tsp-csharp-xunit

name: tsp-csharp-xunit description: "Comprehensive xUnit testing guide for C# projects. Use when writing unit tests, setting up test projects, creating data-driven tests, mocking dependencies, or organizing test suites."

xUnit Testing — TSP Standards

Project Setup

• Test project name: {ProjectName}.Tests
• Required packages: Microsoft.NET.Test.Sdk, xunit, xunit.runner.visualstudio
• Recommended: FluentAssertions, NSubstitute (or Moq)
• Run tests: dotnet test (use --filter for targeted runs)

Test Structure

• No test class attribute required (unlike MSTest/NUnit)
• Use constructor for per-test setup; implement IDisposable for teardown
• Use IClassFixture<T> for shared context within a test class
• Use ICollectionFixture<T> for shared context across test classes
• Use IAsyncLifetime for async setup/teardown

public class OrderServiceTests : IClassFixture<DatabaseFixture>, IDisposable
{
    private readonly DatabaseFixture _db;

    public OrderServiceTests(DatabaseFixture db)
    {
        _db = db;
    }

    [Fact]
    public void CreateOrder_WithValidItems_ReturnsOrderId()
    {
        // Arrange
        var service = new OrderService(_db.Context);
        var items = new[] { new OrderItem("SKU-1", 2) };

        // Act
        var result = service.CreateOrder(items);

        // Assert
        result.Should().BeGreaterThan(0);
    }

    public void Dispose() { /* cleanup */ }
}

Naming Conventions

• Test class: {ClassUnderTest}Tests
• Test method: {Method}_{Scenario}_{ExpectedResult}
• Examples:
  • GetUser_WhenIdIsInvalid_ReturnsNotFound
  • CalculateTotal_WithDiscount_AppliesPercentage
  • SaveAsync_WhenCancelled_ThrowsOperationCancelled

Data-Driven Tests

• [Theory] + [InlineData] for simple inline values
• [Theory] + [MemberData] for method/property-based data
• [Theory] + [ClassData] for reusable data generators
• Use meaningful parameter names that describe the scenario

[Theory]
[InlineData(0, false)]
[InlineData(5, true)]
[InlineData(-1, false)]
public void IsValidQuantity_ReturnsExpected(int quantity, bool expected)
{
    var result = OrderValidator.IsValidQuantity(quantity);

    result.Should().Be(expected);
}

Assertions

• Prefer FluentAssertions when available:
  • result.Should().Be(expected)
  • collection.Should().HaveCount(3).And.Contain(x => x.IsActive)
  • act.Should().ThrowAsync<InvalidOperationException>()
• Built-in xUnit assertions as fallback:
  • Assert.Equal, Assert.True, Assert.Throws<T>
  • Assert.Contains / Assert.DoesNotContain for collections
• One logical behavior per test — multiple assertions are fine if testing one concept

Mocking

• Prefer NSubstitute for readability; Moq is also acceptable
• Mock interfaces, not concrete classes
• Verify interactions only when the interaction IS the behavior being tested
• Avoid over-mocking — if you're mocking more than 3 dependencies, the class may need refactoring

var repo = Substitute.For<IOrderRepository>();
repo.GetByIdAsync(42, Arg.Any<CancellationToken>())
    .Returns(new Order { Id = 42 });

var service = new OrderService(repo);
var result = await service.GetOrderAsync(42, CancellationToken.None);

result.Should().NotBeNull();
result.Id.Should().Be(42);

Test Organization

• Group tests by feature or component
• Use [Trait("Category", "Integration")] for categorization
• Skip tests with Skip = "reason" when conditionally disabled
• Use ITestOutputHelper for diagnostic output (not Console.WriteLine)

Rules

• Tests must run in any order and in parallel — no shared mutable state
• Avoid disk I/O — use in-memory alternatives or mocks
• No Thread.Sleep — use Task.Delay with cancellation tokens in async tests
• Every new or changed public API must have corresponding tests