How to use ContainsWarnings method of org.mockitousage.junitrunner.SilentRunnerTest class

The thenReturn() method signature is

thenReturn(T value, T... values)

So it takes an instance of T, followed by a vararg T..., which is syntactic sugar for an array. So you can use

when(foo.bar()).thenReturn(list.get(0), 
                           list.subList(1, list.size()).toArray(new Foo[]{}));

But a cleaner solution would be to create an Answer implementation that takes a List as argument and answers the next element of the list every time it's used. And then use

when(foo.bar()).thenAnswer(new SequenceAnswer<>(list));

For example:

private static class SequenceAnswer<T> implements Answer<T> {

    private Iterator<T> resultIterator;

    // the last element is always returned once the iterator is exhausted, as with thenReturn()
    private T last;

    public SequenceAnswer(List<T> results) {
        this.resultIterator = results.iterator();
        this.last = results.get(results.size() - 1);
    }

    @Override
    public T answer(InvocationOnMock invocation) throws Throwable {
        if (resultIterator.hasNext()) {
            return resultIterator.next();
        }
        return last;
    }
}

Mockito cannot mock final classes or methods. Try removing the final modifier from the hasInventory method. Or better yet, do not mock the Warehouse class, instead mock the IWarehouse interface, whose methods cannot be final, and presumably defines those used by Order.

In general, it's preferable to mock interfaces, but it's not mandatory.

Not being able to mock final classes or methods is briefly mentioned in the Mockito FAQ, it is due to the runtime class generation technique used for creating the mocks.

OK. So, the contract of the method is the following: Parse the input string as JSON, and send back BAD_REQUEST if it's invalid. If it's valid, create an entity in the datastore with various properties (you know them), and send back OK.

And you need to verify that this contract is fulfilled by the method.

Where does Mockito help here? Well, if you test this method without Mockito, you need a real DataStoreService, and you need to verify that the entity has been created correctly in this real DataStoreService. This is where your test is not a unit test anymore, and this is also where it's too complex to test, too long, and too hard to run because it needs a complex environment.

Mockito can help by mocking the dependency on the DataStoreService: you can create a mock of DataStoreService, and verify that this mock is indeed called with the appropriate entity argument when you call your initialize() method in your test.

To do that, you need to be able to inject the DataStoreService into your object under test. It can be as easy as refactoring your object in the following way:

public class MyRestService {
    private DataStoreService dataStoreService;

    // constructor used on the server
    public MyRestService() {
        this.dataStoreService = DatastoreServiceFactory.getDatastoreService();
    }

    // constructor used by the unit tests
    public MyRestService(DataStoreService dataStoreService) {
        this.dataStoreService = dataStoreService;
    }

    public Response initialize(String DatabaseSchema) {
         ...
         // use this.dataStoreService instead of datastore
    }
}

And now in your test method, you can do:

@Test
public void testInitializeWithGoodInput() {
    DataStoreService mockDataStoreService = mock(DataStoreService.class);
    MyRestService service = new MyRestService(mockDataStoreService);
    String goodInput = "...";
    Response response = service.initialize(goodInput);
    assertEquals(Response.Status.OK, response.getStatus());

    ArgumentCaptor<Entity> argument = ArgumentCaptor.forClass(Entity.class);
    verify(mock).put(argument.capture());
    assertEquals("the correct kind", argument.getValue().getKind());
    // ... other assertions
}

A good technique for getting rid of static calls on 3rd party API is hiding the static call behind an interface.

Let's say you make this interface :

interface IPDOFacade {

    IPDO getContextPDO();
}

and have a default implementation that simply calls the static method on the 3rd party API :

class IPDOFacadeImpl implements IPDOFacade {

    @Override
    public IPDO getContextPDO() {
        return Util.getContextPDO();
    }
}

Then it is simply a matter of injecting a dependency on the interface into MyClass and using the interface, rather than the 3rd party API directly :

public class MyClass {

    private final IPDOFacade ipdoFacade;

    public MyClass(IPDOFacade ipdoFacade) {
        this.ipdoFacade = ipdoFacade;
    }

    private IPDO getIPDO() {
        return ipdoFacade.getContextPDO();
    }

    public String handleIn(Object input) throws Throwable
    {
        String result = "";
        IPDO pdo = getIPDO();

        someImportantBusinessLogic(pdo);

        return result;
    }
    
    ...

}

In your unit test, you can then easily mock your own interface, stub it any way you like and inject it into the unit under test.

This

• avoids the need to make private methods package private.
• makes your tests more readable by avoiding partial mocking.
• applies inversion of control.
• decouples your application from a specific 3rd party library.

I read this question and the issue carefully but not sure if I understood them clearly - From my understanding, it's correct that powermock raise the exception when you pass k and x but only verify k.

Because you are mocking the static method StaticObj.put, when you pass parameter k and x and verify it with

PowerMockito.verifyStatic(times(1));
StaticObj.put("k1", "k1");

PowerMockito.verifyStatic(times(1));
StaticObj.put("x1", "x"); 

This should work. And when you verify parameter k and x with verification for k is commented out.

// PowerMockito.verifyStatic(times(1));
// StaticObj.put("k1", "k1");

PowerMockito.verifyStatic(times(1));
StaticObj.put("x1", "x");

Powermock will get the invocation with put("k1"...) first apparently, so the verification of x will raise an error. Your verification process is sequenced.

There are two testing strategies. One is unit testing, i.e. making sure your service works. The other is integration/end-to-end testing, i.e. making sure everything plays nicely together.

You unit test stuff you own, you integration test everything you have. This is a pretty rough example using just your statement, plus some made up stuff where I can't fill in the blanks.

Unit Tests

Using Mockito

PersonRepository personRepository = mock(PersonRepository.class);

@TestSubject
PersonService personService = new PersonService(): 

@Test
public void unitTest() {
    personService.setPersonRepository(personRepository);
    Person person = new Person(1L);
    Person person2 = new Person(1L);

    when(personRepository.returnPerson(1L)).thenReturn(person2); //expect a fetch, return a "fetched" person;

    personService.deleteFromPerson(person);

    verify(personRepository, times(1)).delete(person2); //pretty sure it is verify after call
}

Using EasyMock...

@Mock
PersonRepository personRepository; //assuming it is autowired

@TestSubject
PersonService personService = new PersonService(): 

@Test
public void unitTest() {
    Person person = new Person(1L);
    Person person2 = new Person(1L);

    EasyMock.expect(personRepository.returnPerson(1L)).andReturn(person2); //expect a fetch, return a "fetched" person;
    personRepository.delete(person2);
    EasyMock.expectLastCall(); //expect a delete for person2 we plan to delete
    replayAll();

    personService.deleteFromPerson(person);

    verifyAll(); //make sure everything was called
}

Yes, this test looks like it is written rigidly, but that's really all you're testing in the unit test, anyways. You want the DB to fetch a Person from the database using an argument, hence why there are two Person objects, and you expect to delete that passed Person object, which is why you expect the call. Simple method yields simple test. You basically want to ensure you're interacting with your repository as you expect. Repository could be broken or null in actual implementation, but that doesn't change the fact that your service is implemented correctly.

Integration Tests

On the other hand, if you want to do an integration test, no mocking is used. Instead, you'll need to wire up everything like a test DB and repo. I'll leave that up to you since there's no reference for implementation.

@Test
public void integrationTestForAddAndDelete() {
    Person person = createDummyPersonForInsertion(); //static method that creates a test Person for you
    Person comparePerson;
    //make sure we haven't added the person yet
    Assert.assertNull(personService.getPerson(person));

    //add the Person
    comparePerson = personService.addPerson(person);
    Assert.assertNotNull(personService.getPerson(person));
    //add a rigorous compare method to make sure contents are the same, i.e. nothing is lost or transmuted incorrectly, ignoring ID if that is autogen
    //alternatively, you can create a unit test just for Person
    Assert.assertEquals(person, comparePerson); 

    //remove the Person
    personService.deleteFromPerson(person);
    Assert.assertNull(personService.getPerson(person));

    //test for exception handling when you try to remove a non-existent person;
    personService.deleteFromPerson(person);

    //test for exception handling when you try to remove null
    personService.deleteFromPerson(null);
}

In this case, you want to make sure your repo actually handles all the calls from the service. You know your service works from the unit test, but does the repo work from the service or have you configured something wrong

Although not a direct answer to your question, however I suggest that you move away from field injection and use constructor injection instead (makes code more readable and testable).

Your code would look like:

class Patient {

  private final Syringe syringe;

  @Inject
  public Patient(Syringe syringe) {
    System.out.println("That hurt like crazy!");
  }

}

Then your test would simply be:

@RunWith(MockitoJUnitRunner.class)
class TestCase {

  @Mock
  Syringe siringeMock;

  Patient patient;

  @Before
  public void setup() {
     patient = new Patient(siringeMock);
  }

}

Update

As suggested by Erik-Karl in the comments, you could use @InjectMocks to get rid of the setup method. The solution works because Mockito will use the appropriate constructor injection (as described here). The code would then look like:

@RunWith(MockitoJUnitRunner.class)
class TestCase {

  @Mock
  Syringe siringeMock;

  @InjectMocks
  Patient patient;

}

I don't see a problem here. With the following code using the Mockito API, I managed to do just that :

public class CodeWithPrivateMethod {

    public void meaningfulPublicApi() {
        if (doTheGamble("Whatever", 1 << 3)) {
            throw new RuntimeException("boom");
        }
    }

    private boolean doTheGamble(String whatever, int binary) {
        Random random = new Random(System.nanoTime());
        boolean gamble = random.nextBoolean();
        return gamble;
    }
}

And here's the JUnit test :

import org.junit.Test;
import org.junit.runner.RunWith;
import org.powermock.api.mockito.PowerMockito;
import org.powermock.core.classloader.annotations.PrepareForTest;
import org.powermock.modules.junit4.PowerMockRunner;
import static org.mockito.Matchers.anyInt;
import static org.mockito.Matchers.anyString;
import static org.powermock.api.mockito.PowerMockito.when;
import static org.powermock.api.support.membermodification.MemberMatcher.method;

@RunWith(PowerMockRunner.class)
@PrepareForTest(CodeWithPrivateMethod.class)
public class CodeWithPrivateMethodTest {

    @Test(expected = RuntimeException.class)
    public void when_gambling_is_true_then_always_explode() throws Exception {
        CodeWithPrivateMethod spy = PowerMockito.spy(new CodeWithPrivateMethod());

        when(spy, method(CodeWithPrivateMethod.class, "doTheGamble", String.class, int.class))
                .withArguments(anyString(), anyInt())
                .thenReturn(true);

        spy.meaningfulPublicApi();
    }
}

how about you create an anonymous exception of type JsonProcessingException

when(mapper.writeValueAsString(any(Object.class))).thenThrow(new JsonProcessingException("Error"){});

The {} braces does the trick. This is much better since it is not confusing to the reader of the test code.