WebAPI的一种单元测试方案

大家是如何对webApi写测试的呢？

1.利用Fiddler直接做请求，观察response的内容。

2.利用Httpclient做请求，断言response的内容。

3.直接调用webApi的action,这种方式的测试跟真实的调用还是有一定差距，不够完美。

接下来我介绍一种webApi的in-memory调用方法，也能够达到对webApi的测试，并且由于是in-memory调用，效率也比较高，非常适写单元测试。本文参考了In memory client, host and integration testing of your Web API service。

一、首先写一个OrderController用来做测试用

public class OrderController : ApiController
{
// GET api/order
public Order Get()
{
return new Order(){Id = 1,Descriptions = "descriptions",Name = "name"};
}

// GET api/order/5
public string Get(int id)
{
return "value";
}

// POST api/order
public Order Post(Order order)
{
return order;
}

// DELETE api/order/5
public void Delete(int id)
{
}
}

二、WebApi的请求过程

webApi的核心是对消息的管道处理，整个核心是有一系列消息处理器(HttpMessageHandler)首尾连接的双向管道，管道头为HttpServer,管道尾为HttpControllerDispatcher,HttpControllerDispatcher负责对controller的激活和action的执行，然后相应的消息逆向流出管道。

所以我们可以利用HttpMessageInvoker将一个请求消息HttpRequestMessage发送到管道中，最后收到的消息HttpResponseMessage就代表一个真实的请求响应。

三、Get请求的测试

[Test]
public void GetTest()
{
string baseAddress = "http://localhost:33203/";

HttpConfiguration config = new HttpConfiguration();
WebApiConfig.Register(config);
config.IncludeErrorDetailPolicy = IncludeErrorDetailPolicy.Always;
HttpServer server = new HttpServer(config);
HttpMessageInvoker messageInvoker = new HttpMessageInvoker(server);
CancellationTokenSource cts = new CancellationTokenSource();
HttpRequestMessage request = new HttpRequestMessage(HttpMethod.Get, baseAddress + "api/order");

using (HttpResponseMessage response = messageInvoker.SendAsync(request, cts.Token).Result)
{
var content = response.Content.ReadAsStringAsync().Result;
var result = JsonConvert.DeserializeObject<Order>(content);

result.Name.Should().Be("name");
}
}

四、Post请求的测试

[Test]
public void PostTest()
{
string baseAddress = "http://localhost:33203/";

HttpConfiguration config = new HttpConfiguration();
WebApiConfig.Register(config);
config.IncludeErrorDetailPolicy = IncludeErrorDetailPolicy.Always;
HttpServer server = new HttpServer(config);
HttpMessageInvoker messageInvoker = new HttpMessageInvoker(server);
CancellationTokenSource cts = new CancellationTokenSource();
HttpRequestMessage request = new HttpRequestMessage(HttpMethod.Post, baseAddress + "api/order");
var order = new Order() { Id = 1, Name = "orderName", Descriptions = "orderDescriptions" };
request.Content = new ObjectContent<Order>(order, new JsonMediaTypeFormatter());
using (HttpResponseMessage response = messageInvoker.SendAsync(request, cts.Token).Result)
{
var content = JsonConvert.SerializeObject(order, new JsonSerializerSettings() { ContractResolver = new CamelCasePropertyNamesContractResolver() });
response.Content.ReadAsStringAsync().Result.Should().Be(content);
}
}

四、重构

可以看到这两个测试大部分的代码是相同的，都是用来发送请求。因此我们提取一个webApiTestBase类，该基类可以提供InvokeGetRequest,InvokePostRequest,InvokePutRequest等方法

public abstract class ApiTestBase
{
public abstract string GetBaseAddress();

protected TResult InvokeGetRequest<TResult>(string api)
{
using (var invoker = CreateMessageInvoker())
{
using (var cts = new CancellationTokenSource())
{
var request = new HttpRequestMessage(HttpMethod.Get, GetBaseAddress() + api);
using (HttpResponseMessage response = invoker.SendAsync(request, cts.Token).Result)
{
var result = response.Content.ReadAsStringAsync().Result;
return JsonConvert.DeserializeObject<TResult>(result);
}
}
}
}

protected TResult InvokePostRequest<TResult, TArguemnt>(string api, TArguemnt arg)
{
var invoker = CreateMessageInvoker();
using (var cts = new CancellationTokenSource())
{
var request = new HttpRequestMessage(HttpMethod.Post, GetBaseAddress() + api);
request.Content = new ObjectContent<TArguemnt>(arg, new JsonMediaTypeFormatter());
using (HttpResponseMessage response = invoker.SendAsync(request, cts.Token).Result)
{
var result = response.Content.ReadAsStringAsync().Result;
return JsonConvert.DeserializeObject<TResult>(result);
}
}
}

private HttpMessageInvoker CreateMessageInvoker()
{
var config = new HttpConfiguration();
WebApiConfig.Register(config);
var server = new HttpServer(config);
var messageInvoker = new HttpMessageInvoker(server);
return messageInvoker;
}
}

有了这个基类，我们写测试只需要重写方法GetBaseAddress()，然后直接调用基类方法并进行断言即可

[TestFixture]
public class OrderApiTests:ApiTestBase
{
public override string GetBaseAddress()
{
return "http://localhost:33203/";
}

[Test]
public void Should_get_order_successfully()
{
var result = InvokeGetRequest<Order>("api/order");

result.Name.Should().Be("name");
result.Descriptions.Should().Be("descriptions");
result.Id.Should().Be(1);
}

[Test]
public void Should_post_order_successfully()
{
var newOrder=new Order(){Name = "newOrder",Id = 100,Descriptions = "new-order-description"};

var result = InvokePostRequest<Order,Order>("api/order", newOrder);

result.Name.Should().Be("newOrder");
result.Id.Should().Be(100);
result.Descriptions.Should().Be("new-order-description");
}
}

是不是干净多了。

这种in-memory的测试方案有什么优点和缺点呢？

优点：

1.模拟真实调用，需要传入api地址即可得到结果，由于整个调用是in-memory的，所有效率很高，很适合集成测试。

2.整个测试时可以调试的，可以直接从单元测试调试进去，如果你写一个httpClient的测试，需要把webApi启动起来，然后。。。麻烦

缺点：我觉得原文作者说的那些缺点都可以忽略不计。