9

Consider the following XML:

<a>
    <b>2</b>
    <c></c>
</a>

I need to deserialize this xml to an object. So, i wrote the following class. 

public class A
{
    [XmlElement("b", Namespace = "")]
    public int? B { get; set; }

    [XmlElement("c", Namespace = "")]
    public int? C { get; set; }

}

Since i'm using nullables, i was expecting that, when deserialing the above xml, i would get an object A with a null C property.

Instead of this, i get an exception telling the document has an error.

John Saunders
  • 160,644
  • 26
  • 247
  • 397
Zé Carlos
  • 3,627
  • 5
  • 43
  • 51

2 Answers2

11

There's a difference between a missing element and a null element.

A missing element, <a><b>2</b></a>. Here C would take whatever default value you specify, using the DefaultValue attribute, or null if there's no explicit default.

A null element <a><b>2</b><c xs:Nil='true'/></a>. Here you will get null.

When you do <a><b>2</b><c></c><a/> the xml serializer will try to parse string.Empty as an integer an will correctly fail.

Since your provider is generating invalid xml you will need to do this, if using the XmlSerializer:

[XmlRoot(ElementName = "a")]
public class A
{
    [XmlElement(ElementName = "b")]
    public int? B { get; set; }

    [XmlElement(ElementName = "c")]
    public string _c { get; set; }

    public int? C
    {
        get
        {
            int retval;

            return !string.IsNullOrWhiteSpace(_c) && int.TryParse(_c, out retval) ? (int?) retval : null;
        }
    }
}

or slightly better using the DataContractSerializer

[DataContract(Name="a")]
public class A1
{
    [DataMember(Name = "b")]
    public int? B { get; set; }

    [DataMember(Name = "c")]
    private string _c { get; set; }

    public int? C
    {
        get
        {
            int retval;

            return !string.IsNullOrWhiteSpace(_c) && int.TryParse(_c, out retval) ? (int?)retval : null;
        }
    }
}

although the DataContractSerializer doesn't support attributes if that's a problem.

Phil
  • 42,255
  • 9
  • 100
  • 100
  • Thanks for your post. I'm aware of what you are telling me. But i don't have control of xml data (it's comming from an external service) . And the sevice provider are returning when in fact it should return a xs:Nil='true', so i have to deal with this. – Zé Carlos Mar 12 '12 at 19:08
  • You will need to make C a string and then have a wrapper property that parses the deserialized string to an int? as required. – Phil Mar 12 '12 at 19:11
  • Thanks. Shouldn't your test be "Strng.IsNullOrEmpty" instead of "IsNullOrWhiteSpace"? You don't need to perform the cast (int?)retval. However, to avoid compile errors, we need to cast (int?)null – Zé Carlos Mar 13 '12 at 10:44
  • I guess you don't need the string.IsNullOrWhitespace or string.IsNullOrEmpty since TryParse is enough. Yes you need the (int?) case when using the ?: operator so both sides evaluate to (int?). – Phil Mar 13 '12 at 10:49
  • It may be worth noting that if your class properties' names exactly match your XML element names (above a property lower case "c" was mapping to upper case property "C" through attributes), redirecting an element (by the XmlElement or DataMember attributes) can cause a conflict if your intended target already matched implicitly by name. Adding an [XmlIgnore] attribute to the target element can clear this up. – John Spiegel Jul 03 '13 at 19:18
9

To deserialize empty tags like 'c' in your example:

    <foo>
        <b>2</b>
        <c></c>
    </foo>

I used this approach. First it removes the null or empty elements from the XML file using LINQ and then it deserialize the new document without the null or empty tags to the Foo class.

    public static Foo ReadXML(string file)
    {
            Foo foo = null;
            XDocument xdoc = XDocument.Load(file);
            xdoc.Descendants().Where(e => string.IsNullOrEmpty(e.Value)).Remove();

            XmlSerializer xmlSer = new XmlSerializer(typeof(Foo));
            using (var reader = xdoc.Root.CreateReader())
            {
                foo = (Foo)xmlSer.Deserialize(reader);
                reader.Close();
            }
            if (foo == null)
                foo = new Foo();

            return foo;
    }

Which will give you default values on the missing properties.

    foo.b = 2;
    foo.c = 0; //for example, if it's an integer

I joined information from this links:

Remove empty XML tags

Use XDocument as the source for XmlSerializer.Deserialize?

Community
  • 1
  • 1
Luis David Sá
  • 91
  • 1
  • 3