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

❗️Mockative 2 introduces a new API that should feel familiar to the Kotlin/JVM developers coming from MockK and Java developers coming from Mockito.

Please see Migrating to Mockative 2 for a migration guide.

Mockative uses KSP to generate mock classes, and as such, it requires adding the KSP plugin in addition to adding the runtime library and symbol processor dependencies.

build.gradle.kts

plugins {
    id("com.google.devtools.ksp")
}

repositories {
    mavenCentral()
}

kotlin {
    sourceSets {
        val commonTest by getting {
            dependencies {
                implementation("io.mockative:mockative:2.2.0")
            }
        }
    }
}

dependencies {
    configurations
        .filter { it.name.startsWith("ksp") && it.name.contains("Test") }
        .forEach {
            add(it.name, "io.mockative:mockative-processor:2.2.0")
        }
}

💡 I have plans to introduce a Gradle plugin that will take care of most of the configuration specified above and below

To utilize Mockative effectively in your Kotlin project, it's essential to configure your Gradle build script to open classes for mocking. Mockative requires classes to be open for subclassing, as Kotlin classes are final by default. This is achieved using the Kotlin all-open compiler plugin.

First, ensure that your project applies the Kotlin all-open plugin. Add the following to your build.gradle.kts file:

plugins {
    kotlin("jvm") // or kotlin("multiplatform") as applicable
    kotlin("plugin.allopen")
}

Create an annotation in your project that will be used to mark classes that should be open, and thus mockable by Mockative. Consider making an annotation class:

package io.github

@Retention(AnnotationRetention.SOURCE)
@Target(AnnotationTarget.CLASS)
annotation class Mockable

Next, configure the allOpen plugin to recognize your custom @Mockable annotation. This step makes all classes annotated with @Mockable open during test, thus allowing Mockative to generate a mock implementation for each of them. An example configuration is shown below:

val taskIsRunningTest = gradle.startParameter.taskNames
  .any { it == "check" || it.startsWith("test") || it.contains("Test") }

if (taskIsRunningTest) {
  allOpen {
    annotation("io.github.Mockable")
  }
}

This configuration should be placed outside of and after the plugins block.

To mock a class with Mockative, annotate the class with your annotation (interfaces are always open). This annotation indicates to Mockative (and the Kotlin compiler, with the above configuration) that the class should be open for subclassing and available for mocking during tests.

import io.github.Mockable

@Mockable
class MyService {
    // ...
}

If you prefer to make all classes in your project open (and thus also mockable by Mockative) without annotating them individually, you can adjust the allOpen configuration to apply to all classes. However, use this approach with caution, as it may have unintended side effects by making all your classes non-final.

allOpen {
    annotation("kotlin.Metadata")
}

To mock a given method on an interface or a class, annotate a property holding the interface- or class type with the @Mock annotation, and assign it to the result of a call to the <T> mock(KClass<T>) function:

class GitHubServiceTests {
    @Mock
    val api = mock(classOf<GitHubAPI>())
}

Then, to stub a function or property on the mock there is 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()

❕The matcher API of Mockative 2 is different that the one of Mockative 1 since we have decided to drop support for stubbing using callable references in favor of the new expression based matcher API. This makes the API of Mockative closer resemble that of other popular mocking frameworks in the Kotlin community.

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.

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(classOf<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:

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.

❕This behaviour is different from Mockative 1.X, where stubbing Unit returning functions was opt-in, where-as it is not opt-out.

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

build.gradle.kts

ksp {
    arg("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(classOf<GitHubAPI>())) { stubsUnitByDefault = false }

Or as needed:

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

class GenericTypeTest {
    @Mock
    val list = mock(classOf<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 @AfterEach function of your tests:

@AfterEach
fun afterEach() {
  // 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)
}