/TestableCombinePublishers

An easy, declarative way to unit test Combine Publishers in Swift

Primary LanguageSwiftMIT LicenseMIT

Testable Combine Publishers

An easy, declarative way to unit test Combine Publishers in Swift

Example Combine Unit Test

About

Combine Publishers are notoriously verbose to unit test. They require you to write complex Combine chains in Swift for each test, keeping track of AnyCancellables, and interweaving XCTestExpectations, fulfillment requirements, and timeouts.

This Swift Package aims to simplify writing unit tests for Combine Publishers by providing a natural spelling of .expect(...) for chaining expectations on the Publisher subject. The resulting PublisherExpectation type collects the various expectations and then provides a way to assert that the expectations are fulfilled by calling .waitForExpectations(timeout: 1)

Under the hood, PublisherExpectation is utilizing standard XCTest framework APIs and forwarding those assertion results to the corresponding lines of code that declared the expectation. This allows you to quickly see which specific expectation, in a chain of expectations, is failing in your unit tests, both in Xcode and in the console output.

Usage

In an XCTestCase, add a new unit test function, as normal, preparing the Publisher test subject to be tested. Utilize any combination of the examples below to validate the behavior of any Publisher in your unit tests.

Examples

For a Publisher that is expected to emit a single value and complete with .finished

func testSingleValueCompletingPublisher() {
    somePublisher
        .expect(someEquatableValue)
        .expectSuccess()
        .waitForExpectations(timeout: 1)
}

For a Publisher that is expected to emit multiple values, but is expected to not complete

func testMultipleValuePersistentPublisher() {
    somePublisher
        .collect(someCount)
        .expect(someEquatableValueArray)
        .expectNoCompletion()
        .waitForExpectations(timeout: 1)
}

For a Publisher that is expected to fail

func testPublisherFailure() {
    somePublisher
        .expectFailure()
        .waitForExpectations(timeout: 1)
}

For a Publisher that is expected to emit a value after being acted upon externally

func testLoadablePublisher() {
    let test = someDataSource.publisher
        .expect(someEquatableValue)
    someDataSource.load()
    test.waitForExpectations(timeout: 1)
}

For a Publisher expected to emit a single value whose Output is not Equatable

func testNonEquatableSingleValue() {
    somePublisher
        .expect({ value in
            if case .loaded(let model) = value, !model.rows.isEmpty { } else {
                XCTFail("Expected loaded and populated model")
            }
        })
        .waitForExpectations(timeout: 1)
}

For a Publisher that should emit a specific non-Equatable Error

func testNonEquatableFailure() {
    somePublisher
        .expectFailure({ failure in 
            switch failure {
            case .noInternet, .airPlaneMode:
                break
            default:
                XCTFail("Expected connectivity error")
            }
        })
        .waitForExpectations(timeout: 1)
}

Available Expectations

Value Expectations

  • expect(_ expected: Output) - Asserts that the provided Equatable value will be emitted by the Publisher
  • expectNot(_ expected: Output) - Asserts that a value will be emitted by the Publisher and that it does NOT match the provided Equatable
  • expect(_ assertion: (Output) -> Void) - Invokes the provided assertion closure on every value emitted by the Publisher. Useful for calling XCTAssert variants where custom evaluation is required

Success Expectations

  • expectSuccess() - Asserts that the Publisher data stream completes with a success status (.finished)

Failure Expectations

  • expectFailure() - Asserts that the Publisher data stream completes with a failure status (.failure(Failure))
  • expectFailure(_ failure: Failure) - Asserts that the provided Equatable Failure type is returned when the Publisher completes
  • expectNotFailure(_ failure: Failure) - Asserts that the Publisher completes with a Failure type which does NOT match the provided Equatable Failure
  • expectFailure(_ assertion: (Failure) -> Void) - Invokes the provided assertion closure on the Failure result's associated Error value of the Publisher. Useful for calling XCTAssert variants where custom evaluation is required

Completion Expectations

  • expectCompletion() - Asserts that the Publisher data stream completes, indifferent of the returned success/failure status
  • expectNoCompletion() - Asserts that the Publisher data stream does NOT complete. ⚠️ This will wait for the full timeout in waitForExpectations(timeout:)
  • expectCompletion(_ assertion: (Completion<Failure>) -> Void) - Invokes the provided assertion closure on the recieveCompletion handler of the Publisher. Useful for calling XCTAssert variants where custom evaluation is required

Upcoming Features

  • Support for working with Schedulers to avoid relying on timeouts

AutomaticallyEquatable

When you have a complex type or type graph that you would like to compare for unit testing purposes, you can use this protocol by extending your type to conform to it. This will drastically reduce the volume of code required to make unit test equality assertions on custom types. It also negates the need to rely on custom Equatable implementations. (Custom Equatable implementations come with the risk that future changes to the type may invalidate the equatable implementation without warning.)

Important Disclosures

This is an imperfect and assuming implementation of Equatable. It should not be used without understanding the following concepts.

The implementation:

  • Is subject to the consequences of observing a calculated property. (ie, code that changes the state of data when a property is observed).
  • Cannot respect custom Equatable implementations of the values being compared or any of the subsequent members. It will use its own comparison logic instead.
  • Skips over members that cannot be reasonably compared, such as closures. These are assumed to be equal.

Usage

You can conform to the protocol with a single line of code:

class Baz {
    let answer: Int
    init(answer: Init) {
        self.answer = answer
    }
}

enum MyCustomType {
    case foo
    case bar(Baz)
}

extension MyCustomType: AutomaticallyEquatable { /*no-op*/ }

Then, you can compare two of MyCustomType using expect(...), ==`, or an XCTest framework equality assertion.

somePublisher
    .expect(MyCustomType.bar(Baz(answer: 42))
    .waitForExpectations(timeout: 1)

// or

XCTAssertEqual(output, MyCustomType.bar(Baz(answer: 42))

// or

print(output == MyCustomType.bar(Baz(answer: 42)))

If you would like to see the comparison result, you can invoke the following directly to get a detailed result object:

switch MyCustomType.compare(foo, bar) {
case .equal:
    break
case .unequal(let difference):
    print(difference.debugDescription)
}