Object position (line, column) in XML after deserialization .NET

Question

How can I get position in the original xml file of an xml tag after deserialization into a .NET object using XmlSerializer ?

Here is an example XML

Answer 1

Another, more simple approach: Let the deserializer do the work.

• Add LineInfo and LinePosition properties to all classes for which you would like to have position information:

  [XmlRoot("Root")]
  public class AddressDetails
  {
      [XmlAttribute]
      public int LineNumber { get; set; }
  
      [XmlAttribute]
      public int LinePosition { get; set; }
      ...
  }
  

  This of course can be done by subclassing.

• Load an XDocument with LoadOptions.SetLineInfo.

• Add LineInfo and LinePosition attributes to all elements:

  foreach (var element in xdoc.Descendants())
  {
       var li = (IXmlLineInfo) element;
       element.SetAttributeValue("LineNumber", li.LineNumber);
       element.SetAttributeValue("LinePosition", li.LinePosition);
   }
  

• Deserializing will populate LineInfo and LinePosition.

Cons:

• Line information only for elements that are deserialized as class, not for simple elements, not for attributes.
• Need to add attributes to all classes.

Answer 2

I tried several approaches. Two of them:

• Implement IXmlSerializable. In principle this would work, but then one would have to implement the complete deserialization mechanism.

• Deserialize using an XmlReader, save a reference to that XmlReader at some static location, and in the constructor of classes that are deserialized, access that XmlReader and retrieve line and position information. Cons: static; relatively complicated.

I did not find a way to get more information than start line and position, but honestly, that’s enough for my use case.

Currently I think I will use the following approach.

• Load the XML file with line information:

  XDocument xdoc = XDocument.Parse(xml, LoadOptions.SetLineInfo);
  

• To all elements in xdoc, add a new attribute (e.g. NewId) that allows to identify the elements, e.g. give the first element a 0, the second a 1, etc.

• Make sure all classes that are deserialized (in your case AddressDetails) will have such a NewId, e.g.:

      [XmlAttribute]
      public int NewId { get; set; }
  

  This can be done by deriving from a common base class, and even when xsd.exe is used to create the classes, this is possible because it will create all classes as partial (e.g. a partial class foo : MyBaseClassElement { } needs to be added somewhere for all classes).

• After deserializing, for each element that has been deserialized as a class object, the XElement from which it has been deserialized can be looked up using NewId. (To speed up the look up, one can use a List<XElement> that contains all XElement such that NewId is the index in that list.)

• Casting the XElement to IXmlLineInfo will give line and position information.

• For elements that have not been deserialized as a class object (e.g. Number in your example) and for attributes, first look up the XElement that contains a NewId (e.g. AdressDetails), then query the element’s or attribute’s line information:

  XElement element = ...;
  XNode descendant = element.Descendants(childElementOrAttributetName).FirstOrDefault();
  return descendant as IXmlLineInfo;
  

Cons:

• Memory usage.
• Need to add attribute to all classes.