How to work reliably with XmlReader

Question

I'm using the XmlReader class, the forward-only reader. A method I'm calling moves the cursor as a side-effect. However, sometimes the method throws an exception, and leaves the cursor somewhere unexpected. How can I handle that?

xml.ReadStartElement("root");

if (xml.IsStartElement("Results"))
{
    try
    {
        results = Results.FromXml(xml);
        // if method successful, it reads past the closing tag of the 'Results' element
    }
    catch
    {
        results = null;
        // I want to manually move the cursor past the closing tag of the 'Results' element.
    }
}

Example document

<root>
    <results>
        <arbitaryxml/>
    </results>*
    <signatures>

If the Results.FromXml method is successful, the cursor gets left at *. However if it fails, it might be left anywhere inside the results element. I want my catch block to make sure the cursor is advanced to *. (NB. The next element isn't always called 'signatures').

I found this quite hard to explain. Please ask if it needs clarification, I can give more examples.

I feel like an exception generally shouldn't be a part of the expected flow of the program. They are supposed to be planned for, yes, but I feel like you should do some checking to prevent the exception from being thrown in the first place. — Gray, Mar 21 '13 at 14:01
On what grounds can it fail? I'd address the cause and not the symptoms. — Grant Thomas, Mar 21 '13 at 14:04

Jim Mischel · Answer 1 · 2013-03-21T18:18:11.817

Take a look at the ReadSubtree method, which will do what you want. Actually, it positions to the end element node, but it accomplishes what you want. Typically, you write something like:

XmlReader resultsReader = reader.ReadSubtree();
while (resultsReader.Read())
{
    // process results node here
}

After ReadSubtree returns, a call to reader.Read will return the </results> EndElement node. So if processing the results throws an exception, you're still in the right place.

If ReadSubtree throws an exception, that's of course unrecoverable. That indicates an error in the XML, which as far as I've been able to determine, isn't reliably recoverable using XmlReader.

Colonel Panic · Accepted Answer · 2016-04-28T13:14:15.967

Jim is right, the ReadSubtree method works, although it's a tad fiddly:

if (xml.IsStartElement("Results"))
{
    // Be careful so that the cursor will be left after the closing tag of the 'Results' element, even if Results.FromXml throws.

    using (XmlReader resultsElement = xml.ReadSubtree())
    {
        resultsElement.Read();
        try 
        {
            results = Results.FromXml(resultsElement);
        }
        catch (Exception e)
        {
            Console.Error.WriteLine("Reading results xml went wrong {0}", e.Message);
        }
    }

    xml.ReadEndElement();
}

How to work reliably with XmlReader

2 Answers2