Moq Testing and Lambda Equivalence

I have been steadily improving my unit tests as my experience with Moq improves, but I have now encountered an issue that has thrown me a little.

To recap the app design, I am using EF Code First for the ORM, accessed via Repository classes that expose (mockable) operations on the repository. I then have a service layer that performs the business operations across multiple repositories.

My unit testing is focusing on the Service Layer methods, with (partially) mocked repositories. My initial tests were pretty simple setup the mocks, perform the action, assert the response. However as I became more familiar with the Moq Verify function I was able to improve the way I tested by actually verifying that my repositories were being accessed with the expected parameters.
An example is that when my service method GetItem(itemId) is called, I expect my repository method
GetSingle<Item>(int)
to be called with a value of itemId.
I can then test using the following
Repository.Setup(x=>x.GetSingle<Item>(It.IsAny<int>()).Returns(new Item());<BR><BR><br>service.GetItem(10);<BR>Repository.Verify(x->x.GetSingle<Item>(10), Times.Once());

The test above ensures that calling GetItem on my service calls the GetSingle on my repository with the expected parameter, and calls it only once. It is a very basic test for a simple method, but is a good example.

The issue is that my repository is a bit more complex than I have shown, where the Get methods actually accept Lambda expressions, so we use
Repository.GetSingle<Item>(x=>x.ItemId == itemId)
instead of just
GetSingle<Item>(itemId);

Now this actually works IF we are using static/const variables in the collection, for example

Item class has a static const ITEM_TYPE_BOX = "Box"
and my service class calls
List<Item> i = Repository.GetAll<Item>(x=>x.ItemTypeName == Item.ITEM_TYPE_BOX)<BR>and my unit test verifies that the repository method was called<BR>Repository.Verify(x=>x.GetAll<Item>(y=>y.ItemTypeName == Item.ITEM_TYPE_BOX, Times.Once())
If we are passing variables to the service method, and that variable is used to create the repository expression, the test fails as the Verify method cannot find a matching execution of the method.
My service method accepts a string itemType
List<Item> i = Repository.GetAll<Item>(x=>x.ItemTypeName = itemType)
and my unit test uses the following verify method
Repository.Verify(x=>x.GetAll<Item>(y=>y.ItemTypeName == "Box", Times.Once());

The test code above fails because while the two lamdba expressions are functionallity identical, they are not expressively equal, so the Moq verify function comparison fails. Drilling down into the issues I have found that the expression specifically includes the namespace of the method that creates the expression when using local types, which means creating the lambda expression in one class, and comparing it to an identical expression created in another class will always fail (you can see this if you create an expression and call ToString() on the expression). The reason it works for constant comparisons is that there are no local variables to compare to.

I cannot remove the dependency on lambdas in my repository as this forms the core of the repository flexibility, but I have identified one way of keeping my unit tests robust while overcoming this issue.
It is possible to expose methods in the service that create an expression object which can be used by both the unit test verify, and the service method. The issue with this is that it is somewhat cumbersome since you need to expose a method for each of the expression combinations your service is using. In many cases this will be relatively straight forward, but it is still a fair bit of effort.

This is pretty disappointing as the unit testing was going quite well up to this point.

The difference between bad code and good code

A colleague asked for some advice today on a project that he inherited (which I am extending with a separate module incidentally).

The issue was related to the usage of Entity Framework in the code that he had to maintain, and he needed some advice on how to proceed.  The problem was that the service layer was calling the repository multiple times, but each repository method was wrapped in a separate unit of work.
e.g.

public void DeleteEntity(int entityID)
{
    using (var context = EntityContext())
    {
        var entity = context.Postings.SingleOrDefault(p => p.entityID == entityID);
        context.Entities.DeleteObject(entity);
        context.SaveChanges();
    }
}
and
public Entity GetPosting(int entityID)

{
    using (var context = EntityContext())
    {
        return context.Entities.FirstOrDefault(p => p.entityID == entityID);
    }
}

This caused two problems for the developer, who needed to perform a complex action in his service that referenced multiple repository calls.

1. He had no control over the transactional scope for the repository methods
2. Each operation was on a separate EF context, so the service could not load and entity, edit it, and then save the changes (unless the repository was designed for disconnected entities, which it wasn't).

From a maintainability and testability point of view this was also a very poor design, as the repository methods created instances of dependency object (the service method also created instances of the repositories, making the services inherently untestable).


The version of this design that I implemented for my component follows a similar service/repository/entity pattern, but is implemented in a far more testable and robust manner.

The first improvement over the legacy design is in the dependency management
My service accepts a context and all required repositories in the constructor, and my repositories accepts a context, which allows for improved maintainability (all dependencies are described) and testability (all dependencies can be mocked).  This also allows us to use dependency injection/IoC to create our object instances.

The second improvement was in the Unit of Work design.
Rather than have each repository method as a single unit of work, the service methods are the units of work, so any action within the service uses the same context (as it is passed as a dependency to the repositories that the service uses), and each service call acts as a Unit of Work, calling SaveChanges at the end of the service to ensure that the changes act under a single transaction.
There are limitations to this design (your public service methods become an atomic transaction and you should not call other public methods from within another method) but for simplicity and maintainability it is a pretty good solution.

Below is a simple example of the design I am using, preserving maintainability, testability, and predictability.  I'm not saying it is necessarily the best code around, but it solves a number of issues that I often see in other developers code.

public class HydrantService
{
  public HydrantService(HydrantsSqlServer context, EFRepository<Hydrant> hydrantRepository, EFRepository<WorkOrder> workOrderRepository, EFRepository<HydrantStatus> hydrantStatusRepository)
  {
    _context = context;
    _hydrantRepository = hydrantRepository;
    _workOrderRepository = workOrderRepository;
    _hydrantStatusRepository = hydrantStatusRepository;
  }
  public void createFaultRecord(WorkOrder order)
  {
    HydrantStatus status = _hydrantStatusRepository.GetSingle<HydrantStatus>(x => x.StatusCode == "Fault"); //_context.HydrantStatuses.Where(x => x.StatusCode == "Fault").FirstOrDefault();
    order.Hydrant.HydrantStatus = status;
    _workOrderRepository.Add(order);
    _context.SaveChanges();
  }
}

  public class EFRepository<T>
 {
 public EFRepository(IDbContext context)
 {
    _context = context;
  }   public virtual ICollection GetAll()

  {
    IQueryable query = _context.Set();
    return query.ToList();
  }
}

Unit Testing, Moq, EF, and Repositories

　
Well, I have just started a small (8-12 week / 1 resource) project using an unfinished version of our in-house framework for some parts of it. In the process I want to ensure that I integrate some key design patterns (null object, repository, and unit of work) and full unit testing on the service implementation. This will hopefully help alleviate the pain of working with DotNetNuke, cross-application dependencies, and webforms.

So my first step was to property expose services from the dependent application as this is a major point of failure in other systems that use this application, which was pretty straight forward as the application design is not too bad. As this is a shared dependency on the DotNetNuke instance, I did not need to expose this as a WCF service, but could easily change it in the future if necessary. The new service interface will help prevent changes in the core application from breaking the dependent application, as any changes will be reflected as build failures in the service class, highlighting this to the developers and ensuring they either make the change to not break the interface, or let all consumers of this service know there is a breaking update and plan appropriate changes. This is a key issue encountered when services and application references are not well defined, and has caused a number of deployment issues at my current client.

The guts of this post however is to discuss my plan for unit testing, and how I had to rethink my previous statement of going ‘repository-less’. I previously discussed the removal of the repository from the framework and using the DbSet functionality in the EF context as the repository pattern. This worked really well, until I decided to do some unit tests.
I decided to use a mocking library in my unit tests specifically to ensure I was performing appropriately isolated tests, and to reduce the impact of managing test data. I had previously looked at Moles (Microsoft stubbing tool), but it always seemed so cumbersome and confusing, so I picked up Moq instead. I really like the Moq usage pattern, and so I thought it would be a good fit.

So, the plan was to use Moq to create mocks of the repository functions that act in predictable and repeatable ways, which means we can run the service and test that the service behaves as we expect.

An example is given below – in this example I created a service to get a list of ‘stations’ from the dependent application. Since I am testing my service, I want to Mock the dependent application service to act predictably, so I can ensure that my service acts the way I want it to (we are not performing end-to-end integration testing, so we don’t want to rely on the dependent application succeeding or failing at this point)

//when we call ‘GetStations’ with a parameter of 0, our mocked service throws an exception – I know the dependent service reacts in this way, so I can ensure this is integrated in my test
_samsServiceMoq.Setup(x => x.GetStations(0)).Throws();
//when we call ‘GetStations’ with a parameter of -1, our mocked service returns no results
_samsServiceMoq.Setup(x => x.GetStations(-1)).Returns(new List());
//when we call ‘GetStations’ with a parameter of 1, our mocked service returns a list with one item in it
_samsServiceMoq.Setup(x => x.GetStations(1)).Returns(new List() { new Unit(){ UnitID = "100" } });
　
UserService target = new
UserService(_samsServiceMoq.Object); //create an instance of my service, and pass in the mocked dependent service
List actual1;
List actual2;
List actual3;
actual1 = target.GetStations(-1); //execute the service method with the specified parameter
actual2 = target.GetStations(1); //execute the service method with the specified parameter
actual3 = target.GetStations(0); //execute the service method with the specified parameter
_samsServiceMoq.VerifyAll();
//check whether the mocked service methods were called in the execution of our tests – this is useful to ensure that your service method is calling the expected mocked method with the expected parameters.
//check the results from the service to ensure they match what you expect (based on the response from the mocked service)
Assert.IsNotNull(actual1);
Assert.IsNotNull(actual2);
Assert.IsNotNull(actual3);
Assert.AreEqual(0, actual1.Count);
Assert.AreEqual(1, actual2.Count);
Assert.AreEqual("100", actual2[0].UnitID);
Assert.AreEqual(0, actual3.Count);

The above example shows how you can configure a test without worrying about the dependent services, so you can test only the functionality in your service. You will also note that the service itself needs to be designed so that all dependencies are passed to the service, instead of created in the service (this is a key point in ensuring testability of components, all dependencies must be passed to the object). If we did not do this, we could never mock the dependent service, which means we would need to set up the test to ensure the dependent service responds appropriately (configure the dependency, and know/configure sample data that the dependency will respond to).
This works really well, I can test my (admittedly very simple) service without caring about configuring the dependent service. However doing the same thing on an EF repository instead of the dependent service does not work so well. The code below should work, but doesn’t due to limitations in EF/C#/Moq.

_hydrantContextMoq.Setup(x=>x.Hydrants).Returns(_hydrantDbSetMoq.Object);
_hydrantDbSetMoq.Setup(x => x.ToList()).Returns(new List() { new Hydrant() });
HydrantService service = new HydrantService(_hydrantContextMoq.Object);
List actual;
actual = service.GetHydrantList();
_hydrantContextMoq.VerifyAll();
_hydrantDbSetMoq.VerifyAll();
Assert.IsTrue(actual.Count == 1);

Here I am mocking my DbContext to return a mocked IDbSet, and mocking the IDbSet.ToList() to return a list of Hydrants with 1 item. This way I can test my service so that calling getHydrantList on my service returns the single length list. Unfortunately, IDbSet.ToList() is not a mockable method (it is actually an extension method) which means it is not possible to set up a mock for this method. Since my service is using this method, I cannot test my service in isolation of the database.

This is where the Repository comes in. Instead of using the IDbSet.ToList() directly, I would use a Repository GetAll() method which abstracts the call to the underlying DbSet method. As the repository is just another dependency on the service, we can mock this instead of the EF IDbSet, and hence have an appropriately testable service. We will also then have the ability to ensure that the repository supports the null object pattern, so a call to the IDbSet that may return null (such as a find() with an invalid key) can return an appropriate null object to the service, so the service, and all clients, know it will never receive a null as the result of a service operation.

So, big backtrack on the framework repository, and big kudos to Moq for making testing easier (at least for my simple examples so far).