Below, I have code for an object that I would like to test. It is in an assembly called Business and I have added the attributes in the AssemblyInfo.cs to make internals visible to the test and rhino mocks which are located in another assembly. When testing the GenerateReport method, I can not fake out the call to ValidateWorkingDirectory when it is "internal" (System.ApplicationException : Must set Working Directory before any method calls.). If I make ValidateWorkingDirectory public, the problem goes away. I thought InternalsVisibleTo would address this issue.
public class MyClass : IMyClass
{
private readonly IMyClassDataProvider _myClassDataProvider;
public virtual string WorkingDirectory { get; set; }
public MyClass(IMyClassDataProvider myClassDataProvider)
{
_myClassDataProvider = myClassDataProvider;
}
internal virtual void ValidateWorkingDirectory()
{
if (string.IsNullOrEmpty(WorkingDirectory))
{
throw new ApplicationException("Must set Working Directory before any method calls.");
}
}
public virtual void GenerateReport(vars)
{
ValidateWorkingDirectory();
InsertData(_myClassDataProvider.GetData(vars), "ReportName");
}
internal virtual void InsertData(DataSet analysis, string fileName)
{
DoSomeStuff();
}
private static void DoSomeStuff()
{
//Whatevs
}
}
//In AssmeblyInfo.cs
[assembly: InternalsVisibleTo("UnitTests.Business")]
[assembly: InternalsVisibleTo("Rhino.Mocks")]
[TestFixture]
public class MyClassTests : TestFixtureBase
{
private MockRepository _mocks;
private IMyClassDataProvider _myClassDataProvider;
private MyClass _myClass;
private var _vars;
[SetUp]
protected void Init()
{
_mocks = new MockRepository();
_myClassDataProvider = _mocks.StrictMock<IMyClassDataProvider >();
_myClass = _mocks.PartialMock<MyClass>(_myClassDataProvider);
_vars = "who cares";
}
[Test]
[ExpectedException(typeof(ApplicationException), ExpectedMessage = "Must set Working Directory before any method calls.")]
public virtual void ShouldThrowAnExceptionIfWorkingDirectoryNotSet()
{
Expect.Call(_myClass.WorkingDirectory).Return(Random.Get<bool>() ? null : string.Empty);
_mocks.ReplayAll();
_myClass.ValidateWorkingDirectory();
_mocks.VerifyAll();
}
[Test]
public virtual void ShouldGenerateReport()
{
DataSet dataSetToReturn = new DataSet();
using (_mocks.Ordered())
{
Expect.Call(() => _myClass.ValidateWorkingDirectory());
Expect.Call(_myClassDataProvider.GetData(vars)).Return(dataSetToReturn);
_myClass.InsertData(dataSetToReturn, "ReportName");
}
_mocks.ReplayAll();
_myClass.GenerateReport(vars);
_mocks.VerifyAll();
}
}
You need to expose your internal members to proxy assembly, not Rhino's assembly itself:
[assembly: InternalsVisibleTo ("DynamicProxyGenAssembly2")]
When a class is mocked, a new class is generated at run-time which is derived from the mocked class. This generated class resides in a separate "temporary" assembly which is called "DynamicProxyGenAssembly2". So, the InternalsVisibleTo attribute needs to be set on the target assembly to allow access to its internal members from the temporary assembly.
This happens to be common misunderstanding, for detailed information on how to use internals visible with Rhino, check this documentation page.
Related
I am using XUnit and need to perform some action before running a test suit. so, I try to use IClassFixture feature of XUnit. but I cannot find a way to inject dependencies into the Fixture class. my code structure is such as below:
public class MyFixture
{
IDependency _dep;
public MyFixture(IDependency dep)
{
_dep = dep;
}
void DoSomeJob()
{
//// some code there
dep.DoSome();
}
}
And this is my test class code:
public class MyTest : IClassFixture<MyFixture>
{
[Fact]
public void test_my_code()
{
////simply just test the code
}
}
but when I run the test I am getting the exception
Xunit.Sdk.TestClassException Class fixture type 'MyFixture' had one or more unresolved constructor
Your Fixture class depends on IDependency dep, which has not been configured. You could use the Fixture class to setup a service provider; However it is not the best solution, as you have to end up using service locator patter such as
serviceProvider.GetRequiredService<T>()
Suggest to use xunit.di, it is an extension built into xunit framework to support constructor dependency injection, which allows us to achieve Inversion of Control (IoC) between test classes and their dependencies.
Install-Package Xunit.Di
To use xunit.di:
Install the xunit.di nuget package
Create a Setup.cs class to configure dependencies, (optional) and inherits the Xunit.Di.Setup.cs
Configure dependencies in the Setup.cs class.
Find full instructions and demos from xunit.di GET-STARTED
Your test project has the following:
Setup class that has a public IServiceProvider, which configures all the dependencies
Test class with constructor injecting the dependencies
Your Setup.cs class looks like below:
private IServiceProvider _services;
private bool _built = false;
private readonly IHostBuilder _defaultBuilder;
public Setup()
{
_defaultBuilder = Host.CreateDefaultBuilder();
}
public IServiceProvider Services => _services ?? Build();
private IServiceProvider Build()
{
if (_built)
throw new InvalidOperationException("Build can only be called once.");
_built = true;
_defaultBuilder.ConfigureServices((context, services) =>
{
services.AddSingleton<TextReaderService>();
services.AddSingleton<IDependency, DependencyImpl>();
// where DependencyImpl implements IDependency
// ... add other services needed
});
_services = _defaultBuilder.Build().Services;
return _services;
}
Then your test class looks like below:
public class MyTest
{
private readonly IDependency _dependency;
public MyTest(IDependency dependency)
{
_dependency = dependency;
}
[Fact]
public void test_my_code()
{
var result = _dependency.DoStuff();
Assert.NotNull(result);
////simply just test the code
}
}
In Visual Studio when you select project or project items in the solution explorer there are times when you might want to add custom properties to the properties window(the window that pops up when you press F4). Also, to fill in the values of those properties I need to add a button to pop up a form so I can collect information from the user at design time.
What is the simplest implementation of this so I can get started?
How would I create a user interface to collect the value some how by using UITypeEditAttribute?
This is the simplest implementation I could come up with.
Since this is an advanced topic, it is implied that you feel comfortable with completing all the steps before you start the implementation(these are all common programming tasks).
If anything is not clear enough just comment and I will try to simplify. Note that this is configured to create a custom property for a Visual C# file within visual studio. When you run or debug your visual studio package followed by clicking any .cs file, the custom property should show in the properties window. The comments provided are required instructions.
Create a Visual Studio package.
Create an interface that implements the custom properties that you would like to add to the properties page.
Create a class that implements the custom property interface and decorate the custom property with attributes.
Create class that implements IExtenderProvider interface and override GetExtender and CanExtend methods.
Create a new class that inherits from UITypeEditor and override GetEditStyle and EditValue methods.
Let's get started.
1. Create Package in visual studio.
Package.cs
// ...
public sealed class ThePackage : Package
{
private DTE2 Host;
private ObjectExtenders _extensionManager;
private MyExtenderProvider _extenderProvider;
protected override void Initialize()
{
Host = (DTE2)Microsoft.VisualStudio.Shell.Package.GetGlobalService(typeof(SDTE));
_extenderProvider = new MyExtenderProvider();
_extenderProviderCookie = Host.ObjectExtenders.RegisterExtenderProvider(VSConstants.CATID.CSharpFileProperties_string,
"MyExtenderProvider", _extenderProvider);
}
protected override void Dispose(bool disposing)
{
Host.ObjectExtenders.UnregisterExtenderProvider(_extenderProviderCookie);
_extenderProvider = null;
base.Dispose(disposing);
}
}
2. Create class that implements your desired custom properties.
[ComVisible(true)] // Important!
public interface IMyDynamicExtender
{
String NewProperty { get; set; }
}
3. Create a class that implements the custom property interface.
[ComVisible(true)] // Important!
public class NewPropertyExtender : IMyDynamicExtender, IDisposable
{
// These attibutes supply the property with some information
// on how to display and which UITypeEditor to use.
[DisplayName("New Property")]
[Category("New")]
[Description("Specifies the new property")]
[Editor(typeof(CustomUiTypeEditor), typeof(UITypeEditor))]
public String NewProperty { get; set; }
private readonly IExtenderSite _extenderSite;
private readonly int _cookie;
private bool _disposed;
public NewPropertyExtender(IExtenderSite extenderSite, int cookie)
{
_extenderSite = extenderSite;
_cookie = cookie;
}
public void Dispose()
{
Dispose(true);
// take the instance off of the finalization queue.
GC.SuppressFinalize(this);
}
private void Dispose(bool disposing)
{
if (_disposed) return;
if (disposing && _cookie != 0)
{
_extenderSite.NotifyDelete(_cookie);
}
_disposed = true;
}
}
4. Create class that implements [IExtenderProvider] interface and override [GetExtender] and [CanExtend] methods.
public class MyExtenderProvider : IExtenderProvider
{
private IMyDynamicExtender _extender;
public object GetExtender(string extenderCatid, string extenderName,
object extendeeObject, IExtenderSite extenderSite,
int cookie)
{
return _extender = CanExtend(extenderCatid, extenderName, extendeeObject) ?
new NewPropertyExtender(extenderSite, cookie) : null;
}
public bool CanExtend(string extenderCatid, string extenderName, object extendeeObject)
{
// Some implementation will be here in the real world.
return true;
}
}
5. Create a new class that inherits from [UITypeEditor] and override [GetEditStyle] and [EditValue] methods.
public class CustomUiTypeEditor : UITypeEditor
{
public override UITypeEditorEditStyle GetEditStyle(ITypeDescriptorContext context)
{
return UITypeEditorEditStyle.Modal;
}
public override object EditValue(ITypeDescriptorContext context, IServiceProvider provider, object value)
{
// Use the result of a dialog or something else here.
return "HELLO WORLD";
}
}
I'm mucking about with reactive extensions and Iv'e hit a snag that I can't for the life of me work out what the cause is.
If I use a .NET 4 console mode app, where everything is static as follows:
using System;
using System.Reactive.Subjects;
using FakeDal;
using FakeDal.Entites;
using RxProducer;
namespace Runner
{
class Program
{
private static readonly Subject<DaftFrog> _subject = new Subject<DaftFrog>();
private static readonly Repository<DaftFrog> _frogRepo = new Repository<DaftFrog>();
static void Main()
{
_subject.Subscribe(RespondToNewData);
}
private static void RespondToNewData(DaftFrog frog)
{
_frogRepo.Save(frog);
}
}
}
DaftFrog is just a test class in my fake DAL class, this is a simple .NET 4 Class library project, the DaftFrog class, is a simple poco with a few fields in, the dal.save method just simply does a console.WriteLine of a field in the DaftFrog object.
Both classes are just simple stand in's for the real things once I get around to making the RX code work.
Anyway, back to the problem, so the code above works fine, and if I do a few
_subject.OnNext(new DaftFrog());
calls, the fake dal class, prints out what I expect and everything works fine...
HOWEVER>....
If I then transport this code as is, to a class library, and then new up that class library from within my "static program" as follows:
using System.Reactive.Subjects;
using FakeDal;
using FakeDal.Entites;
namespace RxProducer
{
public class Producer
{
private readonly Subject<DaftFrog> _subject = new Subject<DaftFrog>();
private readonly Repository<DaftFrog> _frogRepo = new Repository<DaftFrog>();
private int _clock;
public void Start()
{
_subject.Subscribe(RespondToNewData);
}
public void Stop()
{
}
public void Tick()
{
if(_clock % 5 == 0)
{
DaftFrog data = new DaftFrog();
_subject.OnNext(data);
}
_clock++;
}
private void RespondToNewData(DaftFrog frog)
{
_frogRepo.Save(frog);
}
}
}
And then use that class in my program
using System;
using RxProducer;
namespace Runner
{
class Program
{
private static readonly Producer _myProducer = new Producer();
static void Main()
{
_myProducer.Start();
while(!line.Contains("quit"))
{
_myProducer.Tick();
line = Console.ReadLine();
}
_myProducer.Stop();
}
}
}
Then my project fails to compile.
Specifically it fails on the line:
_subject.Subscribe(RespondToNewData);
in the RxProducer class library, mores the point, the error the compiler throws back makes little sense either:
Error 1 The best overloaded method match for 'System.Reactive.Subjects.Subject<FakeDal.Entites.DaftFrog>.Subscribe(System.IObserver<FakeDal.Entites.DaftFrog>)' has some invalid arguments H:\programming\rxtesting\RxProducer\Producer.cs 17 7 RxProducer
Error 2 Argument 1: cannot convert from 'method group' to 'System.IObserver<FakeDal.Entites.DaftFrog>' H:\programming\rxtesting\RxProducer\Producer.cs 17 26 RxProducer
At first I thought that it might have been the static thing, so I made everything in the class library static, and that made no difference at all.
Iv'e really not done much with Rx until now, but I work with C# and VS 99% of the time, so I'm aware that the error is telling me it can't convert a type of some description, I just don't understand why it's telling me that, esp when the code works perfectly in the static program, but not in a class library.
Shawty
UPDATE
Second thoughts, I just know there are going to be those who insist that I post the fakedal and daft frog definitions, even though IMHO they won't be required, but to pacify the hordes of pretenders who will ask here they are :-)
using System;
namespace FakeDal
{
public class Repository<T>
{
public void Save(T entity)
{
Console.WriteLine("Here we write T to the database....");
}
}
}
namespace FakeDal.Entites
{
public class DaftFrog
{
public int Id { get; set; }
public string Name { get; set; }
public bool IsTotalyDaft { get; set; }
}
}
Include using System; into file where you have Producer, this will help to convert RespondToNewData to IObserver<T>.
Sounds like the compiler is having trouble inferring the Action...might be missing a using statement for the relevant extension method. Alternatively, try either of:
_subject.Subscribe ((Action<DaftFrog>) RespondToData);
Or:
var obs = Observer.Create ( I forget the overload );
_subject.Subscribe( obs);
I have a Page which consist of AddPage.xaml and AddPage.xaml.cs. I want to create a generic class AddPage which extends from PhoneApplicationPage to outsource some repetitive code like Save or Cancel.
If I change the base class from PhoneApplicationPage to my new generic class, I get this error: Partial declarations of 'AddPage' must not specify different base classes.
To accomplish this you need to do the following.
First, create your base class
public class SaveCancelPhoneApplicationPage : PhoneApplicationPage
{
protected void Save() { ... }
protected void Cancel() { ... }
}
Then, your AddPage needs to be modified to inherit from the base class. The main places this is needed is within the code (AddPage.xaml.cs) AND within the xaml
Code:
public partial class AddPage : SaveCancelPhoneApplicationPage { ... }
Xaml:
<local:SaveCancelPhoneApplicationPage
x:Class="MyPhone.Namespace.AddPage"
xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
xmlns:local="clr-namespace:MyPhone.Namespace"
<!-- other xaml elements -->
</local:SaveCancelPhoneApplicationPage>
UPDATE: Info added based on comments
If you need to have generic like functionality and you must use the Page to do this (rather than a ViewModel) then you can still do this using generic methods
public abstract class SaveCancelPhoneApplicationPage : PhoneApplicationPage
{
protected override void OnNavigatedTo(blaa,blaa)
{
var obj = CreateMyObject();
obj.DoStuff();
}
// You should know what your objects are,
// don't make it usable by every phone dev out there
protected MyBaseObject MyObject { get; set; }
protected T GetMyObject<T>() where T : MyBaseObject
{
return MyObject as T;
}
}
public class AddPage : SaveCancelPhoneApplicationPage
{
public AddPage()
{
MyObject = new MyAddObject();
}
}
In order to outsource some functions you just declare some add class which does the common work. Having another page doesn't do that work.
public class Add
{
public bool SaveContent(string filename, string content)
{
....//some content
return true;
}
public string ViewContent(string filename)
{
string content="";
.....
return content;
}
}
Add this part of code where you thought it is redundant.
Add obj=new Add();
obj.SaveContent("myfile.txt","Hello.This is my content.");
string content("myfile.txt");
Tell me if this is what you intend or not.
I am using Ninject and the MVC3 extension installed with nuget. My kernel setup code is in the App_Start/NinjectMVC3.cs file. Everything works great in controllers, but I can't figure out how to (properly) bind interfaces in the Global.asax.cs MvcApplication code.
I ended up using a hack (creating a public NinjectMVC3.GetKernel() method that returns bootstrap.kernel). However, that will be deprecated, and there must be a proper way to do this that I am not seeing.
Here is my code:
public class LogFilterAttribute : ActionFilterAttribute
{
private IReportingService ReportingService { get; set; }
public LogFilterAttribute( IReportingService reportingService )
{
this.ReportingService = reportingService;
}
...
}
public class MvcApplication : System.Web.HttpApplication
{
public static void RegisterGlobalFilters( GlobalFilterCollection filters )
{
filters.Add( new HandleErrorAttribute() );
filters.Add( new LogFilterAttribute() );
}
...
protected void Application_Start()
{
...
RegisterGlobalFilters( GlobalFilters.Filters );
// NOTE hack:
var kernel = NinjectMVC3.GetKernel();
var logger = kernel.Get<ILogger>();
var bw = new BackgroundWork(logger);
Application["BackgroundWork"] = bw;
bw.Start();
}
}
There are two interfaces I am interested in. The first is just binding an object to a Global variable (the ILogger for the BackgroundWork).
And the second is for an ActionFilter. I read http://www.planetgeek.ch/2010/11/13/official-ninject-mvc-extension-gets-support-for-mvc3/, but I don't see how it plugs into the actual registration (filter.Add).
I don't want to use the Property Inject if I can avoid it.
Any thoughts on the proper way to do this?
Thanks
MVC 3 introduces the DependencyResolver which is populated into a singleton, and the Ninject extension supports it. You could use that in your MvcApplication class if you need it:
protected void Application_Start()
{
// ...
var logger = DependencyResolver.Current.GetService<ILogger>();
}
Now I should point out that it is unnecessary to do this with action filters. In Ninject.MVC3 you are supposed to use the BindFilter syntax, like so:
// Declare empty attribute
public class MyFilterAttribute : FilterAttribute { }
// Dependency module
public class MyModule : NinjectModule
{
public override void Load()
{
// Other bindings
// ...
this.BindFilter<MyActionFilter>(FilterScope.Action, 1)
.WhenControllerHas<MyFilterAttribute>();
}
}
Note that you have to use this because BindFilter is an extension method, and you also have to reference the Ninject.Web.Mvc.FilterBindingSyntax namespace.