Category: Design Patterns

Design Patterns

Creating a Chatbot using Amazon Lex Service: Day 1 (Initial Bot)

In this tutorial series, we’ll discuss how to create a real-time chatbot using Amazon Lex service and deploy the bot over Facebook messenger.

Advertisements

Introduction

In this tutorial series, we’ll discuss how to create a real-time chatbot using Amazon Lex service and deploy the bot over Facebook messenger. The reader of the tutorial series does not need to have prior knowledge of bot creation as the article will demonstrate the bot creation and deployment step by step using explanatory pictures. Trust me, it is fun and exciting to create your own bot and see it working over the thought of creating a bot. We’ll setup AWS account to leverage Lex Services and Facebook developers account to deploy the bot on your own Facebook page and see it working. Let’s jump directly into the tutorial.

Continue reading “Creating a Chatbot using Amazon Lex Service: Day 1 (Initial Bot)”

Singleton Design Pattern In C# – Part Three (Static vs Singleton)


Introduction

In this series on learning singleton pattern, we learned lazy initialization and eager initialization with practical examples. We also learned why it is necessary to make the singleton class sealed with the sealed keyword. In this article, I’ll try to explain the differences between static and singleton class. We will also see where to use static class and where to use singleton classes.

Continue reading “Singleton Design Pattern In C# – Part Three (Static vs Singleton)”

Singleton Design Pattern In C# – Part 2 (Eager and Lazy Initialization in Singleton)


Introduction

In the previous article on learning singleton pattern, we discussed what is singleton design pattern, when is it needed and what problems it solves. We also discussed how to create a basic singleton class step by step and how to enhance that class to be thread-safe with the help of locking and performance effective with the help of double check locking. In this article, we’ll discuss Lazy initialization, the lazy keyword, why to make singleton class a sealed class, and what are the differences between singleton and static class. Before we start, I strongly recommend you go through my last article.

Continue reading “Singleton Design Pattern In C# – Part 2 (Eager and Lazy Initialization in Singleton)”

Singleton Design Pattern in C# – Part 1


Introduction

I always wanted to write on Singleton design pattern in C#. Though there already are many posts available on Singleton design pattern, I’ll try to cover this topic in the most simplistic and easy to understand way.

Continue reading “Singleton Design Pattern in C# – Part 1”

RESTful Day #6: Request logging and Exception handling/logging in Web APIs using Action Filters, Exception Filters and NLog


Download Source Code from GitHub

Introduction

We have been learning a lot about WebAPI, its uses, implementations, and security aspects since last five articles of the series. This article of the series will explain how we can handle requests and log them for tracking and for the sake of debugging, how we can handle exceptions and log them. We’ll follow a centralized way of handling exceptions in WebAPI and write our custom classes to be mapped to the type of exception that we encounter and log the accordingly. I’ll use NLog to log requests and exceptions as well. We’ll leverage the capabilities of Exception Filters and Action Filters to centralize request logging and exception handling in WebAPI.

Roadmap

The following is the roadmap I have setup to learn WebAPI step by step,

 

I’ll purposely use Visual Studio 2010 and .NET Framework 4.0 because there are few implementations that are very hard to find in .NET Framework 4.0, but I’ll make it easy by showing how we can do it.

Request Logging

Since we are writing web services, we are exposing our end points. We must know where the requests are coming from and what requests are coming to our server. Logging could be very beneficial and helps us in a lot of ways like, debugging, tracing, monitoring and analytics.

We already have an existing design. If you open the solution, you’ll get to see the structure as mentioned below or one can also implement this approach in their existing solution as well.

Setup NLog in WebAPI

NLog serves various purposes but primarily logging. We’ll use NLog for logging into files and windows event as well. You can read more about NLog at http://NLog-project.org/
One can use the sample application that we used in Day#5 or can have any other application as well. I am using the existing sample application that we were following throughout all the parts of this series. Our application structure looks something like:

Step 1: Download NLog Package

Right click WebAPI project and select manage Nuget Packages from the list. When the Nuget Package Manager appears, search for NLog. You’ll get Nlog like shown in image below, just install it to our project.

After adding this you will find following NLog dll referenced in your application.

Step 2: Configuring NLog

To configure NLog with application add following settings in our existing WebAPI web.config file,

ConfigSection –

Configuration Section – I have added the section to configuration and defined the path and format dynamic target log file name, also added the eventlog source to Api Services.

As mentioned in above target path, I have also created to “APILog” folder in the base directory of application.

Now we have configured the NLog in our application, and it is ready to start work for request logging. Note that in the rules section we have defined rules for logging in files as well as in windows events log as well, you can choose both of them or can opt for one too. Let’s start with logging request in application, with action filters –

NLogger Class

Add a folder “Helpers” in the API, which will segregate the application code for readability, better understanding and maintainability.

To start add our main class “NLogger“, which will responsible for all types of errors and info logging, to same Helper folder. Here NLogger class implements ITraceWriter interface, which provides “Trace” method for the service request.
#region Using namespaces.
using System;
using System.Collections.Generic;
using System.Linq;
using System.Web;
using System.Web.Http.Tracing;
using NLog;
using System.Net.Http;
using System.Text;
using WebApi.ErrorHelper;
#endregion

namespace WebApi.Helpers
{
    /// <summary>
    /// Public class to log Error/info messages to the access log file
    /// </summary>
    public sealed class NLogger : ITraceWriter
    {
        #region Private member variables.
        private static readonly Logger ClassLogger = LogManager.GetCurrentClassLogger();

        private static readonly Lazy<Dictionary<TraceLevel, Action<string>>> LoggingMap = new Lazy<Dictionary<TraceLevel, Action<string>>>(() => new Dictionary<TraceLevel, Action<string>> { { TraceLevel.Info, ClassLogger.Info }, { TraceLevel.Debug, ClassLogger.Debug }, { TraceLevel.Error, ClassLogger.Error }, { TraceLevel.Fatal, ClassLogger.Fatal }, { TraceLevel.Warn, ClassLogger.Warn } });
        #endregion

        #region Private properties.
        /// <summary>
        /// Get property for Logger
        /// </summary>
        private Dictionary<TraceLevel, Action<string>> Logger
        {
            get { return LoggingMap.Value; }
        }
        #endregion

        #region Public member methods.
        /// <summary>
        /// Implementation of TraceWriter to trace the logs.
        /// </summary>
        /// <param name="request"></param>
        /// <param name="category"></param>
        /// <param name="level"></param>
        /// <param name="traceAction"></param>
        public void Trace(HttpRequestMessage request, string category, TraceLevel level, Action traceAction)
        {
            if (level != TraceLevel.Off)
            {
                if (traceAction != null && traceAction.Target != null)
                {
                    category = category + Environment.NewLine + "Action Parameters : " + traceAction.Target.ToJSON();
                }
                var record = new TraceRecord(request, category, level);
                if (traceAction != null) traceAction(record);
                Log(record);
            }
        }
        #endregion

        #region Private member methods.
        /// <summary>
        /// Logs info/Error to Log file
        /// </summary>
        /// <param name="record"></param>
        private void Log(TraceRecord record)
        {
            var message = new StringBuilder();

            if (!string.IsNullOrWhiteSpace(record.Message))
                message.Append("").Append(record.Message + Environment.NewLine);

            if (record.Request != null)
            {
                if (record.Request.Method != null)
                    message.Append("Method: " + record.Request.Method + Environment.NewLine);

                if (record.Request.RequestUri != null)
                    message.Append("").Append("URL: " + record.Request.RequestUri + Environment.NewLine);

                if (record.Request.Headers != null && record.Request.Headers.Contains("Token") && record.Request.Headers.GetValues("Token") != null && record.Request.Headers.GetValues("Token").FirstOrDefault() != null)
                    message.Append("").Append("Token: " + record.Request.Headers.GetValues("Token").FirstOrDefault() + Environment.NewLine);
            }

            if (!string.IsNullOrWhiteSpace(record.Category))
                message.Append("").Append(record.Category);

            if (!string.IsNullOrWhiteSpace(record.Operator))
                message.Append(" ").Append(record.Operator).Append(" ").Append(record.Operation);

            
            Logger[record.Level](Convert.ToString(message) + Environment.NewLine);
        }
        #endregion
    }
}

Adding Action Filter

Action filter will be responsible for handling all the incoming requests to our APIs and logging them using NLogger class. We have “OnActionExecuting” method that is implicitly called if we mark our controllers or global application to use that particular filter. So each time any action of any controller will be hit, our “OnActionExecuting” method will execute to log the request.

Step 1: Adding LoggingFilterAttribute class

Create a class LoggingFilterAttribute to “ActionFilters” folder and add following code.
using System;
using System.Collections.Generic;
using System.Linq;
using System.Web;
using System.Web.Http.Filters;
using System.Web.Http.Controllers;
using System.Web.Http.Tracing;
using System.Web.Http;
using WebApi.Helpers;


namespace WebApi.ActionFilters
{
    public class LoggingFilterAttribute : ActionFilterAttribute
    {
        public override void OnActionExecuting(HttpActionContext filterContext)
        {
            GlobalConfiguration.Configuration.Services.Replace(typeof(ITraceWriter), new NLogger());
            var trace = GlobalConfiguration.Configuration.Services.GetTraceWriter();
            trace.Info(filterContext.Request, "Controller : " + filterContext.ControllerContext.ControllerDescriptor.ControllerType.FullName + Environment.NewLine + "Action : " + filterContext.ActionDescriptor.ActionName, "JSON", filterContext.ActionArguments);
        }
    }
}
The LoggingFilterAttribute class derived from ActionFilterAttribute, which is underSystem.Web.Http.Filters and overriding the OnActionExecuting method.
Here I have replaced the default “ITraceWriter” service with our NLogger class instance in the controller’s service container. Now GetTraceWriter() method will return our instance (instance NLogger class) and Info() will call trace() method of our NLogger class.
Note that the code below.
GlobalConfiguration.Configuration.Services.Replace(typeof(ITraceWriter), new NLogger());
is used to resolve dependency between ITaceWriter and NLogger class. Thereafter we use a variable namedtrace to get the instance and trace.Info() is used to log the request and whatever text we want to add along with that request.

Step 2: Registering Action Filter (LoggingFilterAttribute)

In order to register the created action filter to application’s filters, just add a new instance of your action filter toconfig.Filters in WebApiConfig class.
using System.Web.Http;
using WebApi.ActionFilters;

namespace WebApi.App_Start
{
    public static class WebApiConfig
    {
        public static void Register(HttpConfiguration config)
        {
            config.Filters.Add(new LoggingFilterAttribute());
        }
    }
}
Now this action filter is applicable to all the controllers and actions in our project. You may not believe but request logging is done. It’s time to run the application and validate our homework.
Image credit: https://pixabay.com/en/social-media-network-media-54536/

Running the application

Let’s run the application and try to make a call, using token based authorization, we have already covered authorization in day#5. You first need to authenticate your request using login service and then that service will return a token for making calls to other services. Use that token to make calls to other services. For more details you can read day 5 of this series.
Just run the application, we get
We already have our test client added, but for new readers, just go to Manage Nuget Packages, by right clicking WebAPI project and typing WebAPITestClient in searchbox in online packages
You’ll get “A simple Test Client for ASP.NET Web API”, just add it. You’ll get a help controller in Areas-> HelpPage like shown below:
I have already provided the database scripts and data in my previous article, you can use the same.
Append “/help” in the application url, and you’ll get the test client,
GET:
POST:
PUT:
DELETE:
You can test each service by clicking on it. Once you click on the service link, you’ll be redirected to test the service page of that particular service. On that page there is a button Test API in the right bottom corner, just press that button to test your service:
Service for Get All products:
In the below case, I have already generated the token and now I am using it to make call to fetch all the products from the products table in the database.

Here I have called allproducts API, Add the value for parameter Id and “Token” header with its current value and click to get the result:

Now let’s see what happens to our APILog folder in application. Here you find the API log has been created, with the same name we have configured in NLog configuration in web.config file. The log file contains all the supplied details like Timestamp, Method type, URL , Header information (Token), Controller name, action and action parameters. You can also add more details to this log which you deem important for your application.

Logging Done!

Exception Logging

Our logging setup is completed, now we’ll focus on centralizing exception logging as well, so that none of the exception escapes without logging itself. Logging exception is of very high importance, it keeps track of all the exceptions. No matter business or application or system exceptions, all of them have to be logged.

Implementing Exception logging

Step 1: Exception Filter Attribute

Now we will add an action filter in our application for logging the exceptions. For this create a class,GlobalExceptionAttribute to “ActionFilter” folder and add the code below, the class is derived fromExceptionFilterAttribute, which is under System.Web.Http.Filters.
I override the OnException() method and replace the default “ITraceWriter” service with our NLogger class instance in the controller’s service container, same as we have done in Action logging in above section. Now theGetTraceWriter() method will return our instance (instance NLogger class) and Info() will call trace()method of NLogger class.
using System;
using System.Collections.Generic;
using System.Linq;
using System.Web;
using System.Web.Http.Filters;
using System.Web.Http;
using System.Web.Http.Tracing;
using WebApi.Helpers;
using System.ComponentModel.DataAnnotations;
using System.Net.Http;
using System.Net;

namespace WebApi.ActionFilters
{
    /// <summary>
    /// Action filter to handle for Global application errors.
    /// </summary>
    public class GlobalExceptionAttribute : ExceptionFilterAttribute
    {
        public override void OnException(HttpActionExecutedContext context)
        {
            GlobalConfiguration.Configuration.Services.Replace(typeof(ITraceWriter), new NLogger());
            var trace = GlobalConfiguration.Configuration.Services.GetTraceWriter();
            trace.Error(context.Request, "Controller : " + context.ActionContext.ControllerContext.ControllerDescriptor.ControllerType.FullName + Environment.NewLine + "Action : " + context.ActionContext.ActionDescriptor.ActionName, context.Exception);

            var exceptionType = context.Exception.GetType();

            if (exceptionType == typeof(ValidationException))
            {
                var resp = new HttpResponseMessage(HttpStatusCode.BadRequest) { Content = new StringContent(context.Exception.Message), ReasonPhrase = "ValidationException", };
                throw new HttpResponseException(resp);

            }
            else if (exceptionType == typeof(UnauthorizedAccessException))
            {
                throw new HttpResponseException(context.Request.CreateResponse(HttpStatusCode.Unauthorized));
            }
            else
            {
                throw new HttpResponseException(context.Request.CreateResponse(HttpStatusCode.InternalServerError));
            }
        }
    }
}

Step 2: Modify NLogger Class

Our NLogger class is capable to log all info and events, I have done some changes in private method Log() to handle the exceptions
#region Private member methods.
/// <summary>
/// Logs info/Error to Log file
/// </summary>
/// <param name="record"></param>
private void Log(TraceRecord record)
{
var message = new StringBuilder();

if (!string.IsNullOrWhiteSpace(record.Message))
                message.Append("").Append(record.Message + Environment.NewLine);

      if (record.Request != null)
{
       if (record.Request.Method != null)
            message.Append("Method: " + record.Request.Method + Environment.NewLine);

                if (record.Request.RequestUri != null)
                    message.Append("").Append("URL: " + record.Request.RequestUri + Environment.NewLine);

                if (record.Request.Headers != null && record.Request.Headers.Contains("Token") && record.Request.Headers.GetValues("Token") != null && record.Request.Headers.GetValues("Token").FirstOrDefault() != null)
                    message.Append("").Append("Token: " + record.Request.Headers.GetValues("Token").FirstOrDefault() + Environment.NewLine);
            }

            if (!string.IsNullOrWhiteSpace(record.Category))
                message.Append("").Append(record.Category);

            if (!string.IsNullOrWhiteSpace(record.Operator))
                message.Append(" ").Append(record.Operator).Append(" ").Append(record.Operation);

            if (record.Exception != null && !string.IsNullOrWhiteSpace(record.Exception.GetBaseException().Message))
            {
                var exceptionType = record.Exception.GetType();
                message.Append(Environment.NewLine);
                message.Append("").Append("Error: " + record.Exception.GetBaseException().Message + Environment.NewLine);
            }

            Logger[record.Level](Convert.ToString(message) + Environment.NewLine);
        }

Step 3: Modify Controller for Exception

Our application is now ready to run, but there is no exception in our code, so I added a throw exception code inProductController, just the Get(int id) method so that it can throw exception for testing our exception logging mechanism. It will throw an exception if the product is not there in database with the provided id.
  // GET api/product/5
 [GET("productid/{id?}")]
 [GET("particularproduct/{id?}")]
 [GET("myproduct/{id:range(1, 3)}")]
 public HttpResponseMessage Get(int id)
 {
var product = _productServices.GetProductById(id);
      if (product != null)
       return Request.CreateResponse(HttpStatusCode.OK, product);

 throw new Exception("No product found for this id");
      //return Request.CreateErrorResponse(HttpStatusCode.NotFound,   "No product found for this id");
 }

Step 4: Run the application

Run the application and click on Product/all API
Add the parameter id value to 1 and header Token with it’s current value, click on send button to get the result:
Now we can see that the Status is 200/OK, and we also get a product with the provided id in the response body. Let’s see the API log now:
The log has captured the call of Product API, now provide a new product id as parameter, which is not there in database, I am using 12345 as product id and result is:

We can see there is an 500/Internal Server Error now in response status, let’s check the API Log:

Well, now the log has captured both the event and error of same call on the server, you can see call log details and the error with provided error message in the log.

Custom Exception logging

In the above section we have implemented exception logging, but there is default system response and status (i.e. 500/Internal Server Error). It will be always good to have your own custom response and exceptions for your API. That will be easier for client to consume and understand the API responses.

Step 1: Add Custom Exception Classes

Add “Error Helper” folder to application to maintain our custom exception classes separately and add “IApiExceptions” interface to newly created “ErrorHelper” folder –
Add following code the IApiExceptions interface, this will serve as a template for all exception classes, I have added four common properties for our custom classes to maintain Error Code, ErrorDescription, HttpStatus (Contains the values of status codes defined for HTTP) and ReasonPhrase.
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Net;

namespace WebApi.ErrorHelper
{
    /// <summary>
    /// IApiExceptions Interface
    /// </summary>
    public interface IApiExceptions
    {
        /// <summary>
        /// ErrorCode
        /// </summary>
        int ErrorCode { get; set; }
        /// <summary>
        /// ErrorDescription
        /// </summary>
        string ErrorDescription { get; set; }
        /// <summary>
        /// HttpStatus
        /// </summary>
        HttpStatusCode HttpStatus { get; set; }
        /// <summary>
        /// ReasonPhrase
        /// </summary>
        string ReasonPhrase { get; set; }
    }
}
Here, I divided our exceptions in three categories:
  1. API Exceptions – for API level exceptions.
  2. Business Exceptions – for exceptions at business logic level.
  3. Data Exceptions – Data related exceptions.
To implement this create a three new classes ApiException.cs, ApiDataException.cs and ApiBusinessExceptionclasses to same folder which implements IApiExceptions interface with following code to the classes.
#region Using namespaces.
using System;
using System.Net;
using System.Runtime.Serialization;
#endregion


namespace WebApi.ErrorHelper
{
    /// <summary>
    /// Api Exception
    /// </summary>
    [Serializable]
    [DataContract]
    public class ApiException : Exception, IApiExceptions
    {
        #region Public Serializable properties.
        [DataMember]
        public int ErrorCode { get; set; }
        [DataMember]
        public string ErrorDescription { get; set; }
        [DataMember]
        public HttpStatusCode HttpStatus { get; set; }
        
        string reasonPhrase = "ApiException";

        [DataMember]
        public string ReasonPhrase
        {
            get { return this.reasonPhrase; }

            set { this.reasonPhrase = value; }
        }
        #endregion
    }
}
I have initialized ReasonPhrase property with different default values in these classes to differentiate the implementation, you can use implement your custom classes as per your application needs.
The directives applied on class as Serializable and DataContract to make sure that the class defines or implements a data contract is serializable and can be serialize by a serializer.
Note: Add reference of “System.Runtime.Serialization.dll” DLL if you facing any assembly issue.
In the same way add “ApiBusinessException” and “ApiDataException” classes into the same folder, with the following code –
#region Using namespaces.
using System;
using System.Net;
using System.Runtime.Serialization; 
#endregion

namespace WebApi.ErrorHelper
{
    /// <summary>
    /// Api Business Exception
    /// </summary>
    [Serializable]
    [DataContract]
    public class ApiBusinessException : Exception, IApiExceptions
    {
        #region Public Serializable properties.
        [DataMember]
        public int ErrorCode { get; set; }
        [DataMember]
        public string ErrorDescription { get; set; }
        [DataMember]
        public HttpStatusCode HttpStatus { get; set; }

        string reasonPhrase = "ApiBusinessException";

        [DataMember]
        public string ReasonPhrase
        {
            get { return this.reasonPhrase; }

            set { this.reasonPhrase = value; }
        }
        #endregion

        #region Public Constructor.
        /// <summary>
        /// Public constructor for Api Business Exception
        /// </summary>
        /// <param name="errorCode"></param>
        /// <param name="errorDescription"></param>
        /// <param name="httpStatus"></param>
        public ApiBusinessException(int errorCode, string errorDescription, HttpStatusCode httpStatus)
        {
            ErrorCode = errorCode;
            ErrorDescription = errorDescription;
            HttpStatus = httpStatus;
        } 
        #endregion

    }
}

#region Using namespaces.
using System;
using System.Net;
using System.Runtime.Serialization;
#endregion

namespace WebApi.ErrorHelper
{
    /// <summary>
    /// Api Data Exception
    /// </summary>
    [Serializable]
    [DataContract]
    public class ApiDataException : Exception, IApiExceptions
    {
        #region Public Serializable properties.
        [DataMember]
        public int ErrorCode { get; set; }
        [DataMember]
        public string ErrorDescription { get; set; }
        [DataMember]
        public HttpStatusCode HttpStatus { get; set; }

        string reasonPhrase = "ApiDataException";

        [DataMember]
        public string ReasonPhrase
        {
            get { return this.reasonPhrase; }

            set { this.reasonPhrase = value; }
        }

        #endregion

        #region Public Constructor.
        /// <summary>
        /// Public constructor for Api Data Exception
        /// </summary>
        /// <param name="errorCode"></param>
        /// <param name="errorDescription"></param>
        /// <param name="httpStatus"></param>
        public ApiDataException(int errorCode, string errorDescription, HttpStatusCode httpStatus)
        {
            ErrorCode = errorCode;
            ErrorDescription = errorDescription;
            HttpStatus = httpStatus;
        }
        #endregion
    }
}

JSon Serializers

There are some objects need to be serialized in JSON, to log and to transfer through the modules, for this I have add some extension methods to Object class.
For that add “System.Web.Extensions.dll” reference to project and add “JSONHelper” class to Helpers folder, with following code:
#region Using namespaces.
using System.Web.Script.Serialization;
using System.Data;
using System.Collections.Generic;
using System;

#endregion

namespace WebApi.Helpers
{
    public static class JSONHelper
    {
         #region Public extension methods.
        /// <summary>
        /// Extened method of object class, Converts an object to a json string.
        /// </summary>
        /// <param name="obj"></param>
        /// <returns></returns>
        public static string ToJSON(this object obj)
        {
            var serializer = new JavaScriptSerializer();
            try
            {
                return serializer.Serialize(obj);
            }
            catch(Exception ex)
            {
                return "";
            }
        }
         #endregion
    }
}
In the above code “ToJSON()” method is an extension of base Object class, which serializes supplied the object to a JSON string. The method using “JavaScriptSerializer” class which exists in “System.Web.Script.Serialization“.

Modify NLogger Class

For exception handling I have modified the Log() method of NLogger, which will now handle the different API exceptions.
/// <summary>
/// Logs info/Error to Log file
/// </summary>
/// <param name="record"></param>
private void Log(TraceRecord record)
{
var message = new StringBuilder();

      if (!string.IsNullOrWhiteSpace(record.Message))
                message.Append("").Append(record.Message + Environment.NewLine);

            if (record.Request != null)
            {
                if (record.Request.Method != null)
                    message.Append("Method: " + record.Request.Method + Environment.NewLine);

                if (record.Request.RequestUri != null)
                    message.Append("").Append("URL: " + record.Request.RequestUri + Environment.NewLine);

                if (record.Request.Headers != null && record.Request.Headers.Contains("Token") && record.Request.Headers.GetValues("Token") != null && record.Request.Headers.GetValues("Token").FirstOrDefault() != null)
                    message.Append("").Append("Token: " + record.Request.Headers.GetValues("Token").FirstOrDefault() + Environment.NewLine);
            }

            if (!string.IsNullOrWhiteSpace(record.Category))
                message.Append("").Append(record.Category);

            if (!string.IsNullOrWhiteSpace(record.Operator))
                message.Append(" ").Append(record.Operator).Append(" ").Append(record.Operation);

            if (record.Exception != null && !string.IsNullOrWhiteSpace(record.Exception.GetBaseException().Message))
            {
                var exceptionType = record.Exception.GetType();
                message.Append(Environment.NewLine);
                if (exceptionType == typeof(ApiException))
                {
                    var exception = record.Exception as ApiException;
                    if (exception != null)
                    {
                        message.Append("").Append("Error: " + exception.ErrorDescription + Environment.NewLine);
                        message.Append("").Append("Error Code: " + exception.ErrorCode + Environment.NewLine);
                    }
                }
                else if (exceptionType == typeof(ApiBusinessException))
                {
                    var exception = record.Exception as ApiBusinessException;
                    if (exception != null)
                    {
                        message.Append("").Append("Error: " + exception.ErrorDescription + Environment.NewLine);
                        message.Append("").Append("Error Code: " + exception.ErrorCode + Environment.NewLine);
                    }
                }
                else if (exceptionType == typeof(ApiDataException))
                {
                    var exception = record.Exception as ApiDataException;
                    if (exception != null)
                    {
                        message.Append("").Append("Error: " + exception.ErrorDescription + Environment.NewLine);
                        message.Append("").Append("Error Code: " + exception.ErrorCode + Environment.NewLine);
                    }
                }
                else
                    message.Append("").Append("Error: " + record.Exception.GetBaseException().Message + Environment.NewLine);
            }

            Logger[record.Level](Convert.ToString(message) + Environment.NewLine);
        }
The code above checks the exception object of TraceRecord and updates the logger as per the exception type.

Modify GlobalExceptionAttribute

As we have created GlobalExceptionAttribute to handle all exceptions and create response in case of any exception. Now I have added some new code to this in order to enable the GlobalExceptionAttribute class to handle custom exceptions. I am adding only modified method here for your reference .
public override void OnException(HttpActionExecutedContext context)
{
         GlobalConfiguration.Configuration.Services.Replace(typeof(ITraceWriter), new NLogger());
            var trace = GlobalConfiguration.Configuration.Services.GetTraceWriter();
            trace.Error(context.Request, "Controller : " + context.ActionContext.ControllerContext.ControllerDescriptor.ControllerType.FullName + Environment.NewLine + "Action : " + context.ActionContext.ActionDescriptor.ActionName, context.Exception);

            var exceptionType = context.Exception.GetType();

            if (exceptionType == typeof(ValidationException))
            {
                var resp = new HttpResponseMessage(HttpStatusCode.BadRequest) { Content = new StringContent(context.Exception.Message), ReasonPhrase = "ValidationException", };
                throw new HttpResponseException(resp);

            }
            else if (exceptionType == typeof(UnauthorizedAccessException))
            {
                throw new HttpResponseException(context.Request.CreateResponse(HttpStatusCode.Unauthorized, new ServiceStatus() { StatusCode = (int)HttpStatusCode.Unauthorized, StatusMessage = "UnAuthorized", ReasonPhrase = "UnAuthorized Access" }));
            }
            else if (exceptionType == typeof(ApiException))
            {
                var webapiException = context.Exception as ApiException;
                if (webapiException != null)
                    throw new HttpResponseException(context.Request.CreateResponse(webapiException.HttpStatus, new ServiceStatus() { StatusCode = webapiException.ErrorCode, StatusMessage = webapiException.ErrorDescription, ReasonPhrase = webapiException.ReasonPhrase }));
            }
            else if (exceptionType == typeof(ApiBusinessException))
            {
                var businessException = context.Exception as ApiBusinessException;
                if (businessException != null)
                    throw new HttpResponseException(context.Request.CreateResponse(businessException.HttpStatus, new ServiceStatus() { StatusCode = businessException.ErrorCode, StatusMessage = businessException.ErrorDescription, ReasonPhrase = businessException.ReasonPhrase }));
            }
            else if (exceptionType == typeof(ApiDataException))
            {
                var dataException = context.Exception as ApiDataException;
                if (dataException != null)
                    throw new HttpResponseException(context.Request.CreateResponse(dataException.HttpStatus, new ServiceStatus() { StatusCode = dataException.ErrorCode, StatusMessage = dataException.ErrorDescription, ReasonPhrase = dataException.ReasonPhrase }));
            }
            else
            {
                throw new HttpResponseException(context.Request.CreateResponse(HttpStatusCode.InternalServerError));
            }
        }
In the above code I have modified the overrided method OnExeption() and created new Http response exception based on the different exception types.

Modify Product Controller

Now modify the Product controller to throw our custom exception form, please look into the Get method I have modified to throw the APIDataException in case if data is not found and APIException in any other kind of error.
// GET api/product/5
[GET("productid/{id?}")]
[GET("particularproduct/{id?}")]
[GET("myproduct/{id:range(1, 3)}")]
public HttpResponseMessage Get(int id)
{
if (id != null)
      {
       var product = _productServices.GetProductById(id);
            if (product != null)
             return Request.CreateResponse(HttpStatusCode.OK, product);

throw new ApiDataException(1001, "No product found for this id.", HttpStatusCode.NotFound);
      }
      throw new ApiException() { ErrorCode = (int)HttpStatusCode.BadRequest, ErrorDescription = "Bad Request..." };
}

Run the application

Run the application and click on Product/all API:
Add the parameter id value to 1 and header Token with its current value, click on send button to get the result:
Now we can see that the Status is 200/OK, and we also get a product with the provided id in the response body. Lets see the API log now –
The log has captured the call of Product API, now provide a new product id as parameter, which is not there in database, I am using 12345 as product id and result is:
We can see, now there is a custom error status code “1001” and messages “No product found for this id.” And the generic status code “500/Internal Server Error” is now replaced with our supplied code “404/ Not Found”, which is more meaningful for the client or consumer.
Lets see the APILog now:
Well, now the log has captured both the event and error of same call on the server, you can see call log details and the error with provided error message in the log with our custom error code, I have only captured error description and error code, but you can add more details in the log as per your application needs.

Update the controller for new Exception Handling

Following is the code for controllers with implementation of custom exception handling and logging:

Product Controller

using System.Collections.Generic;
using System.Linq;
using System.Net;
using System.Net.Http;
using System.Web.Http;
using AttributeRouting;
using AttributeRouting.Web.Http;
using BusinessEntities;
using BusinessServices;
using WebApi.ActionFilters;
using WebApi.Filters;
using System;
using WebApi.ErrorHelper;

namespace WebApi.Controllers
{
    [AuthorizationRequired]
    [RoutePrefix("v1/Products/Product")]
    public class ProductController : ApiController
    {
        #region Private variable.

        private readonly IProductServices _productServices;

        #endregion

        #region Public Constructor

        /// <summary>
        /// Public constructor to initialize product service instance
        /// </summary>
        public ProductController(IProductServices productServices)
        {
            _productServices = productServices;
        }

        #endregion

        // GET api/product
        [GET("allproducts")]
        [GET("all")]
        public HttpResponseMessage Get()
        {
            var products = _productServices.GetAllProducts();
            var productEntities = products as List ?? products.ToList();
            if (productEntities.Any())
                return Request.CreateResponse(HttpStatusCode.OK, productEntities);
            throw new ApiDataException(1000, "Products not found", HttpStatusCode.NotFound);
        }

        // GET api/product/5
        [GET("productid/{id?}")]
        [GET("particularproduct/{id?}")]
        [GET("myproduct/{id:range(1, 3)}")]
        public HttpResponseMessage Get(int id)
        {
            if (id != null)
            {
                var product = _productServices.GetProductById(id);
                if (product != null)
                    return Request.CreateResponse(HttpStatusCode.OK, product);

                throw new ApiDataException(1001, "No product found for this id.", HttpStatusCode.NotFound);
            }
            throw new ApiException() { ErrorCode = (int)HttpStatusCode.BadRequest, ErrorDescription = "Bad Request..." };
        }

        // POST api/product
        [POST("Create")]
        [POST("Register")]
        public int Post([FromBody] ProductEntity productEntity)
        {
            return _productServices.CreateProduct(productEntity);
        }

        // PUT api/product/5
        [PUT("Update/productid/{id}")]
        [PUT("Modify/productid/{id}")]
        public bool Put(int id, [FromBody] ProductEntity productEntity)
        {
            if (id > 0)
            {
                return _productServices.UpdateProduct(id, productEntity);
            }
            return false;
        }

        // DELETE api/product/5
        [DELETE("remove/productid/{id}")]
        [DELETE("clear/productid/{id}")]
        [PUT("delete/productid/{id}")]
        public bool Delete(int id)
        {
            if (id != null && id > 0)
            {
                var isSuccess = _productServices.DeleteProduct(id);
                if (isSuccess)
                {
                    return isSuccess;
                }
                throw new ApiDataException(1002, "Product is already deleted or not exist in system.", HttpStatusCode.NoContent );
            }
            throw new ApiException() {ErrorCode = (int) HttpStatusCode.BadRequest, ErrorDescription = "Bad Request..."};
        }
    }
}
Now you can see, our application is so rich and scalable that none of the exception or transaction can escapelogging.Once setup is inplaced, now you don’t have to worry about writing code each time for logging or requests and exceptions, but you can relax and focus on business logic only.
Image credit: https://pixabay.com/en/kermit-frog-meadow-daisy-concerns-383370/

Conclusion

In this article we learnt about how to perform request logging and exception logging in WebPI. There could be numerous ways in which you can perform these operations but I tried to present this in as simple way as possible. My approach was to take our enterprise level to next level of development, where developers should not always be worried about exception handling and logging. Our solution provides a generic approach of centralizing the operations in one place; all the requests and exceptions are automatically taken care of. In my new articles, I’ll try to enhance the application by explaining unit testing in WebAPI and OData in WebAPI. You can download the complete source code of this article with packages from GitHub. Happy coding Smile | :)

RESTful Day #3: Resolve dependency of dependencies using Inversion of Control and dependency injection in ASP.NET Web APIs with Unity Container and Managed Extensibility Framework (MEF)


Download Source Code

Introduction

In my last two articles, I explained how to create a RESTful service using ASP.NET Web API working with Entity Framework and resolving dependencies using Unity Container. In this article, I’ll explain how to create a loosely coupled system with Unity Container and MEF (Managed Extensibility Framework) using Inversion of Control. I’ll not be explaining much theory but rather focus more on practical implementations. For the readers who are following this series, they can use their existing solution that they have created till time. For my new readers of this article, I have provided the download link for the previous source code and current source code as well.
For theory and understanding of DI and IOC you can follow the following links: Unity and Inversion of Control(IOC).

Roadmap

Here is my roadmap for learning RESTful APIs,
I’ll purposely use Visual Studio 2010 and .NET Framework 4.0 because there are few implementations that are very hard to find in .NET Framework 4.0, but I’ll make it easy by showing how we can do it.

Existing Design and Problem

We already have an existing design. If you open the solution, you’ll get to see the structure as mentioned below,
We tried to design a loosely coupled architecture in the following way,
  • DataModel (responsible for communication with database) : Only talks to service layer.
  • Services (acts as a business logic layer between REST endpoint and data access) : Communicates between REST endpoint and DataModel.
  • REST API i.e. Controllers: Only talks to services via the interfaces exposed.
But when we tried to resolve the dependency of UnitOfWork from Services, we had to reference
DataModel dll in to our WebAPI project; this violated our system like shown in following image,
In this article we’ll try to resolve dependency (data model) of a dependency (services) from our existing solution. My controller depended on services and my services depended on data model. Now we’ll design an architecture in which components will be independent of each other in terms of object creation and instantiation. To achieve this we’ll make use of MEF(Managed Extensibility Framework) along with Unity Container and reflection.
Image source : http://o5.com/wp-content/uploads/2010/08/dreamstime_15583711-550×371.jpg
Ideally, we should not be having the below code in our Bootstrapper class,
container.RegisterType().RegisterType(new HierarchicalLifetimeManager());

Managed Extensibility Framework (MEF)

You can have a read about Unity from msdn link. I am just quoting some lines from msdn,

“The Managed Extensibility Framework or MEF is a library for creating lightweight, extensible applications. It allows application developers to discover and use extensions with no configuration required. It also lets extension developers easily encapsulate code and avoid fragile hard dependencies. MEF not only allows extensions to be reused within applications, but across applications as well.”
“MEF is an integral part of the .NET Framework 4, and is available wherever the .NET Framework is used. You can use MEF in your client applications, whether they use Windows Forms, WPF, or any other technology, or in server applications that use ASP.NET.”

Creating a Dependency Resolver with Unity and MEF

Open your Visual studio, I am using VS 2010, you can use VS version 2010 or above. Load the solution.
Step 1: Right click solution explorer and add a new project named Resolver,
I have intentionally chosen this name, and you already know it why J
Step 2: Right click Resolver project and click on ManageNugetPackage, in the interface of adding new package, search Unity.MVC3 in online library,
Install the package to your solution.
Step 3: Right click resolver project and add a reference to System.ComponentModel.Composition.
You can find the dll into your GAC.I am using framework 4.0, so referring to the same version dll.
This DLL is the part of MEF and is already installed with .NET Framework 4.0 in the system GAC. This DLL provides classes that are very core of MEF.
Step 4: Just add an interface named IComponent to Resolver project that contains the initialization method named Setup. We’ll try to implement this interface into our Resolver class that we’ll create in our other projects like DataModel, Services and WebApI.
namespace Resolver
{
    /// <summary>
    /// Register underlying types with unity.
    /// </summary>
    public interface IComponent
    {
      
    }
}
Step 5: Before we declare our Setup method, just add one more interface responsible for serving as a contract to register types. I name this interface as IRegisterComponent,
namespace Resolver
{
    /// <summary>
    /// Responsible for registering types in unity configuration by implementing IComponent
    /// </summary>
    public interface IRegisterComponent
    {
        /// <summary>
        /// Register type method
        /// </summary>
        /// <typeparam name="TFrom"></typeparam>
        /// <typeparam name="TTo"></typeparam>
        /// <param name="withInterception"></param>
        void RegisterType(bool withInterception = false) where TTo : TFrom;
 
        /// <summary>
        /// Register type with container controlled life time manager
        /// </summary>
        /// <typeparam name="TFrom"></typeparam>
        /// <typeparam name="TTo"></typeparam>
        /// <param name="withInterception"></param>
        void RegisterTypeWithControlledLifeTime(bool withInterception = false) where TTo : TFrom;
    }
}
In this interface I have declared two methods, one RegisterType and other in to RegisterType with Controlled life time of the object, i.e. the life time of an object will be hierarchal in manner. This is kind of same like we do in Unity.
Step 6: Now declare Setup method on our previously created IComponent interface, that takes instance ofIRegisterComponent as a parameter,
void SetUp(IRegisterComponent registerComponent);
So our IComponent interface becomes,
namespace Resolver
{
    /// <summary>
    /// Register underlying types with unity.
    /// </summary>
    public interface IComponent
    {
        void SetUp(IRegisterComponent registerComponent);
    }
}
Step 6: Now we’ll write a packager or you can say a wrapper over MEF and Unity to register types/ components. This is the core MEF implementation. Create a class named ComponentLoader, and add following code to it,
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using Microsoft.Practices.Unity;
using System.ComponentModel.Composition.Hosting;
using System.ComponentModel.Composition.Primitives;
using System.Reflection;
 
namespace Resolver
{
    public static class ComponentLoader
    {
        public static void LoadContainer(IUnityContainer container, string path, string pattern)
        {
            var dirCat = new DirectoryCatalog(path, pattern);
            var importDef = BuildImportDefinition();
            try
            {
                using (var aggregateCatalog = new AggregateCatalog())
                {
                    aggregateCatalog.Catalogs.Add(dirCat);
 
                    using (var componsitionContainer = new CompositionContainer(aggregateCatalog))
                    {
                        IEnumerable exports = componsitionContainer.GetExports(importDef);
 
                        IEnumerable modules =
                            exports.Select(export => export.Value as IComponent).Where(m => m != null);
 
                        var registerComponent = new RegisterComponent(container);
                        foreach (IComponent module in modules)
                        {
                            module.SetUp(registerComponent);
                        }
                    }
                }
            }
            catch (ReflectionTypeLoadException typeLoadException)
            {
                var builder = new StringBuilder();
                foreach (Exception loaderException in typeLoadException.LoaderExceptions)
                {
                    builder.AppendFormat("{0}\n", loaderException.Message);
                }
 
                throw new TypeLoadException(builder.ToString(), typeLoadException);
            }
        }
 
        private static ImportDefinition BuildImportDefinition()
        {
            return new ImportDefinition(
                def => true, typeof(IComponent).FullName, ImportCardinality.ZeroOrMore, false, false);
        }
    }
 
    internal class RegisterComponent : IRegisterComponent
    {
        private readonly IUnityContainer _container;
 
        public RegisterComponent(IUnityContainer container)
        {
            this._container = container;
            //Register interception behaviour if any
        }
 
        public void RegisterType(bool withInterception = false) where TTo : TFrom
        {
            if (withInterception)
            {
                //register with interception
            }
            else
            {
                this._container.RegisterType();
            }
        }
 
        public void RegisterTypeWithControlledLifeTime(bool withInterception = false) where TTo : TFrom
        {
            this._container.RegisterType(new ContainerControlledLifetimeManager());
        }
    }
}
Step 7: Now our Resolver wrapper is ready. Build the project and add its reference to DataModel, BusinessServices and WebApi project like shown below,

Setup Business Services

We already have added reference of Resolver in BusinessServices project. We agreed to implement IComponent interface in each of our project.
So create a class named DependencyResolver and implement IComponent interface into it, we make use of reflection too to import IComponent type. So add a class and add following code to that DependencyResolverclass,
using System.ComponentModel.Composition;
using DataModel;
using DataModel.UnitOfWork;
using Resolver;
 
namespace BusinessServices
{
    [Export(typeof(IComponent))]
    public class DependencyResolver : IComponent
    {
        public void SetUp(IRegisterComponent registerComponent)
        {
            registerComponent.RegisterType();
 
        }
    }
}
Note that we have implemented SetUp method and in the same method we registered type for my ProductService.
All of the existing code base remains same. We don’t need to touch the IProductServices interface or ProductServices class.

Setup DataModel

We have added Resolver project reference to DataModel project as well. So we’ll try to register the type of UnitOfWork in this project. We proceed in same fashion, just add a DependencyResolver class and implement its Setup method to register type of UnitOfWork. To make the code more readable and standard, I made a change.I just added an interface for UnitOfWork and named it IUnitOfWork. Now my UnitOfWork class derives from this, you can do this exercise in earlier versions of projects we discussed in first two articles.
So my IUnitOfWork contains declaration of a single public method in UnitOfWork,
namespace DataModel.UnitOfWork
{
    public interface IUnitOfWork
    {
        /// <summary>
        /// Save method.
        /// </summary>
        void Save();
    }
}
Now register the type for UnitOfWork in DepenencyResolver class, our class becomes as shown below,
using System.ComponentModel.Composition;
using System.Data.Entity;
using DataModel.UnitOfWork;
using Resolver;
 
namespace DataModel
{
    [Export(typeof(IComponent))]
    public class DependencyResolver : IComponent
    {
        public void SetUp(IRegisterComponent registerComponent)
        {
            registerComponent.RegisterType();
        }
    }
}
Again, no need to touch any existing code of this project.

Setup REST endpoint / WebAPI project

Our 90% of the job is done.
We now need to setup or WebAPI project. We’ll not add any DependencyResolver class in this project. We’ll invert the calling mechanism of layers in Bootstrapper class that we already have, so when you open your bootstrapper class, you’ll get the code something like,
using System.Web.Http;
using System.Web.Mvc;
using BusinessServices;
using DataModel.UnitOfWork;
using Microsoft.Practices.Unity;
using Unity.Mvc3;
 
namespace WebApi
{
    public static class Bootstrapper
    {
        public static void Initialise()
        {
            var container = BuildUnityContainer();
 
            DependencyResolver.SetResolver(new UnityDependencyResolver(container));
 
            // register dependency resolver for WebAPI RC
            GlobalConfiguration.Configuration.DependencyResolver = new Unity.WebApi.UnityDependencyResolver(container);
        }
 
        private static IUnityContainer BuildUnityContainer()
        {
            var container = new UnityContainer();
 
            // register all your components with the container here
            // it is NOT necessary to register your controllers
           
            // e.g. container.RegisterType();       
            container.RegisterType().RegisterType(new HierarchicalLifetimeManager());
 
            return container;
        }
    }
}
Now, we need to change the code base a bit to make our system loosely coupled.Just remove the reference of DataModel from WebAPI project.
We don’t want our DataModel to be exposed to WebAPI project, that was our aim though, so we cut down the dependency of DataModel project now.
Add following code of Bootstrapper class to the existing Bootstarpper class,
using System.Web.Http;
//using DataModel.UnitOfWork;
using Microsoft.Practices.Unity;
using Resolver;
using Unity.Mvc3;
 
namespace WebApi
{
    public static class Bootstrapper
    {
        public static void  Initialise()
        {
            var container = BuildUnityContainer();
 
            System.Web.Mvc.DependencyResolver.SetResolver(new UnityDependencyResolver(container));
 
            // register dependency resolver for WebAPI RC
            GlobalConfiguration.Configuration.DependencyResolver = new Unity.WebApi.UnityDependencyResolver(container);
        }
 
        private static IUnityContainer BuildUnityContainer()
        {
            var container = new UnityContainer();
 
            // register all your components with the container here
            // it is NOT necessary to register your controllers
           
            // e.g. container.RegisterType();       
           // container.RegisterType().RegisterType(new HierarchicalLifetimeManager());
 
            RegisterTypes(container);
 
            return container;
        }
 
        public static void RegisterTypes(IUnityContainer container)
        {
 
            //Component initialization via MEF
            ComponentLoader.LoadContainer(container, ".\\bin", "WebApi.dll");
            ComponentLoader.LoadContainer(container, ".\\bin", "BusinessServices.dll");
 
        }
    }
}
It is kind of redefining Bootstrapper class without touching our existing controller methods. We now don’t even have to register type for ProductServices as well, we already did this in BusinessServices project.
Note that in RegisterTypes method we load components/dlls through reflection making use of ComponentLoader.We wrote two lines, first to load WebAPI.dll and another one to load Business Services.dll.
Had the name of BusinessServicess.dll be WebAPI.Services.dll, then we would have only written one line of code to load both the WebAPI and BusinessService dll like shown below,
ComponentLoader.LoadContainer(container, ".\\bin", "WebApi*.dll");
Yes we can make use of Regex.

Running the application

Just run the application, we get,
We already have our test client added, but for new readers, just go to Manage Nuget Packages, by right clicking WebAPI project and type WebAPITestClient in searchbox in online packages,
You’ll get “A simple Test Client for ASP.NET Web API”, just add it. You’ll get a help controller in Areas-> HelpPage like shown below,
I have already provided the database scripts and data in my previous article, you can use the same.
Append “/help” in the application url, and you’ll get the test client,
You can test each service by clicking on it. Once you click on the service link, you’ll be redirected to test the service page of that particular service. On that page there is a button Test API in the right bottom corner, just press that button to test your service,
Service for GetAllProduct,
For Create a new product,
In database, we get new product,
Update product:
We get in database,
Delete product:
In database:

Advantages of this design

In my earlier articles I focussed more on design flaws, but our current design have emerged with few added advantages,
  1. We got an extensible and loosely coupled application design that can go far with more new components added in the same way.
  2. Registering types automatically through reflection. Suppose we want to register any Interface implementation to our REST endpoint, we just need to load that dll in our Bootstrapper class, or if dll’s are of common suffix names then we just have to place that DLL in bin folder, and that will automatically be loaded at run time.
    Image source: http://a.pragprog.com/magazines/2013-07/images/iStock_000021011143XSmall__1mfk9b__.jpg
  3. Database transactions or any of such module is now not exposed to the service endpoint, this makes our service more secure and maintains the design structure too.

Conclusion

We now know how to use Unity container to resolve dependency and perform inversion of control using MEF too. In my next article I’ll try to explain how we can open multiple endpoints to our REST service and create custom url’s in the true REST fashion in my WebAPI. Till then Happy Coding Smile | :) You can also download the source code from GitHub. Add the required packages, if they are missing in the source code.

Download Source Code

RESTful Day #2: Inversion of control using dependency injection in Web API’s using Unity Container and Bootstrapper


Introduction

My article will explain how we can make our Web API service architecture loosely coupled and more flexible. We already learnt that how we can create a RESTful service using Asp.net Web API and Entity framework in my last article. If you remember we ended up in a solution with a design flaw, we’ll try to overcome that flaw by resolving the dependencies of dependent components. For those who have not followed my previous article, they can learn by having the sample project attached as a test application from my first article.

Image source : https://www.pehub.com/wp-content/uploads/2013/06/independent-300×200.jpg
There are various methods you can use to resolve dependency of components. In my article I’ll explain how to resolve dependency with the help of Unity Container provided by Microsoft’s Unity Application Block.
We’ll not go into very detailed theory, for theory and understanding of DI and IOC you can follow the following links: Unity and Inversion of Control(IOC). We’ll straight away jump into practical implementation.

Roadmap

Our roadmap for learning RESTful APIs remains same,

I’ll purposely use Visual Studio 2010 and .NET Framework 4.0 because there are few implementations that are very hard to find in .NET Framework 4.0, but I’ll make it easy by showing how we can do it.

Existing Design and Problem

We already have an existing design. If you open the solution, you’ll get to see the structure as mentioned below,
The modules are dependent in a way,
There is no problem with the structure, but the way they interact to each other is really problematic. You must have noticed that we are trying to communicate with layers, making the physical objects of classes.
For e.g.
Controller constructor makes an object of Service layer to communicate,
        /// <summary>
        /// Public constructor to initialize product service instance
        /// </summary>
        public ProductController()
        {
            _productServices =new ProductServices();
        }
Service constructor, in turn, makes and object of UnitOfWork to communicate to database,
        /// <summary>
        /// Public constructor.
        /// </summary>
        public ProductServices()
        {
            _unitOfWork = new UnitOfWork();
        }
The problem lies in these code pieces. We see Controller is dependent upon instantiation of Service and Service is dependent upon UnitOfWork to get instantiated. Our Layers should not be that tightly coupled and should be dependent on each other.
The work of creating object should be assigned to someone else. Our layers should not worry about creating objects.
We’ll assign this role to a third party that will be called our container. Fortunately, Unity provides that help to us, to get rid of this dependency problem and invert the control flow by injecting dependency not by creating objects by new but through constructors or properties. There are other methods too, but I am not going into detail.

Introduction to Unity

You can have a read about Unity from this link; I am just quoting some lines,
Image source: http://img.tradeindia.com/fp/1/669/664.jpg
“The Unity Application Block (Unity) is a lightweight, extensible dependency injection container that supports constructor injection, property injection, and method call injection. It provides developers with the following advantages:
  • It provides simplified object creation, especially for hierarchical object structures and dependencies, which simplifies application code.
  • It supports abstraction of requirements; this allows developers to specify dependencies at run time or in configuration and simplify management of crosscutting concerns.
  • It increases flexibility by deferring component configuration to the container.
  • It has a service location capability; this allows clients to store or cache the container. This is especially useful in ASP.NET Web applications where developers can persist the container in the ASP.NET session or application.”

Setup Unity

Open your Visual studio, I am using VS 2010, You can use VS version 2010 or above. Load the solution.
Step 1: browse to Tools-> Library Packet Manager – > Packet manager Console,
We’ll add package for Unity Application Block.
In the left bottom corner of Visual Studio, You’ll find where to write the command.
Type command Unity.MVC3 and choose “WebApi” project before you fire the command.
Step 2: Bootstrapper class
Unity.MVC3 comes with a Bootstrapper class, as soon as you run the command, the Bootstrapper class will be generated in your solution->WebAPI project,
using System.Web.Http;
using System.Web.Mvc;
using BusinessServices;
using DataModel.UnitOfWork;
using Microsoft.Practices.Unity;
using Unity.Mvc3;

namespace WebApi
{
    public static class Bootstrapper
    {
        public static void Initialise()
        {
            var container = BuildUnityContainer();

            DependencyResolver.SetResolver(new UnityDependencyResolver(container));

            // register dependency resolver for WebAPI RC
            GlobalConfiguration.Configuration.DependencyResolver = new Unity.WebApi.UnityDependencyResolver(container);
        }

        private static IUnityContainer BuildUnityContainer()
        {
            var container = new UnityContainer();

            // register all your components with the container here
            // it is NOT necessary to register your controllers
            
            // e.g. container.RegisterType();        
            container.RegisterType().RegisterType(new HierarchicalLifetimeManager());

            return container;
        }
    }
}
This class comes with an initial configuration to setup your container. All the functionality is inbuilt, we only need to specify the dependencies that we need to resolve in the “BuildUnityContainer“, like it says in the commented statement,
            // register all your components with the container here
            // it is NOT necessary to register your controllers
            
            // e.g. container.RegisterType();
Step 3: Just specify the components below these commented lines that we need to resolve. In our case, it’sProductServices and UnitOfWork, so just add,
container.RegisterType().RegisterType(new HierarchicalLifetimeManager());
“HierarchicalLifetimeManager” maintains the lifetime of the object and child object depends upon parent object’s lifetime.
If you don’t find “UnitOfWork“, just add reference to DataModel project in WebAPI project.
So our Bootstrapper class becomes,
   public static class Bootstrapper
    {
        public static void Initialise()
        {
            var container = BuildUnityContainer();

            DependencyResolver.SetResolver(new UnityDependencyResolver(container));

            // register dependency resolver for WebAPI RC
            GlobalConfiguration.Configuration.DependencyResolver = new Unity.WebApi.UnityDependencyResolver(container);
        }

        private static IUnityContainer BuildUnityContainer()
        {
            var container = new UnityContainer();

            // register all your components with the container here
            // it is NOT necessary to register your controllers
            
            // e.g. container.RegisterType();        
            container.RegisterType().RegisterType(new HierarchicalLifetimeManager());

            return container;
        }
Like this we can also specify other dependent objects in BuildUnityContainerMethod.
Step 4: Now we need to call the Initialise method of Bootstrapper class. Note , we need the objects as soon as our modules load, therefore we require the container to do its work at the time of application load, therefore go to Global.asax file and add one line to call Initialise method, since this is a static method, we can directly call it using class name,
Bootstrapper.Initialise();
Our global.asax becomes,
using System.Linq;
using System.Web.Http;
using System.Web.Mvc;
using System.Web.Optimization;
using System.Web.Routing;
using Newtonsoft.Json;
using WebApi.App_Start;

namespace WebApi
{
    // Note: For instructions on enabling IIS6 or IIS7 classic mode, 
    // visit http://go.microsoft.com/?LinkId=9394801

    public class WebApiApplication : System.Web.HttpApplication
    {
        protected void Application_Start()
        {
            AreaRegistration.RegisterAllAreas();

            WebApiConfig.Register(GlobalConfiguration.Configuration);
            FilterConfig.RegisterGlobalFilters(GlobalFilters.Filters);
            RouteConfig.RegisterRoutes(RouteTable.Routes);
            BundleConfig.RegisterBundles(BundleTable.Bundles);
            //Initialise Bootstrapper
            Bootstrapper.Initialise();

            //Define Formatters
            var formatters = GlobalConfiguration.Configuration.Formatters;
            var jsonFormatter = formatters.JsonFormatter;
            var settings = jsonFormatter.SerializerSettings;
            settings.Formatting = Formatting.Indented;
            // settings.ContractResolver = new CamelCasePropertyNamesContractResolver();
            var appXmlType = formatters.XmlFormatter.SupportedMediaTypes.FirstOrDefault(t => t.MediaType == "application/xml");
            formatters.XmlFormatter.SupportedMediaTypes.Remove(appXmlType);

            //Add CORS Handler
            GlobalConfiguration.Configuration.MessageHandlers.Add(new CorsHandler());
        }
    }
}
Half of the job is done. We now need to touchbase our controller and Service class constructors to utilize the instances already created for them at application load.

Setup Controller

We have already set up unity in our application. There are various methods in which we can inject dependency, like constructor injection, property injection, via service locator. I am here using Constructor Injection, because I find it best method to use with Unity Container to resolve dependency.
Just go to your ProductController, you find your constructor written as,
        /// <summary>
        /// Public constructor to initialize product service instance
        /// </summary>
        public ProductController()
        {
            _productServices =new ProductServices();
        }
Just add a parameter to your constructor that takes your ProductServices reference, like we did below
       /// <summary>
        /// Public constructor to initialize product service instance
        /// </summary>
        public ProductController(IProductServices productServices)
        {
            _productServices = productServices;
        }
And initialize your “productServices” variable with the parameter. In this case when the constructor of the controller is called, It will be served with pre-instantiated service instance, and does not need to create an instance of the service, our unity container did the job of object creation.

Setup Services

For services too, we proceed in a same fashion. Just open your ProductServices class, we see the dependency of UnitOfWork here as,
        /// <summary>
        /// Public constructor.
        /// </summary>
        public ProductServices()
        {
            _unitOfWork = new UnitOfWork();
        }
Again, we perform the same steps ,and pass a parameter of type UnitOfWork to our constructor,
Our code becomes,
        /// <summary>
        /// Public constructor.
        /// </summary>
        public ProductServices(UnitOfWork unitOfWork)
        {
            _unitOfWork = unitOfWork;
        }
Here also we’ll get the pre instantiated object on UnitOfWork. So service does need to worry about creating objects. Remember we did .RegisterType() in Bootstrapper class.
We have now made our components independent.
Image source: https://codeteddycom.files.wordpress.com/2015/05/d91c4-h1.jpg

Running the application

Our job is almost done.We need to run the application, Just hit F5. To our surprise we’ll end up in an error page,
Do you remember we added a test client to our project to test our API in my first article. That test client have a controller too, we need to override its settings to make our application work.Just go to Areas->HelpPage->Controllers->HelpController in WebAPI project like shown below,
Comment out the existing constructors and add a Configuration property like shown below,
      //Remove constructors and existing Configuration property.

        //public HelpController()
        //    : this(GlobalConfiguration.Configuration)
        //{
        //}

        //public HelpController(HttpConfiguration config)
        //{
        //    Configuration = config;
        //}

        //public HttpConfiguration Configuration { get; private set; }

        /// <summary>
        /// Add new Configuration Property
        /// </summary>
        protected static HttpConfiguration Configuration
        {
            get { return GlobalConfiguration.Configuration; }
        }
Our controller code becomes,
using System;
using System.Web.Http;
using System.Web.Mvc;
using WebApi.Areas.HelpPage.Models;

namespace WebApi.Areas.HelpPage.Controllers
{
    /// <summary>
    /// The controller that will handle requests for the help page.
    /// </summary>
    public class HelpController : Controller
    {
        //Remove constructors and existing Configuration property.

        //public HelpController()
        //    : this(GlobalConfiguration.Configuration)
        //{
        //}

        //public HelpController(HttpConfiguration config)
        //{
        //    Configuration = config;
        //}

        //public HttpConfiguration Configuration { get; private set; }

        /// <summary>
        /// Add new Configuration Property
        /// </summary>
        protected static HttpConfiguration Configuration
        {
            get { return GlobalConfiguration.Configuration; }
        }

        public ActionResult Index()
        {
            return View(Configuration.Services.GetApiExplorer().ApiDescriptions);
        }

        public ActionResult Api(string apiId)
        {
            if (!String.IsNullOrEmpty(apiId))
            {
                HelpPageApiModel apiModel = Configuration.GetHelpPageApiModel(apiId);
                if (apiModel != null)
                {
                    return View(apiModel);
                }
            }

            return View("Error");
        }
    }
}
Just run the application, we get,
We alreay have our test client added, but for new readers, I am just again explaining on how to add a test client to our API project.
Just go to Manage Nuget Packages, by right clicking WebAPI project and type WebAPITestClient in searchbox in online packages,
You’ll get “A simple Test Client for ASP.NET Web API”, just add it. You’ll get a help controller in Areas-> HelpPage like shown below,
I have already provided the database scripts and data in my previous article, you can use the same.
Append “/help” in the application url, and you’ll get the test client,
You can test each service by clicking on it.
Service for GetAllProduct,
For Create a new product,
In database, we get new product,
Update product:
We get in database,
Delete product:
In database:
Job done.
Image source: http://codeopinion.com/wp-content/uploads/2015/02/injection.jpg

Design Flaws

What if I say there are still flaws in this design, the design is still not loosely coupled.
Do you remember what we decided while writing our first application?
Our API talks to Services and Services talk to DataModel. We’ll never allow DataModel talk to APIs for security reasons. But did you notice that when we were registering the type in Bootstrapper class, we also registered the type of UnitOfWork that means we added DataModel as a reference to our API project. This is a design breach. We tried to resolve dependency of a dependency by violating our design and compromising security.
In my next article, we’ll overcome this situation, we’ll try to resolve dependency and its dependency without violating our design and compromising security. In fact we’ll make it more secure and loosely coupled.
In my next article we’ll make use of Managed Extensibility Framework(MEF) to achieve the same.

Conclusion

We now know how to use Unity container to resolve dependency and perform inversion of control.
But still there are some flaws in this design. In my next article, I’ll try to make the system more strong. Till then Happy Coding  :-). You can also download the source code from GitHub. Add the required packages, if they are missing in the source code.