Develop Reference

Xml List Serialization and Node Type Names

Ive come across multiple questions and answers on here but none specific to my situation.
I have a class 'Entity' with multiple classes that extend off of it. I want the serialization to hit the list and understand and use the type of each item for the node name.
Now, I can use what is commented out (define each array item in the main class and define the name of such by using [XmlArrayItem("Subclass1", typeof(subclass1)] but I want to keep all definitions in their subclass and I will be having too many subclasses to define everything in the main entity class...Is there anyway to achieve this?
I have tried using [XmlType(TypeName="...")] for the subclasses and so on but that did not work.
[Serializable]
[XmlInclude(typeof(Subclass1))]
[XmlRoot("Entity")]
public class Entity{
[XmlArray("CausedBy")]
//[XmlArrayItem("Subclass1", typeof(subclass1))]
//[XmlArrayItem("Sublcass2", typeof(Subclass2))]
public List<Entity> CausedBy { get; set; }
}
[Serializable]
[XmlRoot("Subclass1")]
[XmlInclude(typeof(Subclass2))]
public class Subclass1:Entity{
//Code...
}
[Serializable]
[XmlRoot("Subclass2")]
public class Subclass2:Subclass1{
//Code...
}
Serializing the above code after creating an entity and adding a Subclass1 and Subclass2 to the list 'CausedBy' class results in the following:
<Entity>
<CausedBy>
<Entity ... xsi:type="SubClass1" />
<Entity ... xsi:type="SubClass2" />
</CausedBy>
<Entity>
I would like the output to show:
<Entity>
<CausedBy>
<SubClass1 .../>
<SubClass2 .../>
</CausedBy>
<Entity>

Since I totally failed to read the question to begin with, here's a new answer (it's a bit of a tl;dr, so you can always skip to the end and follow the link):
It isn't possible to get the built in serializer class to work because you don't wish to add the attributes that it needs to be able to operate. Your only option is to seralize the class yourself, however, this need not be as tedious as it sounds; I had a similar issue a few years ago with DataGridView in virtual mode and produced a generic virtualizer that could be used to virtualize the data for display; it used a custom attribute:
[AttributeUsage(AttributeTargets.Property, AllowMultiple = false)]
public sealed class showColumnAttribute : System.Attribute
{
///<summary>Optional display format for column</summary>
public string Format;
///<summary>Optional Header string for column<para>Defaults to propety name</para></summary>
public string Title;
///<summary>Optional column edit flag - defaults to false</summary>
public bool ReadOnly;
///<summary>Optional column width</summary>
public int Width;
///<summary>
///Marks public properties that are to be displayed in columns
///</summary>
public showColumnAttribute()
{
Format = String.Empty;
Title = String.Empty;
ReadOnly = false;
Width = 0;
}
}
And a constructor:
///<summary>
///Extracts the properties of the supplied type that are to be displayed
///<para>The type must be a class or an InvalidOperationException will be thrown</para>
///</summary>
public Virtualiser(Type t)
{
if (!t.IsClass)
throw new InvalidOperationException("Supplied type is not a class");
List<VirtualColumnInfo> definedColumns = new List<VirtualColumnInfo>();
PropertyInfo[] ps = t.GetProperties();
MethodInfo mg, ms;
for (int i = 0; i < ps.Length; i++)
{
Object[] attr = ps[i].GetCustomAttributes(true);
if (attr.Length > 0)
{
foreach (var a in attr)
{
showColumnAttribute ca = a as showColumnAttribute;
if (ca != null)
{
mg = ps[i].GetGetMethod();
if (mg != null)
{
ms = ps[i].GetSetMethod();
definedColumns.Add
(
new VirtualColumnInfo
(
ps[i].Name, ca.Width, ca.ReadOnly, ca.Title == String.Empty ? ps[i].Name : ca.Title,
ca.Format, mg, ms
)
);
}
break;
}
}
}
}
if (definedColumns.Count > 0)
columns = definedColumns.ToArray();
}
This extracts the public properties of the class and supplies marked items to the DataGridView as columns together with a header, format, etc.
The effect of all of this (and the rest of the missing code) was that any type could be virtualized in a dataGridView simply by tagging public properties and calling the virtualizer once for a given type:
#region Virtualisation
static readonly Virtualiser Virtual = new Virtualiser(typeof(UserRecord));
[XmlIgnore] // just in case!
public static int ColumnCount { get { return Virtual.ColumnCount; } }
public static VirtualColumnInfo ColumnInfo(int column)
{
return Virtual.ColumnInfo(column);
}
public Object GetItem(int column)
{
return Virtual.GetItem(column, this);
}
/*
** The supplied item should be a string - it is up to this method to supply a valid value to the property
** setter (this is the simplest place to determine what this is and how it can be derived from a string).
*/
public void SetItem(int column, Object item)
{
String v = item as String;
int t = 0;
if (v == null)
return;
switch (Virtual.GetColumnPropertyName(column))
{
case "DisplayNumber":
if (!int.TryParse(v, out t))
t = 0;
item = t;
break;
}
try
{
Virtual.SetItem(column, this, item);
}
catch { }
}
#endregion
The number of columns, their properties and order can be specified automatically by creating a number of public properties derived from the class data:
#region Display columns
[showColumn(ReadOnly = true, Width = 100, Title = "Identification")]
public String DisplayIdent
{
get
{
return ident;
}
set
{
ident = value;
}
}
[showColumn(Width = 70, Title = "Number on Roll")]
public int DisplayNumber
{
get
{
return number;
}
set
{
number = value;
}
}
[showColumn(Width = -100, Title = "Name")]
public string DisplayName
{
get
{
return name == String.Empty ? "??" : name;
}
set
{
name = value;
}
}
#endregion
This would virtualize any class for dataGridView to display and edit data and I used it many times over the years and the extraction of properties to display is exactly what is required for XML serialization, indeed, it has a lot of the same characteristics.
I was going to adapt this method to do the same job for XML serialization but someone has already done it at https://www.codeproject.com/script/Articles/ViewDownloads.aspx?aid=474453, I hope you can make use of this method to solve your problem.

This works for me:
public Form1()
{
InitializeComponent();
}
private void Form1_Load(object sender, EventArgs e)
{
Entity entity = new Entity();
entity.CausedBy = new List<Entity>();
entity.CausedBy.Add(new Subclass1());
entity.CausedBy.Add(new Subclass2());
entity.CausedBy.Add(new Subclass2());
entity.CausedBy.Add(new Subclass1());
entity.CausedBy.Add(new Subclass1());
entity.Save(Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.MyDocuments), "Test.txt"));
}
}
[Serializable]
[XmlRoot("Entity")]
public class Entity
{
[XmlArray("CausedBy")]
[XmlArrayItem("SubClass1", typeof(Subclass1))]
[XmlArrayItem("SubClass2", typeof(Subclass2))]
public List<Entity> CausedBy { get; set; }
}
[Serializable]
[XmlRoot("Subclass1")]
public class Subclass1 : Entity
{
[XmlIgnore]
String t = DateTime.Now.ToShortDateString();
public String SubClass1Item { get { return "Test1 " + t; } set { } }
}
[Serializable]
[XmlRoot("Subclass2")]
public class Subclass2 : Entity
{
[XmlIgnore]
String t = DateTime.Now.ToString();
public String SubClass2Item { get { return "Test2 " + t; } set { } }
}
It produces:
<Entity xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema">
<CausedBy>
<SubClass1>
<SubClass1Item>Test1 20/09/2017</SubClass1Item>
</SubClass1>
<SubClass2>
<SubClass2Item>Test2 20/09/2017 01:06:55</SubClass2Item>
</SubClass2>
<SubClass2>
<SubClass2Item>Test2 20/09/2017 01:06:55</SubClass2Item>
</SubClass2>
<SubClass1>
<SubClass1Item>Test1 20/09/2017</SubClass1Item>
</SubClass1>
<SubClass1>
<SubClass1Item>Test1 20/09/2017</SubClass1Item>
</SubClass1>
</CausedBy>
</Entity>

Looping to Create and Add New Objects to ArrayList

Edit to save you from reading through this whole post
tldr: an object's fields should not be static unless you want all instances of that object to have the same value for that field
I'm trying to create and populate an ArrayList of Blog objects. I do know the generic way do this:
create ArrayList of Blogs
loop (some condition)
create new Blog
add this Blog to AL
However, when I attempt to do so within the while(datareader.read()) loop, all of the elements in the ArrayList are exactly the same Blog. Specifically, I end up with an ArrayList filled with multiple pointers to the very last Blog object from the database table. Here is my code:
public static ArrayList AllBlogs()
{
SqlDataReader dr = anonPage.ExecuteReader("SELECT * FROM Kristina_Blogs");
ArrayList allBlogs = new ArrayList();
if (dr.HasRows)
{
while (dr.Read())
{
Blog b = new Blog();
//grab a row from Kristina_Blogs and assign those attributes to b
b.setTitle(dr["title"].ToString());
b.setMessage(dr["message"].ToString());
b.setId(dr["id"]);
allBlogs.Add(b);
}
}
dr.Close();
return allBlogs;
}
As I said before, the result of this is an ArrayList filled with pointers to the very last blog from the Kristina_Blogs table. I imagine the ArrayList allBlogs looks like [b, b, b, ... b] and therefore they ALL get updated when I say b.setTitle() etc. But how can this be the case if I am creating a NEW Blog object at the beginning of each iteration?
Here is some extra info that you don't have to read but it might clear up some confusion about the structure of the problem:
Blog object has id, title, and message fields and their respective getter/setters
Kristina_Blogs is a table representing these blogs with columns for id, title, message
The suggestions say to include a tag for my DB engine but I can't find a tag for it: Microsoft SQL Server Management Studio
This code works perfectly when I use an ArrayList of Strings instead of Blogs
Edit: Including the code from Blog class
public class Blog
{
public App myApp;
public static string Title;
public static string Message;
public static int Id;
//constructors
public Blog() { }
public Blog(App App) { this.myApp = App; }
//all getters and setters look like this
public string getTitle() { return Title; }
public void setTitle(string t) { Title = t; }
}

The main problem you have, as I mentioned in comments is your member variables are static, so when you set the value, they change in all instances. you should change your code this way:
public class Blog
{
public int Id { get; set; }
public string Title { get; set; }
public string Message { get; set; }
}
And fill your list this way, don't forget to add using System.Linq;:
var result = new List<Blog>();
var connection = #"your connection string";
var command = "SELECT * FROM Kristina_Blogs";
var adapter = new System.Data.SqlClient.SqlDataAdapter(command, connection);
var dataTable = new DataTable();
//Get data
adapter.Fill(dataTable);
dataTable.Rows.Cast<DataRow>().ToList()
.ForEach(row =>
{
var b = new Blog();
b.Id = row.Field<int>("Id");
b.Title = row.Field<string>("Title");
b.Message = row.Field<string>("Message");
result.Add(b);
});
return result;
Note:
When you create a member static, it is shared between all instances of that calss.
In C# you can use property to get or set values, you don't need to setX or setY, when you get the value of a property, the get code of that property will execute and when you assign a value to a property the set part of it will execute. you can define properties this way:
Property:
private int id;
public int Id
{
get
{
return id;
}
set
{
id = value;
}
}
or more simple:
public int Id { get; set; }

All of the fields in your Blog class are static, meaning they're shared between all object instances. You want them to be instance field (meaning not static) so that each object has its own copy of each of those values.

Remove the static attributes from your class:
public class Blog
{
public App myApp;
public String Title;
public String Message;
public int Id;
//constructors
public Blog() { }
public Blog(App App) { this.myApp = App; }
//all getters and setters look like this
public String getTitle() { return Title; }
public String getMessage() { return Message; }
public void setTitle(String t) { Title = t; }
public void setMessage(String m) { Message = m; }
}
When you use static variables, all instances of an object will contain the same values in those variables. By removing the static keyword, you are allowing different instances of the object to hold different values.
Now, every time you create a blog object, that object's Title and Message etc, will contain its own information.

I would make a quick method to prevent null value from throwing error
public static string GetSafeString(SqlDataReader reader, int index)
{
if (!reader.IsDBNull(index))
return reader.GetString(index);
else
return string.Empty;
}
Replace this code:
while (dr.Read())
{
Blog b = new Blog();
//grab a row from Kristina_Blogs and assign those attributes to b
b.setTitle(dr["title"].ToString());
b.setMessage(dr["message"].ToString());
b.setId(dr["id"]);
allBlogs.Add(b);
}
With This Code:
while (dr.Read())
{
Blog b = new Blog();
//grab a row from Kristina_Blogs and assign those attributes to b
b.setId(dr.GetInt32(0));
b.setTitle(GetSafeString(dr, 1);
b.setMessage(GetSafeString(dr, 2);
allBlogs.Add(b);
}
Where the number is the index of field in the record and assuming "id" is an integer. Also consider moving creation of "Blog" object outside of loop and just change values.

How to bind an object to a list?

This is not a windows form app, I'm not sure I'm using the correct terminology. I'm trying to bind an object to a list so when the object is modified outside of the list, those changes are reflected in the list. I'm not entirely sure how to start, my searches just keep returning "winform" answers to a datasource but this is not what I want. Here is what I have so far:
You can copy this code into a console app if you wish to test it. Notice the foreach that loops through go.Getcomponents() does not show the names because I don't think the object modified is still referenced when pulled out of the list. Essentially I'm trying to modify the object outside the list but when that object is modified the object in the list is also modified.
It is important that it can be serialized because the GameObject will be transferred across a network and data within it will be read by a server.
class Test
{
public void TestStart()
{
GameObject go = new GameObject(); //create GameObject
Dog dog = go.AddComponent<Dog>(); //Add a Dog component to the GameObject
dog.name = "Fluffy"; //name the dog fluffy, this should be reflected in the GenericComponent list of GameObject
Dog dog2 = go.AddComponent<Dog>();
dog2.name = "Fuzzy";
//loop through all dog components in GameObject go, doesn't print dog names :(
foreach (Dog dg in go.GetComponents<Dog>())
{
Console.WriteLine(dg.name);
}
Console.ReadLine();
}
}
[Serializable]
public class GameObject
{
List<GenericComponent<Object>> componentList = new List<GenericComponent<Object>>();
//returns first found in list.
public T GetComponent<T>()
{
return (T)componentList.Find(c => c.component.GetType() == typeof(T)).component;
}
//add a component to component list.
public T AddComponent<T>()
{
GenericComponent<Object> newComponent = new GenericComponent<Object>();//;(T)Activator.CreateInstance(typeof(T));
newComponent.component = (T)Activator.CreateInstance(typeof(T));
componentList.Add(newComponent);
return (T)Activator.CreateInstance(typeof(T));
}
//returns all components of type T
public List<T> GetComponents<T>()
{
List<T> r = new List<T>();
for (int i = 0; i < componentList.Count; i++)
{
if (componentList[i].component.GetType() == typeof(T))
{
r.Add((T)componentList[i].component);
}
}
return r;
}
}
[Serializable]
public class GenericComponent<T> where T : new()
{
public T component;
public GenericComponent()
{
component = new T();
}
}
[Serializable]
public class Dog
{
public string name = "";
public Dog() { }
public Dog(string name)
{
this.name = name;
}
}

In your AddComponent method, you are adding one component and then returning another one. Instead, return the same one you added:
public T AddComponent<T>()
{
GenericComponent<Object> newComponent = new GenericComponent<Object>();
newComponent.component = (T)Activator.CreateInstance(typeof(T));
componentList.Add(newComponent);
return (T)newComponent.component;
}

How to make XmlSerializer.Deserialize handle DefaultAttribute properly?

There seems to be a bug / inconsistency in the Microsoft XmlSerializer: If you have a property marked with a System.ComponentModel.DefaultValue attribute, this does not get serialized. Fair enough - this could be seen as an expected behavior.
The problem is the same attribute is not respected when deserializing. The code below illustrates the issue.
Question is how could I bypass this? I have potentially hundreds of business classes with default values used in the UI tier (Views), so default value initialization in constructor is not an option. It has to be something generic. I could create a completely new default attribute, but it seems like duplicate work. Do you see a way to override the XmlSerializer behavior or should I use just another serializer that does the job better?
The example code:
public class DefaultValueTestClass
{
[System.ComponentModel.DefaultValue(10000)]
public int Foo { get; set; }
}
[TestMethod]
public void SimpleDefaultValueTest()
{
// Create object and set the property value TO THE DEFAULT
var before = new DefaultValueTestClass();
before.Foo = 10000;
// Serialize => xml
var serializer = new System.Xml.Serialization.XmlSerializer(typeof(DefaultValueTestClass));
string xml;
using (var stream = new System.IO.StringWriter())
{
serializer.Serialize(stream, before);
xml = stream.ToString();
}
// Deserialize the same object
DefaultValueTestClass after;
using (var reader = new System.IO.StringReader(xml))
{
after = (DefaultValueTestClass)serializer.Deserialize(reader);
}
// before.Foo = 10000
// after.Foo = 0
Assert.AreEqual(before.Foo, after.Foo);
}

It is your job to implement the defaults; [DefaultValue] merely says "this is the default, you don't need to worry about this" - it doesn't apply it. This applies not just to XmlSerializer, but to the core System.ComponentModel API to which [DefaultValue] belongs (which drives things like the bold / not-bold in PropertyGrid, etc)
Basically:
public class DefaultValueTestClass
{
public DefaultValueTestClass()
{
Foo = 10000;
}
[DefaultValue(10000)]
public int Foo { get; set; }
}
will work in the way you expect. If you want it to serialize whether or not it is that particular value, then the correct implementations is:
public class DefaultValueTestClass
{
public DefaultValueTestClass()
{
Foo = 10000;
}
public int Foo { get; set; }
}
If you want to preserve the [DefaultValue], but want it to always serialize, then:
public class DefaultValueTestClass
{
[DefaultValue(10000)]
public int Foo { get; set; }
[Browsable(false), EditorBrowsable(EditorBrowsableState.Never)]
public bool ShouldSerializeFoo() { return true; }
}
Where ShouldSerialize* is another pattern from System.ComponentModel that is recognised by several serializers.
And here's some UI code to show that XmlSerializer is actually doing exactly the same things that the UI code (built on System.ComponentModel) has always done:
using System.ComponentModel;
using System.Windows.Forms;
static class Program
{
[System.STAThread]
static void Main()
{
Application.EnableVisualStyles();
using (var form = new Form())
using (var grid = new PropertyGrid())
{
grid.Dock = DockStyle.Fill;
var obj = new DefaultValueTestClass
{ // TODO - try with other numbers to
// see bold / not bold
Foo = 10000
};
// note in the grid the value is shown not-bold; that is
// because System.ComponentModel is saying
// "this property doesn't need to be serialized"
// - or to show it more explicitly:
var prop = TypeDescriptor.GetProperties(obj)["Foo"];
bool shouldSerialize = prop.ShouldSerializeValue(obj);
// ^^^ false, because of the DefaultValueAttribute
form.Text = shouldSerialize.ToString(); // win title
grid.SelectedObject = obj;
form.Controls.Add(grid);
Application.Run(form);
}
}
}
public class DefaultValueTestClass
{
[System.ComponentModel.DefaultValue(10000)]
public int Foo { get; set; }
}

Is it possible to modify the attribute of a property at runtime?

Is it possible to modify the attribute of a property at runtime?
let's say I have some class:
public class TheClass
{
[TheAttribute]
public int TheProperty { get; set; }
}
Is there a way to do this?
if (someCondition)
{
// disable attribute. Is this possible and how can this be done?
}

No this is not possible. You cannot modify attribute values from metadata, or metadata in general, at runtime
Strictly speaking the above is not true. There are certain APIs which do allow allow for some metadata generation and modification. But they are very scenario specific, (ENC, profiling, debugging) and should not be used in general purpose programs.

It depends; from a reflection perspective: no. You can't. But if you are talking about attributes used by System.ComponentModel in things like data-binding, they you can use TypeDescriptor.AddAttributes to append extra attributes. Or other customer models involving custom descriptors. So it depends on the use-case.
In the case of xml serialization, it gets more interesting. Firstly, we can use fun object models:
using System;
using System.Xml.Serialization;
public class MyData
{
[XmlAttribute]
public int Id { get; set; }
[XmlAttribute]
public string Name { get; set; }
[XmlIgnore]
public bool NameSpecified { get; set; }
static void Main()
{
var ser = new XmlSerializer(typeof(MyData));
var obj1 = new MyData { Id = 1, Name = "Fred", NameSpecified = true };
ser.Serialize(Console.Out, obj1);
Console.WriteLine();
Console.WriteLine();
var obj2 = new MyData { Id = 2, Name = "Fred", NameSpecified = false };
ser.Serialize(Console.Out, obj2);
}
}
The bool {name}Specified {get;set;} pattern (along with bool ShouldSerialize{name}()) is recognised and used to control which elements to include.
Another alternative is to use the non-default ctor:
using System;
using System.Xml.Serialization;
public class MyData
{
[XmlAttribute]
public int Id { get; set; }
public string Name { get; set; }
static void Main()
{
var obj = new MyData { Id = 1, Name = "Fred" };
XmlAttributeOverrides config1 = new XmlAttributeOverrides();
config1.Add(typeof(MyData),"Name",
new XmlAttributes { XmlIgnore = true});
var ser1 = new XmlSerializer(typeof(MyData),config1);
ser1.Serialize(Console.Out, obj);
Console.WriteLine();
Console.WriteLine();
XmlAttributeOverrides config2 = new XmlAttributeOverrides();
config2.Add(typeof(MyData), "Name",
new XmlAttributes { XmlIgnore = false });
var ser2 = new XmlSerializer(typeof(MyData), config2);
ser2.Serialize(Console.Out, obj);
}
}
Note though that if you use this second approach you need to cache the serializer instance, as it emits an assembly every time you do this. I find the first approach simpler...

Attributes are baked into code at compilation time. The only way you can define new attributes at run time is to generate new code at runtime (using Reflection.Emit, for example). But you cannot change the attributes of existing code.

You can put Boolean variable in the class to disable/enable the property instead of disabling it at run time.

You might want to look at this http://social.msdn.microsoft.com/Forums/en-US/csharpgeneral/thread/5b0d356d-d006-43ff-bfcd-aa90dd8de6db
And Dave Morton's explanation on this blog http://blog.codinglight.com/2008/10/changing-attribute-parameters-at.html

Sounds like you want to consider implementing IXmlSerializable

You can implement IDataErrorInfo, then check range in Validate method.
public string this[string property] {
get { return Validate(property); }
}
public string Error { get; }
protected virtual string Validate(string property) {
var propertyInfo = this.GetType().GetProperty(property);
var results = new List<ValidationResult>();
var result = Validator.TryValidateProperty(
propertyInfo.GetValue(this, null),
new ValidationContext(this, null, null) {
MemberName = property
},
results);
if (!result) {
var validationResult = results.First();
return validationResult.ErrorMessage;
}
return string.Empty;
}
In sub class
protected override string Validate(string property) {
Debug.WriteLine(property);
if (property == nameof(YourProperty)) {
if (_property > 5) {
return "_property out of range";
}
}
return base.Validate(property);
}