Avoiding Duplicate Boilerplate Code in WCF Service

Question

I'm responsible for maintaining a web service project in c#. I have one service class with a bunch of methods that look a lot like this:

 [OperationContract]
 [WebInvoke(Method = "POST", UriTemplate = "v1/Login", ResponseFormat = WebMessageFormat.Json)]
 public string Login()
 {
 try
 {
 JObject incomingRequestJson = retrieveJson();
 JObject returningJson = new JObject();
 string accountId = incomingRequestJson.SelectToken("accountId", true).ToString();
 string username = incomingRequestJson.SelectToken("userName", true).ToString();
 string password = incomingRequestJson.SelectToken("password", true).ToString();
 LoginResult result = readerBLL.Login(accountId, username, password);
 returningJson.Add(new JProperty("userSessionId", result.userSessionId));
 returningJson.Add(new JProperty("friendlyUserName", result.friendlyUserName));
 return returningJson.ToString(Newtonsoft.Json.Formatting.None);
 }
 catch (Exception ex)
 {
 throw new WebFaultException<string>(ex.Message.ToString(), HttpStatusCode.OK);
 }
 }

Note that retrieveJson() grabs a JSON string from the HttpContext and runs JObject.Parse() on it. Note also that the consumer of this service is a Flex app, so I do need the WebFaultException business with the "OK" response code (The networking component in Flex treats connection timeouts and all HTTP error codes exactly the same - no further data other than

I don't like the way I have the try-catch block repeated in every method, so I tried making the following method:

private object invokeBusinessLogic(Func<JObject, object> f)
 {
 try
 {
 return f(retrieveJson());
 }
 catch (Exception ex)
 {
 WebFaultException<string> wfe = new WebFaultException<string>(ex.Message.ToString(), HttpStatusCode.OK);
 throw wfe;
 }
 }

Then my service methods can look like this:

public object Login()
{
 LoginResult result = (LoginResult) invokeBusinessLogic(readerBLL.Login);
 returningJson.Add(new JProperty("userSessionId", result.userSessionId));
 returningJson.Add(new JProperty("friendlyUserName", result.friendlyUserName));
 return returningJson.ToString(Newtonsoft.Json.Formatting.None);
}

However, this has the problem that the use of JSON to carry the request data gets exposed to the business layer. Is there a better way to avoid the duplicate try-catch blocks than this invokeBusinessLogic idea?

• svick: Why are you parsing and creating the JSONs instead of using serialization? I think that would simplify your code quite a lot.
• Brian: Historical reasons. I'd love to change it, but then the clients of this service would need to be updated. That's something for version 2 of this service.
• svick: Why would that affect the clients? You could still serialize to the same JSON.
• Brian: Oh, perhaps I misunderstood you. I thought doing serialization would mean function signatures like public LoginResult Login() in the service layer -- and that would break my clients. The current code actually sends an escaped string rather than a real JSON string :/

Answer 1

I'm not sure why your change all of a sudden would expose data to the business layer when it didn't before.

If you have a pattern of methods which look like this:

public string MyServiceCall()
{
 try
 {
 JObject incomingRequestJson = retrieveJson();
 JObject returningJson = new JObject();
 // do stuff
 return returningJson.ToString(Newtonsoft.Json.Formatting.None);
 }
 catch (Exception ex)
 {
 throw new WebFaultException<string>(ex.Message.ToString(), HttpStatusCode.OK);
 }
}

Then you can simply add a helper method to your service implementation:

private string ProcessServiceCall(Func<JObject, JObject> processor)
{
 try
 {
 JObject incomingRequestJson = retrieveJson();
 JObject returningJson = processor(incomingRequestJson);
 return returningJson.ToString(Newtonsoft.Json.Formatting.None);
 }
 catch (Exception ex)
 {
 throw new WebFaultException<string>(ex.Message.ToString(), HttpStatusCode.OK);
 }
}

and your service call then becomes:

public string MyServiceCall()
{
 ProcessServiceCall(incomingRequestJson => {
 string accountId = incomingRequestJson.SelectToken("accountId", true).ToString();
 string username = incomingRequestJson.SelectToken("userName", true).ToString();
 string password = incomingRequestJson.SelectToken("password", true).ToString();
 JObject result = new JObject();
 LoginResult result = readerBLL.Login(accountId, username, password);
 result.Add(new JProperty("userSessionId", result.userSessionId));
 result.Add(new JProperty("friendlyUserName", result.friendlyUserName));
 return result;
 }
}

This should be a simple refactoring on your service class and not expose any more json to the business logic than before.

• Brian: Thanks, this is perfect. I hadn't thought of doing an anonymous function to pass to ProcessServiceCall.
• Kiquenet: Maybe reuse source code (common logic, implementation) for WebAPI, WCF Service (JSON), WCF Service (SOAP-XML)? IMHO, better samples for minimize learning curve are real applications with full source code and good patterns