Mocking for Kotlin/Native and Kotlin Multiplatform using the Kotlin Symbol Processing API (KSP). Notable features include:

• Concise, non-intrusive, type-safe API
• Mocking of classes and interfaces
• Supports both values and matchers when during stubbing verification
• Supports implicit stubbing of functions returning Unit

Add the io.mockative plugin and dependency to your build.gradle.kts file:

plugins {
  id("io.mockative") version "3.0.0"
}

kotlin {
    commonMain {
        dependencies {
            implementation("io.mockative:mockative:3.0.0")
        }
    }
}

Make sure you specify the following plugin repositories in your build.settings.kts file:

pluginManagement {
    repositories {
        gradlePluginPortal()
        google()
        mavenCentral()
    }
}

Then add the following to your gradle.properties file:

#KSP
ksp.useKSP2=true

If you're having the following error:

KSP2: KtInvalidLifetimeOwnerAccessException: Access to invalid KtAlwaysAccessibleLifetimeToken: PSI has changed since creation

Please disable incremental processing for KSP by adding the following to your gradle.properties file:

ksp.incremental=false

Mockative 3 code generation should "just work", but if you're looking for more control over when Mockative code generation is enabled or disabled, please read Controlling generation of mocks in Mockative 3.

To mock an interface or class with Mockative, annotate the type with the @Mockable annotation. This annotation tells Mockative to generate a mock implementation of the annotated type, which can be used in tests.

import io.mockative.Mockable

@Mockable
class MyService {
    // ...
}

The Kotlin all-open plugin will automatically be applied to any class annotated with @Mockable.

Sometimes we would like to mock types that are external to our module in our tests. To do so, pass a list of types you'd like to mock in any @Mockable annotation in your main module. If you add the list of types to the @Mockable annotation of a type you'd also like to mock by itself, you must specify the type itself in the list of types to mock. By not specifying any types in the annotation, a mock for the annotated type itself is generated.

import io.mockative.Mockable

@Mockable(MyService::class, Clock::class)
class MyService {
    // ...
}

Obtaining an instance of a mock is as ease as calling the <T> mock(KClass<T>) function:

class GitHubServiceTests {
    val api = mock(of<GitHubAPI>())
}

Then, to stub a function or property on the mock you have a couple of options:

To begin stubbing a function you may simply pass the values to the function call inside a block passed to the every or coEvery (when stubbing a suspend function) functions:

// Stub a blocking function
every { githubApi.getRepository("mockative/mockative") }
    .invokes { response }

// Stub a `suspend` function (notice the use of `coEvery`)
coEvery { repositoryMapper.mapResponseToRepository(response) }
    .invokes { repository }

// Stub a property getter
every { repository.maintainer }
    .returns("Nillerr")

// Stub a property setter (these are stubbed by default)
every { repository.stars = repository.stars + 1 }
    .doesNothing()

Sometimes when stubbing a function we're faced with difficulties providing a specified value for one or more of the parameters of our expectation. In such cases we can use the matcher API to specify the values our stub accepts:

// Assuming we want to stub the function in this S3Client:
interface S3Client {
    suspend fun <T> getObject(input: GetObjectRequest, block: suspend (GetObjectResponse) -> T): T
}

// Providing a value for the `block` parameter is difficult, so we can use the `any()` matcher to
// specify that this expectation will match any value passed to that parameter. When we use one 
// matcher, we must use matchers for every parameter, and as such we must use the `eq(value)` 
// matcher to specify that we're expecting a specific request.
coEvery { s3Client.getObject<File>(eq(request), any()) }
    .returnsMany(expected)

Also note how we're explicitly specifying the type parameter. We could do the same by explicitly specifying it in the call to any() like this:

any<suspend (GetObjectResponse) -> File>()

But that would be significantly more verbose that the alternative.

❕Mockative has limited support for the matcher API when targeting Kotlin/Wasm (WASI).

You may want to provide a function as an argument to another function, where you would like to be able to record invocations on that function. To do that, Mockative includes the interfaces Fun[N] and CoFun[N] (for suspend functions), each declaring a single function invoke that you can pass as a mock to other functions in order to stub and verify invocations on it:

// Declare the mock function as a property in your test class
@Mock
val block = mock(of<Fun1<GetObjectResponse, File>>())

// Stub the mock function
every { block.invoke(response) }
    .returns(file)

// Call something that calls the mock function
s3Client.getObject(request, block::invoke)

// Verify the call to the mock function
verify { block.invoke(response) }
    .wasInvoked(exactly = once)

The following functions are available to provide a stub implementation for every expectation:

In order to provide familiarity to developers coming from MockK, who prefer using infix notation, Mockative also supports infix notation for the invokes, returns, and throws functions:

every { api.getUsers() } returns users

When the return type of the function or property being stubbed is Unit, the following additional then function is available:

Mockative automatically stubs functions returning Unit, based on the idea that such functions are typically used for verification rather than stubbing, and stubbing them could thus be considered boilerplate, while they are trivially automatically stubbed.

You can opt out of this behaviour on the project level through your build.gradle.kts file:

build.gradle.kts

mockative {
    stubsUnitByDefault = false
}

Alternatively, you can opt out (or opt-in if you've opted out on the project level), using the configure(mock, block) function either inline:

@Mock
val api = configure(mock(of<GitHubAPI>())) { stubsUnitByDefault = false }

Or as needed:

@Mock
val api = mock(of<GitHubAPI>())

@Test
fun test() {
    configure(api) { stubsUnitByDefault = false }
}

The configuration is stored on the mock instance. It must be configured before receiving any invocations.

When mocking a generic type use the <T> of(): KClass<T> function to retain the type arguments when passed to the <T> mock(KClass<T>) function. You can use the of function regardless of whether you're mocking a generic or non-generic type.

class GenericTypeTest {
    @Mock
    val list = mock(of<List<String>>())
}

Verification of invocations on mocks is supported through the verify and coVerify functions using either values or matchers as per the stubbing (every) API:

// Verify a `suspend` function (notice the use of `coVerify`)
coVerify { githubApi.getRepository("mockative/mockative") }
    .wasNotInvoked()

// Verify a blocking function
verify { repositoryMapper.mapResponseToRepository(response) }
    .wasInvoked(atLeast = 1)

// Verify a property getter
verify { repository.maintainer }
    .wasInvoked(atLeast = once, atMost = 6)

// Verify a property setter
verify { repository.stars = repository.stars + 1 }
    .wasInvoked(exactly = 1)

// Verify a suspend function (notice the use of `coVerify`)
coVerify { s3Client.getObject<File>(eq(request), any()) }
    .wasInvoked(exactly = once)

In addition to verification of expectations there's also validation of the use of expectations, which, if not used, can lead to both "under verification" and "over mocking" resulting in inaccurate tests. To handle such cases Mockative provides 2 functions you can use on your mocks, usually in the @AfterTest function of your tests:

@AfterTest
fun afterTest() {
    // Verifies that all expectations were verified through a call to `verify { ... }.wasInvoked()`.
    verifyNoUnverifiedExpectations(githubApi)

    // Verifies that the mock has no expectations that weren't invoked at least once.
    verifyNoUnmetExpectations(s3Client)
}