在xunit中模拟不同端点的HttpClient

Question

我正在嘲笑HttpClient按照这个SO建议进行单元测试。它对单个端点起到了预期的作用，我想知道如何为不同的端点调整它。请参阅以下示例：

class Agent
{
private HttpClient _client;
private string _baseUri = "http://example.com/";
public Agent(HttpClient client)
{ _client = client; }
public bool Run()
{
var res1 = client.GetAsync(new Uri(_baseUri, "endpoint1"));
var res2 = client.GetAsync(new Uri(_baseUri, "endpoint2"));
return res1 == res2
}
}
// In the test method:
var responseMessage = new HttpResponseMessage
{
StatusCode = HttpStatusCode.OK,
Content = new StringContent("test_return")
};
var mock = new Mock<HttpMessageHandler>();
mock.Protected()
.Setup<Task<HttpResponseMessage>>(
"SendAsync",
ItExpr.IsAny<HttpRequestMessage>(),
ItExpr.IsAny<CancellationToken>())
.ReturnsAsync(responseMessage);
client = new HttpClient(mock.Object);
var agent = new Agent(client);
var resp = agent.Run();
Assert.True(resp)

在上面的例子中，resp将始终是true，因为作为模拟的结果，Run方法中来自两个端点的响应将相等。

Answer 1

我认为您应该设置两个SendAsync调用，如下所示。

// response to /endpoint1
var responseMessage1 = new HttpResponseMessage
{
StatusCode = HttpStatusCode.OK,
Content = new StringContent("test_return")
};

// response to /endpoint2
var responseMessage2 = new HttpResponseMessage
{
StatusCode = HttpStatusCode.OK,
Content = new StringContent("test_other_return")
};

var mock = new Mock<HttpMessageHandler>();

// mock a call to /endpoint1
mock.Protected()
.Setup<Task<HttpResponseMessage>>(
"SendAsync",
ItExpr.Is<HttpRequestMessage>(
m => m.RequestUri.AbsolutePath.Contains(
"endpoint1")), 
ItExpr.IsAny<CancellationToken>())
.ReturnsAsync(responseMessage1);

// mock a call to /endpoint2
mock.Protected()
.Setup<Task<HttpResponseMessage>>(
"SendAsync",
ItExpr.Is<HttpRequestMessage>(
m => m.RequestUri.AbsolutePath.Contains(
"endpoint2")), 
ItExpr.IsAny<CancellationToken>())
.ReturnsAsync(responseMessage2);

var client = new HttpClient(mock.Object);

var agent = new Agent(client);
var resp = agent.Run();