1

I am quite new at C# and also rhino mocks. I searched and found similar topics with my question but couldnt find a proper solution.

I am trying to understand if the private method is called or not in my unit test. I am using rhino mock, read many files about it, some of them just say that change the access specifier of the method from private to public, but I can not change the source code. I tried to link source file to my test project but it doesnt change. 

  public void calculateItems()
    {
        var result = new Result(fileName, ip, localPath, remotePath);

        calculateItems(result, nameOfString);
    }

    private void calculateItems(Result result, string nameOfString )

As you see from the code above, I have two methods have exactly same name, calculateItems, but public one has no parameter, private one has two parameters. I am trying to understand when I called public one in my unittest, is private method called? 

    private CalculateClass sut;
    private Result result; 

    [SetUp]
    public void Setup()
    {
      result = MockRepository.GenerateStub<Result>();
      sut = new CalculateClass();
    }

    [TearDown]
    public void TearDown()
    {

    }

    [Test]
    public void test()
    {     
        sut.Stub(stub => stub.calculateItems(Arg<Result>.Is.Anything, Arg<string>.Is.Anything));

        sut.calculateItems();

        sut.AssertWasCalled(stub => stub.calculateItems(Arg<Result>.Is.Anything, Arg<string>.Is.Anything));
    }

In my unittest, I am taking such an error which says "No overload method for calculateItems take two arguments". Is there a way to test it without any changing in source code?

Uwe Keim
  • 39,551
  • 56
  • 175
  • 291
SSanc
  • 13
  • 4

3 Answers3

5

You're testing the wrong thing. Private methods are private. They are of no concern to consuming code, and unit tests are consuming code like any other.

In your tests you test and validate the outward facing functionality of the component. Its inner implementation details aren't relevant to the tests. All the tests care about is whether the invoked operation produces the expected results.

So the question you must ask yourself is... What are the expected results when invoking this operation?:

calculateItems()

It doesn't return anything, so what does it do? What state does it modify in some way? That is what your test needs to observe, not the implementation details but the observable result. (And if the operation has no observable result, then there's no difference between "passed" or "failed" so there's nothing to test.)

We can't see the details of your code, but it's possible that the observable result is coupled to another component entirely. If that's the case then that other component is a dependency for this operation and the goal of the unit test is to mock that dependency so the operation can be tested independently of the dependency. The component may then need to be modified so that a dependency is provided rather than internally controlled. (This is referred to as the Dependency Inversion Principle.)

Also of note...

but I can not change the source code

That's a separate problem entirely. If you truly can't change the source code, then the value of these tests is drastically reduced and possibly eliminated entirely. If a test fails, what can you do about it? Nothing. Because you can't change the code. So what are you testing?

Keep in mind that it's not only possible but unfortunately very common for programmers to write code which can't be meaningfully unit tested. If this code was provided to you by someone else and you are forbidden to change it for some non-technical reason, then it will be the responsibility of that someone else to correct the code. "Correcting" may include "making it possible to meaningfully unit test". (Or, honestly, they should be unit testing it. Not you.)

David
  • 208,112
  • 36
  • 198
  • 279
0

If your public method calls your private one then the same thing will happen in your tests. Tests are nothing more than code that can be run and debugged and you can try that so see what happens.

Private methods can't be tested directly but they can be tested via their public callers which is what you are doing, so it's all good. Whether it's a good idea to have a setup like this well, that's a different story entirely but I am not going into that now.

Now, let's discuss what you are actually testing.

Unit tests should not have deep knowledge of the code they test. The reason is that you should have inputs and outputs and you shouldn't care what happens in between.

If you refactor the code and eliminate the private method then your test would break, even if your inputs and outputs to your public method remain the same. That's not a good position to be in, this is what we call brittle tests.

So add your functional tests around the public method, verify that you get hat you expect and don't worry whether it calls your private method or not.

Andrei Dragotoniu
  • 6,155
  • 3
  • 18
  • 32
0

When you say you need to know whether your private methods are called, this can have two different interpretations:

  1. You want to ensure that the private method is called within one particular test, making it a success criterion for that very test.

  2. You want to know if the private method is called at all, by any of your test cases. You might be interested in this because you want to be sure if the private method is covered by your test suite, or as you said, just to form an understanding of what is actually going on in your code.

Regarding the second interpretation: If you want to understand what is going on in the code, a good approach is to use a debugger and just step through the code to see what function is called. As I am not a C# expert here, I can not recommend any specific debugging tool, but finding some recommendations about this on the web should not be difficult. This approach would fulfill your requirements not to require changes to the source code

Another possibility, in particular if you are interested in whether your private function is covered by the tests, is to use a test coverage tool for C#. The coverage tool would show you whether or not the private method was called or not. Again, this would not require to make any changes to the source code.

Regarding the first interpretation of your question: If you want to test that some privat function is called as part of your test's success criterion, you preferrably do this with tests that use the public API. Then, in these tests, you should be able to judge if the private function is called because of the effect that the private function has on the test result.

And, in contrast to other opinions, you should test the implementation. The primary goal of unit-testing is to find the bugs in the code. Different implementations have different bugs. This is why people also use coverage tools, to see if they have covered the code of their implementation. And, coverage is not enough, you also need to check boundary cases of expressions etc. Certainly, having maintainable tests and tests that do not break unnecessarily in case of refactorings are good goals (why testing through the public API is typically a good approach - but not always), but they are secondary goals compared to the goal to find all bugs.

Dirk Herrmann
  • 5,550
  • 1
  • 21
  • 47