Serialization Attributes

In the previous section we learned how to save and restore an object with the XmlSerializer, yet we haven’t actually looked at the output of the Serialize() method. We set out to write a class to obtain an XML representation that would match the car element in the XML document; the generated XML was supposed to contain a car element with attributes for make, model and year of the car. Now let’s compare the output of the XmlSerializer to the earlier format. The following listing shows the two side-by-side.

Close, but no cigar! All the data encapsulated by the Car object is present in the XML document, yet the structure is not quite what needed. The output document starts with a Car element following the name of the class of the serialized object. All fields of the class are represented as elements in the generated XML document, but we really wanted them to be attributes. What would we do if we had to send XML to a receiver who can only process the data in attributes? Could we still serialize with the XmlSerializer? Yes, absolutely. In fact that is what the XmlSerializer was designed for: map classes to arbitrary XML formats.

During the introduction of this chapter we mentioned the important role .NET metadata attributes are playing in XML serialization. Now we will get to know the specific attributes that control how classes are mapped to XML types and we will learn applying them, too. Table 9.3 lists the XML serialization attributes available in System.Runtime.Serialization namespace of the .NET framework

Table 9.3 XML Serialization Attributes control Class-To-XML Type Mappings. Each attribute customizes how the XmlSerializer maps a class, field or property to an XML document. The attributes can also declare types that are not explicitly referenced in a source file.

Let’s see how which of these attributes we can apply to our Car class to get the desired output from the XmlSerializer. First we attach an XmlRootAttribute attribute to change the name of the root element to “car” with a lower case ‘c’. The new the name of the root element is passed to the constructor of the attribute.

Second, we would like to map the fields to attributes rather than elements. We can accomplish this by applying an XmlAttribute to each field. Like the XmlRoot, this attribute also has a constructor which takes the name to use for  the generated XML attribute as a parameter. 

using System.Xml.Serialization;

[XmlRoot("car")]

public class Car

{

  public Car(){}

  [XmlAttribute("make")]

  public string Make;

  [XmlAttribute("model")]

  public string Model;

  [XmlAttribute("year")]

  public int Year;

}

Now when we call Serialize() the XmlSerializer will create an XML document file of the format we wanted:

<?xml version="1.0" encoding="utf-8" ?>

<car […] make="Ford" model="Explorer" year="1997" />

There is another metadata attribute very similar to the two we just heard about so it needs to be mentioned here. Similar to the XmlRoot, the XmlElement maps the member name to another XML element name. While the XmlRoot can only be applied to classes, the XmlAttribute and the XmlElement are only valid on class members.

Each attribute offers further properties to fine-tune the how the XmlSerializer maps an item to XML. Table 9.4 shows common properties of many attributes and explains their use. You can also find a complete reference for the metadata attributes and their properties in the appendix.

Table 9.4: Common properties exposed by many Xml serialization attributes. These properties fine-tune the behavior of the attributes. They define XML namespaces, limit the scope of the attribute to a certain type, specify the corresponding type in an XSD schema, control the form of the generated XML and define whether or not the create XML for null references.