这是我的合同:
[ServiceContract]
public interface IMyServiceContract {
[OperationContract]
OperationResponse1 Operation1(OperationRequest1 req);
[OperationContract]
OperationResponse2 Operation2(OperationRequest2 req);
}
OperationRequest1 和 OperationRequest2 都继承自 BaseOperationRequest,后者包含进入服务的所有请求的凭据信息:
[MessageContract]
public abstract class BaseOperationRequest {
[MessageHeader(MustUnderstand = true)]
public Guid Token { get; set; }
[MessageHeader(MustUnderstand = true)]
public string Password { get; set; }
private User User { get; set; }
}
OperationResponse1 和 OperationResponse2 都继承自一个基类:
[MessageContract]
public abstract class BaseOperationResponse {
[MessageHeader(MustUnderstand = true)]
public bool Success { get; set; }
[MessageHeader(MustUnderstand = true)]
public ServiceErrors ErrorCode { get; set; }
}
ErrorCode 是一个枚举。
正如您在请求中看到的那样,我有两个消息头和一个内部对象,它不会作为 SOAP 消息的一部分反序列化。这样做的原因是我想在我的服务实现处理它之前将此对象注入到请求中。每个操作实现都将使用这个对象,我不希望每个操作对我的数据层进行两次调用。
我想使用 WCF 可扩展性(通过属性)来执行两项任务:
- 验证请求用户。
- 使用复杂/复合业务对象在传入请求类上填充“用户”,以在每个操作中使用。
我调查了 IOperationInvoker、IDispatchMessageFormatter 和 IDispatchMessageInspector,但我没有发现它们中的任何一个都足够合适。
仅供参考,这是我的服务的原始示例实现,没有任何花哨的 WCF 可扩展性(或我的存储库/数据层调用):
public class MyService: IMyServiceContract {
public OperationResponse1 Operation1(OperationRequest1 req) {
if(req.Token == new Guid("GUID VALUE") && req.Password == "password") {
// Perform some actions....
return new OperationResponse1 {
Success = true
}
} else {
return new OperationResponse1 {
Success = false,
Error = "You are not authenticated"
}
}
}
public OperationResponse2 Operation2(OperationRequest2 req) {
if(req.Token == new Guid("GUID VALUE") && req.Password == "password") {
// Perform some actions....
return new OperationResponse2 {
Success = true
}
} else {
return new OperationResponse2 {
Success = false,
Error = "You are not authenticated"
}
}
}
}
IOperationInvoker 似乎是最合适的扩展点,但我不太清楚如何“取消”操作并覆盖对客户端的响应。这是我要去的地方:
/// <summary>
/// Provides an invoker that can be used to authenticate a BaseOperationRequest message.
/// </summary>
public class UserAuthenticationInvoker : IOperationInvoker {
/// <summary>
/// The original operation invoker.
/// </summary>
private IOperationInvoker _originalInvoker;
/// <summary>
/// The injected User service, for authentication.
/// </summary>
[Inject]
public IUserService UserService { get; set; }
public UserAuthenticationInvoker(IOperationInvoker originalInvoker) {
_originalInvoker = originalInvoker;
}
#region Implementation of IOperationInvoker {
public object[] AllocateInputs() {
return _originalInvoker.AllocateInputs();
}
public object Invoke(object instance, object[] inputs, out object[] outputs) {
// Validate base request
if(!(inputs[0] is BaseOperationRequest)) {
throw new ArgumentException("The request object must inherit from BaseOperationRequest in order for User authentication to take place.");
}
// Get BaseOperationRequest
var req = (BaseOperationRequest)inputs[0];
// Authenticate the User
var authResult = UserService.AuthenticateUser(new AuthenticateUserRequest {
Token = req.Token,
Password = req.Password
});
if(authResult.Success) {
// This is where I get stuck - do I need to modify "outputs"? If so, how do I tell the invoker that I want a particular response to be returned, and to cancel the rest of the operation?
return _originalInvoker.Invoke(instance, inputs, outputs);
}
return _originalInvoker.Invoke(instance, inputs, out outputs);
}
public IAsyncResult InvokeBegin(object instance, object[] inputs, AsyncCallback callback, object state) {
throw new NotImplementedException("The operation cannot be invoked asynchronously.");
}
public object InvokeEnd(object instance, out object[] outputs, IAsyncResult result) {
throw new NotImplementedException("The operation cannot be invoked asynchronously.");
}
public bool IsSynchronous {
get { return true; }
}
#endregion
}