Develop Reference

How to test void method with Mockito's doAnswer

I am new to Mockito and trying to understand how to use doAnswer in order to test a void method.
Here's my class with the onDestroy method to test:
public class TPresenter implements TContract.Presenter {
private CompositeSubscription viewSubscription;
//.......
#Override public void onCreate(.......) {
this.viewSubscription = new CompositeSubscription();
//.......
}
#Override public void onDestroy() {
if(viewSubscription != null && !viewSubscription.isUnsubscribed()) {
viewSubscription.unsubscribe();
}
}
Now I want to write a test for onDestroy() namely to verify that after executing onDestroy the subscription is unsubscribed. I found several examples to use doAnswer for testing void methods, for example here, and also here but I do not understand them.
Please show how to test the method onDestroy.

The normal way how you could test your onDestroy() would be based on viewSubscription being a mocked object. And then you would do something like:
#Test
public testOnDestroyWithoutUnsubscribe() {
when(mockedSubscription.isUnsubscribed()).thenReturn(false);
//... trigger onDestroy()
verifyNoMoreInteractions(mockedSubscription);
}
#Test
public testOnDestroyWithUnsubscribe() {
when(mockedSubscription.isUnsubscribed()).thenReturn(true);
//... trigger onDestroy()
verify
verify(mockedSubscription, times(1)).unsubscribe();
}
In other words: you create a mocked object, and you configure it to take both paths that are possible. Then you verify that the expected actions took place (or not, that is what the first test case does: ensure you do not unsubscribe).
Of course, you can't test the "subscription object is null" case (besides making it null, and ensuring that no NPE gets thrown when triggering the onDestroy()!
Given the comment by the OP: one doesn't necessarily have to use mocking here. But when you want to test a void method, your options are pretty limited. You have to observe side effects somehow!
If you can get a non-mocked viewSubscription instance to do that, fine, then do that. But if not, then somehow inserting a mocked instance is your next best choice. How to do the "dependency injection" depends on the exact context, such as the mocking/testing frameworks you are using.

Testing void methods in your main class under test is not a problem as does not require doAnswer.
Here is an example of how could you go about testing the call to unsubscribe.
import static org.mockito.Mockito.never;
import static org.mockito.Mockito.verify;
import static org.mockito.Mockito.when;
#RunWith(MockitoJUnitRunner.class)
public class TPresenterTest {
#InjectMocks
private TPresenter target = new TPresenter();
#Mock
private CompositeSubscription viewSubscription;
#Test
public void onDestroyShouldUnsubscribeWhenSubscriptionNotNullAndUnsubscribed() {
when(viewSubscription.isUnsubscribed()).thenReturn(false);
target.onDestroy();
verify(viewSubscription).unsubscribe();
}
#Test
public void onDestroyShouldNotUnsubscribeWhenSubscriptionNotNullAndNotUnsubscribed() {
when(viewSubscription.isUnsubscribed()).thenReturn(true);
target.onDestroy();
verify(viewSubscription, never()).unsubscribe();
}
}

As I mentioned in my comment to #GhostCat 's answer, my example is in fact un-testable because of the "new" instance of CompositeSubscription class. I would have to re-factor it and #GhostCat 's comment to his/her answer shows a way to do it.

How do I abstract away dependencies in Android library code?

Here is my scenario.
I have an android activity in which I want to abstract my I/O dependencies. The dependencies are represented by this interface (edited for brevity and simplicity):
public interface ITimeDataServer {
TimeRecord[] get(int userID);
void save(TimeRecord record);
}
What I want is for my activity to be able to call these interface methods, and leave the implementation to be supplied by the calling code. (Pretty standard, I think).
ITimeDataServer myServer;
int myUserID;
void loadRecords() {
TimeRecord[] records = myServer.get(myUserID);
// etc...
}
My difficulty is, how can I ensure that myServer gets set?
This seems like a common problem, but I can't find a clean solution.
My first thought would be that myServer would be passed in through the constructor, but Android activities aren't really instantiated with constructors.
I've come up with several solutions, but they're all icky in some way:
Icky Solution 1
Create a static method to launch the activity class which takes an ITimeDataServer parameter and stores it in a static variable from which the activity can access it:
private static ITimeDataSource theDataSource;
public static void launch(Activity currentActivity, ITimeDataSource dataSource) {
theDataSource = dataSource;
Intent intent = new Intent(currentActivity, MainActivity.class);
currentActivity.startActivity(intent);
}
This is icky because (a) the data source is static and not actually associated with the instance, and (b) a consumer could initiate the activity by the standard activity API rather than this static method, which will cause NullPointerException.
Icky Solution 2
I can create a Provider class which provides a singleton instance of ITimeDataSource, which needs to be initialized by the calling library before use:
public class TimeDataSourceProvider {
private static ITimeDataSource myDataSource = null;
public void initialize(ITimeDataSource dataSource) {
myDataSource = dataSource;
}
public ITimeDataSource get() {
if (myDataSource == null)
throw new NullPointerException("TimeDataSourceProvider.initialize() must be called before .get() can be used.");
else
return myDataSource;
}
}
This seems a little less icky, but it's still a little icky because the activity's dependency is not obvious, and since there may be many paths to launch it, it's highly possible that some of them would forget to call TimeDataSourceProvider.initialize().
Icky solution 3
As a variation on #2, create a static IODependencyProvider class which must be initialized with ALL dependencies on app startup.
public class IODependencyProvider {
static ITimeDataSource myTimeData;
static IScheduleDataSource myScheduleData; // etc
public static void initialize(ITimeDataSource timeData, IScheduleDataSource scheduleData /* etc */) {
myTimeData = timeData;
myScheduleData = scheduleData;
//etc
}
public static ITimeDataSource getTimeData() {
if (myTimeData == null)
throw new NullPointerException("IODependencyProvider.initialize() must be called before the getX() methods can be used.");
else
return myTimeData;
}
// getScheduleData(), etc
}
This seems superior to #1 and #2 since a failure to initialize would be much harder to sneak by, but it also creates interdependencies among the data types that otherwise need not exist.
...and other icky variations on that theme.
The common themes that make these solutions crappy:
the need to use static fields to pass non-serializable information to an activity
the lack of ability to enforce initialization of those static fields (and subsequent haphazardness)
inability to clearly identify an activity's dependencies (due to reliance on statics)
What's a nooby Android developer to do?

As long as these dependencies implement Parcelable correctly, you should be able to add them to your intent, then unparcel them as ITimeDataServer and get the correct class.

I found a nice solution here, in the least-loved answer.
I define the library activity as abstract and with no default constructor, but a constructor that takes an interface, like so:
public abstract class TimeActivity extends AppCompatActivity {
private ITimeDataSource myTimeDataSource;
public TimeActivity(#NonNull ITimeDataSource dataSource) {
myTimeDataSource = dataSource;
}
#Override
protected void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
setContentView(R.layout.activity_time);
// do stuff with myTimeDataSource!
}
}
Then, the calling code can create a concrete subclass with its chosen implementation that does have a parameterless constructor. No static members, easy-peasy!
This allows you to abstract and inject all sorts of crazy behaviours! Woooo!
(Note that the concrete subclass activity needs to be manually added to AndroidManifest.xml, like all activities, or the app will crash when it tries to launch.)

Mock objects in Android not passed as parameters

I am trying to test a Fragment I've created in Android. I have complete control of the code, so I can change it as I see fit. The issue is that I'm not sure what design pattern I'm missing to make it reasonable.
I am looking for a way to mock objects in Android that are not passed as parameters. This question suggests that anything you might want to mock should be written to be passed as a parameter.
This makes sense for some situations, but I can't figure out how to get it working on Android, where some of this isn't possible. With a Fragment, for example, you're forced to let much of the heavy lifting be done in callback methods. How can I get my mocked objects into the Fragment?
For example, in this ListFragment I need to retrieve an array of things to display to the user. The things I'm displaying need to be retrieved dynamically and added to a custom adapter. It currently looks as follows:
public class MyFragment extends ListFragment {
private List<ListItem> mList;
void setListValues(List<ListItem> values) {
this.mList = values;
}
List<ListItem> getListValues() {
return this.mList;
}
#Override
public void onCreateView(LayoutInflater i, ViewGroup vg, Bundle b) {
// blah blah blah
}
#Override
public void onViewCreated(View view, Bundle savedInstanceState) {
this.setListValues(ListFactory.getListOfDynamicValues());
CustomAdapter adapter = new CustomAdapter(
getActivity(),
R.layout.row_layout,
this.getListValues());
this.setListAdapter(adapter);
}
}
I'm trying to do this using Mockito and Robolectric.
This is the beginning of my robolectric test case:
public class MyFragmentTest {
private MyFragment fragment;
#Before
public void setup() {
ListItem item1 = mock(ListItem.class);
ListItem item2 = mock(ListItem.class);
when(item1.getValue()).thenReturn("known value 1");
when(item2.getValue()).thenReturn("known value 2");
List<ListItem> mockList = new ArrayList<ListItem>();
mockList.add(item1);
mockList.add(item2);
MyFragment real = new MyFragment();
this.fragment = spy(real);
when(this.fragment.getValueList()).thenReturn(mockList);
startFragment();
}
}
This feels so very wrong. This section from the mockito api points out that you shouldn't have to do partial mocks like this very frequently unless you're dealing with legacy code.
Further, I'm not actually able to mock out the CustomAdapter class using this approach.
What is the right way to do this sort of thing? Am I structuring things incorrectly in my Fragment classes? I suppose I might be able to add a bunch of package-private setters, but this still doesn't feel right.
Can someone shed some light on this? I'm happy to do rewrites, I just want to know some good patterns for dealing with the state in my Fragments and how I can make them testable.

I ended up creating my own solution to this. My approach was to add another level of indirection to each my calls that create or set an object.
First, let me point out that I couldn't actually get Mockito to work reliably with Fragment or Activity objects. It was somewhat hit or miss, but especially with trying to create Mockito Spy objects, some lifecycle methods appeared to not be called. I think this is related to gotcha number 2 shown here. Perhaps this is due to the ways that Android uses reflection to recreate and instantiate activities and fragments? Note that I was NOT incorrectly holding onto the reference, as it warns of, but interacting only with the Spy, as indicated.
So, I wasn't able to mock Android objects that required lifecycle methods be invoked by the framework.
My solution was to create to more types of methods in my Activity and Fragment methods. These methods are:
getters (getX()) that return the field named X.
retrievers (retrieveX()) that do some sort of work to get an object.
creators (createMyFragment()) that create objects by calling new. Similar to the retrievers.
Getters have whatever visibility you need. Mine are usually public or private.
Retrievers and creators are package private or protected, allowing you to override them in your test packages but not making them generally available. The idea behind these methods is that you can subclass your regular objects with stub objects and inject in known values during testing. You could also just mock out those methods if Mockito mocks/spies are working for you.
Taken in toto, the test would look something like the following.
Here is the fragment from my original question, modified to use the above approach. This is in the normal project:
package org.myexample.fragments
// imports
public class MyFragment extends ListFragment {
private List<ListItem> mList;
void setListValues(List<ListItem> values) {
this.mList = values;
}
List<ListItem> getListValues() {
return this.mList;
}
#Override
public void onCreateView(LayoutInflater i, ViewGroup vg, Bundle b) {
// blah blah blah
}
#Override
public void onViewCreated(View view, Bundle savedInstanceState) {
this.setListValues(this.retrieveListItems());
CustomAdapter adapter = this.createCustomAdapter();
this.setListAdapter(adapter);
}
List<ListItem> retrieveListItems() {
List<Item> result = ListFactory.getListOfDynamicValues();
return result;
}
CustomAdapter createCustomAdapter() {
CustomAdapter result = new CustomAdapter(
this.getActivity();
R.layout.row_layout,
this.getListValues());
return result;
}
}
When I test this object, I want to be able to control what gets passed around. My first thought was to use a Spy, replacing the return values of retrieveListItems() and createCustomAdapter() with my known values. However, like I said above, I wasn't able to get Mockito spies to behave when working with fragments. (Especially ListFragments--I had mixed success with other types, but don't trust it.) So, we are going to subclass this object. In the test project, I have the following. Note that your method visibility in your real class must allow subclasses to override, so it needs to be package private and in the same package or protected. Note that I am overriding the retriever and creator, returning instead static variables that my tests will set.
package org.myexample.fragments
// imports
public class MyFragmentStub extends MyFragment {
public static List<ListItem> LIST = null;
public static CustomAdapter ADAPTER = null;
/**
* Resets the state for the stub object. This should be called
* in the teardown methods of your test classes using this object.
*/
public static void resetState() {
LIST = null;
ADAPTER = null;
}
#Override
List<ListItem> retrieveListItems() {
return LIST_ITEMS;
}
#Override
CustomAdapter createCustomAdapter() {
return CUSTOM_ADAPTER;
}
}
In the same package in my test project I have the actual test of the fragment. Note that while I'm using Robolectric, this should work with whatever test framework you're using. The #Before annotation becomes less useful, as you need to update your static state for individual tests.
package org.myexample.fragments
// imports
#RunWith(RobolectricTestRunner.class)
public class MyFragmentTest {
public MyFragment fragment;
public Activity activity;
#After
public void after() {
// Very important to reset the state of the object under test,
// as otherwise your tests will affect each other.
MyFragmentStub.resetState();
}
private void setupState(List<ListItem> testList, CustomAdapter adapter) {
// Set the state you want the fragment to use.
MyFragmentStub.LIST = testList;
MyFragmentStub.ADAPTER = adapter;
MyFragmentStub stub = new MyFragmentStub();
// Start and attach the fragment using Robolectric.
// This method doesn't call visible() on the activity, though so
// you'll have to do that yourself.
FragmentTestUtil.startFragment(stub);
Robolectric.ActivityController.of(stub.getActivity()).visible();
this.fragment = stub;
this.activity = stub.getActivity();
}
#Test
public void dummyTestWithKnownValues() {
// This is a test that does nothing other than show you how to use
// the stub.
// Create whatever known values you want to test with.
List<ListItem> list = new ArrayList<ListItem>();
CustomAdapter adapter = mock(CustomAdapter.class);
this.setupState(list, adapter);
// android fest assertions
assertThat(this.fragment).isNotNull();
}
}
This is definitely more verbose than using a mocking framework. However, it works even with Android's life cycle. If I'm testing an Activity, I'll also often include a static boolean BUILD_FRAGMENTS variable. If true, I'll go call through to super in the appropriate methods or return a known fragment as appropriate. In this way I'm able to inject my test objects and play nice with the Android life cycle.

Verify a static method was called by another static method in PowerMock

I have a Tool class with two static methods, doSomething(Object) and callDoSomething(). The names are intuitive in that callDoSomething delegates its call to doSomething(Object);
public class Tool
{
public static void doSomething( Object o )
{
}
public static void callDoSomething()
{
doSomething( new Object());
}
}
I have a Test class for Tool and I'd like to verify if doSomething(Object) was called (I want to do Argument Matching too in the future)
#RunWith( PowerMockRunner.class )
#PrepareForTest( { Tool.class } )
public class ToolTest
{
#Test
public void toolTest()
{
PowerMockito.mockStatic( Tool.class );
Tool.callDoSomething();// error!!
//Tool.doSomething();// this works! it gets verified!
PowerMockito.verifyStatic();
Tool.doSomething( Mockito.argThat( new MyArgMatcher() ) );
}
class MyArgMatcher extends ArgumentMatcher<Object>
{
#Override
public boolean matches( Object argument )
{
return true;
}
}
}
Verify picks up doSomething(Object) if it's called directly. I've commented this code out above. Verify does NOT pick up doSomething(Object) when using callDoSomething, (this is the code shown above). This is my error log when running the code above:
Wanted but not invoked tool.doSomething(null);
However, there were other interactions with this mock.
at org.powermock.api.mockito.internal.invocation.MockitoMethodInvocationControl.performIntercept(MockitoMethodInvocationControl.java:260)
at org.powermock.api.mockito.internal.invocation.MockitoMethodInvocationControl.invoke(MockitoMethodInvocationControl.java:192)
at org.powermock.core.MockGateway.doMethodCall(MockGateway.java:105)
at org.powermock.core.MockGateway.methodCall(MockGateway.java:60)
at Tool.doSomething(Tool.java)
at ToolTest.toolTest(ToolTest.java:22)
... [truncated]
I'd like to avoid making any changes to the Tool class. My question is, how can I verify doSomething(Object) was called from callDoSomething(), as well as perform some argument matching on doSomething's param

It sounds like you want to use a static spy (partial mock). The section of the PowerMock documentation that talks about mocking static has a note in the second bullet that could be easily missed:
(use PowerMockito.spy(class) to mock a specific method)
Note, in your example you're not actually mocking the behavior, just verifying the method is called. There's a subtle but important difference. If you don't want doSomething(Object) to be called you'd need to do something like this:
#Test
public void toolTest() {
PowerMockito.spy(Tool.class); //This will call real methods by default.
//This will suppress the method call.
PowerMockito.doNothing().when(Tool.class);
Tool.doSomething(Mockito.argThat( new MyArgMatcher() ));
Tool.callDoSomething();
//The rest isn't needed since you're already mocking the behavior
//but you can still leave it in if you'd like.
PowerMockito.verifyStatic();
Tool.doSomething(Mockito.argThat( new MyArgMatcher() ));
}
If you still want the method to fire though, just remove the two lines for doNothing(). (I added a simple System.out.println("do something " + o); to my version of Tool.java as an additional verification of doNothing().)

You can do your validation with this:
public class Tool{
public static boolean isFromCallDoSomethingMethod= false;
public static void doSomething(Object o){
}
public static void callDoSomething() {
doSomething(new Object());
isFromCallDoSomethingMethod= true;
}
}
You can do the verification as:
if(Tool.isFromCallDoSomethingMethod){
//you called doSomething() from callDoSomething();
}
REMEMBER
Don't forget to do the validation if you call the doSomething() from another way that is not from callDoSomething(), you can do this by ussing Tool.isFromCallDoSomethingMethod = false
Is this what you want?

How does exactly custom Shadow objects work in Robolectric?

If I write a custom Shadow for my Activity, and registering it with RobolectricTestRunner, will the framework intercept the Activity with my custom Shadow whenever it's started?
Thanks.

The short answer is no.
Robolectric is selective about what classes it intercepts and instruments. At the time of this writing, the only classes that will be instrumented must have a fully qualified classname match one of these selectors:
android.*
com.google.android.maps.*
org.apache.http.impl.client.DefaultRequestDirector
The whole reason for Robolectric's existence is that the classes provided in the Android SDK jar throw exceptions when invoked in a JVM (i.e. not on an emulator or device). Your application's Activity has source that is not 'hostile' (it probably does not throw exceptions when the methods or constructors are invoked). Robolectric's intended purpose is to allow you to put your application's code under test, which would otherwise not be possible due to the way the SDK is written. Some of the other reasons why Robolectric was created were:
The SDK does not always have methods that would allow you to query the state of the Android objects manipulated by your application's code. Shadows can be written to provide access to this state.
Many of the classes and methods in the Android SDK are final and/or private or protected, making it difficult to create the dependencies needed by your application code that would otherwise be available to your application code.
The code could clearly be changed to shadow any class. There has been talk in the past about extracting the shadowing features into a standalone library, to assist writing tests using some other test-hostile api.
Why do you want to shadow your Activity?

This has significantly changed with Robolectric 2. You can specify custom shadows in the configuration instead of writing your own TestRunner.
For example:
#Config(shadows = {ShadowAudioManager.class, ShadowContextWrapper.class})

Yes, if you subclass the RobolectricTestRunner, add a custom package to the constructor and load your Shadow classes in the bindShadowClasses method. No need to use the android.* package trick.
(Note: this is with robolectric-1.1)
There are a number of hooks provided in the RobolectricTestRunner#setupApplicationState that you can override.
Here's my implementation of the RobolectricTestRunner.
import org.junit.runners.model.InitializationError;
import com.android.testFramework.shadows.ShadowLoggerConfig;
import com.xtremelabs.robolectric.Robolectric;
import com.xtremelabs.robolectric.RobolectricTestRunner;
public class RoboRunner extends RobolectricTestRunner {
public RoboRunner(Class<?> clazz) throws InitializationError {
super(clazz);
addClassOrPackageToInstrument("package.you're.creating.shadows.of");
}
#Override
protected void bindShadowClasses() {
super.bindShadowClasses(); // as you can see below, you really don't need this
Robolectric.bindShadowClass(ShadowClass.class);
}
}
More methods you can subclass (from RobolectricTestRunner.class)
/**
* Override this method to bind your own shadow classes
*/
protected void bindShadowClasses() {
}
/**
* Override this method to reset the state of static members before each test.
*/
protected void resetStaticState() {
}
/**
* Override this method if you want to provide your own implementation of Application.
* <p/>
* This method attempts to instantiate an application instance as specified by the AndroidManifest.xml.
*
* #return An instance of the Application class specified by the ApplicationManifest.xml or an instance of
* Application if not specified.
*/
protected Application createApplication() {
return new ApplicationResolver(robolectricConfig).resolveApplication();
}
Here's where they're called in the Robolectric TestRunner:
public void setupApplicationState(final RobolectricConfig robolectricConfig) {
setupLogging();
ResourceLoader resourceLoader = createResourceLoader(robolectricConfig);
Robolectric.bindDefaultShadowClasses();
bindShadowClasses();
resourceLoader.setLayoutQualifierSearchPath();
Robolectric.resetStaticState();
resetStaticState();
DatabaseConfig.setDatabaseMap(this.databaseMap);//Set static DatabaseMap in DBConfig
Robolectric.application = ShadowApplication.bind(createApplication(), resourceLoader);
}

As an update, I have been able to create shadows of my own classes, as long as am careful to bind the shadow class before any possible loader acts on that class. So, per the instructions, in the RoboRunner I did:
#Override protected void bindShadowClasses() {
Robolectric.bindShadowClass(ShadowLog.class);
Robolectric.bindShadowClass(ShadowFlashPlayerFinder.class);
}
Did I mention that I'm cheating a bit? The original answer above is (of course) correct. So I use this for my real class:
package android.niftyco;
public class FlashPlayerFinder {
.. .
And my mock (shadow) is in back in my test package, as one might expect:
package com.niftyco.android.test;
#Implements(FlashPlayerFinder.class)
public class ShadowFlashPlayerFinder {
#RealObject private FlashPlayerFinder realFPF;
public void __constructor(Context c) {
//note the construction
}
#Implementation
public boolean isFlashInstalled() {
System.out.print("Let's pretend that Flash is installed\n");
return(true);
}
}

Might be late, but from here: org.robolectric.bytecode.Setup, you might find further detail about what classes are instrumented.
public boolean shouldInstrument(ClassInfo classInfo) {
if (classInfo.isInterface() || classInfo.isAnnotation() || classInfo.hasAnnotation(DoNotInstrument.class)) {
return false;
}
// allow explicit control with #Instrument, mostly for tests
return classInfo.hasAnnotation(Instrument.class) || isFromAndroidSdk(classInfo);
}
public boolean isFromAndroidSdk(ClassInfo classInfo) {
String className = classInfo.getName();
return className.startsWith("android.")
|| className.startsWith("libcore.")
|| className.startsWith("dalvik.")
|| className.startsWith("com.android.internal.")
|| className.startsWith("com.google.android.maps.")
|| className.startsWith("com.google.android.gms.")
|| className.startsWith("dalvik.system.")
|| className.startsWith("org.apache.http.impl.client.DefaultRequestDirector");
}