Microsoft should have implemented something snappy for INotifyPropertyChanged, like in the automatic properties, just specify {get; set; notify;}
I think it makes a lot of sense to do it. Or are there any complications to do it?
Can we ourselves implement something like 'notify' in our properties. Is there a graceful solution for implementing INotifyPropertyChanged in your class or the only way to do it is by raising the PropertyChanged event in each property.
If not can we write something to auto-generate the piece of code to raise PropertyChanged event?
Without using something like postsharp, the minimal version I use uses something like:
public class Data : INotifyPropertyChanged
{
// boiler-plate
public event PropertyChangedEventHandler PropertyChanged;
protected virtual void OnPropertyChanged(string propertyName)
{
PropertyChangedEventHandler handler = PropertyChanged;
if (handler != null) handler(this, new PropertyChangedEventArgs(propertyName));
}
protected bool SetField<T>(ref T field, T value, string propertyName)
{
if (EqualityComparer<T>.Default.Equals(field, value)) return false;
field = value;
OnPropertyChanged(propertyName);
return true;
}
// props
private string name;
public string Name
{
get { return name; }
set { SetField(ref name, value, "Name"); }
}
}
Each property is then just something like:
private string name;
public string Name
{
get { return name; }
set { SetField(ref name, value, "Name"); }
}
which isn't huge; it can also be used as a base-class if you want. The bool return from SetField tells you if it was a no-op, in case you want to apply other logic.
or even easier with C# 5:
protected bool SetField<T>(ref T field, T value,
[CallerMemberName] string propertyName = null)
{...}
which can be called like this:
set { SetField(ref name, value); }
with which the compiler will add the "Name" automatically.
C# 6.0 makes the implementation easier:
protected void OnPropertyChanged([CallerMemberName] string propertyName = null)
{
PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
}
...and now with C#7:
protected void OnPropertyChanged(string propertyName)
=> PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
protected bool SetField<T>(ref T field, T value,[CallerMemberName] string propertyName = null)
{
if (EqualityComparer<T>.Default.Equals(field, value)) return false;
field = value;
OnPropertyChanged(propertyName);
return true;
}
private string name;
public string Name
{
get => name;
set => SetField(ref name, value);
}
And, with C# 8 and Nullable reference types, it would look like this:
public event PropertyChangedEventHandler? PropertyChanged;
protected void OnPropertyChanged(string propertyName) => PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
protected bool SetField<T>(ref T field, T value, [CallerMemberName] string propertyName = "")
{
if (EqualityComparer<T>.Default.Equals(field, value)) return false;
field = value;
OnPropertyChanged(propertyName);
return true;
}
private string name;
public string Name
{
get => name;
set => SetField(ref name, value);
}
As of .Net 4.5 there is finally an easy way to do this.
.Net 4.5 introduces a new Caller Information Attributes.
private void OnPropertyChanged<T>([CallerMemberName]string caller = null) {
// make sure only to call this if the value actually changes
var handler = PropertyChanged;
if (handler != null) {
handler(this, new PropertyChangedEventArgs(caller));
}
}
It's probably a good idea to add a comparer to the function as well.
EqualityComparer<T>.Default.Equals
More examples here and here
Also see Caller Information (C# and Visual Basic)
I really like Marc's solution, but I think it can be slightly improved to avoid using a "magic string" (which doesn't support refactoring). Instead of using the property name as a string, it's easy to make it a lambda expression :
private string name;
public string Name
{
get { return name; }
set { SetField(ref name, value, () => Name); }
}
Just add the following methods to Marc's code, it will do the trick :
protected virtual void OnPropertyChanged<T>(Expression<Func<T>> selectorExpression)
{
if (selectorExpression == null)
throw new ArgumentNullException("selectorExpression");
MemberExpression body = selectorExpression.Body as MemberExpression;
if (body == null)
throw new ArgumentException("The body must be a member expression");
OnPropertyChanged(body.Member.Name);
}
protected bool SetField<T>(ref T field, T value, Expression<Func<T>> selectorExpression)
{
if (EqualityComparer<T>.Default.Equals(field, value)) return false;
field = value;
OnPropertyChanged(selectorExpression);
return true;
}
BTW, this was inspired by this blog post.
There's also Fody which has a AddINotifyPropertyChangedInterface add-in, which lets you write this:
[AddINotifyPropertyChangedInterface]
public class Person
{
public string GivenNames { get; set; }
public string FamilyName { get; set; }
}
...and at compile time injects property changed notifications.
I think people should pay a little more attention to performance; it really does impact the UI when there are a lot of objects to be bound (think of a grid with 10,000+ rows), or if the object's value changes frequently (real-time monitoring app).
I took various implementation found here and elsewhere and did a comparison; check it out perfomance comparison of INotifyPropertyChanged implementations.
Here is a peek at the result
I introduce a Bindable class in my blog at http://timoch.com/blog/2013/08/annoyed-with-inotifypropertychange/
Bindable uses a dictionary as a property bag. It's easy enough to add the necessary overloads for a subclass to manage its own backing field using ref parameters.
No magic string
No reflection
Can be improved to suppress the default dictionary lookup
The code:
public class Bindable : INotifyPropertyChanged {
private Dictionary<string, object> _properties = new Dictionary<string, object>();
/// <summary>
/// Gets the value of a property
/// </summary>
/// <typeparam name="T"></typeparam>
/// <param name="name"></param>
/// <returns></returns>
protected T Get<T>([CallerMemberName] string name = null) {
Debug.Assert(name != null, "name != null");
object value = null;
if (_properties.TryGetValue(name, out value))
return value == null ? default(T) : (T)value;
return default(T);
}
/// <summary>
/// Sets the value of a property
/// </summary>
/// <typeparam name="T"></typeparam>
/// <param name="value"></param>
/// <param name="name"></param>
/// <remarks>Use this overload when implicitly naming the property</remarks>
protected void Set<T>(T value, [CallerMemberName] string name = null) {
Debug.Assert(name != null, "name != null");
if (Equals(value, Get<T>(name)))
return;
_properties[name] = value;
OnPropertyChanged(name);
}
public event PropertyChangedEventHandler PropertyChanged;
protected virtual void OnPropertyChanged([CallerMemberName] string propertyName = null) {
PropertyChangedEventHandler handler = PropertyChanged;
if (handler != null) {
handler(this, new PropertyChangedEventArgs(propertyName));
}
}
}
It can be used like this:
public class Contact : Bindable {
public string FirstName {
get { return Get<string>(); }
set { Set(value); }
}
}
I haven't actually had a chance to try this myself yet, but next time I'm setting up a project with a big requirement for INotifyPropertyChanged I'm intending on writing a Postsharp attribute that will inject the code at compile time. Something like:
[NotifiesChange]
public string FirstName { get; set; }
Will become:
private string _firstName;
public string FirstName
{
get { return _firstname; }
set
{
if (_firstname != value)
{
_firstname = value;
OnPropertyChanged("FirstName")
}
}
}
I'm not sure if this will work in practice and I need to sit down and try it out, but I don't see why not. I may need to make it accept some parameters for situations where more than one OnPropertyChanged needs to be triggered (if, for example, I had a FullName property in the class above)
Currently I'm using a custom template in Resharper, but even with that I'm getting fed up of all my properties being so long.
Ah, a quick Google search (which I should have done before I wrote this) shows that at least one person has done something like this before here. Not exactly what I had in mind, but close enough to show that the theory is good.
Yes, better way certainly exists.
Here it is:
Step by step tutorial shrank by me, based on this useful article.
Create new project
Install castle core package into the project
Install-Package Castle.Core
Install mvvm light libraries only
Install-Package MvvmLightLibs
Add two classes in project:
NotifierInterceptor
public class NotifierInterceptor : IInterceptor
{
private PropertyChangedEventHandler handler;
public static Dictionary<String, PropertyChangedEventArgs> _cache =
new Dictionary<string, PropertyChangedEventArgs>();
public void Intercept(IInvocation invocation)
{
switch (invocation.Method.Name)
{
case "add_PropertyChanged":
handler = (PropertyChangedEventHandler)
Delegate.Combine(handler, (Delegate)invocation.Arguments[0]);
invocation.ReturnValue = handler;
break;
case "remove_PropertyChanged":
handler = (PropertyChangedEventHandler)
Delegate.Remove(handler, (Delegate)invocation.Arguments[0]);
invocation.ReturnValue = handler;
break;
default:
if (invocation.Method.Name.StartsWith("set_"))
{
invocation.Proceed();
if (handler != null)
{
var arg = retrievePropertyChangedArg(invocation.Method.Name);
handler(invocation.Proxy, arg);
}
}
else invocation.Proceed();
break;
}
}
private static PropertyChangedEventArgs retrievePropertyChangedArg(String methodName)
{
PropertyChangedEventArgs arg = null;
_cache.TryGetValue(methodName, out arg);
if (arg == null)
{
arg = new PropertyChangedEventArgs(methodName.Substring(4));
_cache.Add(methodName, arg);
}
return arg;
}
}
ProxyCreator
public class ProxyCreator
{
public static T MakeINotifyPropertyChanged<T>() where T : class, new()
{
var proxyGen = new ProxyGenerator();
var proxy = proxyGen.CreateClassProxy(
typeof(T),
new[] { typeof(INotifyPropertyChanged) },
ProxyGenerationOptions.Default,
new NotifierInterceptor()
);
return proxy as T;
}
}
Create your view model, for example:
-
public class MainViewModel
{
public virtual string MainTextBox { get; set; }
public RelayCommand TestActionCommand
{
get { return new RelayCommand(TestAction); }
}
public void TestAction()
{
Trace.WriteLine(MainTextBox);
}
}
Put bindings into xaml:
<TextBox Text="{Binding MainTextBox}" ></TextBox>
<Button Command="{Binding TestActionCommand}" >Test</Button>
Put line of code in code-behind file MainWindow.xaml.cs like this:
DataContext = ProxyCreator.MakeINotifyPropertyChanged<MainViewModel>();
Enjoy.
Attention!!! All bounded properties should be decorated with
keyword virtual because they used by castle proxy for overriding.
A very AOP-like approach is to inject the INotifyPropertyChanged stuff onto an already instantiated object on the fly. You can do this with something like Castle DynamicProxy. Here is an article that explains the technique:
Adding INotifyPropertyChanged to an existing object
It's 2022. Now there's an official solution.
Use the MVVM source generators in Microsoft MVVM Toolkit.
This
[ObservableProperty]
private string? name;
will generate:
private string? name;
public string? Name
{
get => name;
set
{
if (!EqualityComparer<string?>.Default.Equals(name, value))
{
OnNameChanging(value);
OnPropertyChanging();
name = value;
OnNameChanged(value);
OnPropertyChanged();
}
}
}
// Property changing / changed listener
partial void OnNameChanging(string? value);
partial void OnNameChanged(string? value);
protected void OnPropertyChanging([CallerMemberName] string? propertyName = null)
{
PropertyChanging?.Invoke(this, new PropertyChangingEventArgs(propertyName));
}
protected void OnPropertyChanged([CallerMemberName] string? propertyName = null)
{
PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
}
It supports .NET standard 2.0 and .NET >= 5.0.
Look here : http://dotnet-forum.de/blogs/thearchitect/archive/2012/11/01/die-optimale-implementierung-des-inotifypropertychanged-interfaces.aspx
It's written in German, but you can download the ViewModelBase.cs. All the comments in the cs-File are written in English.
With this ViewModelBase-Class it is possible to implement bindable properties similar to the well known Dependency Properties :
public string SomeProperty
{
get { return GetValue( () => SomeProperty ); }
set { SetValue( () => SomeProperty, value ); }
}
Let me introduce my own approach called Yappi.
It belongs to Runtime proxy|derived class generators, adding new functionality to an existing object or type, like Caste Project's Dynamic Proxy.
It allows to implement INotifyPropertyChanged once in base class, and then declare derived classes in following style, still supporting INotifyPropertyChanged for new properties:
public class Animal:Concept
{
protected Animal(){}
public virtual string Name { get; set; }
public virtual int Age { get; set; }
}
Complexity of derived class or proxy construction can be hidden behind the following line:
var animal = Concept.Create<Animal>.New();
And all INotifyPropertyChanged implementation work can be done like this:
public class Concept:INotifyPropertyChanged
{
//Hide constructor
protected Concept(){}
public static class Create<TConcept> where TConcept:Concept
{
//Construct derived Type calling PropertyProxy.ConstructType
public static readonly Type Type = PropertyProxy.ConstructType<TConcept, Implementation<TConcept>>(new Type[0], true);
//Create constructing delegate calling Constructor.Compile
public static Func<TConcept> New = Constructor.Compile<Func<TConcept>>(Type);
}
public event PropertyChangedEventHandler PropertyChanged;
protected void OnPropertyChanged(PropertyChangedEventArgs eventArgs)
{
var caller = PropertyChanged;
if(caller!=null)
{
caller(this, eventArgs);
}
}
//define implementation
public class Implementation<TConcept> : DefaultImplementation<TConcept> where TConcept:Concept
{
public override Func<TBaseType, TResult> OverrideGetter<TBaseType, TDeclaringType, TConstructedType, TResult>(PropertyInfo property)
{
return PropertyImplementation<TBaseType, TDeclaringType>.GetGetter<TResult>(property.Name);
}
/// <summary>
/// Overriding property setter implementation.
/// </summary>
/// <typeparam name="TBaseType">Base type for implementation. TBaseType must be TConcept, and inherits all its constraints. Also TBaseType is TDeclaringType.</typeparam>
/// <typeparam name="TDeclaringType">Type, declaring property.</typeparam>
/// <typeparam name="TConstructedType">Constructed type. TConstructedType is TDeclaringType and TBaseType.</typeparam>
/// <typeparam name="TResult">Type of property.</typeparam>
/// <param name="property">PropertyInfo of property.</param>
/// <returns>Delegate, corresponding to property setter implementation.</returns>
public override Action<TBaseType, TResult> OverrideSetter<TBaseType, TDeclaringType, TConstructedType, TResult>(PropertyInfo property)
{
//This code called once for each declared property on derived type's initialization.
//EventArgs instance is shared between all events for each concrete property.
var eventArgs = new PropertyChangedEventArgs(property.Name);
//get delegates for base calls.
Action<TBaseType, TResult> setter = PropertyImplementation<TBaseType, TDeclaringType>.GetSetter<TResult>(property.Name);
Func<TBaseType, TResult> getter = PropertyImplementation<TBaseType, TDeclaringType>.GetGetter<TResult>(property.Name);
var comparer = EqualityComparer<TResult>.Default;
return (pthis, value) =>
{//This code executes each time property setter is called.
if (comparer.Equals(value, getter(pthis))) return;
//base. call
setter(pthis, value);
//Directly accessing Concept's protected method.
pthis.OnPropertyChanged(eventArgs);
};
}
}
}
It is fully safe for refactoring, uses no reflection after type construction and fast enough.
Whilst there are obviously lots of ways to do this, with the exception of the AOP magic answers, none of the answers seem to look at setting a Model's property directly from the view model without having a local field to reference.
The issue is you can't reference a property. However, you can use an Action to set that property.
protected bool TrySetProperty<T>(Action<T> property, T newValue, T oldValue, [CallerMemberName] string propertyName = null)
{
if (EqualityComparer<T>.Default.Equals(oldValue, newValue))
{
return false;
}
property(newValue);
RaisePropertyChanged(propertyName);
return true;
}
This can be used like the following code extract.
public int Prop {
get => model.Prop;
set => TrySetProperty(x => model.Prop = x, value, model.Prop);
}
Check out this BitBucket repo for a full implementation of the method and a few different ways of achieving the same result, including a method that uses LINQ and a method that uses reflection. Do note that these methods are slower performance wise.
All these answer are very nice.
My solution is using the code snippets to do the job.
This uses the simplest call to PropertyChanged event.
Save this snippet and use it as you use 'fullprop' snippet.
the location can be found at 'Tools\Code Snippet Manager...' menu at Visual Studio.
<?xml version="1.0" encoding="utf-8" ?>
<CodeSnippets xmlns="http://schemas.microsoft.com/VisualStudio/2005/CodeSnippet">
<CodeSnippet Format="1.0.0">
<Header>
<Title>inotifypropfull</Title>
<Shortcut>inotifypropfull</Shortcut>
<HelpUrl>http://ofirzeitoun.wordpress.com/</HelpUrl>
<Description>Code snippet for property and backing field with notification</Description>
<Author>Ofir Zeitoun</Author>
<SnippetTypes>
<SnippetType>Expansion</SnippetType>
</SnippetTypes>
</Header>
<Snippet>
<Declarations>
<Literal>
<ID>type</ID>
<ToolTip>Property type</ToolTip>
<Default>int</Default>
</Literal>
<Literal>
<ID>property</ID>
<ToolTip>Property name</ToolTip>
<Default>MyProperty</Default>
</Literal>
<Literal>
<ID>field</ID>
<ToolTip>The variable backing this property</ToolTip>
<Default>myVar</Default>
</Literal>
</Declarations>
<Code Language="csharp">
<![CDATA[private $type$ $field$;
public $type$ $property$
{
get { return $field$;}
set {
$field$ = value;
var temp = PropertyChanged;
if (temp != null)
{
temp(this, new PropertyChangedEventArgs("$property$"));
}
}
}
$end$]]>
</Code>
</Snippet>
</CodeSnippet>
</CodeSnippets>
You can modify the call as you like (to use the above solutions)
Based on the answer by Thomas which was adapted from an answer by Marc I've turned the reflecting property changed code into a base class:
public abstract class PropertyChangedBase : INotifyPropertyChanged
{
public event PropertyChangedEventHandler PropertyChanged;
protected void OnPropertyChanged(string propertyName)
{
PropertyChangedEventHandler handler = PropertyChanged;
if (handler != null)
handler(this, new PropertyChangedEventArgs(propertyName));
}
protected void OnPropertyChanged<T>(Expression<Func<T>> selectorExpression)
{
if (selectorExpression == null)
throw new ArgumentNullException("selectorExpression");
var me = selectorExpression.Body as MemberExpression;
// Nullable properties can be nested inside of a convert function
if (me == null)
{
var ue = selectorExpression.Body as UnaryExpression;
if (ue != null)
me = ue.Operand as MemberExpression;
}
if (me == null)
throw new ArgumentException("The body must be a member expression");
OnPropertyChanged(me.Member.Name);
}
protected void SetField<T>(ref T field, T value, Expression<Func<T>> selectorExpression, params Expression<Func<object>>[] additonal)
{
if (EqualityComparer<T>.Default.Equals(field, value)) return;
field = value;
OnPropertyChanged(selectorExpression);
foreach (var item in additonal)
OnPropertyChanged(item);
}
}
Usage is the same as Thomas' answer except that you can pass additional properties to notify for. This was necessary to handle calculated columns which need to be refreshed in a grid.
private int _quantity;
private int _price;
public int Quantity
{
get { return _quantity; }
set { SetField(ref _quantity, value, () => Quantity, () => Total); }
}
public int Price
{
get { return _price; }
set { SetField(ref _price, value, () => Price, () => Total); }
}
public int Total { get { return _price * _quantity; } }
I have this driving a collection of items stored in a BindingList exposed via a DataGridView. It has eliminated the need for me to do manual Refresh() calls to the grid.
I created an Extension Method in my base Library for reuse:
public static class INotifyPropertyChangedExtensions
{
public static bool SetPropertyAndNotify<T>(this INotifyPropertyChanged sender,
PropertyChangedEventHandler handler, ref T field, T value,
[CallerMemberName] string propertyName = "",
EqualityComparer<T> equalityComparer = null)
{
bool rtn = false;
var eqComp = equalityComparer ?? EqualityComparer<T>.Default;
if (!eqComp.Equals(field,value))
{
field = value;
rtn = true;
if (handler != null)
{
var args = new PropertyChangedEventArgs(propertyName);
handler(sender, args);
}
}
return rtn;
}
}
This works with .Net 4.5 because of CallerMemberNameAttribute.
If you want to use it with an earlier .Net version you have to change the method declaration from: ...,[CallerMemberName] string propertyName = "", ... to ...,string propertyName, ...
Usage:
public class Dog : INotifyPropertyChanged
{
public event PropertyChangedEventHandler PropertyChanged;
string _name;
public string Name
{
get { return _name; }
set
{
this.SetPropertyAndNotify(PropertyChanged, ref _name, value);
}
}
}
I keep this around as a snippet. C# 6 adds some nice syntax for invoking the handler.
// INotifyPropertyChanged
public event PropertyChangedEventHandler PropertyChanged;
private void Set<T>(ref T property, T value, [CallerMemberName] string propertyName = null)
{
if (EqualityComparer<T>.Default.Equals(property, value) == false)
{
property = value;
PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
}
}
I have just found ActiveSharp - Automatic INotifyPropertyChanged, I have yet to use it, but it looks good.
To quote from it's web site...
Send property change notifications
without specifying property name as a
string.
Instead, write properties like this:
public int Foo
{
get { return _foo; }
set { SetValue(ref _foo, value); } // <-- no property name here
}
Note that there is no need to include the name of the property as a string. ActiveSharp reliably and correctly figures that out for itself. It works based on the fact that your property implementation passes the backing field (_foo) by ref. (ActiveSharp uses that "by ref" call to identify which backing field was passed, and from the field it identifies the property).
If you are using dynamics in .NET 4.5 you don't need to worry about INotifyPropertyChanged.
dynamic obj = new ExpandoObject();
obj.Name = "John";
if Name is bound to some control it just works fine.
Another combined solution is using StackFrame:
public class BaseViewModel : INotifyPropertyChanged
{
public event PropertyChangedEventHandler PropertyChanged;
protected void Set<T>(ref T field, T value)
{
MethodBase method = new StackFrame(1).GetMethod();
field = value;
Raise(method.Name.Substring(4));
}
protected void Raise(string propertyName)
{
var temp = PropertyChanged;
if (temp != null)
{
temp(this, new PropertyChangedEventArgs(propertyName));
}
}
}
Usage:
public class TempVM : BaseViewModel
{
private int _intP;
public int IntP
{
get { return _intP; }
set { Set<int>(ref _intP, value); }
}
}
I resolved in This Way (it's a little bit laboriouse, but it's surely the faster in runtime).
In VB (sorry, but I think it's not hard translate it in C#), I make this substitution with RE:
(?<Attr><(.*ComponentModel\.)Bindable\(True\)>)( |\r\n)*(?<Def>(Public|Private|Friend|Protected) .*Property )(?<Name>[^ ]*) As (?<Type>.*?)[ |\r\n](?![ |\r\n]*Get)
with:
Private _${Name} As ${Type}\r\n${Attr}\r\n${Def}${Name} As ${Type}\r\nGet\r\nReturn _${Name}\r\nEnd Get\r\nSet (Value As ${Type})\r\nIf _${Name} <> Value Then \r\n_${Name} = Value\r\nRaiseEvent PropertyChanged(Me, New ComponentModel.PropertyChangedEventArgs("${Name}"))\r\nEnd If\r\nEnd Set\r\nEnd Property\r\n
This transofrm all code like this:
<Bindable(True)>
Protected Friend Property StartDate As DateTime?
In
Private _StartDate As DateTime?
<Bindable(True)>
Protected Friend Property StartDate As DateTime?
Get
Return _StartDate
End Get
Set(Value As DateTime?)
If _StartDate <> Value Then
_StartDate = Value
RaiseEvent PropertyChange(Me, New ComponentModel.PropertyChangedEventArgs("StartDate"))
End If
End Set
End Property
And If I want to have a more readable code, I can be the opposite just making the following substitution:
Private _(?<Name>.*) As (?<Type>.*)[\r\n ]*(?<Attr><(.*ComponentModel\.)Bindable\(True\)>)[\r\n ]*(?<Def>(Public|Private|Friend|Protected) .*Property )\k<Name> As \k<Type>[\r\n ]*Get[\r\n ]*Return _\k<Name>[\r\n ]*End Get[\r\n ]*Set\(Value As \k<Type>\)[\r\n ]*If _\k<Name> <> Value Then[\r\n ]*_\k<Name> = Value[\r\n ]*RaiseEvent PropertyChanged\(Me, New (.*ComponentModel\.)PropertyChangedEventArgs\("\k<Name>"\)\)[\r\n ]*End If[\r\n ]*End Set[\r\n ]*End Property
With
${Attr} ${Def} ${Name} As ${Type}
I throw to replace the IL code of the set method, but I can't write a lot of compiled code in IL... If a day I write it, I'll say you!
Here is a Unity3D or non-CallerMemberName version of NotifyPropertyChanged
public abstract class Bindable : MonoBehaviour, INotifyPropertyChanged
{
private readonly Dictionary<string, object> _properties = new Dictionary<string, object>();
private static readonly StackTrace stackTrace = new StackTrace();
public event PropertyChangedEventHandler PropertyChanged;
/// <summary>
/// Resolves a Property's name from a Lambda Expression passed in.
/// </summary>
/// <typeparam name="T"></typeparam>
/// <param name="property"></param>
/// <returns></returns>
internal string GetPropertyName<T>(Expression<Func<T>> property)
{
var expression = (MemberExpression) property.Body;
var propertyName = expression.Member.Name;
Debug.AssertFormat(propertyName != null, "Bindable Property shouldn't be null!");
return propertyName;
}
#region Notification Handlers
/// <summary>
/// Notify's all other objects listening that a value has changed for nominated propertyName
/// </summary>
/// <param name="propertyName"></param>
internal void NotifyOfPropertyChange(string propertyName)
{
OnPropertyChanged(new PropertyChangedEventArgs(propertyName));
}
/// <summary>
/// Notifies subscribers of the property change.
/// </summary>
/// <typeparam name="TProperty">The type of the property.</typeparam>
/// <param name="property">The property expression.</param>
internal void NotifyOfPropertyChange<TProperty>(Expression<Func<TProperty>> property)
{
var propertyName = GetPropertyName(property);
NotifyOfPropertyChange(propertyName);
}
/// <summary>
/// Raises the <see cref="PropertyChanged" /> event directly.
/// </summary>
/// <param name="e">The <see cref="PropertyChangedEventArgs" /> instance containing the event data.</param>
internal void OnPropertyChanged(PropertyChangedEventArgs e)
{
var handler = PropertyChanged;
if (handler != null)
{
handler(this, e);
}
}
#endregion
#region Getters
/// <summary>
/// Gets the value of a property
/// </summary>
/// <typeparam name="T"></typeparam>
/// <param name="name"></param>
/// <returns></returns>
internal T Get<T>(Expression<Func<T>> property)
{
var propertyName = GetPropertyName(property);
return Get<T>(GetPropertyName(property));
}
/// <summary>
/// Gets the value of a property automatically based on its caller.
/// </summary>
/// <typeparam name="T"></typeparam>
/// <returns></returns>
internal T Get<T>()
{
var name = stackTrace.GetFrame(1).GetMethod().Name.Substring(4); // strips the set_ from name;
return Get<T>(name);
}
/// <summary>
/// Gets the name of a property based on a string.
/// </summary>
/// <typeparam name="T"></typeparam>
/// <param name="name"></param>
/// <returns></returns>
internal T Get<T>(string name)
{
object value = null;
if (_properties.TryGetValue(name, out value))
return value == null ? default(T) : (T) value;
return default(T);
}
#endregion
#region Setters
/// <summary>
/// Sets the value of a property whilst automatically looking up its caller name.
/// </summary>
/// <typeparam name="T"></typeparam>
/// <param name="value"></param>
internal void Set<T>(T value)
{
var propertyName = stackTrace.GetFrame(1).GetMethod().Name.Substring(4); // strips the set_ from name;
Set(value, propertyName);
}
/// <summary>
/// Sets the value of a property
/// </summary>
/// <typeparam name="T"></typeparam>
/// <param name="value"></param>
/// <param name="name"></param>
internal void Set<T>(T value, string propertyName)
{
Debug.Assert(propertyName != null, "name != null");
if (Equals(value, Get<T>(propertyName)))
return;
_properties[propertyName] = value;
NotifyOfPropertyChange(propertyName);
}
/// <summary>
/// Sets the value of a property based off an Expression (()=>FieldName)
/// </summary>
/// <typeparam name="T"></typeparam>
/// <param name="value"></param>
/// <param name="property"></param>
internal void Set<T>(T value, Expression<Func<T>> property)
{
var propertyName = GetPropertyName(property);
Debug.Assert(propertyName != null, "name != null");
if (Equals(value, Get<T>(propertyName)))
return;
_properties[propertyName] = value;
NotifyOfPropertyChange(propertyName);
}
#endregion
}
This code enables you to write property backing fields like this:
public string Text
{
get { return Get<string>(); }
set { Set(value); }
}
Furthermore, in resharper if you create a pattern/search snippet you can then also automate you're workflow by converting simple prop fields into the above backing.
Search Pattern:
public $type$ $fname$ { get; set; }
Replace Pattern:
public $type$ $fname$
{
get { return Get<$type$>(); }
set { Set(value); }
}
I have written an article that helps with this (https://msdn.microsoft.com/magazine/mt736453). You can use the SolSoft.DataBinding NuGet package. Then you can write code like this:
public class TestViewModel : IRaisePropertyChanged
{
public TestViewModel()
{
this.m_nameProperty = new NotifyProperty<string>(this, nameof(Name), null);
}
private readonly NotifyProperty<string> m_nameProperty;
public string Name
{
get
{
return m_nameProperty.Value;
}
set
{
m_nameProperty.SetValue(value);
}
}
// Plus implement IRaisePropertyChanged (or extend BaseViewModel)
}
Benefits:
base class is optional
no reflection on every 'set value'
can have properties that depend on other properties, and they all automatically raise the appropriate events (article has an example of this)
I came up with this base class to implement the observable pattern, pretty much does what you need ("automatically" implementing the set and get). I spent line an hour on this as prototype, so it doesn't have many unit tests, but proves the concept. Note it uses the Dictionary<string, ObservablePropertyContext> to remove the need for private fields.
public class ObservableByTracking<T> : IObservable<T>
{
private readonly Dictionary<string, ObservablePropertyContext> _expando;
private bool _isDirty;
public ObservableByTracking()
{
_expando = new Dictionary<string, ObservablePropertyContext>();
var properties = this.GetType().GetProperties(BindingFlags.Public | BindingFlags.Instance).ToList();
foreach (var property in properties)
{
var valueContext = new ObservablePropertyContext(property.Name, property.PropertyType)
{
Value = GetDefault(property.PropertyType)
};
_expando[BuildKey(valueContext)] = valueContext;
}
}
protected void SetValue<T>(Expression<Func<T>> expression, T value)
{
var keyContext = GetKeyContext(expression);
var key = BuildKey(keyContext.PropertyName, keyContext.PropertyType);
if (!_expando.ContainsKey(key))
{
throw new Exception($"Object doesn't contain {keyContext.PropertyName} property.");
}
var originalValue = (T)_expando[key].Value;
if (EqualityComparer<T>.Default.Equals(originalValue, value))
{
return;
}
_expando[key].Value = value;
_isDirty = true;
}
protected T GetValue<T>(Expression<Func<T>> expression)
{
var keyContext = GetKeyContext(expression);
var key = BuildKey(keyContext.PropertyName, keyContext.PropertyType);
if (!_expando.ContainsKey(key))
{
throw new Exception($"Object doesn't contain {keyContext.PropertyName} property.");
}
var value = _expando[key].Value;
return (T)value;
}
private KeyContext GetKeyContext<T>(Expression<Func<T>> expression)
{
var castedExpression = expression.Body as MemberExpression;
if (castedExpression == null)
{
throw new Exception($"Invalid expression.");
}
var parameterName = castedExpression.Member.Name;
var propertyInfo = castedExpression.Member as PropertyInfo;
if (propertyInfo == null)
{
throw new Exception($"Invalid expression.");
}
return new KeyContext {PropertyType = propertyInfo.PropertyType, PropertyName = parameterName};
}
private static string BuildKey(ObservablePropertyContext observablePropertyContext)
{
return $"{observablePropertyContext.Type.Name}.{observablePropertyContext.Name}";
}
private static string BuildKey(string parameterName, Type type)
{
return $"{type.Name}.{parameterName}";
}
private static object GetDefault(Type type)
{
if (type.IsValueType)
{
return Activator.CreateInstance(type);
}
return null;
}
public bool IsDirty()
{
return _isDirty;
}
public void SetPristine()
{
_isDirty = false;
}
private class KeyContext
{
public string PropertyName { get; set; }
public Type PropertyType { get; set; }
}
}
public interface IObservable<T>
{
bool IsDirty();
void SetPristine();
}
Here's the usage
public class ObservableByTrackingTestClass : ObservableByTracking<ObservableByTrackingTestClass>
{
public ObservableByTrackingTestClass()
{
StringList = new List<string>();
StringIList = new List<string>();
NestedCollection = new List<ObservableByTrackingTestClass>();
}
public IEnumerable<string> StringList
{
get { return GetValue(() => StringList); }
set { SetValue(() => StringIList, value); }
}
public IList<string> StringIList
{
get { return GetValue(() => StringIList); }
set { SetValue(() => StringIList, value); }
}
public int IntProperty
{
get { return GetValue(() => IntProperty); }
set { SetValue(() => IntProperty, value); }
}
public ObservableByTrackingTestClass NestedChild
{
get { return GetValue(() => NestedChild); }
set { SetValue(() => NestedChild, value); }
}
public IList<ObservableByTrackingTestClass> NestedCollection
{
get { return GetValue(() => NestedCollection); }
set { SetValue(() => NestedCollection, value); }
}
public string StringProperty
{
get { return GetValue(() => StringProperty); }
set { SetValue(() => StringProperty, value); }
}
}
Other things you may want to consider when implementing these sorts of properties is the fact that the INotifyPropertyChang *ed *ing both use event argument classes.
If you have a large number of properties that are being set then the number of event argument class instances can be huge, you should consider caching them as they are one of the areas that a string explosion can occur.
Take a look at this implementation and explanation of why it was conceived.
Josh Smiths Blog
An idea using reflection:
class ViewModelBase : INotifyPropertyChanged {
public event PropertyChangedEventHandler PropertyChanged;
bool Notify<T>(MethodBase mb, ref T oldValue, T newValue) {
// Get Name of Property
string name = mb.Name.Substring(4);
// Detect Change
bool changed = EqualityComparer<T>.Default.Equals(oldValue, newValue);
// Return if no change
if (!changed) return false;
// Update value
oldValue = newValue;
// Raise Event
if (PropertyChanged != null) {
PropertyChanged(this, new PropertyChangedEventArgs(name));
}//if
// Notify caller of change
return true;
}//method
string name;
public string Name {
get { return name; }
set {
Notify(MethodInfo.GetCurrentMethod(), ref this.name, value);
}
}//method
}//class
Use this
using System;
using System.ComponentModel;
using System.Reflection;
using System.Reflection.Emit;
using System.Runtime.Remoting.Messaging;
using System.Runtime.Remoting.Proxies;
public static class ObservableFactory
{
public static T Create<T>(T target)
{
if (!typeof(T).IsInterface)
throw new ArgumentException("Target should be an interface", "target");
var proxy = new Observable<T>(target);
return (T)proxy.GetTransparentProxy();
}
}
internal class Observable<T> : RealProxy, INotifyPropertyChanged, INotifyPropertyChanging
{
private readonly T target;
internal Observable(T target)
: base(ImplementINotify(typeof(T)))
{
this.target = target;
}
public override IMessage Invoke(IMessage msg)
{
var methodCall = msg as IMethodCallMessage;
if (methodCall != null)
{
return HandleMethodCall(methodCall);
}
return null;
}
public event PropertyChangingEventHandler PropertyChanging;
public event PropertyChangedEventHandler PropertyChanged;
IMessage HandleMethodCall(IMethodCallMessage methodCall)
{
var isPropertySetterCall = methodCall.MethodName.StartsWith("set_");
var propertyName = isPropertySetterCall ? methodCall.MethodName.Substring(4) : null;
if (isPropertySetterCall)
{
OnPropertyChanging(propertyName);
}
try
{
object methodCalltarget = target;
if (methodCall.MethodName == "add_PropertyChanged" || methodCall.MethodName == "remove_PropertyChanged"||
methodCall.MethodName == "add_PropertyChanging" || methodCall.MethodName == "remove_PropertyChanging")
{
methodCalltarget = this;
}
var result = methodCall.MethodBase.Invoke(methodCalltarget, methodCall.InArgs);
if (isPropertySetterCall)
{
OnPropertyChanged(methodCall.MethodName.Substring(4));
}
return new ReturnMessage(result, null, 0, methodCall.LogicalCallContext, methodCall);
}
catch (TargetInvocationException invocationException)
{
var exception = invocationException.InnerException;
return new ReturnMessage(exception, methodCall);
}
}
protected virtual void OnPropertyChanged(string propertyName)
{
var handler = PropertyChanged;
if (handler != null) handler(this, new PropertyChangedEventArgs(propertyName));
}
protected virtual void OnPropertyChanging(string propertyName)
{
var handler = PropertyChanging;
if (handler != null) handler(this, new PropertyChangingEventArgs(propertyName));
}
public static Type ImplementINotify(Type objectType)
{
var tempAssemblyName = new AssemblyName(Guid.NewGuid().ToString());
var dynamicAssembly = AppDomain.CurrentDomain.DefineDynamicAssembly(
tempAssemblyName, AssemblyBuilderAccess.RunAndCollect);
var moduleBuilder = dynamicAssembly.DefineDynamicModule(
tempAssemblyName.Name,
tempAssemblyName + ".dll");
var typeBuilder = moduleBuilder.DefineType(
objectType.FullName, TypeAttributes.Public | TypeAttributes.Interface | TypeAttributes.Abstract);
typeBuilder.AddInterfaceImplementation(objectType);
typeBuilder.AddInterfaceImplementation(typeof(INotifyPropertyChanged));
typeBuilder.AddInterfaceImplementation(typeof(INotifyPropertyChanging));
var newType = typeBuilder.CreateType();
return newType;
}
}
}
I use the following extension method (using C# 6.0) to make the INPC implemenation as easy as possible:
public static bool ChangeProperty<T>(this PropertyChangedEventHandler propertyChanged, ref T field, T value, object sender,
IEqualityComparer<T> comparer = null, [CallerMemberName] string propertyName = null)
{
if (comparer == null)
comparer = EqualityComparer<T>.Default;
if (comparer.Equals(field, value))
{
return false;
}
else
{
field = value;
propertyChanged?.Invoke(sender, new PropertyChangedEventArgs(propertyName));
return true;
}
}
The INPC implementation boils down to (you can either implement this every time or create a base class):
public class INPCBaseClass: INotifyPropertyChanged
{
public event PropertyChangedEventHandler PropertyChanged;
protected bool changeProperty<T>(ref T field, T value,
IEqualityComparer<T> comparer = null, [CallerMemberName] string propertyName = null)
{
return PropertyChanged.ChangeProperty(ref field, value, this, comparer, propertyName);
}
}
Then write your properties like this:
private string testProperty;
public string TestProperty
{
get { return testProperty; }
set { changeProperty(ref testProperty, value); }
}
NOTE: You can omit the [CallerMemberName] declaration in the extension method, if you want, but I wanted to keep it flexible.
If you have properties without a backing field you can overload changeProperty:
protected bool changeProperty<T>(T property, Action<T> set, T value,
IEqualityComparer<T> comparer = null, [CallerMemberName] string propertyName = null)
{
bool ret = changeProperty(ref property, value, comparer, propertyName);
if (ret)
set(property);
return ret;
}
An example use would be:
public string MyTestProperty
{
get { return base.TestProperty; }
set { changeProperty(base.TestProperty, (x) => { base.TestProperty = x; }, value); }
}
I realize this question already has a gazillion answers, but none of them felt quite right for me. My issue is I don't want any performance hits and am willing to put up with a little verbosity for that reason alone. I also don't care too much for auto properties either, which led me to the following solution:
public abstract class AbstractObject : INotifyPropertyChanged
{
public event PropertyChangedEventHandler PropertyChanged;
public void OnPropertyChanged(string propertyName)
{
PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
}
protected virtual bool SetValue<TKind>(ref TKind Source, TKind NewValue, params string[] Notify)
{
//Set value if the new value is different from the old
if (!Source.Equals(NewValue))
{
Source = NewValue;
//Notify all applicable properties
foreach (var i in Notify)
OnPropertyChanged(i);
return true;
}
return false;
}
public AbstractObject()
{
}
}
In other words, the above solution is convenient if you don't mind doing this:
public class SomeObject : AbstractObject
{
public string AnotherProperty
{
get
{
return someProperty ? "Car" : "Plane";
}
}
bool someProperty = false;
public bool SomeProperty
{
get
{
return someProperty;
}
set
{
SetValue(ref someProperty, value, "SomeProperty", "AnotherProperty");
}
}
public SomeObject() : base()
{
}
}
Pros
No reflection
Only notifies if old value != new value
Notify multiple properties at once
Cons
No auto properties (you can add support for both, though!)
Some verbosity
Boxing (small performance hit?)
Alas, it is still better than doing this,
set
{
if (!someProperty.Equals(value))
{
someProperty = value;
OnPropertyChanged("SomeProperty");
OnPropertyChanged("AnotherProperty");
}
}
For every single property, which becomes a nightmare with the additional verbosity ;-(
Note, I do not claim this solution is better performance-wise compared to the others, just that it is a viable solution for those who don't like the other solutions presented.
I suggest to use ReactiveProperty.
This is the shortest method except Fody.
public class Data : INotifyPropertyChanged
{
// boiler-plate
...
// props
private string name;
public string Name
{
get { return name; }
set { SetField(ref name, value, "Name"); }
}
}
instead
public class Data
{
// Don't need boiler-plate and INotifyPropertyChanged
// props
public ReactiveProperty<string> Name { get; } = new ReactiveProperty<string>();
}
(DOCS)
I am passing information between a SQL database and a PLC using 3rd party OPC libraries.
There are essentially two transactions.
Information passed from the PLC to the SQL server is statically typed. Very specific data is captured by the PLC and passed to the SQL database.
Information passed from the SQL server to the PLC is dynamically typed and may be limited to a single property or hundreds.
ITransaction.cs
public interface ITransaction : INotifyPropertyChanged
{
short Response { get; set; }
bool Request { get; set; }
void Reset();
}
BaseTransaction.cs
internal abstract class BaseTransaction<T> : IDisposable
where T : class, INotifyPropertyChanged
{
private T _opcClient;
protected T OpcClient
{
get { return _opcClient; }
set
{
if (_opcClient != value)
{
OnOpcClientChanging();
_opcClient = value;
OnOpcClientChanged();
}
}
}
protected abstract void OnOpcClientPropertyChanged(object sender, PropertyChangedEventArgs e);
private void OnOpcClientChanged()
{
if (_opcClient != null)
{
_opcClient.PropertyChanged += OnOpcClientPropertyChanged;
OpcManager = new OpcManager(_opcClient);
}
}
private void OnOpcClientChanging()
{
if (_opcClient != null)
_opcClient.PropertyChanged -= OnOpcClientPropertyChanged;
}
}
StaticTransaction.cs
internal abstract class StaticTransaction<T> : BaseTransaction<T>
where T : class, ITransaction, new()
{
public StaticTransaction()
{
OpcClient = new T();
}
protected override void OnOpcClientPropertyChanged(object sender, PropertyChangedEventArgs e)
{
switch (e.PropertyName)
{
case "Response":
ProcessResponse(OpcClient.Response);
break;
case "Request":
ProcessRequest(OpcClient.Request);
break;
}
}
}
DynamicTransaction.cs
internal abstract class DynamicTransaction : BaseTransaction<ExpandoObject>
{
protected new dynamic OpcClient
{
get { return base.OpcClient as dynamic; }
}
public DynamicTransaction()
{
dynamic opcClient = new ExpandoObject();
opcClient.Request = false;
opcClient.Response = 0;
// Access database, use IDictionary interface to add properties to ExpandoObject.
opcClient.Reset = new Action(Reset);
base.OpcClient = opcClient;
}
protected override void OnOpcClientPropertyChanged(object sender, PropertyChangedEventArgs e)
{
switch (e.PropertyName)
{
case "Response":
ProcessResponse(OpcClient.Response);
break;
case "Request":
ProcessRequest(OpcClient.Request);
break;
}
}
private void Reset()
{
// Use IDictionary interface to reset dynamic properties to defaults.
OpcClient.Request = false;
OpcClient.Response = 0;
}
}
As shown both StaticTransaction and DynamicTransaction have identical implementations of OnOpcClientPropertyChanged among other methods not shown. I would like to bring OnOpcClientPropertyChanged and the other methods into the base class but am prevented from doing so because the base class is unaware of the Response and Request properties found in the OpcClient. Can I bring the interface ITransaction into the base class somehow and still accommodate the dynamic implementation?
You can subclass DynamicObject (which acts just like ExpandoObject) and make your own version that implements ITransaction. This lets you move the ITransaction constraint up to the base class.
BaseTransaction.cs
internal abstract class BaseTransaction<T> : IDisposable where T : class, ITransaction
{
private T _opcClient;
protected T OpcClient
{
get { return _opcClient; }
set
{
if (_opcClient != value)
{
OnOpcClientChanging();
_opcClient = value;
OnOpcClientChanged();
}
}
}
private void OnOpcClientPropertyChanged(object sender, PropertyChangedEventArgs e)
{
switch (e.PropertyName)
{
case "Response":
ProcessResponse(OpcClient.Response);
break;
case "Request":
ProcessRequest(OpcClient.Request);
break;
}
}
protected abstract void ProcessResponse(short opcClientResponse);
protected abstract void ProcessRequest(bool opcClientRequest);
private void OnOpcClientChanged()
{
if (_opcClient != null)
{
_opcClient.PropertyChanged += OnOpcClientPropertyChanged;
OpcManager = new OpcManager(_opcClient);
}
}
private void OnOpcClientChanging()
{
if (_opcClient != null)
_opcClient.PropertyChanged -= OnOpcClientPropertyChanged;
}
}
StaticTransaction.cs
internal abstract class StaticTransaction<T> : BaseTransaction<T>
where T : class, ITransaction, new()
{
public StaticTransaction()
{
OpcClient = new T();
}
}
DynamicTransactionObject.cs
internal class DynamicTransactionObject : DynamicObject, ITransaction, IDictionary<string, object>
{
private readonly Dictionary<string, object> _data = new Dictionary<string, object>();
public DynamicTransactionObject()
{
//Set initial default values for the two properties to populate the entries in the dictionary.
_data[nameof(Response)] = default(short);
_data[nameof(Request)] = default(bool);
}
public short Response
{
get
{
return (short)_data[nameof(Response)];
}
set
{
if (Response.Equals(value))
return;
_data[nameof(Response)] = value;
OnPropertyChanged();
}
}
public bool Request
{
get
{
return (bool)_data[nameof(Request)];
}
set
{
if (Request.Equals(value))
return;
_data[nameof(Request)] = value;
OnPropertyChanged();
}
}
public override IEnumerable<string> GetDynamicMemberNames()
{
return _data.Keys;
}
public override bool TryGetMember(GetMemberBinder binder, out object result)
{
return _data.TryGetValue(binder.Name, out result);
}
public override bool TrySetMember(SetMemberBinder binder, object value)
{
object oldValue;
_data.TryGetValue(binder.Name, out oldValue)
_data[binder.Name] = value;
if(!Object.Equals(oldValue, value)
OnPropertyChanged(binder.Name);
return true;
}
public event PropertyChangedEventHandler PropertyChanged;
protected virtual void OnPropertyChanged([CallerMemberName] string propertyName = null)
{
PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
}
#region IDictionary<string,object> members
IEnumerator<KeyValuePair<string, object>> IEnumerable<KeyValuePair<string, object>>.GetEnumerator()
{
return _data.GetEnumerator();
}
IEnumerator IEnumerable.GetEnumerator()
{
return ((IEnumerable)_data).GetEnumerator();
}
void ICollection<KeyValuePair<string, object>>.Add(KeyValuePair<string, object> item)
{
((ICollection<KeyValuePair<string, object>>)_data).Add(item);
}
void ICollection<KeyValuePair<string, object>>.Clear()
{
_data.Clear();
}
bool ICollection<KeyValuePair<string, object>>.Contains(KeyValuePair<string, object> item)
{
return ((ICollection<KeyValuePair<string, object>>)_data).Contains(item);
}
void ICollection<KeyValuePair<string, object>>.CopyTo(KeyValuePair<string, object>[] array, int arrayIndex)
{
((ICollection<KeyValuePair<string, object>>)_data).CopyTo(array, arrayIndex);
}
bool ICollection<KeyValuePair<string, object>>.Remove(KeyValuePair<string, object> item)
{
return ((ICollection<KeyValuePair<string, object>>)_data).Remove(item);
}
int ICollection<KeyValuePair<string, object>>.Count
{
get { return _data.Count; }
}
bool ICollection<KeyValuePair<string, object>>.IsReadOnly
{
get { return ((ICollection<KeyValuePair<string, object>>)_data).IsReadOnly; }
}
bool IDictionary<string, object>.ContainsKey(string key)
{
return _data.ContainsKey(key);
}
void IDictionary<string, object>.Add(string key, object value)
{
_data.Add(key, value);
}
bool IDictionary<string, object>.Remove(string key)
{
return _data.Remove(key);
}
bool IDictionary<string, object>.TryGetValue(string key, out object value)
{
return _data.TryGetValue(key, out value);
}
object IDictionary<string, object>.this[string key]
{
get { return _data[key]; }
set { _data[key] = value; }
}
ICollection<string> IDictionary<string, object>.Keys
{
get { return _data.Keys; }
}
ICollection<object> IDictionary<string, object>.Values
{
get { return _data.Values; }
}
#endregion
}
DynamicTransaction.cs
internal abstract class DynamicTransaction : BaseTransaction<DynamicTransactionObject>
{
protected new dynamic OpcClient
{
get { return base.OpcClient as dynamic; }
}
public DynamicTransaction()
{
var opcClient = new DynamicTransactionObject();
// Access database, use IDictionary<string,object> interface to add properties to DynamicObject.
base.OpcClient = opcClient;
}
}
I want to automatically show every IList as expandable in my PropertyGrid (By "expandable", I obviously mean that the items will be shown).
I don't want to use attributes on each list (Once again, I want it to work for EVERY IList)
I tried to achive it by using a custom PropertyDescriptor and an ExpandableObjectConverter. It works, but after I delete items from the list, the PropertyGrid is not being refreshed, still displaying the deleted items.
I tried to use ObservableCollection along with raising OnComponentChanged, and also RefreshProperties attribute, but nothing worked.
This is my code:
public class ExpandableCollectionPropertyDescriptor : PropertyDescriptor
{
private IList _collection;
private readonly int _index = -1;
internal event EventHandler RefreshRequired;
public ExpandableCollectionPropertyDescriptor(IList coll, int idx) : base(GetDisplayName(coll, idx), null)
{
_collection = coll
_index = idx;
}
public override bool SupportsChangeEvents
{
get { return true; }
}
private static string GetDisplayName(IList list, int index)
{
return "[" + index + "] " + CSharpName(list[index].GetType());
}
private static string CSharpName(Type type)
{
var sb = new StringBuilder();
var name = type.Name;
if (!type.IsGenericType)
return name;
sb.Append(name.Substring(0, name.IndexOf('`')));
sb.Append("<");
sb.Append(string.Join(", ", type.GetGenericArguments()
.Select(CSharpName)));
sb.Append(">");
return sb.ToString();
}
public override AttributeCollection Attributes
{
get
{
return new AttributeCollection(null);
}
}
public override bool CanResetValue(object component)
{
return true;
}
public override Type ComponentType
{
get
{
return _collection.GetType();
}
}
public override object GetValue(object component)
{
OnRefreshRequired();
return _collection[_index];
}
public override bool IsReadOnly
{
get { return false; }
}
public override string Name
{
get { return _index.ToString(); }
}
public override Type PropertyType
{
get { return _collection[_index].GetType(); }
}
public override void ResetValue(object component)
{
}
public override bool ShouldSerializeValue(object component)
{
return true;
}
public override void SetValue(object component, object value)
{
_collection[_index] = value;
}
protected virtual void OnRefreshRequired()
{
var handler = RefreshRequired;
if (handler != null) handler(this, EventArgs.Empty);
}
}
.
internal class ExpandableCollectionConverter : ExpandableObjectConverter
{
public override object ConvertTo(ITypeDescriptorContext context, CultureInfo culture, object value, Type destType)
{
if (destType == typeof(string))
{
return "(Collection)";
}
return base.ConvertTo(context, culture, value, destType);
}
public override PropertyDescriptorCollection GetProperties(ITypeDescriptorContext context, object value, Attribute[] attributes)
{
IList collection = value as IList;
PropertyDescriptorCollection pds = new PropertyDescriptorCollection(null);
for (int i = 0; i < collection.Count; i++)
{
ExpandableCollectionPropertyDescriptor pd = new ExpandableCollectionPropertyDescriptor(collection, i);
pd.RefreshRequired += (sender, args) =>
{
var notifyValueGivenParentMethod = context.GetType().GetMethod("NotifyValueGivenParent", BindingFlags.NonPublic | BindingFlags.Instance);
notifyValueGivenParentMethod.Invoke(context, new object[] {context.Instance, 1});
};
pds.Add(pd);
}
// return the property descriptor Collection
return pds;
}
}
And I use it for all ILists with the following line:
TypeDescriptor.AddAttributes(typeof (IList), new TypeConverterAttribute(typeof(ExpandableCollectionConverter)));
Some Clarifications
I want the grid to automatically update when I change the list. Refreshing when another property changes, does not help.
A solution that works, is a solution where:
If you expand the list while it is empty, and then add items, the grid is refreshed with the items expanded
If you add items to the list, expand it, and then remove items (without collapsing), the grid is refreshed with the items expanded, and not throwing ArgumentOutOfRangeException because it is trying to show items that were deleted already
I want this whole thing for a configuration utility. Only the PropertyGrid should change the collections
IMPORTANT EDIT:
I did manage to make the expanded collections update with Reflection, and calling NotifyValueGivenParent method on the context object when the PropertyDescriptor GetValue method is called (when RefreshRequired event is raised):
var notifyValueGivenParentMethod = context.GetType().GetMethod("NotifyValueGivenParent", BindingFlags.NonPublic | BindingFlags.Instance);
notifyValueGivenParentMethod.Invoke(context, new object[] {context.Instance, 1});
It works perfectly, except it causes the event to be raised infinite times, because calling NotifyValueGivenParent causes a reload of the PropertyDescriptor, and therfore, raising the event, and so on.
I tried to solve it by adding a simple flag that will prevent the reloading if it is already reloading, but for some reason NotifyValueGivenParent behaves asynchronously, and therefore the reloading happens after the flag is turned off.
Maybe it is another direction to explore. The only problem is the recursion
There is no need for using ObservableCollection. You can modify your descriptor class as follows:
public class ExpandableCollectionPropertyDescriptor : PropertyDescriptor
{
private IList collection;
private readonly int _index;
public ExpandableCollectionPropertyDescriptor(IList coll, int idx)
: base(GetDisplayName(coll, idx), null)
{
collection = coll;
_index = idx;
}
private static string GetDisplayName(IList list, int index)
{
return "[" + index + "] " + CSharpName(list[index].GetType());
}
private static string CSharpName(Type type)
{
var sb = new StringBuilder();
var name = type.Name;
if (!type.IsGenericType)
return name;
sb.Append(name.Substring(0, name.IndexOf('`')));
sb.Append("<");
sb.Append(string.Join(", ", type.GetGenericArguments()
.Select(CSharpName)));
sb.Append(">");
return sb.ToString();
}
public override bool CanResetValue(object component)
{
return true;
}
public override Type ComponentType
{
get { return this.collection.GetType(); }
}
public override object GetValue(object component)
{
return collection[_index];
}
public override bool IsReadOnly
{
get { return false; }
}
public override string Name
{
get { return _index.ToString(CultureInfo.InvariantCulture); }
}
public override Type PropertyType
{
get { return collection[_index].GetType(); }
}
public override void ResetValue(object component)
{
}
public override bool ShouldSerializeValue(object component)
{
return true;
}
public override void SetValue(object component, object value)
{
collection[_index] = value;
}
}
Instead of the ExpandableCollectionConverter I would derive the CollectionConverter class, so you can still use the ellipsis button to edit the collection in the old way (so you can add/remove items if the collection is not read-only):
public class ListConverter : CollectionConverter
{
public override bool GetPropertiesSupported(ITypeDescriptorContext context)
{
return true;
}
public override PropertyDescriptorCollection GetProperties(ITypeDescriptorContext context, object value, Attribute[] attributes)
{
IList list = value as IList;
if (list == null || list.Count == 0)
return base.GetProperties(context, value, attributes);
var items = new PropertyDescriptorCollection(null);
for (int i = 0; i < list.Count; i++)
{
object item = list[i];
items.Add(new ExpandableCollectionPropertyDescriptor(list, i));
}
return items;
}
}
And I would use this ListConverter on the properties where I want to see expandable list. Of course, you can register the type converter generally as you do in your example, but that overrides everything, which might not be overall intended.
public class MyClass
{
[TypeConverter(typeof(ListConverter))]
public List<int> List { get; set; }
public MyClass()
{
List = new List<int>();
}
[RefreshProperties(RefreshProperties.All)]
[Description("Change this property to regenerate the List")]
public int Count
{
get { return List.Count; }
set { List = Enumerable.Range(1, value).ToList(); }
}
}
Important: The RefreshProperties attribute should be defined for the properties that change other properties. In this example, changing the Count replaces the whole list.
Using it as propertyGrid1.SelectedObject = new MyClass(); produces the following result:
I don't want it to refresh when other property refreshes. I want it to refresh when the list is changed. I add items to the list, expand it, add more items, but the items are not updated
This is a typical misuse of PropertyGrid. It is for configuring a component, and not for reflecting the concurrent changes on-the-fly by an external source. Even wrapping the IList into an ObservableCollection will not help you because it is used only by your descriptor, while the external source manipulates directly the underlying IList instance.
What you can still do is an especially ugly hack:
public class ExpandableCollectionPropertyDescriptor : PropertyDescriptor
{
// Subscribe to this event from the form with the property grid
public static event EventHandler CollectionChanged;
// Tuple elements: The owner of the list, the list, the serialized content of the list
// The reference to the owner is a WeakReference because you cannot tell the
// PropertyDescriptor that you finished the editing and the collection
// should be removed from the list.
// Remark: The references here may survive the property grid's life
private static List<Tuple<WeakReference, IList, byte[]>> collections;
private static Timer timer;
public ExpandableCollectionPropertyDescriptor(ITypeDescriptorContext context, IList collection, ...)
{
AddReference(context.Instance, collection);
// ...
}
private static void AddReference(object owner, IList collection)
{
// TODO:
// - serialize the collection into a byte array (BinaryFormatter) and add it to the collections list
// - if this is the first element, initialize the timer
}
private static void Timer_Elapsed(object sender, ElapsedEventArgs e)
{
// TODO: Cycle through the collections elements
// - If WeakReference is not alive, remove the item from the list
// - Serialize the list again and compare the result to the last serialized content
// - If there a is difference:
// - Update the serialized content
// - Invoke the CollectionChanged event. The sender is the owner (WeakReference.Target).
}
}
Now you can use it like this:
public class Form1 : Form
{
MyObject myObject = new MyObject();
public MyForm()
{
InitializeComponent();
ExpandableCollectionPropertyDescriptor.CollectionChanged += CollectionChanged();
propertyGrid.SelectedObject = myObject;
}
private void CollectionChanged(object sender, EventArgs e)
{
if (sender == myObject)
propertyGrid.SelectedObject = myObject;
}
}
But honestly, I would not use it at all. It has serious flaws:
What if a collection element is changed by the PropertyGrid, but the timer has not updated the last external change yet?
The implementer of the IList must be serializable
Ridiculous performance overhead
Though using weak references may reduce memory leaks, it does not help if the objects to edit have longer life cycle than the editor form, because they will remain in the static collection
Putting it all together, this works:
Here is the class with the lists that we will put an instance of in our property grid. Also to demonstrate usage with a list of a complex object, I have the NameAgePair class.
public class SettingsStructure
{
public SettingsStructure()
{
//To programmatically add this to properties that implement ILIST for the naming of the edited node and child items:
//[TypeConverter(typeof(ListConverter))]
TypeDescriptor.AddAttributes(typeof(IList), new TypeConverterAttribute(typeof(ListConverter)));
//To programmatically add this to properties that implement ILIST for the refresh and expansion of the edited node
//[Editor(typeof(CollectionEditorBase), typeof(System.Drawing.Design.UITypeEditor))]
TypeDescriptor.AddAttributes(typeof(IList), new EditorAttribute(typeof(CollectionEditorBase), typeof(UITypeEditor)));
}
public List<string> ListOfStrings { get; set; } = new List<string>();
public List<string> AnotherListOfStrings { get; set; } = new List<string>();
public List<int> ListOfInts { get; set; } = new List<int>();
public List<NameAgePair> ListOfNameAgePairs { get; set; } = new List<NameAgePair>();
}
public class NameAgePair
{
public string Name { get; set; } = "";
public int Age { get; set; } = 0;
public override string ToString()
{
return $"{Name} ({Age})";
}
}
Here is the ListConverter class to handle making the child nodes.
public class ListConverter : CollectionConverter
{
public override bool GetPropertiesSupported(ITypeDescriptorContext context)
{
return true;
}
public override PropertyDescriptorCollection GetProperties(ITypeDescriptorContext context, object value, Attribute[] attributes)
{
IList list = value as IList;
if (list == null || list.Count == 0)
return base.GetProperties(context, value, attributes);
var items = new PropertyDescriptorCollection(null);
for (int i = 0; i < list.Count; i++)
{
object item = list[i];
items.Add(new ExpandableCollectionPropertyDescriptor(list, i));
}
return items;
}
public override object ConvertTo(ITypeDescriptorContext pContext, CultureInfo pCulture, object value, Type pDestinationType)
{
if (pDestinationType == typeof(string))
{
IList v = value as IList;
int iCount = (v == null) ? 0 : v.Count;
return $"({iCount} Items)";
}
return base.ConvertTo(pContext, pCulture, value, pDestinationType);
}
}
Here is the ExpandableCollectionPropertyDescriptor class for the individual items.
public class ExpandableCollectionPropertyDescriptor : PropertyDescriptor
{
private IList _Collection;
private readonly int _Index;
public ExpandableCollectionPropertyDescriptor(IList coll, int idx) : base(GetDisplayName(coll, idx), null)
{
_Collection = coll;
_Index = idx;
}
private static string GetDisplayName(IList list, int index)
{
return "[" + index + "] " + CSharpName(list[index].GetType());
}
private static string CSharpName(Type type)
{
var sb = new StringBuilder();
var name = type.Name;
if (!type.IsGenericType) return name;
sb.Append(name.Substring(0, name.IndexOf('`')));
sb.Append("<");
sb.Append(string.Join(", ", type.GetGenericArguments().Select(CSharpName)));
sb.Append(">");
return sb.ToString();
}
public override bool CanResetValue(object component)
{
return true;
}
public override Type ComponentType
{
get { return this._Collection.GetType(); }
}
public override object GetValue(object component)
{
return _Collection[_Index];
}
public override bool IsReadOnly
{
get { return false; }
}
public override string Name
{
get { return _Index.ToString(CultureInfo.InvariantCulture); }
}
public override Type PropertyType
{
get { return _Collection[_Index].GetType(); }
}
public override void ResetValue(object component)
{
}
public override bool ShouldSerializeValue(object component)
{
return true;
}
public override void SetValue(object component, object value)
{
_Collection[_Index] = value;
}
}
And then the CollectionEditorBase class for refreshing the property grid after the collection editor is closed.
public class CollectionEditorBase : CollectionEditor
{
protected PropertyGrid _PropertyGrid;
private bool _ExpandedBefore;
private int _CountBefore;
public CollectionEditorBase(Type type) : base(type) { }
public override object EditValue(ITypeDescriptorContext context, IServiceProvider provider, object value)
{
//Record entry state of property grid item
GridItem giThis = (GridItem)provider;
_ExpandedBefore = giThis.Expanded;
_CountBefore = (giThis.Value as IList).Count;
//Get the grid so later we can refresh it on close of editor
PropertyInfo piOwnerGrid = provider.GetType().GetProperty("OwnerGrid", BindingFlags.Instance | BindingFlags.NonPublic | BindingFlags.Public);
_PropertyGrid = (PropertyGrid)piOwnerGrid.GetValue(provider);
//Edit the collection
return base.EditValue(context, provider, value);
}
protected override CollectionForm CreateCollectionForm()
{
CollectionForm cf = base.CreateCollectionForm();
cf.FormClosing += delegate (object sender, FormClosingEventArgs e)
{
_PropertyGrid.Refresh();
//Because nothing changes which grid item is the selected one, expand as desired
if (_ExpandedBefore || _CountBefore == 0) _PropertyGrid.SelectedGridItem.Expanded = true;
};
return cf;
}
protected override object CreateInstance(Type itemType)
{
//Fixes the "Constructor on type 'System.String' not found." when it is an empty list of strings
if (itemType == typeof(string)) return string.Empty;
else return Activator.CreateInstance(itemType);
}
}
Now the usage produces:
And performing various operations produces:
You can tweak it to operate like you like.