the parent element.
            // However, this position is irrelevant if the element is unique under its parent:
            XmlNodeList siblings = parentElement.SelectNodes(element.Name);
            if (siblings != null && siblings.Count > 1) // There's more than 1 element with the same name
            {
                int position = 1;
                foreach (XmlElement sibling in siblings)
                {
                    if (sibling == element)
                        break;

                    position++;
                }

                path = path + "[" + position + "]";
            }

            // Climbing up to the parent elements:
            path = parentElement.GetXPath_SequentialIteration() + path;
        }

        return path;
    }

The test case

    private static void Measure(string functionName, int iterations, Action implementation)
    {
        Stopwatch watch = new Stopwatch();
        watch.Start();

        for (int i = 0; i < iterations; i++)
        {
            implementation();
        }

        watch.Stop();
        Console.WriteLine("{0}: {1}ms", functionName, watch.ElapsedMilliseconds);
    }

    private static void Main(string[] args)
    {
        XmlDocument doc = new XmlDocument();
        doc.Load(@"location of some large and complex XML file");

        string referenceXPath = "/vps/vendorProductSets/vendorProductSet/product[100]/prodName/locName";

        Measure("UsingPreviousSiblings", 10000,
                () =>
                    {
                        XmlElement target = doc.SelectSingleNode(referenceXPath) as XmlElement;
                        Debug.Assert(referenceXPath == target.GetXPath_UsingPreviousSiblings());
                    });

        Measure("SequentialIteration", 10000,
                () =>
                {
                    XmlElement target = doc.SelectSingleNode(referenceXPath) as XmlElement;
                    Debug.Assert(referenceXPath == target.GetXPath_SequentialIteration());
                });
    }