I am really struggling to find a WPF validation pattern that is easily maintainable. I have been doing some web work, and I was really impressed at how easy validation is in AngularJS. So I may have unrealistic expectations of what WPF will do. The solution I have now seems like it has a bunch of garbage. Here is the XAML I have right now. I don't like having to create the DataResource, and it seems far too verbose. Any suggestions?
<local:DataResource x:Key="RequireFcpaGovernmentRelationsText" BindingTarget="{Binding Vendor.RequireFcpaGovernmentRelationsText}" />
<TextBox MaxLength="50" Width="350" HorizontalAlignment="Left">
<Binding Path="Vendor.FcpaGovernmentRelationsText" UpdateSourceTrigger="PropertyChanged">
<Binding.ValidationRules>
<local:RequiredFieldRule ValidationStep="UpdatedValue" RequireIf="{local:DataResourceBinding DataResource={StaticResource RequireFcpaGovernmentRelationsText}}" RequiredMessage="Please specify government relations."/>
</Binding.ValidationRules>
</Binding>
</TextBox>
Unfortunately, WPF is very verbose by nature. It isn't really fair to compare it to a web technology. The DataResource is needed because of the way that WPF manages the DataContext. We have started doing more and more web project because WPF just doesn't keep up.
This question comes up regularly, and things also change in the .NET world, so let's come up with a current solution that, unlike many snippets here and there, works and is easily maintanable. It will use INotifyDataErrorInfo that's appeared in .NET 4.5.
Make sure you have a base class that you inherit all your data classes from. Make this base class implement INotifyDataErrorInfo:
public abstract class ObjectBase<T> : INotifyDataErrorInfo, INotifyPropertyChanged {
public event PropertyChangedEventHandler PropertyChanged;
public event EventHandler<DataErrorsChangedEventArgs> ErrorsChanged;
private readonly Dictionary<string, List<string>> StoredErrors = new();
protected void OnPropertyChanged(string propertyName) {
PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
if (!string.IsNullOrEmpty(propertyName))
Validate(propertyName);
}
Using INotifyPropertyChanged is not strictly required but you probably want to use that, anyway, and it's easy to combine both.
protected void OnErrorsChanged(string propertyName) =>
ErrorsChanged?.Invoke(this, new DataErrorsChangedEventArgs(propertyName));
public bool HasErrors => StoredErrors.Any();
public IEnumerable GetErrors(string propertyName) => StoredErrors.GetValueOrDefault(propertyName);
protected void AddError(string propertyName, string error) {
if (!StoredErrors.ContainsKey(propertyName))
StoredErrors[propertyName] = new List<string>();
if (!StoredErrors[propertyName].Contains(error)) {
StoredErrors[propertyName].Add(error);
OnErrorsChanged(propertyName);
}
}
protected void ClearErrors(string propertyName) {
if (StoredErrors.ContainsKey(propertyName)) {
StoredErrors.Remove(propertyName);
OnErrorsChanged(propertyName);
}
}
You have two approaches for the actual validation logic, you can select either or even combine them. You can add code to your properties that check for your validation conditions and set the errors if those conditions are not met. Or, you can use attributes on your properties to make it automatic. Only add these extra functions if you plan to use the second approach as well:
public void ValidateAll() {
foreach (var prop in GetType().GetProperties(BindingFlags.Public | BindingFlags.Instance | BindingFlags.Static))
Validate(prop.Name);
}
protected void Validate(string propertyName) {
ClearErrors(propertyName);
var prop = GetType().GetProperty(propertyName, BindingFlags.Public | BindingFlags.Instance | BindingFlags.Static);
var validationAttributes = prop?.GetCustomAttributes(typeof(ValidationAttribute)) ?? new List<Attribute>();
if (validationAttributes.Any()) {
var validationResults = new List<ValidationResult>();
object value = prop.GetValue(this);
if (!Validator.TryValidateProperty(value, new ValidationContext(this, null, null) { MemberName = propertyName }, validationResults)) {
foreach (var attributeValidationResult in validationResults)
AddError(propertyName, attributeValidationResult.ErrorMessage);
}
}
}
That's a bit of a boilerplate all right, but it only needs to go into your base class and the actual data classes will use it very simply:
public class Person : ObjectBase<Person> {
private string name;
[Required(ErrorMessage = "Name is required")]
public string Name {
get => name;
set {
name = value;
OnPropertyChanged(nameof(Name));
}
}
private string address;
public string Address {
get => address;
set {
address = value;
OnPropertyChanged(nameof(Address));
ClearErrors(nameof(Address));
if (address.IsEmpty()) //your own functions
AddError(nameof(Address), "Address is required");
if (address.IsInvalid()) //your own functions
AddError(nameof(Address), "Invalid address");
}
}
You can observe the two approaches here. Either you use attributes like Required, StringLength, etc (WPF has quite a few of them) or you can create your own by inheriting ValidationAttribute and providing an IsValid() override. Or you can use simple conditional logic to set errors. Or you can mix those approaches, both in the same data class, or even in the same property -- whichever you find easier to use and more maintainable.
This is already mostly automatic, but using fields in dialogs might require a little bit more plumbing. Use a CanExecute callback with your OK button command handler:
<CommandBinding Command="{StaticResource DoOk}" Executed="Ok" CanExecute="CanOk" />
private void CanOk(object sender, CanExecuteRoutedEventArgs e) =>
e.CanExecute = !YourData?.HasErrors ?? false;
private void Ok(object sender, ExecutedRoutedEventArgs e) {
YourData.ValidateAll();
if (!YourData.HasErrors) {
DialogResult = true;
}
}
This only allows OK to be enabled if there are no validation errors, and makes sure that the dialog can't be finished even if it's enabled but there are errors (which is usually the case right after opening the dialog but before actually editing the fields).
Related
So I looked at this link before:
http://blogs.msdn.com/b/delay/archive/2009/10/26/creating-something-from-nothing-developer-friendly-virtual-file-implementation-for-net.aspx
The class works flawlessly for Files, but it doesn't support directory's etc, does anyone have any idea how I can change the class to support it, I"m no pinvoke whiz. I've tried a million different things, overriding some code to do File Copy and Directory creation of my drop source into the TEMP directory and attempting to trigger a FileDrop, but this locks up the app entirely.
This leads me to believe that there must be a better way to enable directory structure creation as well.
The main part of the Drag and Drop operation is the DragDrop.DoDragDrop method. From the DragDrop.DoDragDrop Method page on MSDN:
public static DragDropEffects DoDragDrop(
DependencyObject dragSource,
Object data,
DragDropEffects allowedEffects
)
Of particular interest is the data parameter:
A data object that contains the data being dragged.
Notice how this parameter is of type Object, so it's completely up to you as to what object you use in the operation. Now I'm not sure what code you found from the page that you linked to, but if I were trying to drag and drop files and folders, I wouldn't need special classes to do it for me.
The simplest way to do that is to just pass the file and/or folder paths instead of the actual data. The control that the data is dropped on can access the data using the file paths just as easily as the drag source. You should be able to locate the DragDrop.DoDragDrop method from your code and easily adapt that code.
If you want to do Drag and Drop operations in the correct way, then I'd recommend that you take a look at the Drag and Drop Overview page on MSDN. It fully explains what to do and provides several code examples.
To implement drag and drop in MVVM without much experience in WPF, you can refer to 5 steps; I will outline these...
Step 1: Attached Behaviours
Add a new class to your project called Behaviours. It should look like this...
public static class Behaviours
{
#region DandBehaviour
public static readonly DependencyProperty DandBehaviourProperty =
DependencyProperty.RegisterAttached("DandBehaviour", typeof(ICommand), typeof(Behaviours),
new FrameworkPropertyMetadata(null,
FrameworkPropertyMetadataOptions.None,
OnDandBehaviourChanged));
public static ICommand GetDandBehaviour(DependencyObject d)
{
return (ICommand)d.GetValue(DandBehaviourProperty);
}
public static void SetDandBehaviour(DependencyObject d, ICommand value)
{
d.SetValue(DandBehaviourProperty, value);
}
private static void OnDandBehaviourChanged(DependencyObject d, DependencyPropertyChangedEventArgs e)
{
Grid g = d as Grid;
if (g != null)
{
g.Drop += (s, a) =>
{
ICommand iCommand = GetDandBehaviour(d);
if (iCommand != null)
{
if (iCommand.CanExecute(a.Data))
{
iCommand.Execute(a.Data);
}
}
};
}
else
{
throw new ApplicationException("Non grid");
}
}
#endregion
}
This class implements an attached dependency property which can be reached from both your Xaml and your View Model. It hooks the "Drop" event and invokes a command on it.
Step 2: Instrument the Xaml
In this step you need to add the name space to the Xaml so that it can find the behaviours class in Step 1. It looks something like this...
xmlns:b="clr-namespace:DdMvvm"
This statement assigns the alias 'b' to the behaviours. Then you tell the WPF root window to accept drops...
AllowDrop="true"
Then you can add the behaviour to your logical tree thusly...
<Grid AllowDrop="True" b:Behaviours.DandBehaviour="{Binding DandCommand}">
<DockPanel Background="Bisque" AllowDrop="True"/>
</Grid>
Step 3: Add command support
Download Josh Smith's 'Relay Command' from http://msdn.microsoft.com/en-us/magazine/dd419663.aspx For completeness purposes, it is given here as...
public class RelayCommand : ICommand
{ //http://msdn.microsoft.com/en-us/magazine/dd419663.aspx
public RelayCommand(Action<object> execute, Predicate<object> canExecute)
{
_execute = execute;
_canExecute = canExecute;
}
public bool CanExecute(object parameter)
{
return _canExecute(parameter);
}
public event EventHandler CanExecuteChanged
{
add { CommandManager.RequerySuggested += value; }
remove { CommandManager.RequerySuggested -= value; }
}
public void Execute(object parameter)
{
_execute(parameter);
}
private readonly Action<object> _execute;
private readonly Predicate<object> _canExecute;
}
Step 4: Write the View Model
A View Model looks like this...
public class ViewModel : INotifyPropertyChanged
{
public ICommand DandCommand { get; set; }
public ViewModel()
{
DandCommand = new RelayCommand(ExecuteDandCommand, CanExecuteDandCommand);
}
private void ExecuteDandCommand(object obj)
{
if (obj != null)
{
IDataObject ido = obj as IDataObject;
if (ido != null)
{
var fileDrop = ido.GetData(DataFormats.FileDrop, true);
var filesOrDirectories = fileDrop as String[];
if (filesOrDirectories != null && filesOrDirectories.Length > 0)
{
foreach (string fullPath in filesOrDirectories)
{
if (Directory.Exists(fullPath))
{
Console.WriteLine(#"{0} is a directory", fullPath);
}
else if (File.Exists(fullPath))
{
Console.WriteLine(#"{0} is a file", fullPath);
}
else
{
Console.WriteLine(#"{0} is not a file and not a directory", fullPath);
}
}
}
}
}
}
private bool CanExecuteDandCommand(object obj)
{
return true;
}
#region INotifyPropertyChanged Implementation
public event PropertyChangedEventHandler PropertyChanged;
protected virtual void OnPropertyChanged(string name)
{
var handler = System.Threading.Interlocked.CompareExchange(ref PropertyChanged, null, null);
if (handler != null)
{
handler(this, new PropertyChangedEventArgs(name));
}
}
#endregion
}
This VM implements a command (called DandCommand) which will be fired by the attached behaviour (remember Step 1?). The 'juicy part' of the VM is the bit which dereferences the drag-and-drop payload. In this particular VM, the code dereferences the data and finds out if it is a file or a directory. It then prints a diagnostic to the console. You can change this part to load images, or internet links, or what-ever can be dropped.
Step 5: Wiring the data context
This is done by different developers in different ways (for industrial apps, lots of people like to use Prism and Unity, but that's ott for a simple how-to like this post). The most straight-forward approach is to change your View to look like this...
public MainWindow()
{
InitializeComponent();
DataContext = new ViewModel();
}
This code sets the window's data context to your VM so that the binding can take place.
Those steps give you a good starting point for MVVM and drag-and-drop and getting the whole thing to work. The big payback in using MVVM for these things is the clean separation and compartmentalisation that you get. I.e., the VM can be unit-tested without having to instantiate the WPF binding engine.
iv'e got a CompositeCommand exposed globally in my startup project
public static class Commands
{
public static readonly CompositeCommand DiceRolledCommand = new CompositeCommand();
}
in a ControlLibrary referenced by my startup project iv'e got a Control which has a DelegateCommand ,
each instance of this Control has to register it's Command with the globally exposed DiceRolledCommand.
what wold be the best practice of doing so :
here are 3 idea's of which the first 2 i don't like because they are a kinda of hack , where you take some programming component (dp) and alter it's use for your benefit , resulting in poor code and design .
1)
a regular decadency property of type CompositeCommand which will be set with DiceRolledCommand
and on it's CallBack register MyControl's DelegateCommand (OnDiceRolledCommand) .
public class MyControl : Control
{
public DelegateCommand<Tuple<int, int>> OnDiceRolledCommand { get; private set; }
public CompositeCommand GlobalDiceRolledCommand
{
get { return (CompositeCommand)GetValue(GlobalDiceRolledCommandProperty); }
set { SetValue(GlobalDiceRolledCommandProperty, value); }
}
public static readonly DependencyProperty GlobalDiceRolledCommandProperty =
DependencyProperty.Register("GlobalDiceRolledCommand", typeof(CompositeCommand), typeof(MyControl), new UIPropertyMetadata(null,GlobalDiceRolledCommandPropertyChanged));
private static void GlobalDiceRolledCommandPropertyChanged(DependencyObject d, DependencyPropertyChangedEventArgs e)
{
var myControl= d as MyControl ;
var compoisteCommand = e.NewValue as CompositeCommand;
compoisteCommand.RegisterCommand(myControl.OnDiceRolledCommand);
}
}
<local:MyControl GlobalDiceRolledCommand="{x:Static local:Commands.DiceRolledCommand}"/>
i don't like this approach since it's a kind of manipulation where a Dependency Property is used has a Complex logical setter .
2) i could also do the same as in (1) using a third party class with an attached property which will register the OnDiceRolledCommand in an attached property's CallBack
public static class Commands
{
public static readonly CompositeCommand DiceRolledCommand = new CompositeCommand();
public static ICommand GetRegisterToDiceRolledCommand(DependencyObject obj)
{
return (ICommand)obj.GetValue(RegisterToDiceRolledCommandProperty);
}
public static void SetRegisterToDiceRolledCommand(DependencyObject obj, ICommand value)
{
obj.SetValue(RegisterToDiceRolledCommandProperty, value);
}
public static readonly DependencyProperty RegisterToDiceRolledCommandProperty =
DependencyProperty.RegisterAttached("RegisterToDiceRolledCommand", typeof(ICommand), typeof(Commands), new UIPropertyMetadata(null,OnRegisterToDiceRolledCommandProperty);
private static void OnRegisterToDiceRolledCommandProperty(DependencyObject d , DependencyPropertyChangedEventArgs e)
{
var commandToRegister = e.newValue as DelegateCommand;
DiceRolledCommand.RegisterCommand(commandToRegister );
}
}
<local:MyContorl local:Commands.RegisterToDiceRolledCommand="{Binding OnDiceRolledCommand , RelativeSource={RelativeSource Self}}"/>
i also don't like this approach for the same reason as 1 ..
3) passing the composite command as a parameter to constructor , this approach is better since it keeps
the initializing logic in the constructor where it should be , i just can't figure out how to pass
an argument to a contractor through XAML , i'm not sure if it's even possible .
public class MyControl : Control
{
public MyControl(CompositeCommand globalDiceRolledCommand)
{
.........
globalDiceRolledCommand.Register(OnDiceRolledCommand);
}
}
<local:MyControl ..... >
Some how pass parameters to contractor in order to create the element in XAML
</local:MyControl>
to summarize :
A) any thoughts about (1) and (2) .
B) thoughts of how to accomplish 3 , and if it seems like good design .
C) Any good pattern of accomplishing this scenario.
thanks in advance .
Whenever I use Global Commands like that they are usually defined in either an Infrastructure class library which every library can reference. Or they are defined in a consuming core library that each module could reference directly.
I wrote a lot of this up in a Code Project article
Part 2 here
I just tried to subscribe to WPF property change events using C++/CLI. I didn't expect this to get difficult.
First I tried to subscribe to a specific property of some window (IsMouseDirectlyOver) and finally succeeded with following code:
void MyClass::DependencyPropertyChanged(Object^ sender, DependencyPropertyChangedEventArgs args)
{
Debug::WriteLine("DependencyPropertyChanged: "+sender->ToString()+", "+args.Property->Name);
}
window->IsMouseDirectlyOverChanged += gcnew DependencyPropertyChangedEventHandler(this, &MyClass::DependencyPropertyChanged);
Then I tried to subscribe to any property changes of an object (which is most important to me because my final code must be able to handle property changes by property names). I totally failed on this.
I tried various things but nothing worked. I could not find any C++/CLI examples but according to documentation and C# examples the following seemed to be the most sensible code to me:
window->PropertyChanged += gcnew PropertyChangedEventHandler(this, &MyClass::PropertyChanged);
void MyClass::PropertyChanged(Object^ sender, PropertyChangedEventArgs^ args)
{
...
}
But the compiler tells me by error C2039 that 'PropertyChangedEvent' is no element of 'System::Windows::Window'.
How can I achieve what I want?
Al mentioned in the comments, your code doesn't work, because there is no PropertyChanged event on Window, it's as simple as that.
What you can do instead is to override the OnPropertyChanged() method, which is present on a Window. In your override, you can do anything you want, including raising PropertyChanged (don't forget to create that event first).
I had a look on the snoop sources. I modified it and wrote a very, very basic example that works:
String^ ownerPropertyName = "IsActive";
DependencyObject^ propertyOwner = window;
DependencyPropertyDescriptor^ ownerPropertyDescriptor = DependencyPropertyDescriptor::FromName(ownerPropertyName, propertyOwner->GetType(), propertyOwner->GetType());
DependencyProperty^ ownerProperty = ownerPropertyDescriptor->DependencyProperty;
Type^ ownerPropertyType = ownerProperty->PropertyType;
DependencyProperty^ myProperty = DependencyProperty::Register(ownerPropertyName, ownerPropertyType, GetType(), gcnew PropertyMetadata(gcnew PropertyChangedCallback(&MyClass::BoundPropertyChangedCallback)));
Binding^ myPropertyToOwnerPropertyBinding = gcnew Binding(ownerPropertyName);
myPropertyToOwnerPropertyBinding->Mode = BindingMode::OneWay;
myPropertyToOwnerPropertyBinding->Source = propertyOwner;
BindingOperations::SetBinding(this, myProperty, myPropertyToOwnerPropertyBinding);
And:
static void BoundPropertyChangedCallback(DependencyObject^ me, DependencyPropertyChangedEventArgs args)
{
Debug::WriteLine("BoundPropertyChangedCallback: "+args.OldValue+", "+args.NewValue+", "+args.Property->Name);
}
Looks pretty complicated to me. I have no idea if that binding stuff is really necessary. In fact this can even subscribe to properties that do not have events (like IsMouseOver) and can operate on objects that do not implement INotifyPropertyChanged (like Window). And it does not need any switch/case for properties.
The class PropertyDescriptor (or the derived DependencyPropertyDescriptor) provides a mechanism to add a property change handler by their AddValueChanged method:
DependencyPropertyDescriptor^ propertyDescriptor = DependencyPropertyDescriptor::FromName(
"ActualWidth", component->GetType(), component->GetType());
propertyDescriptor->AddValueChanged(component, gcnew EventHandler(ActualWidthChanged));
...
static void ActualWidthChanged(Object^ component, EventArgs^ e)
{
...
}
Unfortunately the handler doesn't get passed the changed property, so i guess you would have to add different handlers for all properties you want to monitor.
EDIT: You might implement something like the code shown below that uses an anonymous delegate to pass the property name to an appropriate handler. Note however that this is C#, and to my understanding this can't be done in C++/CLI, since there it does not support managed lambdas. Mayby you could wrap a helper class like this in a separate assembly and use it from your C++/CLI code.
public delegate void PropertyChangedHandler(object component, string propertyName);
public static class DependencyPropertyDescriptorExt
{
public static void AddPropertyChangedHandler(
this object component, string propertyName, PropertyChangedHandler handler)
{
var propertyDescriptor = DependencyPropertyDescriptor.FromName(
propertyName, component.GetType(), component.GetType());
propertyDescriptor.AddValueChanged(component, (o, e) => handler(o, propertyName));
}
}
Now you could write and use such a PropertyChangedHandler like this:
this.AddPropertyChangedHandler("ActualHeight", PropertyChanged);
...
private void PropertyChanged(object component, string propertyName)
{
...
}
I've been messing around with Expressions - and I may have gone beyond my capabilities - but here goes... I've implemented 'type-safe' INotifyPropertyChanged implementation (an example is here), but gone a bit farther and included changetracking:
public abstract BaseViewModel<TEntity>:INotifyPropertyChanged
{
private readonly IBaseChangeTracker<TEntity> _changeTracker;
protected void OnPropertyChanged<T>(Expression<Func<T>> property, T value)
{
_changeTracker.AddChange(property, value);
OnPropertyChanged(property);
}
protected virtual void OnPropertyChanged<TProperty>(Expression<Func<TProperty>> property)
{
if (PropertyChanged == null) return;
PropertyChanged(this, new PropertyChangedEventArgs(property.GetMemberInfo().Name));
}
public event PropertyChangedEventHandler PropertyChanged;
}
public abstract class BaseChangeTracker<TEntity>:IBaseChangeTracker<TEntity>
{
private readonly IDictionary<Expression, object> _changes = new Dictionary<Expression, object>();
public void AddChange<T>(Expression<Func<T>> expression, T newValue)
{
_changes.Add(expression, newValue);
}
public void ApplyChanges(TEntity entity)
{
foreach (var change in _changes)
{
var property = typeof(TEntity).GetProperty(change.Key.GetMemberInfo().Name);
property.SetValue(entity, change.Value, null);
}
}
public virtual void CopyCurrentState(TEntity entity)
{
_changes.Clear();
}
public virtual void ResetEntity(TEntity entity)
{
_changes.Clear();
}
public bool HasUnsavedChanges
{
get { return _changes.Any(); }
}
}
This may seem a bit excessive - each entity will have it it's own ChangeTracker keeping the original state of the entity when loaded and can reset it back to these, but the idea is that if there is a concurrency conflict when it tries to save the updated entity, I reload the entity from the database and run it through .ApplyChanges and try to save it again. This will remove about 95% of my concurrency problems... if it works. My tests show that for limited entities it works, but that is with simple property-changes.
Known issue:
I have yet to find an elegant way of handling collections.
What else am I missing - or are there obvious flaws in my design?
I realize that this is not a direct answer to your question, but you might consider using the Command Pattern to implement an undo/redo stack.
Encapsulating changes in commands is a very tidy way to cycle/re-cycle through changes, with the added benefits of (1) a nice feature that adds value to the application, (2) you can wrap many actions in any given command, like raising event change notifications for databinding support in both the do and undo directions.
Additionally, managing collection changes is no more or less challenging than simple property updates.
Specific to the code you posted, the OnPropertyChanged implementation will never raise the PropertyChanged event because you call return after the if() statement and then again in bare brackets (these do not correspond to the if condition).
if (PropertyChanged == null) return; // this returns based on if
{
return; // this returns no matter what
}
Additionally, it seems that the user won't ever see any changes in the UI. The values aren't updated until ApplyChanges is called, and when it is there is no PropertyChanged event. (I might not be following your code correctly, but just looking over it this seems to be the case).
In MVVM it is normal to connect View to the ViewModel with data binding.
Therefore if the name of a properties changes on one of the Model objects that is databound to there is no compiler error.
When the compiler will not stop a bug, the next thing I think of is “UnitTest”, However
How do you unit test this without
spending forever writing a GUI test?
Is there a system that will check that all the properties that are bound to is valid, (without having to run the UI) that I can call in a unit test?
I am looking for something that will take the view, and then loop over all WPF controls, for each WPF control it will look at all the binding and check if they are valid.
By the way there have been a few good questions about how to make OnPropertyChanged safe, and/or how to test it (But done of these get down to the level of a WPF view.)
How to make Databinding type safe and support refactoring
Automatically INotifyPropertyChanged
workarounds for nameof() operator in C#: typesafe databinding
A Fluent Interface For Testing INotifyPropertyChanged
Automatic Class Tester will test all simple proptites and INotifyPropertyChanged
I have put a bounty on this question, as someone must have thought hard about the problem and come up with soltions.
I think I've come up with something that may work using simple reflection, and adapting some code I've used in the past (the code for the FindBindingsRecursively method was written by Martin Bennedik's as part of his Enterprise WPF Validation Control):
[TestMethod]
public void CheckWpfBindingsAreValid()
{
// instansiate the xaml view and set DataContext
var yourView = new YourView();
yourView.DataContext = YourViewModel;
FindBindingsRecursively(yourView,
delegate(FrameworkElement element, Binding binding, DependencyProperty dp)
{
var type = yourView.DataContext.GetType();
// check that each part of binding valid via reflection
foreach (string prop in binding.Path.Path.Split('.'))
{
PropertyInfo info = type.GetProperty(prop);
Assert.IsNotNull(info);
type = info.PropertyType;
}
});
}
private delegate void FoundBindingCallbackDelegate(FrameworkElement element, Binding binding, DependencyProperty dp);
private void FindBindingsRecursively(DependencyObject element, FoundBindingCallbackDelegate callbackDelegate)
{
// See if we should display the errors on this element
MemberInfo[] members = element.GetType().GetMembers(BindingFlags.Static |
BindingFlags.Public |
BindingFlags.FlattenHierarchy);
foreach (MemberInfo member in members)
{
DependencyProperty dp = null;
// Check to see if the field or property we were given is a dependency property
if (member.MemberType == MemberTypes.Field)
{
FieldInfo field = (FieldInfo)member;
if (typeof(DependencyProperty).IsAssignableFrom(field.FieldType))
{
dp = (DependencyProperty)field.GetValue(element);
}
}
else if (member.MemberType == MemberTypes.Property)
{
PropertyInfo prop = (PropertyInfo)member;
if (typeof(DependencyProperty).IsAssignableFrom(prop.PropertyType))
{
dp = (DependencyProperty)prop.GetValue(element, null);
}
}
if (dp != null)
{
// Awesome, we have a dependency property. does it have a binding? If yes, is it bound to the property we're interested in?
Binding bb = BindingOperations.GetBinding(element, dp);
if (bb != null)
{
// This element has a DependencyProperty that we know of that is bound to the property we're interested in.
// Now we just tell the callback and the caller will handle it.
if (element is FrameworkElement)
{
callbackDelegate((FrameworkElement)element, bb, dp);
}
}
}
}
// Now, recurse through any child elements
if (element is FrameworkElement || element is FrameworkContentElement)
{
foreach (object childElement in LogicalTreeHelper.GetChildren(element))
{
if (childElement is DependencyObject)
{
FindBindingsRecursively((DependencyObject)childElement, callbackDelegate);
}
}
}
}
Really good question. Voted it up. I would like to know the answer too.
One of the best practices I know (suggested by Josh Smith, thanks Gishu for pointing to this) is having base view model class to check in the OnPropertyChanged() method whether property really exists. E.g.:
abstract class ViewModelBase
{
[Conditional("DEBUG")]
public void VerifyPropertyName(string propertyName)
{
// Verify that the property name matches a real,
// public, instance property on this object.
if (TypeDescriptor.GetProperties(this)[propertyName] == null)
{
if (this.ThrowOnInvalidPropertyName)
throw new ArgumentException(propertyName);
string msg = "Invalid property name: " + propertyName;
Debug.Fail(msg);
}
}
protected void OnPropertyChanged(string propertyName)
{
VerifyPropertyName(propertyName);
PropertyChangedEventHandler handler = this.PropertyChanged;
if (handler != null)
{
var e = new PropertyChangedEventArgs(propertyName);
handler(this, e);
}
}
}
But this wouldn't help you to find spelling problems in XAML. Hmm... I don't know any existing solution for this. Maybe guys from WPF Disciples could suggest something. I think I would go with PresentationTraceSources.DataBindingSource and add to his Listners collection instance of TextWriterTraceListener and then monitor output. As soon as we get an error or warning on our radar we should fail a test.
Found good example: WPF Snippet - Detecting Binding Errors
Hope this helps. At least a bit :).
Cheers, Anvaka.
I know that this is not the direct answer to your question.
If you know the name of the control element, that you expect to be bound against you can do something like the test below (using nunit). This is the rough version. But here you use expressions and explicitly test that the property is in a binding
[Test]
public void TestBindings()
{
TestBinding<IndividualSolutionViewModel, string>(x => x.Name, "Name", TextBlock.TextProperty);
}
private void TestBinding<TViewModel,TResult>(Expression<Func<TViewModel, TResult>> property, string elementName,
DependencyProperty dependencyProperty)
{
string memberName = ExpressionHelper.GetPropertyName(property); // f.ex v => v.Name will return Name
TestBinding(memberName, elementName, dependencyProperty);
}
private void TestBinding(string memberName, string elementInControlName, DependencyProperty dependencyProperty)
{
//object viewModel = new IndividualSolutionViewModel();
var view = new IndividualSolutionView();
//Assert.That(view.DataContext, Is.EqualTo(viewModel));
var element = view.FindName(elementInControlName);
Assert.That(element, Is.Not.Null, string.Format("Unable to find the element {0} in view {1}", elementInControlName, view.Name));
Assert.That(element, Is.InstanceOf(typeof(DependencyObject)));
var binding = BindingOperations.GetBinding(element as DependencyObject, dependencyProperty);
Assert.That(binding, Is.Not.Null, string.Format("Could not find a binding for the control {0}", elementInControlName));
Assert.That(binding.Path.Path, Is.EqualTo(memberName));
}
Ps. You have to add this to the app.config
<configSections>
<sectionGroup name="NUnit">
<section type="System.Configuration.NameValueSectionHandler"
name="TestRunner"></section>
</sectionGroup>
</configSections>
<NUnit>
<TestRunner>
<add value="STA" key="ApartmentState"></add>
</TestRunner>
</NUnit>
There is also this possibility, that might give you some ideas. THe gist of the idea is property names that you would be binding to are exposed as static string properties. If a binding property name changed you would get a compilation error.
I have not had the opportunity to actually test this technique myself - but it does look interesting:
http://www.codeproject.com/Articles/42036/Project-Metadata-Generation-using-T4
As Anvaka points out, using a base class for your view model that checks property names can help avoid this particular problem (though it won't tell you when your VM class does its own property-change notification and ignores the method in the base class, not that I've ever seen anything like that happen in my code).
And you can (and should) instrument your code so that things that aren't working fail in a way that's visible to you. The thing that's kind of paradoxical about this is that if you know what things may fail and you watch them, they won't, because the fact that you're watching them will keep you from making the mistakes that lead them to fail (like writing a template selector that doesn't always return a template).
But fundamentally, the view is the UI, so I would be pretty surprised to find methods of testing it that weren't also methods for testing the UI.