testing/io.neuralheads.kmpworker.testing/FakeKmpWorker

FakeKmpWorker

Test double for KmpWorker that executes tasks immediately in-process.

No WorkManager or BGTaskScheduler involved — runs handlers synchronously in the calling coroutine context.

Features:

Immediate execution (no background scheduling delay)
Tracks all enqueued and cancelled task IDs
Supports simulating failures via failureCount
Supports retry scenario simulation
Fully observable via observe and observeAll
Supports both register() and registerWithContext()
Group cancellation via cancelByTag()

Usage:

val fakeWorker = FakeKmpWorker()

fakeWorker.register("sync") { repository.sync() }
fakeWorker.enqueue(TaskRequest(id = "sync", type = TaskType.OneTime))

assertEquals(TaskState.Success, fakeWorker.lastStateFor("sync"))

Constructors

FakeKmpWorker

constructor()

Properties

cancelledTags

val cancelledTags: MutableList<String>

All tags that have been cancelled via cancelByTag.

cancelledTasks

val cancelledTasks: MutableList<String>

All task IDs that have been cancelled.

enqueuedTasks

val enqueuedTasks: MutableList<TaskRequest>

All task requests that have been enqueued.

failureCount

val failureCount: MutableMap<String, Int>

Map of task ID → number of times it should fail before succeeding. Use to simulate retry scenarios.

Functions

allStatesFor

fun allStatesFor(taskId: String): List<TaskState>

Returns all emitted states for a given task ID in order.

cancel

open suspend override fun cancel(taskId: String)

cancelBatch

open suspend fun cancelBatch(taskIds: List<String>)

cancelByTag

open suspend override fun cancelByTag(tag: String)

clearExecutionHistory

open suspend fun clearExecutionHistory()

enqueue

open suspend override fun enqueue(request: TaskRequest)

enqueueBatch

open suspend fun enqueueBatch(requests: List<TaskRequest>)

enqueueChain

open suspend fun enqueueChain(chain: TaskChain)

open suspend override fun enqueueChain(chain: TaskChain, policy: ChainPolicy)

Executes a TaskChain step-by-step in-process (no background scheduling). Ideal for unit tests — each step runs immediately and synchronously. Chain-level Success/Failed states are emitted to both states and TaskMonitor.

executionCountFor