Csharp Testing Standards

C# Testing Standards

Description

Defines the testing standards, patterns, and conventions for all C# unit and integration test projects. Rules cover test framework usage, naming, structure, mocking, assertions, and parameterization. Apply these rules uniformly across all test projects.

---

1. Framework & Tooling

• Test framework: NUnit 3
• Mocking: Moq
• Coverage: coverlet.collector
• Test runner: Microsoft.NET.Test.Sdk + NUnit3TestAdapter

Core NUnit attributes in use:

| Attribute | Purpose | |---|---| | [TestFixture] | Marks the test class | | [Test] | Marks a single test method | | [SetUp] | Runs before each test | | [TearDown] | Runs after each test | | [OneTimeSetUp] | Runs once before all tests in the fixture | | [TestCase] | Parameterized test inline values |

---

2. Test Class Structure

2.1 Visibility

Test classes are internal. They do not need to be public – NUnit discovers them via the test runner regardless.

[TestFixture]
internal class OrderServiceTests { }

2.2 System Under Test Field

Name the field under test _sut (system under test) and initialize it in [SetUp]:

private OrderService _sut = null!;

2.3 Mock Fields

Declare all mocks as class-level fields initialized in [SetUp]:

private Mock _orderRepositoryMock = null!;
private Mock> _loggerMock = null!;

2.4 SetUp Method

• Initialize all mocks and the SUT in the [SetUp] method.
• Configure happy-path default behaviors here so individual tests only override what they specifically need.
• Keep [SetUp] focused – it should not contain assertions or complex logic.

[SetUp]
public void Setup()
{
    _orderRepositoryMock = new Mock();
    _orderRepositoryMock
        .Setup(r => r.GetByIdAsync(It.IsAny(), It.IsAny()))
        .ReturnsAsync((Order?)null);

    _sut = new OrderService(
        _orderRepositoryMock.Object,
        NullLogger.Instance);
}

---

3. Test Method Naming

3.1 Pattern

MethodName_WhenCondition_ThenExpectedBehavior

• MethodName – the method being tested (must match the implementation exactly).
• WhenCondition – the specific input state or pre-condition being exercised.
• ThenExpectedBehavior – the exact outcome that is asserted.

// ✅ Correct
public async Task GetByIdAsync_WhenOrderDoesNotExist_ThenReturnsNull()
public async Task CreateAsync_WhenRequestIsValid_ThenPersistsAndReturnsSuccess()
public void Validate_WhenAmountIsNegative_ThenReturnsInvalidResult()

// ❌ Wrong – vague and non-descriptive
public async Task TestGetOrder()
public async Task CreateOrder_Success()

3.2 Async Test Methods

All async test methods that are async Task must also follow the Async suffix rule:

// ✅ Correct
public async Task CreateAsync_WhenRequestIsValid_ThenReturnsSuccess()

// ❌ Wrong
public async Task Create_WhenRequestIsValid_ThenReturnsSuccess()

---

4. Arrange / Act / Assert

Every test body must follow the three-section AAA structure, with explicit comments:

[Test]
public async Task GetByIdAsync_WhenOrderExists_ThenReturnsOrder()
{
    // Arrange
    var orderId = Guid.NewGuid();
    var order = new Order { Id = orderId, CustomerName = "Alice" };
    _orderRepositoryMock
        .Setup(r => r.GetByIdAsync(orderId, It.IsAny()))
        .ReturnsAsync(order);

    // Act
    var result = await _sut.GetByIdAsync(orderId, It.IsAny());

    // Assert
    Assert.That(result, Is.Not.Null);
    Assert.That(result!.CustomerName, Is.EqualTo("Alice"));
}

---

5. Assertions

5.1 Constraint Model

Always use Assert.That(actual, constraint) – the NUnit constraint model. Avoid the classic assertion API:

// ✅ Correct – constraint model
Assert.That(result, Is.Not.Null);
Assert.That(result.IsValid, Is.True);
Assert.That(result.Message, Is.Null);
Assert.That(items, Has.Length.EqualTo(2));
Assert.That(items, Is.EquivalentTo(expected));

// ❌ Avoid – classic API
Assert.IsNotNull(result);
Assert.IsTrue(result.IsValid);
Assert.AreEqual(2, items.Length);

5.2 Exception Assertions

Use Assert.ThrowsAsync for async methods that are expected to throw:

Assert.ThrowsAsync(
    () => _sut.CreateAsync(null!, It.IsAny()));

5.3 Mock Verification

Use .Verify() only when asserting that a side-effecting call was (or was not) made. Do not use .Verify() as a substitute for return value assertions:

// ✅ Correct – asserting a save was triggered exactly once
_orderRepositoryMock.Verify(
    r => r.SaveAsync(It.IsAny(), It.IsAny()),
    Times.Once);

// ✅ Correct – asserting a call was never made
_orderRepositoryMock.Verify(
    r => r.DeleteAsync(It.IsAny(), It.IsAny()),
    Times.Never);

---

6. Parameterized Tests

Use [TestCase] for boundary values (null, empty, whitespace, zero, negative) rather than duplicating test logic:

[TestCase(null!)]
[TestCase("")]
[TestCase("   ")]
public async Task CreateAsync_WhenCustomerNameIsEmpty_ThenReturnsFailure(string customerName)
{
    // Arrange
    var request = new CreateOrderRequest { CustomerName = customerName };

    // Act
    var result = await _sut.CreateAsync(request, It.IsAny());

    // Assert
    Assert.That(result.IsValid, Is.False);
}

For multiple varying inputs, prefer [TestCaseSource] over stacking many [TestCase] attributes:

private static IEnumerable InvalidAmounts()
{
    yield return new TestCaseData(0m).SetName("Zero");
    yield return new TestCaseData(-1m).SetName("Negative");
    yield return new TestCaseData(decimal.MinValue).SetName("MinValue");
}

[TestCaseSource(nameof(InvalidAmounts))]
public void Validate_WhenAmountIsInvalid_ThenReturnsInvalidResult(decimal amount) { ... }

---

7. Mocking Best Practices

7.1 Mock Only What You Control

Only mock types that your code directly depends on (interfaces, abstract classes in your own codebase). Never mock types owned by third-party libraries or the .NET runtime.

7.2 Loose vs Strict Mocks

Use the default loose mock behavior (MockBehavior.Default). Use MockBehavior.Strict only when you need to assert that no unexpected calls are made.

7.3 Use It.IsAny() in SetUp

Configure broad default behaviours in [SetUp] using It.IsAny(). Narrow down to exact arguments only in tests that specifically need it:

// SetUp – broad default
_repositoryMock
    .Setup(r => r.GetByIdAsync(It.IsAny(), It.IsAny()))
    .ReturnsAsync((Order?)null);

// Individual test – specific input
_repositoryMock
    .Setup(r => r.GetByIdAsync(specificId, It.IsAny()))
    .ReturnsAsync(specificOrder);

7.4 Use NullLogger in Tests

Never mock ILogger. Use NullLogger.Instance from Microsoft.Extensions.Logging.Abstractions:

_sut = new OrderService(_repositoryMock.Object, NullLogger.Instance);

---

8. Test Organization

8.1 Test Project Naming

Each production project has a corresponding test project with .Tests appended to the name:

MyApp.Core        → MyApp.Core.Tests
MyApp.DataAccess  → MyApp.DataAccess.Tests

8.2 Folder Mirroring

Mirror the folder structure of the production project inside the test project. Each production class has a corresponding Tests.cs file in the same relative folder.

8.3 Test Data Builders

Extract repeated object construction into dedicated builder classes or static factory methods to keep tests readable:

internal static class OrderBuilder
{
    public static Order Build(Guid? id = null, string customerName = "Default Customer")
    {
        return new Order
        {
            Id = id ?? Guid.NewGuid(),
            CustomerName = customerName,
            TotalAmount = 100m,
        };
    }
}

---

Quick Reference Checklist

Before submitting any test file, verify:

• [ ] Test class is internal with [TestFixture]
• [ ] SUT is named _sut and initialized in [SetUp]
• [ ] All mocks declared as fields and initialized in [SetUp]
• [ ] Test method follows MethodName_WhenCondition_ThenExpectedBehavior
• [ ] Every test body has // Arrange, // Act, // Assert sections
• [ ] Assertions use Assert.That(...) constraint model
• [ ] ILogger is not mocked – NullLogger.Instance used instead
• [ ] [TestCase] used for boundary values, not duplicate test methods
• [ ] .Verify() used only for side-effect assertions, not return value checks

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: linuxchata
• Source: linuxchata/ai-playbook
• License: MIT

Install and usage instructions live in the source repository linked above.