我已经创建了一个包，Moq。ILogger，使测试ILogger扩展更容易。

实际上，您可以使用类似下面这样更接近实际代码的代码。

loggerMock.VerifyLog(c => c.LogInformation(
                 "Index page say hello", 
                 It.IsAny<object[]>());

它不仅更容易编写新的测试，而且维护也没有成本。

回购可以在这里找到，也有一个nuget包(Install-Package ILogger.Moq)。

我也在博客上用一个现实生活中的例子解释了这一点。

简而言之，假设你有以下代码:

public class PaymentsProcessor
{
    private readonly IOrdersRepository _ordersRepository;
    private readonly IPaymentService _paymentService;
    private readonly ILogger<PaymentsProcessor> _logger;

    public PaymentsProcessor(IOrdersRepository ordersRepository, 
        IPaymentService paymentService, 
        ILogger<PaymentsProcessor> logger)
    {
        _ordersRepository = ordersRepository;
        _paymentService = paymentService;
        _logger = logger;
    }

    public async Task ProcessOutstandingOrders()
    {
        var outstandingOrders = await _ordersRepository.GetOutstandingOrders();
        
        foreach (var order in outstandingOrders)
        {
            try
            {
                var paymentTransaction = await _paymentService.CompletePayment(order);
                _logger.LogInformation("Order with {orderReference} was paid {at} by {customerEmail}, having {transactionId}", 
                                       order.OrderReference, 
                                       paymentTransaction.CreateOn, 
                                       order.CustomerEmail, 
                                       paymentTransaction.TransactionId);
            }
            catch (Exception e)
            {
                _logger.LogWarning(e, "An exception occurred while completing the payment for {orderReference}", 
                                   order.OrderReference);
            }
        }
        _logger.LogInformation("A batch of {0} outstanding orders was completed", outstandingOrders.Count);
    }
}

然后您可以编写一些测试，例如

[Fact]
public async Task Processing_outstanding_orders_logs_batch_size()
{
    // Arrange
    var ordersRepositoryMock = new Mock<IOrdersRepository>();
    ordersRepositoryMock.Setup(c => c.GetOutstandingOrders())
        .ReturnsAsync(GenerateOutstandingOrders(100));

    var paymentServiceMock = new Mock<IPaymentService>();
    paymentServiceMock
        .Setup(c => c.CompletePayment(It.IsAny<Order>()))
        .ReturnsAsync((Order order) => new PaymentTransaction
        {
            TransactionId = $"TRX-{order.OrderReference}"
        });

    var loggerMock = new Mock<ILogger<PaymentsProcessor>>();

    var sut = new PaymentsProcessor(ordersRepositoryMock.Object, paymentServiceMock.Object, loggerMock.Object);

    // Act
    await sut.ProcessOutstandingOrders();

    // Assert
    loggerMock.VerifyLog(c => c.LogInformation("A batch of {0} outstanding orders was completed", 100));
}

[Fact]
public async Task Processing_outstanding_orders_logs_order_and_transaction_data_for_each_completed_payment()
{
    // Arrange
    var ordersRepositoryMock = new Mock<IOrdersRepository>();
    ordersRepositoryMock.Setup(c => c.GetOutstandingOrders())
        .ReturnsAsync(GenerateOutstandingOrders(100));

    var paymentServiceMock = new Mock<IPaymentService>();
    paymentServiceMock
        .Setup(c => c.CompletePayment(It.IsAny<Order>()))
        .ReturnsAsync((Order order) => new PaymentTransaction
        {
            TransactionId = $"TRX-{order.OrderReference}"
        });

    var loggerMock = new Mock<ILogger<PaymentsProcessor>>();

    var sut = new PaymentsProcessor(ordersRepositoryMock.Object, paymentServiceMock.Object, loggerMock.Object);

    // Act
    await sut.ProcessOutstandingOrders();

    // Assert
    loggerMock.VerifyLog(logger => logger.LogInformation("Order with {orderReference} was paid {at} by {customerEmail}, having {transactionId}",
        It.Is<string>(orderReference => orderReference.StartsWith("Reference")),
        It.IsAny<DateTime>(),
        It.Is<string>(customerEmail => customerEmail.Contains("@")),
        It.Is<string>(transactionId => transactionId.StartsWith("TRX"))),
      Times.Exactly(100));
}

[Fact]
public async Task Processing_outstanding_orders_logs_a_warning_when_payment_fails()
{
    // Arrange
    var ordersRepositoryMock = new Mock<IOrdersRepository>();
    ordersRepositoryMock.Setup(c => c.GetOutstandingOrders())
        .ReturnsAsync(GenerateOutstandingOrders(2));

    var paymentServiceMock = new Mock<IPaymentService>();
    paymentServiceMock
        .SetupSequence(c => c.CompletePayment(It.IsAny<Order>()))
        .ReturnsAsync(new PaymentTransaction
        {
            TransactionId = "TRX-1",
            CreateOn = DateTime.Now.AddMinutes(-new Random().Next(100)),
        })
        .Throws(new Exception("Payment exception"));

    var loggerMock = new Mock<ILogger<PaymentsProcessor>>();

    var sut = new PaymentsProcessor(ordersRepositoryMock.Object, paymentServiceMock.Object, loggerMock.Object);

    // Act
    await sut.ProcessOutstandingOrders();

    // Assert
    loggerMock.VerifyLog(c => c.LogWarning(
                 It.Is<Exception>(paymentException => paymentException.Message.Contains("Payment exception")), 
                 "*exception*Reference 2"));
}

前面已经提到过，您可以将其模拟为任何其他接口。

var logger = new Mock<ILogger<QueuedHostedService>>();

到目前为止一切顺利。

好的一点是，您可以使用Moq来验证某些调用是否已经执行。例如，在这里，我检查日志已经被一个特定的异常调用。

logger.Verify(m => m.Log(It.Is<LogLevel>(l => l == LogLevel.Information), 0,
            It.IsAny<object>(), It.IsAny<TaskCanceledException>(), It.IsAny<Func<object, Exception, string>>()));

在使用Verify时，重点是针对ILooger接口中的真实Log方法，而不是扩展方法进行验证。

仅仅创建一个虚拟的ILogger对于单元测试来说没有多大价值。您还应该验证是否进行了日志记录调用。您可以使用Moq注入一个模拟ILogger，但验证调用可能有点棘手。本文将深入探讨如何使用Moq进行验证。

下面是文章中一个非常简单的例子:

_loggerMock.Verify(l => l.Log(
LogLevel.Information,
It.IsAny<EventId>(),
It.IsAny<It.IsAnyType>(),
It.IsAny<Exception>(),
(Func<It.IsAnyType, Exception, string>)It.IsAny<object>()), Times.Exactly(1));

它验证是否记录了信息消息。但是，如果我们想验证关于消息的更复杂的信息，比如消息模板和命名属性，这就变得更加棘手了:

_loggerMock.Verify
(
    l => l.Log
    (
        //Check the severity level
        LogLevel.Error,
        //This may or may not be relevant to your scenario
        It.IsAny<EventId>(),
        //This is the magical Moq code that exposes internal log processing from the extension methods
        It.Is<It.IsAnyType>((state, t) =>
            //This confirms that the correct log message was sent to the logger. {OriginalFormat} should match the value passed to the logger
            //Note: messages should be retrieved from a service that will probably store the strings in a resource file
            CheckValue(state, LogTest.ErrorMessage, "{OriginalFormat}") &&
            //This confirms that an argument with a key of "recordId" was sent with the correct value
            //In Application Insights, this will turn up in Custom Dimensions
            CheckValue(state, recordId, nameof(recordId))
    ),
    //Confirm the exception type
    It.IsAny<NotImplementedException>(),
    //Accept any valid Func here. The Func is specified by the extension methods
    (Func<It.IsAnyType, Exception, string>)It.IsAny<object>()),
    //Make sure the message was logged the correct number of times
    Times.Exactly(1)
);

我相信您可以用其他模拟框架做同样的事情，但是ILogger接口确保了这很困难。

使用NullLogger -什么都不做的极简日志记录器。

public interface ILoggingClass
{
   public void LogCritical(Exception exception);
}

public class LoggingClass : ILoggingClass
{
    private readonly ILogger<LoggingClass> logger;

    public LoggingClass(ILogger<LoggingClass> logger) =>
            this.logger = logger;

    public void LogCritical(Exception exception) =>
        this.logger.LogCritical(exception, exception.Message);
}

在测试方法的使用中，

ILogger<LoggingClass> logger = new NullLogger<LoggingClass>();
LoggingClass loggingClass = new LoggingClass(logger);

并将loggingClass传递给服务进行测试。

像其他答案建议的那样，传递模拟ILogger很容易，但是验证调用实际上是对logger进行的突然变得更加困难。原因是大多数调用实际上并不属于ILogger接口本身。

因此，大多数调用都是调用接口唯一Log方法的扩展方法。原因似乎是，如果您只有一个而不是很多重载，那么可以更容易地实现接口，从而归结为相同的方法。

当然，缺点是突然间很难验证呼叫是否已发出，因为您应该验证的呼叫与您所发出的呼叫非常不同。有一些不同的方法可以解决这个问题，我发现用于模拟框架的自定义扩展方法将使其更容易编写。

下面是我用NSubstitute做的一个方法的例子:

public static class LoggerTestingExtensions
{
    public static void LogError(this ILogger logger, string message)
    {
        logger.Log(
            LogLevel.Error,
            0,
            Arg.Is<FormattedLogValues>(v => v.ToString() == message),
            Arg.Any<Exception>(),
            Arg.Any<Func<object, Exception, string>>());
    }

}

下面是它的用法:

_logger.Received(1).LogError("Something bad happened");   

它看起来就像您直接使用了这个方法，这里的技巧是我们的扩展方法获得了优先级，因为它在名称空间上比原来的方法“更接近”，所以它将被使用。

不幸的是，它并没有给出我们想要的100%，即错误消息将不会那么好，因为我们不直接检查字符串，而是检查涉及字符串的lambda，但95%总比没有好:)此外，这种方法将使测试代码

P.S.对于Moq，可以使用为Mock<ILogger<T>>编写扩展方法的方法来验证，以实现类似的结果。

P.P.S.这在。net Core 3中不再工作，查看这个线程了解更多细节:https://github.com/nsubstitute/NSubstitute/issues/597#issuecomment-573742574

使用一个使用ITestOutputHelper(来自xunit)的自定义记录器来捕获输出和日志。下面是一个只将状态写入输出的小示例。

public class XunitLogger<T> : ILogger<T>, IDisposable
{
    private ITestOutputHelper _output;

    public XunitLogger(ITestOutputHelper output)
    {
        _output = output;
    }
    public void Log<TState>(LogLevel logLevel, EventId eventId, TState state, Exception exception, Func<TState, Exception, string> formatter)
    {
        _output.WriteLine(state.ToString());
    }

    public bool IsEnabled(LogLevel logLevel)
    {
        return true;
    }

    public IDisposable BeginScope<TState>(TState state)
    {
        return this;
    }

    public void Dispose()
    {
    }
}

在单元测试中使用它

public class BlogControllerTest
{
  private XunitLogger<BlogController> _logger;

  public BlogControllerTest(ITestOutputHelper output){
    _logger = new XunitLogger<BlogController>(output);
  }

  [Fact]
  public void Index_ReturnAViewResult_WithAListOfBlog()
  {
    var mockRepo = new Mock<IDAO<Blog>>();
    mockRepo.Setup(repo => repo.GetMany(null)).Returns(GetListBlog());
    var controller = new BlogController(_logger,mockRepo.Object);
    // rest
  }
}