Context is King

Customize the SimpleMembership in ASP.NET MVC 4.0

2012-11-15T04:29:00Z

As we know, .NET 4.5 have come up to us, and come along with a lot of new interesting features as well. Visual Studio 2012 was also introduced some days ago. They made us feel very happy with cool improvement along with us. Performance when loading code editor is very good at the moment (immediate after click on the solution). I explore some of cool features at these days. Some of them like Json.NET integrated in ASP.NET MVC 4.0, improvement on asynchronous action, new lightweight theme on Visual Studio, supporting very good on mobile development, improvement on authentication…

I reviewed them, and found out that in this version of .NET Microsoft was not only developed new feature that suggest from community but also focused on improvement performance of existing features or components. Besides that, they also opened source more projects, like Entity Framework, Reactive Extensions, ASP.NET Web Stack… At the moment, I feel Microsoft want to open source more and more their projects.

Today, I am going to dive in deep on new SimpleMembership model. It is really good because in this security model, Microsoft actually focus on development needs. As we know, in the past, they introduce some of provider supplied for coding security like MembershipProvider, RoleProvider… I don’t need to talk but everyone that have ever used it know that they were actually hard to use, and not easy to maintain and unit testing. Why? Because every time you inherit it, you need to override all methods inside it. Some people try to abstract it by introduce more method with virtual keyword, and try to implement basic behavior, so in the subclass we only need to override the method that need for their business. But to me, it’s only the way to work around. ASP.NET team and Web Matrix knew about it, so they built the new features based on existing components on .NET framework. And one of component that comes to us is SimpleMembership and SimpleRole. They implemented the Façade pattern on the top of those, and called it is WebSecurity. In the web, we can call WebSecurity anywhere we want, and make a call to inside wrapper of it.

I read a lot of them on web blog, on technical news, on MSDN as well. Matthew Osborn had an excellent article about it at his blog. Jon Galloway had an article like this at here. He analyzed why old membership provider not fixed well to ASP.NET MVC and how to get over it.

Those are very good to me. It introduced to me about how to doing SimpleMembership on it, how to doing it on new ASP.NET MVC web application. But one thing, those didn’t tell me was how to doing it on existing security model (that mean we already had Users and Roles on legacy system, and how we can integrate it to this system), that’s a reason I will introduce it today. I have spent couples of hours to see what’s inside this, and try to make one example to clarify my concern. And it’s lucky that I can make it working well.

The first thing, we need to create new ASP.NET MVC application on Visual Studio 2012. We need to choose Internet type for this web application. ASP.NET MVC actually creates all needs components for the basic membership and basic role. The cool feature is DoNetOpenAuth come along with it that means we can log-in using facebook, twitter or Windows Live if you want. But it’s only for LocalDb, so we need to change it to fix with existing database model on SQL Server. The next step we have to make SimpleMembership can understand which database we use and show it which column need to point to for the ID and UserName. I really like this feature because SimpleMembership on need to know about the ID and UserName, and they don’t care about rest of it.
I assume that we have an existing database model like

So we will point it in code like

The codes for it, we put on InitializeSimpleMembershipAttribute like

    [AttributeUsage(AttributeTargets.Class | AttributeTargets.Method, AllowMultiple = false, Inherited = true)]
    public sealed class InitializeSimpleMembershipAttribute : ActionFilterAttribute
    {
        private static SimpleMembershipInitializer _initializer;
        private static object _initializerLock = new object();
        private static bool _isInitialized;
 
        public override void OnActionExecuting(ActionExecutingContext filterContext)
        {
            // Ensure ASP.NET Simple Membership is initialized only once per app start
            LazyInitializer.EnsureInitialized(ref _initializer, ref _isInitialized, ref _initializerLock);
        }
 
        private class SimpleMembershipInitializer
        {
            public SimpleMembershipInitializer()
            {
                try
                {
                    WebSecurity.InitializeDatabaseConnection("DefaultDb", "User", "Id", "UserName", autoCreateTables: true);
                }
                catch (Exception ex)
                {
                    throw new InvalidOperationException("The ASP.NET Simple Membership database could not be initialized. For more information, please see http://go.microsoft.com/fwlink/?LinkId=256588", ex);
                }
            }
        }
    }

And decorating it in the AccountController as below

    [Authorize]
    [InitializeSimpleMembership]
    public class AccountController : Controller

In this case, assuming that we need to override the ValidateUser to point this to existing User database table, and validate it. We have to add one more class like

    public class CustomAdminMembershipProvider : SimpleMembershipProvider
    {
        // TODO: will do a better way
        private const string SELECT_ALL_USER_SCRIPT = "select * from [dbo].[User]private where UserName = '{0}'";
 
        private readonly IEncrypting _encryptor;
 
        private readonly SimpleSecurityContext _simpleSecurityContext;
 
        public CustomAdminMembershipProvider(SimpleSecurityContext simpleSecurityContext)
            : this(new Encryptor(), new SimpleSecurityContext("DefaultDb"))
        {
        }
 
        public CustomAdminMembershipProvider(IEncrypting encryptor, SimpleSecurityContext simpleSecurityContext)
        {
            _encryptor = encryptor;
            _simpleSecurityContext = simpleSecurityContext;
        }
 
        public override bool ValidateUser(string username, string password)
        {
            if (string.IsNullOrEmpty(username))
            {
                throw new ArgumentException("Argument cannot be null or empty", "username");
            }
 
            if (string.IsNullOrEmpty(password))
            {
                throw new ArgumentException("Argument cannot be null or empty", "password");
            }
 
            var hash = _encryptor.Encode(password);
 
            using (_simpleSecurityContext)
            {
                var users =
                    _simpleSecurityContext.Users.SqlQuery(
                        string.Format(SELECT_ALL_USER_SCRIPT, username));
 
                if (users == null && !users.Any())
                {
                    return false;
                }
 
                return users.FirstOrDefault().Password == hash;
            }
        }
    }

SimpleSecurityDataContext at here

    public class SimpleSecurityContext : DbContext
    {
        public DbSet<User> Users { get; set; }
 
        public SimpleSecurityContext(string connStringName) :
            base(connStringName)
        {
            this.Configuration.LazyLoadingEnabled = true;
            this.Configuration.ProxyCreationEnabled = false;
        }
 
        protected override void OnModelCreating(DbModelBuilder modelBuilder)
        {
            base.OnModelCreating(modelBuilder);
            
            modelBuilder.Configurations.Add(new UserMapping());
        }
    }

And Mapping for User as below

    public class UserMapping : EntityMappingBase<User>
    {
        public UserMapping()
        {
            this.Property(x => x.UserName);
            this.Property(x => x.DisplayName);
            this.Property(x => x.Password);
            this.Property(x => x.Email);
 
            this.ToTable("User");
        }
    }

One important thing, you need to modify the web.config to point to our customize SimpleMembership

   <membership defaultProvider="AdminMemberProvider" userIsOnlineTimeWindow="15">
      <providers>
        <clear/>
        <add name="AdminMemberProvider" type="CIK.News.Web.Infras.Security.CustomAdminMembershipProvider, CIK.News.Web.Infras" />
      </providers>
    </membership>
    <roleManager enabled="false">
      <providers>
        <clear />
        <add name="AdminRoleProvider" type="CIK.News.Web.Infras.Security.AdminRoleProvider, CIK.News.Web.Infras" />
      </providers>
    </roleManager>

The good thing at here is we don’t need to modify the code on AccountController. We only need to modify on SimpleMembership and Simple Role (if need).
Now build all solutions, run it. We should see a screen like this

If I login to Twitter button at the bottom of this page, we will be transfer to twitter authentication page

You have to waiting for a moment

Afterwards it will transfer you back to your admin screen

You can find all source codes at my MSDN code. I will really happy if you guys feel free to put some comments as below. It will be helpful to improvement my code in the future. Thank for all your readings.

Skinny controller in ASP.NET MVC 4

2012-06-28T11:06:00Z

Rails community are always inspire a lot of best ideas. I really love this community by the time. One of these is "Fat models and skinny controllers". I have spent a lot of time on ASP.NET MVC, and really I did some miss-takes, because I made the controller so fat. That make controller is really dirty and very hard to maintain in the future. It is violate seriously SRP principle and KISS as well. But how can we achieve that in ASP.NET MVC? That question is really clear after I read "Manning ASP.NET MVC 4 in Action". It is simple that we can separate it into ActionResult, and try to implementing logic and persistence data inside this. In last 2 years, I have read this from Jimmy Bogard blog, but in that time I never had a consideration about it. That's enough for talking now.

I just published a sample on ASP.NET MVC 4, implemented on Visual Studio 2012 RC at here. I used EF framework at here for implementing persistence layer, and also use 2 free templates from internet to make the UI for this sample.

In this sample, I try to implementing the simple magazine website that managing all articles, categories and news. It is not finished at all in this time, but no problems, because I just show you about how can we make the controller skinny. And I wanna hear more about your ideas.

The first thing, I am abstract the base ActionResult class like this:

    public abstract class MyActionResult : ActionResult, IEnsureNotNull
    {
        public abstract void EnsureAllInjectInstanceNotNull();
    }
 
    public abstract class ActionResultBase<TController> : MyActionResult where TController : Controller
    {
        protected readonly Expression<Func<TController, ActionResult>> ViewNameExpression;
        protected readonly IExConfigurationManager ConfigurationManager;
 
        protected ActionResultBase (Expression<Func<TController, ActionResult>> expr)
            : this(DependencyResolver.Current.GetService<IExConfigurationManager>(), expr)
        {
        }
 
        protected ActionResultBase(
            IExConfigurationManager configurationManager,
            Expression<Func<TController, ActionResult>> expr)
        {
            Guard.ArgumentNotNull(expr, "ViewNameExpression");
            Guard.ArgumentNotNull(configurationManager, "ConfigurationManager");
 
            ViewNameExpression = expr;
            ConfigurationManager = configurationManager;
        }
 
        protected ViewResult GetViewResult<TViewModel>(TViewModel viewModel)
        {
            var m = (MethodCallExpression)ViewNameExpression.Body;
            if (m.Method.ReturnType != typeof(ActionResult))
            {
                throw new ArgumentException("ControllerAction method '" + m.Method.Name + "' does not return type ActionResult");
            }
 
            var result = new ViewResult
            {
                ViewName = m.Method.Name
            };
 
            result.ViewData.Model = viewModel;
 
            return result;
        }
 
        public override void ExecuteResult(ControllerContext context)
        {
            EnsureAllInjectInstanceNotNull();
        }
    }

I also have an interface for validation all inject objects. This interface make sure all inject objects that I inject using Autofac container are not null. The implementation of this as below

    public interface IEnsureNotNull
    {
        void EnsureAllInjectInstanceNotNull();
    }

Afterwards, I am just simple implementing the HomePageViewModelActionResult class like this

    public class HomePageViewModelActionResult<TController> : ActionResultBase<TController> where TController : Controller
    {
        #region variables & ctors
 
        private readonly ICategoryRepository _categoryRepository;
        private readonly IItemRepository _itemRepository;
 
        private readonly int _numOfPage;
 
        public HomePageViewModelActionResult(Expression<Func<TController, ActionResult>> viewNameExpression)
            : this(viewNameExpression,
                   DependencyResolver.Current.GetService<ICategoryRepository>(),
                   DependencyResolver.Current.GetService<IItemRepository>())
        {
        }
 
        public HomePageViewModelActionResult(
            Expression<Func<TController, ActionResult>> viewNameExpression,
            ICategoryRepository categoryRepository,
            IItemRepository itemRepository)
            : base(viewNameExpression)
        {
            _categoryRepository = categoryRepository;
            _itemRepository = itemRepository;
 
            _numOfPage = ConfigurationManager.GetAppConfigBy("NumOfPage").ToInteger();
        }
 
        #endregion
 
        #region implementation
 
        public override void ExecuteResult(ControllerContext context)
        {
            base.ExecuteResult(context);
 
            var cats = _categoryRepository.GetCategories();
 
            var mainViewModel = new HomePageViewModel();
            var headerViewModel = new HeaderViewModel();
            var footerViewModel = new FooterViewModel();
            var mainPageViewModel = new MainPageViewModel();
 
            headerViewModel.SiteTitle = "Magazine Website";
            if (cats != null && cats.Any())
            {
                headerViewModel.Categories = cats.ToList();
                footerViewModel.Categories = cats.ToList();
            }
 
            mainPageViewModel.LeftColumn = BindingDataForMainPageLeftColumnViewModel();
            mainPageViewModel.RightColumn = BindingDataForMainPageRightColumnViewModel();
 
            mainViewModel.Header = headerViewModel;
            mainViewModel.DashBoard = new DashboardViewModel();
            mainViewModel.Footer = footerViewModel;
            mainViewModel.MainPage = mainPageViewModel;
 
            GetViewResult(mainViewModel).ExecuteResult(context);
        }
 
        public override void EnsureAllInjectInstanceNotNull()
        {
            Guard.ArgumentNotNull(_categoryRepository, "CategoryRepository");
            Guard.ArgumentNotNull(_itemRepository, "ItemRepository");
            Guard.ArgumentMustMoreThanZero(_numOfPage, "NumOfPage");
        }
 
        #endregion
 
        #region private functions
 
        private MainPageRightColumnViewModel BindingDataForMainPageRightColumnViewModel()
        {
            var mainPageRightCol = new MainPageRightColumnViewModel();
 
            mainPageRightCol.LatestNews = _itemRepository.GetNewestItem(_numOfPage).ToList();
            mainPageRightCol.MostViews = _itemRepository.GetMostViews(_numOfPage).ToList();
 
            return mainPageRightCol;
        }
 
        private MainPageLeftColumnViewModel BindingDataForMainPageLeftColumnViewModel()
        {
            var mainPageLeftCol = new MainPageLeftColumnViewModel();
 
            var items = _itemRepository.GetNewestItem(_numOfPage);
 
            if (items != null && items.Any())
            {
                var firstItem = items.First();
 
                if (firstItem == null)
                    throw new NoNullAllowedException("First Item".ToNotNullErrorMessage());
 
                if (firstItem.ItemContent == null)
                    throw new NoNullAllowedException("First ItemContent".ToNotNullErrorMessage());
 
                mainPageLeftCol.FirstItem = firstItem;
 
                if (items.Count() > 1)
                {
                    mainPageLeftCol.RemainItems = items.Where(x => x.ItemContent != null && x.Id != mainPageLeftCol.FirstItem.Id).ToList();
                }
            }
 
            return mainPageLeftCol;
        }
 
        #endregion
    }

Final step, I get into HomeController and add some line of codes like this

    [Authorize]
    public class HomeController : BaseController
    {
        [AllowAnonymous]
        public ActionResult Index()
        {
            return new HomePageViewModelActionResult<HomeController>(x=>x.Index());
        }
 
        [AllowAnonymous]
        public ActionResult Details(int id)
        {
            return new DetailsViewModelActionResult<HomeController>(x => x.Details(id), id);
        }
 
        [AllowAnonymous]
        public ActionResult Category(int id)
        {
            return new CategoryViewModelActionResult<HomeController>(x => x.Category(id), id);
        }
    }

As you see, the code in controller is really skinny, and all the logic I move to the custom ActionResult class. Some people said, it just move the code out of controller and put it to another class, so it is still hard to maintain. Look like it just move the complicate codes from one place to another place. But if you have a look and think it in details, you have to find out if you have code for processing all logic that related to HttpContext or something like this. You can do it on Controller, and try to delegating another logic (such as processing business requirement, persistence data,...) to custom ActionResult class.

Tell me more your thinking, I am really willing to hear all of its from you guys.

All source codes can be find out at here.

BDD with Specflow, Moq and Microsoft Unit Testing framework

2011-08-15T16:13:00Z

BDD (Behavior Driven Development) is very popular in these days. I know a lot of companies used BDD in our projects. So I also had some curios about it and I applied for my current project. I used Specflow, Moq and Microsoft Testing framework for working. I did it and found out some things very interested in. In this post I will not focus on what it BDD? What is Mock, stub or how to write a unit testing? Readers can easy find on internet, just googling it and everything will be okay. I just focus on the some best things I found when I implemented the BDD on my code. I used AAA (Arrange-Act-Assert) and GWT (Given-When-Then) patterns in my code. They are clearly and easily to understand and become standard when you write unit testing and mocking.

Now we discuss about a problem when we write unit testing and how to get rid of it. So what's problem? When we write unit testing and try to arrange a lot of things to testing. And the code will be messy and a line of code will become unmanageable, so it is hard to maintain at will. I will give you a simple sample. Assuming we have a small application for get all articles from database. To keep it simple I will not use ORM for communicate and mapping object between Object Model and Database. So the article will have a lot of information to populating. Some of them are Category, Article Information, Author Information, Article content, Article Media Item, Keyword, Topic and Sub topics. And we need to populate all of this, look like a complicate scenario.

We try to make it work. And it has many repositories inside the article manager class. So that mean your mock objects will equal with number of repositories in article manager class. What's happen with your test class? You will declare an object for mock and in its constructor; you will mock it and set it to parameter constructor for article manager class. Like this

var dummyRepositoryMock1 = new Mock<IDummyRepository1>();

var dummyRepositoryMock2 = new Mock<IDummyRepository2>();

var dummyRepositoryMock3 = new Mock<IDummyRepository3>();

var dummyRepositoryMock4 = new Mock<IDummyRepository4>();

var dummyRepositoryMock5 = new Mock<IDummyRepository5>();

var dummyRepositoryMock6 = new Mock<IDummyRepository6>();

// prepare the stub objects for return

var retObject1 = ...;

var retObject2 = ...;

var retObject3 = ...;

var retObject4 = ...;

var retObject5 = ...;

var retObject6 = ...;

dummyRepository1.Setup(framework=>framework.SomeMethodCall(It.IsAny<int>())).Returns(retObjects1);

dummyRepository2.Setup(framework=>framework.SomeMethodCall(It.IsAny<int>())).Returns(retObjects2);

dummyRepository3.Setup(framework=>framework.SomeMethodCall(It.IsAny<int>())).Returns(retObjects3);

dummyRepository4.Setup(framework=>framework.SomeMethodCall(It.IsAny<int>())).Returns(retObjects4);

dummyRepository5.Setup(framework=>framework.SomeMethodCall(It.IsAny<int>())).Returns(retObjects5);

dummyRepository6.Setup(framework=>framework.SomeMethodCall(It.IsAny<int>())).Returns(retObjects6);

and the constructor of Article manager class will be

var articleManager = new ArticleManager(

dummyRepositoryMock1.Object,

dummyRepositoryMock2.Object,

dummyRepositoryMock3.Object,

dummyRepositoryMock4.Object,

dummyRepositoryMock5.Object,

dummyRepositoryMock6.Object);

So how can we optimize a line of code for this processing? I write the abstract class like this

    public abstract class AutoMockObject
    {
        private readonly IDictionary<Type, dynamic> _mocks;
 
        protected AutoMockObject()
        {
            _mocks = new Dictionary<Type, dynamic>();
        }
 
        public virtual void Register<TObject>(Action<Mock<TObject>> action) where TObject : class
        {
            var mockObject = new Mock<TObject>();
            _mocks.Add(typeof(TObject), mockObject);
            action(mockObject);
        }
 
        public virtual TObject Resolve<TObject>() where TObject : class
        {
            dynamic mock;
 
            return !_mocks.TryGetValue(typeof(TObject), out mock) ? null : mock.Object;
        }
    }

Make sure you test class should be inherited from AutoMockObject. And finally I just mock it with a bit effort as below

            Register<IArticleRepository>(mock =>
                                             {
                                                 mock.Setup(framework => framework.GetAll()).Returns(GetAllArticleStub.ArticleList);
                                                 mock.Setup(framework => framework.GetKeywordsByArticleId(It.IsAny<int>())).Returns(GetAllArticleStub.KeywordList);
                                                 mock.Setup(framework => framework.GetTopicsByArticleId(It.IsAny<int>(), false)).Returns(GetAllArticleStub.TopicList);
                                                 mock.Setup(framework => framework.GetTopicsByArticleId(It.IsAny<int>(), true)).Returns(GetAllArticleStub.TopicList);
                                             });
 
            Register<IArticleCategoryRepository>(mock => mock.Setup(framework => framework.GetById(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleCategory));
            Register<IArticleInfoRepository>(mock => mock.Setup(framework => framework.GetArticleInformationById(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleInformation));
            Register<IAuthorInfoRepository>(mock => mock.Setup(framework => framework.GetAuthorInformationById(It.IsAny<int>())).Returns(GetAllArticleStub.AuthorInformation));
            Register<IArticleContentRepository>(mock => mock.Setup(framework => framework.GetSummaryByArticleId(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleContentList));
            Register<IArticleMediaItemRepository>(mock =>
                                                      {
                                                          mock.Setup(framework => framework.GetArticleMediaItemByArticleId(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleMediaItemList);
                                                          mock.Setup(framework => framework.GetArticleMediaItemAltTextByArticleMediaItemId(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleMediaItemAltTextList);
                                                      });
 
            _articleManager = new ArticleManager(
                    Resolve<IArticleRepository>(),
                    Resolve<IArticleCategoryRepository>(),
                    Resolve<IArticleContentRepository>(),
                    Resolve<IReferenceRepository>(),
                    Resolve<IArticleInfoRepository>(),
                    Resolve<IAuthorInfoRepository>(),
                    Resolve<IPublishingArticleRepository>(),
                    Resolve<IArticleMediaItemRepository>()
                );

Look like simple and elegant, isn't it? Just a little code for get rid of messy and redundant code.

Now I will show you how can I make all solutions working.

The first thing I will write a feature like this

Feature: Get all articles from Repository
	In order to get all articles
	As a cms user
	So that I can get and process in all articles

@GetAllArticles
Scenario: Get All articles
	Given I don't input any parametters	
	When I call the GetAllArticle function from service
	Then the result should be a list of article

@GetAllArticlesWithError
Scenario: Get All articles with error
	Given I dont input any parametters
	When I call the GetAllArticle function from service
	Then the result should be a exception

Add App.config and add some line of code as below:

  <configSections>
    <section
       name="specFlow"
       type="TechTalk.SpecFlow.Configuration.ConfigurationSectionHandler, TechTalk.SpecFlow"/>
  </configSections>
  <specFlow>
    <unitTestProvider name="MsTest" />
  </specFlow>

And after that Specflow will generate the code for Microsoft Testing framework (default it will generate for NUnit testing) as below

// ------------------------------------------------------------------------------
//  <auto-generated>
//      This code was generated by SpecFlow (http://www.specflow.org/).
//      SpecFlow Version:1.7.0.0
//      SpecFlow Generator Version:1.7.0.0
//      Runtime Version:4.0.30319.431
// 
//      Changes to this file may cause incorrect behavior and will be lost if
//      the code is regenerated.
//  </auto-generated>
// ------------------------------------------------------------------------------
#region Designer generated code
namespace xxx.xxx.xxx
{
    using TechTalk.SpecFlow;
    
    
    [System.CodeDom.Compiler.GeneratedCodeAttribute("TechTalk.SpecFlow", "1.7.0.0")]
    [System.Runtime.CompilerServices.CompilerGeneratedAttribute()]
    [Microsoft.VisualStudio.TestTools.UnitTesting.TestClassAttribute()]
    public partial class GetAllArticlesFromRepositoryFeature
    {
        
        private static TechTalk.SpecFlow.ITestRunner testRunner;
        
#line 1 "GetAllArticles.feature"
#line hidden
        
        [Microsoft.VisualStudio.TestTools.UnitTesting.ClassInitializeAttribute()]
        public static void FeatureSetup(Microsoft.VisualStudio.TestTools.UnitTesting.TestContext testContext)
        {
            testRunner = TechTalk.SpecFlow.TestRunnerManager.GetTestRunner();
            TechTalk.SpecFlow.FeatureInfo featureInfo = new TechTalk.SpecFlow.FeatureInfo(new System.Globalization.CultureInfo("en-US"), "Get all articles from Repository", "In order to get all articles\r\nAs a cms user\r\nSo that I can get and process in all" +
                    " articles", ProgrammingLanguage.CSharp, ((string[])(null)));
            testRunner.OnFeatureStart(featureInfo);
        }
        
        [Microsoft.VisualStudio.TestTools.UnitTesting.ClassCleanupAttribute()]
        public static void FeatureTearDown()
        {
            testRunner.OnFeatureEnd();
            testRunner = null;
        }
        
        [Microsoft.VisualStudio.TestTools.UnitTesting.TestInitializeAttribute()]
        public virtual void TestInitialize()
        {
            if (((TechTalk.SpecFlow.FeatureContext.Current != null) 
                        && (TechTalk.SpecFlow.FeatureContext.Current.FeatureInfo.Title != "Get all articles from Repository")))
            {
                xxx.xxx.xxx.GetAllArticlesFromRepositoryFeature.FeatureSetup(null);
            }
        }
        
        [Microsoft.VisualStudio.TestTools.UnitTesting.TestCleanupAttribute()]
        public virtual void ScenarioTearDown()
        {
            testRunner.OnScenarioEnd();
        }
        
        public virtual void ScenarioSetup(TechTalk.SpecFlow.ScenarioInfo scenarioInfo)
        {
            testRunner.OnScenarioStart(scenarioInfo);
        }
        
        public virtual void ScenarioCleanup()
        {
            testRunner.CollectScenarioErrors();
        }
        
        [Microsoft.VisualStudio.TestTools.UnitTesting.TestMethodAttribute()]
        [Microsoft.VisualStudio.TestTools.UnitTesting.DescriptionAttribute("Get All articles")]
        [Microsoft.VisualStudio.TestTools.UnitTesting.TestPropertyAttribute("FeatureTitle", "Get all articles from Repository")]
        public virtual void GetAllArticles()
        {
            TechTalk.SpecFlow.ScenarioInfo scenarioInfo = new TechTalk.SpecFlow.ScenarioInfo("Get All articles", new string[] {
                        "GetAllArticles"});
#line 7
this.ScenarioSetup(scenarioInfo);
#line 8
 testRunner.Given("I don\'t input any parametters");
#line 9
 testRunner.When("I call the GetAllArticle function from service");
#line 10
 testRunner.Then("the result should be a list of article");
#line hidden
            this.ScenarioCleanup();
        }
        
        [Microsoft.VisualStudio.TestTools.UnitTesting.TestMethodAttribute()]
        [Microsoft.VisualStudio.TestTools.UnitTesting.DescriptionAttribute("Get All articles with error")]
        [Microsoft.VisualStudio.TestTools.UnitTesting.TestPropertyAttribute("FeatureTitle", "Get all articles from Repository")]
        public virtual void GetAllArticlesWithError()
        {
            TechTalk.SpecFlow.ScenarioInfo scenarioInfo = new TechTalk.SpecFlow.ScenarioInfo("Get All articles with error", new string[] {
                        "GetAllArticlesWithError"});
#line 13
this.ScenarioSetup(scenarioInfo);
#line 14
 testRunner.Given("I dont input any parametters");
#line 15
 testRunner.When("I call the GetAllArticle function from service");
#line 16
 testRunner.Then("the result should be a exception");
#line hidden
            this.ScenarioCleanup();
        }
    }
}
#endregion

Look like terrible and hard to understand, itsn't it? No problem, don't care about it. It will be fine Don't scare it :)

The first thing, I try to write some interface to make a skeleton for solution

    public interface IArticleRepository : IRepositoryEntityBase<Article>
    {
        ...
    }

    public interface IArticleCategoryRepository : IRepositoryEntityBase<ArticleCategory>
    {
 
    }

    public interface IArticleInfoRepository : IRepositoryEntityBase<ArticleInfo>
    {
        ...
    }

    public interface IArticleContentRepository : IRepositoryEntityBase<ArticleContent>
    {
        ....}

    public interface IArticleMediaItemRepository : IRepositoryEntityBase<MediaItem>
    {
        
    }

and an interface for article manager

    public interface IArticleManager
    {

}

Certainly, we should have article manager instance class

    public class ArticleManager : IArticleManager
    {
        private readonly IArticleContentRepository _articleContentRepository;
        private readonly IArticleInfoRepository _articleInfoRepository;
        private readonly IArticleRepository _articleRepository;
        private readonly IAuthorInfoRepository _authorInfoRepository;
        private readonly IArticleCategoryRepository _categoryRepository;
        private readonly IPublishingArticleRepository _publishingArticleRepository;
        private readonly IReferenceRepository _referenceRepository;
        private readonly IArticleMediaItemRepository _articleMediaItemRepository;
 
         
        public ArticleManager(IArticleRepository articleRepository
                              , IArticleCategoryRepository categoryRepository
                              , IArticleContentRepository articleContentRepository
                              , IReferenceRepository referenceRepository
                              , IArticleInfoRepository articleInfoRepository
                              , IAuthorInfoRepository authorInfoRepository
                              , IPublishingArticleRepository publishingArticleRepository
                              , IArticleMediaItemRepository articleMediaItemRepository)
        {
            _articleRepository = articleRepository;
            _authorInfoRepository = authorInfoRepository;
            _articleInfoRepository = articleInfoRepository;
            _referenceRepository = referenceRepository;
            _articleContentRepository = articleContentRepository;
            _categoryRepository = categoryRepository;
            _publishingArticleRepository = publishingArticleRepository;
            _articleMediaItemRepository = articleMediaItemRepository;
        }

....................

}

Note: I use Unity for dependency injection framework, but I will not mention it this post, because it is very easy to achive.

After that I add 2 step definitions for get all article and get all article with an exception should be throw out

    [Binding]
    public class GetAllArticleSteps : AutoMockObject
    {
        private IArticleManager _articleManager;
        private IEnumerable<Article> _result;
 
        [Given(@"I don't input any parametters")]
        [StepScope(Tag = "GetAllArticles")]
        public void GivenIDontInputAnyParamatters()
        {
            Register<IArticleRepository>(mock =>
                                             {
                                                 mock.Setup(framework => framework.GetAll()).Returns(GetAllArticleStub.ArticleList);
                                                 mock.Setup(framework => framework.GetKeywordsByArticleId(It.IsAny<int>())).Returns(GetAllArticleStub.KeywordList);
                                                 mock.Setup(framework => framework.GetTopicsByArticleId(It.IsAny<int>(), false)).Returns(GetAllArticleStub.TopicList);
                                                 mock.Setup(framework => framework.GetTopicsByArticleId(It.IsAny<int>(), true)).Returns(GetAllArticleStub.TopicList);
                                             });
 
            Register<IArticleCategoryRepository>(mock => mock.Setup(framework => framework.GetById(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleCategory));
            Register<IArticleInfoRepository>(mock => mock.Setup(framework => framework.GetArticleInformationById(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleInformation));
            Register<IAuthorInfoRepository>(mock => mock.Setup(framework => framework.GetAuthorInformationById(It.IsAny<int>())).Returns(GetAllArticleStub.AuthorInformation));
            Register<IArticleContentRepository>(mock => mock.Setup(framework => framework.GetSummaryByArticleId(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleContentList));
            Register<IArticleMediaItemRepository>(mock =>
                                                      {
                                                          mock.Setup(framework => framework.GetArticleMediaItemByArticleId(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleMediaItemList);
                                                          mock.Setup(framework => framework.GetArticleMediaItemAltTextByArticleMediaItemId(It.IsAny<int>())).Returns(GetAllArticleStub.ArticleMediaItemAltTextList);
                                                      });
 
            _articleManager = new ArticleManager(
                    Resolve<IArticleRepository>(),
                    Resolve<IArticleCategoryRepository>(),
                    Resolve<IArticleContentRepository>(),
                    Resolve<IReferenceRepository>(),
                    Resolve<IArticleInfoRepository>(),
                    Resolve<IAuthorInfoRepository>(),
                    Resolve<IPublishingArticleRepository>(),
                    Resolve<IArticleMediaItemRepository>()
                );
        }
 
        [When(@"I call the GetAllArticle function from service")]
        [StepScope(Tag = "GetAllArticles")]
        public void WhenICallTheGetAllArticleFunctionFromTheService()
        {
            _result = _articleManager.GetAllArticle();
        }
 
        [Then(@"the result should be a list of article")]
        [StepScope(Tag = "GetAllArticles")]
        public void ThenTheResultShouldBe()
        {
            Assert.IsNotNull(_result);
            Assert.IsTrue(_result.Count() == 1);
        }
    }

    [Binding]
    public class GetAllArticleWithErrorSteps
    {
        private IArticleManager _articleManager;
        private IEnumerable<Article> _result;
 
        [Given(@"I dont input any parametters")]
        [StepScope(Tag = "GetAllArticlesWithError")]
        public void GivenIDontInputAnyParamettersWithError()
        {
            _articleManager = new ArticleManager(null, null, null, null, null, null, null, null);
        }
 
        [When(@"I call the GetAllArticle function from service")]
        [StepScope(Tag = "GetAllArticlesWithError")]
        public void WhenICallTheGetAllArticleFunctionFromServiceWithError()
        {
            try
            {
                _result = _articleManager.GetAllArticle();
            }
            catch (Exception e)
            {
                ScenarioContext.Current.Add("NullException_GetAllArticleMethod", e);
            }
        }
 
        [Then("the result should be a exception")]
        [StepScope(Tag = "GetAllArticlesWithError")]
        public void ThenTheResultShouldBe()
        {
            Assert.IsTrue(ScenarioContext.Current.ContainsKey("NullException_GetAllArticleMethod"));
        }
    }

This is just a simple scenario for demo. Nothing is complex here.

Now try to run all unit testings inside Visual Studio 2010, it will be can run because you don't have an implementation for article business class. Now we add some code for it.

        public IEnumerable<Article> GetAllArticle()
        {
            IEnumerable<Article> articles = _articleRepository.GetAll();
 
            foreach (var article in articles)
            {
                SetReferenceData(article);
            }
 
            return articles;
        }

        private void SetReferenceData(Article article)
        {
            Guard.ArgumentNotNull(article, "article");
 
            article.Category = _categoryRepository.GetById(article.CategoryId);
            article.ArticleInformation = _articleInfoRepository.GetArticleInformationById(article.ArticleInformationId);
            article.AuthorInformation = _authorInfoRepository.GetAuthorInformationById(article.AuthorInformationId);
 
            article.Keywords = _articleRepository.GetKeywordsByArticleId(article.ArticleId).ToList();
            article.Topics = _articleRepository.GetTopicsByArticleId(article.ArticleId, false).ToList();
            article.SubTopics = _articleRepository.GetTopicsByArticleId(article.ArticleId, true).ToList();
 
            List<ArticleContent> summaryContent = _articleContentRepository.GetSummaryByArticleId(article.ArticleId).ToList();
            if (summaryContent.Count > 1)
            {
                var baseContent = summaryContent.Find(ac => ac.LocaleCode == article.LocaleCode);
                summaryContent.Remove(baseContent);
                summaryContent.Insert(0, baseContent);
            }
            article.ArticleContents = summaryContent;
 
            var mediaItems = _articleMediaItemRepository.GetArticleMediaItemByArticleId(article.ArticleId);
 
            foreach (var articleMediaItem in mediaItems)
            {
                articleMediaItem.ArticleMediaItemAltTexts = _articleMediaItemRepository.GetArticleMediaItemAltTextByArticleMediaItemId(
                                                                                                        articleMediaItem.ArticleMediaItemId);
            }
 
            article.MediaItems = mediaItems.ToList();
        }

Make sure you add all methods for all repositories, try to run test again. And now it pass. How about your thinking? All comment for discussion are welcome. Happy coding and see you next post.

WCF Web API with Web Activator package

2011-07-27T16:24:00Z

In these days, I had read some blogs and a lot of post in internet about WCF Web API. And I also have curious about it. It is not new technology. And after I watched some postcast. I recognize it is a good one that I must to researching. RESTfull is a hot topic and have a lot of argues these days. WCF Web API is built on top of WCF service. How does it good? I don't know we just follow and analyze it later. Do you agree with me?

As you know, if you host WCF service to a IIS host, and you want to use IoC inside it. You must to customize some classes in WCF factory to do it, or you can add Global.asax file and try to override the start method for register all things you need. When .NET 4 came to us, it brought with a new technology for start application, that mean you don't need to customize WCF factory or add Global.asax file. As me, we only need to customize WCF when we have a special case on it, for example we need to filter action, AOP on WCF for logging, exception management. So what is technical I use for? Hm...hm. The first thing we need to config the WCF service first. We just open Visual studio and add WCF servce project. It's easy, isn't it? So I will don't mention this step here. Second step is add some packages from NuGet command. And I also don't mention about how do you add this with NuGet? You can find on google for this. A few of packages that you need to add are WebApi.Core, WebActivator, Autofac, CommonServiceLocator. You can see we add WebActivator pakage, this is a package that we need for start the application. And after that I add a class called AppStart.cs. Content of it is here:

[assembly: WebActivator.PreApplicationStartMethod(typeof(AppStart), "Start")]

namespace SampleService
{
    public static class AppStart
    {
        public static void Start()
        {
            var builder = new ContainerBuilder();
            builder.RegisterType<SampleResouce>();
            builder.RegisterType<ResourceProvider>().As<IResourceProvider>();

            var container = builder.Build();

            ServiceLocator.SetLocatorProvider(() => new AutofacServiceLocator(container));

            var configuration = HttpHostConfiguration.Create().SetResourceFactory((t, i, r) => container.Resolve(t), null);

            RouteTable.Routes.MapServiceRoute<SampleResouce>("categories", configuration);
        }
    }
}

I use PreApplicationStart method in Web Activator for start application. I set up Autofac container, register it and assign it to Microsoft Common Service Locator. And filnally I map it to WCF Web API for register Routing service. That's it.

Because I use Autofac for IoC container, so I add a class with named is ObjectFactory.cs:

    internal static class ObjectFactory
    {
        public static TService GetService<TService>()
        {
            return ServiceLocator.Current.GetInstance<TService>();
        }
    }

This class is simple. If you want more methods for resolve the object. Please at here. I just keep it simple.

To demostrate about dependency injection, I add a stub class for providing all resources for WCF service.

    internal interface IResourceProvider
    {
        List<string> GetAllString();
    }

    internal class ResourceProvider : IResourceProvider
    {
        public List<string> GetAllString()
        {
            return new List<string> { "Books", "Fruit" };
        }
    }

Finally, we have a WCF service with web api like this:

    [ServiceContract]
    internal class SampleResouce
    {
        private readonly IResourceProvider _resourceProvider;
 
        public SampleResouce()
            : this(ObjectFactory.GetService<IResourceProvider>())
        {
        }
 
        public SampleResouce(IResourceProvider resourceProvider)
        {
            _resourceProvider = resourceProvider;
        }
 
        [WebGet(UriTemplate = "")]
        public List<string> GetAllString()
        {
            return _resourceProvider.GetAllString();
        }
 

        [WebGet(UriTemplate = "{item}")]
        public string GetString(string item)
        {
            return item;
        }
    }

You can see I have a method for get all strings and one method for get string by name. Because it is RESTful so you can access it in your url. You only need to type like this http://localhost:xxxx/categories or http://localhost:xxxx/categories/books and see what happen. (xxxx is a port that your application is host in) Look good and clean, isn't it? Hope my post will make you happy about register and use WCF Web API and Web Activator. Do you have any suggestion for improve my post?

You can get a source code at here

Razor Themed View Engine for multi-themes site

2011-06-09T17:24:00Z

All sources code for this post can be find at here.

I have just updated the source codes and the post with the suggest from gbitdiff. Thanks for your comment.

Have you ever implemented a multi-themes site using Razor view engine? Yes it is really good to do that. Specially, you can do it and make it work well with Razor. I spent a lot of time to investigate this problem. After searched on Google, I had found some of links very useful about this topic. But all of them also implemented in ASP.NET MVC 2.0 and custom a WebformViewEngine. So I know I must work with it from scratch. But no problem, I decided to analyses some examples at here and here. They are good samples for my work. And I can re-use the ViewEngine from them, only change from WebformViewEngine to RazorViewEngine.
What components do we need to make your site using multi themes? We only need css, images and certainly views for make its working. Basic idea is customize the ViewEngine and when the ASP.NET MVC want to find a image, css, view or partial view, we will point to the valid one in theme folder. That's it. After 4 hours to implemented it, I very happy because it can work very well. And this is my result for default theme:

red theme

and green theme

this is a project structure for multi-themes

As you see, I use the path Content\themes\... is same with the organisation of ASP.NET MVC 3.0, and the path ~\Views\... contain all views. And now I will show you some important line of code to implemented it.
First thing is RazorThemeViewEngine

    public class RazorThemeViewEngine : RazorViewEngine
    {
        public RazorThemeViewEngine()
        {
            MasterLocationFormats = new[] {
                "~/Views/{2}/{1}/{0}.cshtml",
                "~/Views/{2}/{1}/{0}.vbhtml",
                "~/Views/{2}/Shared/{0}.cshtml",
                "~/Views/{2}/Shared/{0}.vbhtml"
            };
            ViewLocationFormats = new[] {
                "~/Views/{2}/{1}/{0}.cshtml",
                "~/Views/{2}/{1}/{0}.vbhtml",
                "~/Views/{2}/Shared/{0}.cshtml",
                "~/Views/{2}/Shared/{0}.vbhtml"
            };
            PartialViewLocationFormats = new[] {
                "~/Views/{2}/{1}/{0}.cshtml",
                "~/Views/{2}/{1}/{0}.vbhtml",
                "~/Views/{2}/Shared/{0}.cshtml",
                "~/Views/{2}/Shared/{0}.vbhtml"
            };
        }
 
        protected override bool FileExists(ControllerContext controllerContext, string virtualPath)
        {
            try
            {
                return File.Exists(controllerContext.HttpContext.Server.MapPath(virtualPath));
            }
            catch (HttpException exception)
            {
                if (exception.GetHttpCode() != 0x194)
                {
                    throw;
                }
                return false;
            }
            catch
            {
                return false;
            }
        }
 
        public override ViewEngineResult FindView(ControllerContext controllerContext, string viewName, string masterName, bool useCache)
        {
            string[] strArray;
            string[] strArray2;
 
            if (controllerContext == null)
            {
                throw new ArgumentNullException("controllerContext");
            }
            if (string.IsNullOrEmpty(viewName))
            {
                throw new ArgumentException("viewName must be specified.", "viewName");
            }
 
            var themeName = GetThemeToUse(controllerContext);
 
            var requiredString = controllerContext.RouteData.GetRequiredString("controller");
 
            var viewPath = GetPath(controllerContext, ViewLocationFormats, "ViewLocationFormats", viewName, themeName, requiredString, "View", useCache, out strArray);
            var masterPath = GetPath(controllerContext, MasterLocationFormats, "MasterLocationFormats", masterName, themeName, requiredString, "Master", useCache, out strArray2);
 
            if (!string.IsNullOrEmpty(viewPath) && (!string.IsNullOrEmpty(masterPath) || string.IsNullOrEmpty(masterName)))
            {
                return new ViewEngineResult(CreateView(controllerContext, viewPath, masterPath), this);
            }
            return new ViewEngineResult(strArray.Union(strArray2));
        }
 
        public override ViewEngineResult FindPartialView(ControllerContext controllerContext, string partialViewName, bool useCache)
        {
            string[] strArray;
            if (controllerContext == null)
            {
                throw new ArgumentNullException("controllerContext");
            }
            if (string.IsNullOrEmpty(partialViewName))
            {
                throw new ArgumentException("partialViewName must be specified.", "partialViewName");
            }
 
            var themeName = GetThemeToUse(controllerContext);
 
            var requiredString = controllerContext.RouteData.GetRequiredString("controller");
            var partialViewPath = GetPath(controllerContext, PartialViewLocationFormats, "PartialViewLocationFormats", partialViewName, themeName, requiredString, "Partial", useCache, out strArray);
            return string.IsNullOrEmpty(partialViewPath) ? new ViewEngineResult(strArray) : new ViewEngineResult(CreatePartialView(controllerContext, partialViewPath), this);
        }
 
        private static string GetThemeToUse(ControllerContext controllerContext)
        {
            var themeName = controllerContext.HttpContext.Items["themeName"] as string ?? "Default";
 
            return themeName;
        }
 
        private static readonly string[] _emptyLocations;
 
        private string GetPath(ControllerContext controllerContext, string[] locations, string locationsPropertyName, string name, string themeName, string controllerName, string cacheKeyPrefix, bool useCache, out string[] searchedLocations)
        {
            searchedLocations = _emptyLocations;
            if (string.IsNullOrEmpty(name))
            {
                return string.Empty;
            }
            if ((locations == null) || (locations.Length == 0))
            {
                throw new InvalidOperationException("locations must not be null or emtpy.");
            }
 
            bool flag = IsSpecificPath(name);
            string key = CreateCacheKey(cacheKeyPrefix, name, flag ? string.Empty : controllerName, themeName);
            if (useCache)
            {
                var viewLocation = ViewLocationCache.GetViewLocation(controllerContext.HttpContext, key);
                if (viewLocation != null)
                {
                    return viewLocation;
                }
            }
            return !flag ? GetPathFromGeneralName(controllerContext, locations, name, controllerName, themeName, key, ref searchedLocations)
                        : GetPathFromSpecificName(controllerContext, name, key, ref searchedLocations);
        }
 
        private static bool IsSpecificPath(string name)
        {
            var ch = name[0];
            if (ch != '~')
            {
                return (ch == '/');
            }
            return true;
        }
 
        private string CreateCacheKey(string prefix, string name, string controllerName, string themeName)
        {
            return string.Format(
                CultureInfo.InvariantCulture,
                ":ViewCacheEntry:{0}:{1}:{2}:{3}:{4}",
                new object[] { GetType().AssemblyQualifiedName, prefix, name, controllerName, themeName });
        }
 
        private string GetPathFromGeneralName(ControllerContext controllerContext, string[] locations, string name, string controllerName, string themeName, string cacheKey, ref string[] searchedLocations)
        {
            var virtualPath = string.Empty;
            searchedLocations = new string[locations.Length];
            for (var i = 0; i < locations.Length; i++)
            {
                var str2 = string.Format(CultureInfo.InvariantCulture, locations[i], new object[] { name, controllerName, themeName });
 
                if (FileExists(controllerContext, str2))
                {
                    searchedLocations = _emptyLocations;
                    virtualPath = str2;
                    ViewLocationCache.InsertViewLocation(controllerContext.HttpContext, cacheKey, virtualPath);
                    return virtualPath;
                }
                searchedLocations[i] = str2;
            }
            return virtualPath;
        }
 
        private string GetPathFromSpecificName(ControllerContext controllerContext, string name, string cacheKey, ref string[] searchedLocations)
        {
            var virtualPath = name;
            if (!FileExists(controllerContext, name))
            {
                virtualPath = string.Empty;
                searchedLocations = new[] { name };
            }
            ViewLocationCache.InsertViewLocation(controllerContext.HttpContext, cacheKey, virtualPath);
            return virtualPath;
        }
    }

so lucky because RazorViewEngine is also same with WebformViewEngine, so it is not hard to convert to. :D
Second is ThemeControllerBase

    public class ThemeControllerBase : Controller
    {
        protected override void Execute(System.Web.Routing.RequestContext requestContext)
        {
            var themeName = ConfigurationManager.AppSettings["ThemeName"];
            var defaultTheme = ConfigurationManager.AppSettings["DefaultTheme"];
 
            if (requestContext.HttpContext.Items[themeName] == null)
            {
                //first time load
                requestContext.HttpContext.Items[themeName] = requestContext.HttpContext.Request.Cookies.Get("theme").Value;
            }
            else
            {
                requestContext.HttpContext.Items[themeName] = defaultTheme;
 
                var previewTheme = requestContext.RouteData.GetRequiredString("theme");
 
                if (!string.IsNullOrEmpty(previewTheme))
                {
                    requestContext.HttpContext.Items[themeName] = previewTheme;
                }
            }
 
            base.Execute(requestContext);
        }
    }

and the DropDownList partial view for choose a list of themes:

@{ var list = ThemeSample.Helpers.ThemeHelper.GetSelectedList(ViewContext.RequestContext.HttpContext.Items["themeName"].ToString());
}
@Html.DropDownList("Theme", list)
<script language="javascript" type="text/javascript">
    $(function () {
        $('#Theme').change(function () {
 
            $.post('@Url.Action("Index", "ThemeDropdown")', { theme: $(this).val() }, function (result) {
                $('body').html('');
                $('body').html(result);
            });
        });
    });
</script>

controller for it, after user choose the theme on DropDownList, it will submit to the action in this controller.

    public class ThemeDropdownController : Controller
    {
        [HttpPost]
        public ActionResult Index(string theme)
        {
            var themeName = ConfigurationManager.AppSettings["themeName"];
 
            ControllerContext.RequestContext.HttpContext.Items[themeName] = theme;
            var themeCookie = new HttpCookie("theme", theme);
            HttpContext.Response.Cookies.Add(themeCookie);
 
            const string controller = "Home";
            const string action = "Index";
 
            return Redirect(string.Format("~/{0}/{1}/{2}", theme, controller, action));
        }
    }

theme helper class

    public static class ThemeHelper
    {
        public static IList<SelectListItem> GetSelectedList(string selectedTheme)
        {
            // TODO: should get from web.config or database
            var list = new List<SelectListItem>
                            {new SelectListItem {Text = "Default", Value = "Default"},
                            new SelectListItem {Text = "Red", Value = "Red"},
                            new SelectListItem {Text = "Green", Value = "Green"}};
 
            foreach (var selectListItem in list.Where(selectListItem => selectListItem.Value.Equals(selectedTheme)))
            {
                selectListItem.Selected = true;
            }
 
            return list;
        }
 
        public static string GetCssWithTheme(ViewContext viewContext)
        {
            var themeName = ConfigurationManager.AppSettings["themeName"];
 
            return string.Format("~/Content/themes/{0}/Site.css", viewContext.HttpContext.Items[themeName]);
        }
    }

some config in Global.asax

            ViewEngines.Engines.Clear();
            ViewEngines.Engines.Add(new RazorThemeViewEngine());

Notes: In this post I did not write the helper for find images, but it is really easy to implemented. How about your thinking in this topic? I really want to know about that. Leave some messages and we will get more clearly about that. Happy coding and see you next post!

dynamic, Why not?

2011-06-01T13:28:00Z

In these days, people begin to expertise some new features in .NET 4.0. And I am same to them as well. One of features I like and write to this post is dynamic keyword. It is not new, at least in PHP, Python, Ruby... But in .NET world, I think it is new and ship to .NET 4.0 as a default. And some of projects are ported to .NET based on DLR as IronPython, IronRuby. We can use dynamic keyword in anywhere in our project. How do you think about it? So dynamite, isn’t it? :D. Some of people are didn’t like it, some else are love it. How about you? With me, I only try to utilize a new feature in .NET 4.0 for make my work is easier.

Today I obstacle issue when I try to implement an error management in my application. All things will not be hard to get if I didn’t separate all base classes to one assembly and try to reference it from my ASP.NET MVC solution. It is meaning all base controller, base view model and some of helpers, extensions are live in one assembly (called it was WebInfrastructure). And try avoiding acyclic dependency between WebInfrastructure and Web; certainly we will only reference WebInfrastructure from Web and no reverse reference direction from Web in WebInfrastructure. And now if we have one class MySettings in Web and want to the WebInfrastructure can use it. How can we achieve this? This is a place that the dynamic come to.
In WebInfrastructure project, I have
+ AuthorizedController

    [Authorize]
    [HandleError]
    public class AuthorizedController : Controller
    {
        public AuthorizedController(dynamic mySettings)
        {
            MySettings = mySettings;
        }
 
        ......, etc
 
        /// <summary>
        /// Gets or sets the settings.
        /// Its should be dynamic because this class inside the Infrastructure project that not reference to Web project
        /// so it cannot get MySetting from this class
        /// </summary>
        /// <value>The settings.</value>
        protected dynamic MySettings { get; set; }
 
        protected override void OnException(ExceptionContext filterContext)
        {
            if (filterContext == null)
                return;
 
            // Get exception from filter context
            var ex = filterContext.Exception ?? new Exception("No further information exists.");
 
            filterContext.ExceptionHandled = true;
 
            var data = new ErrorViewModel();
 
            if (MySettings != null)
            {
                var isShowErrorMessage = MySettings.IsShowErrorMessage;
 
                if (isShowErrorMessage)
                {
                    data.ErrorMessage = HttpUtility.HtmlEncode(ex.Message);
                    data.TheException = ex;
                }
                else
                {
                    // TODO: try to use log4net to log error here
                    // LogException(ex);
                }
            }
 
            filterContext.Result = View("Error", data);
        }
    }

+ ErrorViewModel

    public class ErrorViewModel
    {
        public string ErrorMessage { get; set; }
 
        public Exception TheException { get; set; }
    }

In Web project
+ MySettings, it is simple, and just read the config from web.config

    public class MySettings : ISectionConfiguration
    {
        public void Add(string sectionName, string sectionValue, List<KeyValuePair<string, string>> valuePairs)
        {
            //throw new NotImplementedException();
        }     
 
        public const string SECTION_NAME = "mysettings";
 
        ...., etc
 
        public bool IsShowErrorMessage { get; set; }
    }

+ Error view

<asp:Content ID="errorTitle" ContentPlaceHolderID="TitleContent" runat="server">
    Error
</asp:Content>
<asp:Content ID="errorContent" ContentPlaceHolderID="MainContent" runat="server">
    <h2>
        Oops: The system got a critical error. Please contact with system administrator
        for resolve this error.
    </h2>
    <% if (!string.IsNullOrEmpty(Model.ErrorMessage))
       {%>
    <div>
        <h3>
            Error message:
        </h3>
        <p>
            <%= Model.ErrorMessage%>
        </p>
    </div>
    <%
       }%>
    <% if (Model.TheException != null)
       {%>
    <div>
        <h3>
            Stack trace:
        </h3>
        <p style="color: #ff0000;">
            <%= Model.TheException.StackTrace %>
        </p>
    </div>
    <div>
        <h3>
            Source of error:
        </h3>
        <p>
            <%= Model.TheException.Source %>
        </p>
    </div>
    <%
       }%>
</asp:Content>

+ Controller

    public class HomeController : AuthorizedController
    {
        public HomeController(MySettings mySettings)
            : base(mySettings)
        {
        }
 
        public ActionResult Index()
        {
            ViewData["Message"] = "Welcome to ASP.NET MVC!";
 
            return View();
        }
 
        public ActionResult About()
        {
            return View();
        }
    }

Notes: We used Unity for IoC container in this post. so its configuration look like this:

var mySettings = SectionConfiguration.GetInstance<MySettings>(MySettings.SECTION_NAME);

ObjectFactory.RegisterInstance(mySettings);

And in run-time, Unity will inject all dependency instances into constructors that have most parameters. As here is

        public HomeController(MySettings mySettings)
            : base(mySettings)
        {
        }

Because I need the IsShowMessageError fields in MySettings, but I cannot reference it from WebInfratructure. So I must use the dynamic at this layer for MySetttings, this concrete implementation will be recover when app started (run-time). It’s really cool, isn’t it? But sometimes, I think it is violate the acyclic dependency. I don’t know. So I really love to hear a feedback from you.

Fluent Interface for WebRequest class

2011-05-17T21:55:00Z

All source codes can be found as http://fluentwebrequest.codeplex.com..

In past, I usually used WebRequest for request some data from remote host. I also used it for request JSON data from Google APIs, some RSS, ATOM data from a few hosts. Everything is good for me. I only put some codes like this:

            var buffer = Encoding.ASCII.GetBytes("q=Queen Elizabeth II&video=on&audio=on&text=on");
 
            var webReq = (HttpWebRequest)WebRequest.Create("http://www.bbc.co.uk/search/news/");
            
            webReq.Method = "POST";
          
            webReq.ContentType = "application/x-www-form-urlencoded";
            
            webReq.ContentLength = buffer.Length;
            
            var postData = webReq.GetRequestStream();
            
            postData.Write(buffer, 0, buffer.Length);
 
            postData.Close();
            
            var webResp = (HttpWebResponse)webReq.GetResponse();
           
            var answer = webResp.GetResponseStream();
 
            var result = string.Empty;
 
            if (answer != null)
            {
                using (var tempStream = new StreamReader(answer))
                {
                    result = tempStream.ReadToEnd();
                }
            }

This code for POST method, and with GET method, we usually use it like this:

            const string getVars = "?q=Queen Elizabeth II&video=on&audio=on&text=on";
 
            var webReq = (HttpWebRequest)WebRequest.Create(string.Format("http://www.bbc.co.uk/search/news/{0}", getVars));
      
            webReq.Method = "GET";
            
            var webResp = (HttpWebResponse)webReq.GetResponse();
            
            var answer = webResp.GetResponseStream();
            var result = string.Empty;
 
            if (answer != null)
            {
                var tempStream = new StreamReader(answer);
                result = tempStream.ReadToEnd();
            }

It will work well, no problems. WebRequest class is very cool. But we can do it better with Fluent Interface, easy to read and more semantic. Fluent Interface is a good choice for this case, we can make this code really interesting in. After took 4 hours to consider and finally I also work on it. This is just for fun, I promise with you about that. I love beautiful code and how to make the code can express itself. How do you feel about this code?

          var result = RequestSubmitter.GetInstance()
                           .WithUrl("http://www.bbc.co.uk/search/news/")
                           .Item.WithNameIs("q").ValueIs("Queen Elizabeth II")
                           .ConnectWith.Item.WithNameIs("video").ValueIs("on")
                           .ConnectWith.Item.WithNameIs("audio").ValueIs("on")
                           .ConnectWith.Item.WithNameIs("text").ValueIs("on")
                           .WithRequestTypeIs(RequestMethodType.Post)
                           .BuildUrl()
                           .GetDataFromUrl();

It is quite Fluent Interface. You can config all things by code and you can read it well. It is only a tiny project. For your curious about this project, I will show you some classes inside it. So for easy to you to imagine, I will paste the class diagram at here:

All a red arrows is a flow for Fluent Interface that I made. As you see, it flow from IRequestSubmitter => IRequestDataBuilder => IRequestItem => IRequestItemName => IRequestType => IAndConnection => IRequestDataBuilder and finally it will loop between IRequestDataBuilder and IRequestItem. It is simple, isn't it? Don't have any secret at here.

And the GET and POST method request, I built it like this:

I made a builder for GET and POST, and finally create a factory class for create these objects. This is enough simple and easy to understand, I think so.

Conclusion, I want to talk once again "it is only for fun and better code". You can find some examples and source code for this project at http://fluentwebrequest.codeplex.com/. And don't forget put some your comments about this project. Thanks for your read and happy coding.

Clean and clear configuration with StructureMap

2011-01-26T17:16:00Z

All sources of this post can be found at my git.

Today, I will continue to sharing some knowledge that I know about StructureMap on Multi-layers application. So I think everyone also know about Multi-layers application, in past we often used the 3 layers (Presentation, Business Logic and Data Access), but by the time maybe we had more than 3 layers as well. In past, when I used Spring.NET framework for IoC container, I must make many configuration in XML file (who knew about XML hell???), it is really terrible. Why? Because it is not a strong type, so very easy to got a error when config an object and also hard to find cause of error in XML file. But now we have DSL for config objects and also have a Fluent Interface for working with config tasks. Some of containers also try to migrate to this way. And I think it is really good to work with. So one question here, how do we config on the multi-layers? Somebody will answer that we will config all instances on the Presentation layer or Service layer in WCF service. Yes, that's fine. And we shall config all instances in these layers.

But when I studied a Divide and Conquer algorithm at University. It is really good for some situation in my life. And in config object for IoC container, I also apply it as well, because I feel that we shall be easy to read and manage object configurations if this config file is small and less code. I did not like some application that config file is very long (maybe more than 1000 line of codes), I called it is code-config hell. A long time ago, I tried to find a best solution for this. And now some of IoCs have supported the feature that permit you to config on many modules and after that it will merge your config to its main container. You can reference to StructureMap, AutoFac, MEF for this. In StructureMap, we have a Registry class; it looks like a part of DSL config for your application. And you can call ObjectFactory to scan alll registries in your application. I love this features. You can imagine that we shall register all configs on modules and after that pass the namspace of these modules to ObjectFactory for scanning. Every object is dynamic. And in AutoFac also have Module class for this purpose. The important thing is code config file will short and clear to read. It make you easy to maintain application in the future (I used to open a code config files in Unity and tried to a instance config inside it, it took me a lot of time). Almost people also know that coding is only 10% and maintain is taking you 90% effort and money you got from maintaining application.

It is enough to talk about the theory and the reasons. Now I will focus on the way that we usually work with StructureMap. And try to make the configuration is well when application boot. Do you know Bootstrapper? Yeah it is great for this situation. I will use it for this post. And I also integrated with Automapper. Wating and see how do it works.

The first thing we need to define the contract for BootStrapper. I have an abstract class for BootStrapper as below:

    public abstract class BootStrapperBase : RootObject
    {
        protected IConfigurator Configurator;
 
        protected IList<string> AssemblyNames;
 
        protected BootStrapperBase(IList<string> assemblyNames, IConfigurator configurator)
        {
            AssemblyNames = assemblyNames;
            Configurator = configurator;
        }
 
        public virtual void BootAllConfig()
        {
        }
    }

And I also have a interface for some tasks that will run inside BootStrapper (one of it, I will implement for run new Profile for Automapper):

    public interface IBootstrapperTask
    {
        void Execute();
    }

And I must have a BootStrapper concrete class for run as:

    public class MyBootStrapper : BootStrapperBase
    {
        private static volatile MyBootStrapper _instance;
        private static readonly object SyncRoot = new Object();
 
        /// <summary>
        /// http://msdn.microsoft.com/en-us/library/ff650316.aspx
        /// </summary>
        public static MyBootStrapper GetInstance(IList<string> assemblies, IConfigurator configurator)
        {
            Contract.Assert(assemblies != null, "List of Registries is null");
            Contract.Assert(assemblies.Count > 0, "List of Registries must be larger than zero");
            Contract.Assert(configurator != null, "Configuratorr instance is null");
 
            if (_instance == null)
            {
                lock (SyncRoot)
                {
                    if (_instance == null)
                        _instance = new MyBootStrapper(assemblies, configurator);
                }
            }
 
            return _instance;
        }
 
        private MyBootStrapper() : this(null, null) { }
 
        private MyBootStrapper(IList<string> assemblyNames, IConfigurator configurator)
            : base(assemblyNames, configurator)
        {
        }
 
        public override void BootAllConfig()
        {
            RegisterRegistry();
            RunBootstrapperTasks();
        }
 
        private void RegisterRegistry()
        {
            CatchExceptionHelper.TryCatchAction(() =>
            {
                ObjectFactory.Initialize(x =>
                {
                    var coreRegistry =
                        (Registry)Activator.CreateInstance(typeof(CoreRegistry));
 
                    x.AddRegistry(coreRegistry);
 
                    x.Scan(scan =>
                    {
                        scan.LookForRegistries();
 
                        foreach (var an in
                            AssemblyNames.Where(an => string.Compare(an, typeof(CoreRegistry).Name) != 0))
                        {
                            scan.Assembly(an);
                        }
 
                        scan.AddAllTypesOf<IBootstrapperTask>();
                        scan.AddAllTypesOf(typeof(IConsumer<>));
                        scan.WithDefaultConventions();
                    });
                });
 
                ObjectFactory.AssertConfigurationIsValid();
            }, Logger);
        }
 
        private static void RunBootstrapperTasks()
        {
            var tasks = MyObjectFactory.GetAllInstances<IBootstrapperTask>();
 
            if (tasks == null) return;
            foreach (var task in tasks)
            {
                task.Execute();
            }
        }
    }

As you see it have a RegisterRegistry method for scan all Registry classes based on list of assemblies that I inject from outside. I also have a RunBootStrapperTasks method for run all the BootStrapper tasks that get from StructureMap container.

Now I will show you the CoreRegistry class as below:

    public class CoreRegistry : Registry
    {
        private readonly ILog _logger;
 
        public CoreRegistry()
            : this(new MyLogger())
        {
        }
 
        public CoreRegistry(ILog logger)
        {
            _logger = logger;
            RegisterAllComponents();
        }
 
        private void RegisterAllComponents()
        {
            CatchExceptionHelper.TryCatchAction(
                () =>
                {
                    For<ILog>().Use<MyLogger>();
                    For<IEventPublisher>().Use<EventPublisher>();
                    For<ISubscriptionService>().Use<EventSubscriptions>();
                    For<IMappingEngine>().Use(() => Mapper.Engine);
                    For<IConfigurator>().Use<Configurator>();
 
                    For<IAttributeScanEngine>().Use<AutoAttributeScanEngine>();
                    For<IDirectoryTask>().Use<DirectoryWrapper>();
                }, _logger);
        }
    }

CoreRegistry class only contains all instances that need for core application, such as ILog, IEventPublisher, ISubscriptionService, IMappingEngine, IConfigurator...

Now I must implemented a class for wrapping StructureMap's ObjectFactory as here:

    public static class MyObjectFactory
    {
        private static readonly ILog Logger;
 
        static MyObjectFactory()
        {
            Logger = new MyLogger();
        }
 
        public static object GetInstance(Type type)
        {
            return LogWriting(() => ObjectFactory.GetInstance(type));
        }
 
        public static T GetInstance<T>()
        {
            return LogWriting(ObjectFactory.GetInstance<T>);
        }
 
        public static object GetNamedInstance(Type type, string name)
        {
            return LogWriting(() => ObjectFactory.GetNamedInstance(type, name));
        }
 
        public static T GetNamedInstance<T>(string name)
        {
            return LogWriting(() => ObjectFactory.GetNamedInstance<T>(name));
        }
 
        public static IList<T> GetAllInstances<T>()
        {
            return LogWriting(ObjectFactory.GetAllInstances<T>);
        }
 
        static T LogWriting<T>(Func<T> func)
        {
            try
            {
                return func();
            }
            catch (Exception ex)
            {
                Logger.Error(ex.Message);
                throw;
            }
        }
    }

I only wrap some main functions of ObjectFactory.

Some unit testing for test the config file as below:

    [TestClass]
    public class BootstrapperTesting
    {
        protected IList<string> _assemblyNames;
        protected IConfigurator _configurator;
 
        [TestInitialize]
        public void InitTestCase()
        {
            _assemblyNames = new List<string>
                              {
                                  "ConfORMSample.Core"                               
                              };
 
            _configurator = new Configurator();
 
            MyBootStrapper.GetInstance(_assemblyNames, _configurator).BootAllConfig();
        }
 
        [TestMethod]
        public void Can_Get_Logger_From_Container()
        {
            var logger = MyObjectFactory.GetInstance<ILog>();
            Assert.IsNotNull(logger);
        }
 
        [TestMethod]
        public void Can_Get_Mapping_Engine_From_Container()
        {
            var mappingEngine = MyObjectFactory.GetInstance<ILog>();
            Assert.IsNotNull(mappingEngine);
        }
 
        [TestMethod]
        public void Can_Get_Directory_Wrapper_From_Container()
        {
            var directoryWrapper = MyObjectFactory.GetInstance<IDirectoryTask>();
            Assert.IsNotNull(directoryWrapper);
        }
 
        [TestCleanup]
        public void CleanUpTestCase()
        {
            GC.SuppressFinalize(_configurator);
        }
    }

After that, you can write many Registry classes for you application. As here I just write one Registry class for Repository:

    public class RepositoryRegistry : Registry
    {
        public RepositoryRegistry()
        {
            For<ISpecification<News>>().Use<NewsAddedSpecification>();
            For<INewsRepository>().Use<NewsRepository>();
            For<ICategoryRepository>().Use<CategoryRepository>();
            For<IFootballResultRepository>().Use<FootballResultRepository>();
        }
    }

And when you boot your application you only need config as here:

            _assemblyNames = new List<string>
                              {
                                  "ConfORMSample.Core",
                                  "ConfORMSample.Repository"
                              };
 
            _configurator = new Configurator();
 
            MyBootStrapper.GetInstance(_assemblyNames, _configurator).BootAllConfig();

Wrap up: Really I don't talk this is a best way to do with StructureMap. Some people shall not agree with this. So tell me you’re thinking by leave some message below this post. I will be happy to know more about that. Happy coding!

AOP with StructureMap Container

2011-01-25T14:33:00Z

Source code for this post can be found in my git.

The first thing I want to say is what is a AOP? So we need to know about the definition of it, yes starting now.

Aspect Oriented Programming (AOP) was originally developed in the late 1980’s at the Xerox Palo Alto Research Center (PARC). All software that is developed has concerns. A concern is a goal, concept or area of interest. In most software there are two types of concerns, core concerns and system level concerns. Previous development methodologies such as Object Oriented Programming (OOP) have done a good job of encapsulating core concerns into modules or classes. The issue with Object Oriented Programming however occurs when concerns need to span more than one module or class. These system level concerns are also known as cross cutting concerns. Common examples of cross cutting concerns are logging, security and tracing. Using Object Oriented Programming these concerns would be coded and added to each class and module where required, making the resulting code harder to read, difficult to maintain, understand and reuse.

AOP however handles these concerns without having to modify each class or module affected. AOP complements Object Oriented Programming by facilitating modularity and pulling the implementation for cross cutting concerns into a single unit.

An implementation of AOP will require the following:

Joinpoints - well-defined points within a program, for example method calls, method entry, method exit and exception handling blocks.
Pointcuts - collections of one or more joinpoints that will be acted upon.
Advice or Introductions – Advice is the code that is executed based on a set of specific joinpoints and a specific context. Introductions provide the ability to affect the static code of a program with the introduction of additional attributes or methods.
Weaver – responsible for taking the information specified for joinpoints, pointcuts, advice / introductions and the system code, to create a functional system by incorporating the AOP code into the standard code.

We usually have 4 levels for AOP in enterprise system, include: Authentication, Audit, Logging, Error Handling. And in this post I just work on 3 levels are Audit, Logging and Error Handling.

I really love StructureMap and some of case, I usually find some solution for implemented interceptor on StructureMap (really it is a AOP), but in internet have a little post about it, and some post is not clear for me. So I decide to write this post today for make clear some thing in StructureMap interceptor. In the post about interceptor for StructureMap at here and here, Jeremy Miller explained very detail about make some intercept on object creation and enrich this object with other behavior. These was very good, but I want to make a intercept on the dynamic proxy of Castle project. I think it is a best way to custom the method call. It is really a magic on code. I like it. Really I am a newbie on StructureMap (just finding and working on it for 1.5 years). So I think many things I must learn to improve my code on StructureMap.

So what do you need on IoC container? It will manage object life cycle, intercept and permit you to make something miracle, and also make your code ease to testing. Just thinking that seen to be have a invisible person to help you to manage objects in your application (just joking). I used to apply many IoC containers in my working, such as Spring.NET, Casle Windsor, Unity, StructureMap, AutoFac, MEF, LinFu, NInject, but maybe I only like StructureMap (I don't know the reason, but I like it). And now I like the interceptor on IoC Container. The magic thing is you can register your object to the IoC Container and let him to make the proxy for that object. So we must be using the Dynamic Proxy from Castle project. It is really good for this task. Why do you need a proxy for you object? You can google this purpose, I only talk that we don't want to show real object before we make sure that everything is valid.

I talked many things about it, but no-one can see my work, but no problems, really need to explain about it before you see a code.

The first thing we need to download the StructureMap and Dynamic Proxy, after that, we have already enough to work on it.

Because Convention over Configuration (CoC) is very popular nowaday, maybe got some ideas from Ruby on Rails (RoR). In this post I will make some CoC for config the code for scan all attributes that need to intercepting.

We need a engine that will scan all concerns about audit, logging, exception management. So we shall have a contract for it as below:

    public interface IAttributeScanEngine
    {
        void Run(IInvocation invocation, Type type);
 
        IEnumerable<Type> AutoFindAttribute(string path);
    }

The idea is I will find all class that implemented by IAttributeScanTask interface and run it. So if you write some concern that implemented IAttributeScanTask, it also will run, but don't need to open the class and add the config for your concern (CoC). And the contract for IAttributeScanTask is:

    public interface IAttributeScanTask
    {
        void Run(IInvocation invocation, bool isFirstCall);
 
        bool AttributeRecognize(ICustomAttributeProvider methodInfo);
    }

Run method will be calling the concrete method, and make some intercept on this method. An AttributeRecognize method is using for find the attribute that we shall decorate on the method that we want to intercepting. That's fine. Now we have the implemented for it at below:

    public class AutoAttributeScanEngine : IAttributeScanEngine
    {
        private readonly IDirectoryTask _directoryTask;
        private readonly ILog _logger;
        private bool _firstcall = true;
 
        public AutoAttributeScanEngine()
        {
        }
 
        public AutoAttributeScanEngine(IDirectoryTask directoryTask, ILog logger)
        {
            _directoryTask = directoryTask;
            _logger = logger;
        }
 
        public void Run(IInvocation invocation, Type type)
        {
            var concreteObject = (IAttributeScanTask)Activator.CreateInstance(type, _logger);
            concreteObject.Run(invocation, _firstcall);
            _firstcall = false;
        }
 
        public virtual IEnumerable<Type> AutoFindAttribute(string path = "")
        {
            IList<Type> result = new List<Type>();
 
            if (string.IsNullOrEmpty(path))
            {
                path = _directoryTask.GetCurrentDirectory();
            }
 
            var assemblies = _directoryTask.GetFiles(path, "*.dll");
 
            if (assemblies.Length > 0)
            {
                AddRange(result, FindAllAssemblies(assemblies));
            }
 
            return result;
        }
 
        private static IEnumerable<Type> FindAllAssemblies(IEnumerable<string> paths)
        {
            IList<Type> result = new List<Type>();
 
            foreach (var assembly in paths)
            {
                var realAssembly = Assembly.LoadFrom(assembly);
 
                if (realAssembly == null)
                    throw new NullReferenceException();
 
                var allTypes = realAssembly.GetTypes();
 
                if (allTypes.Length > 0)
                {
                    AddRange(result, FindAllTypes(allTypes));
                }
            }
 
            return result;
        }
 
        private static IEnumerable<Type> FindAllTypes(IEnumerable<Type> types)
        {
            IList<Type> result = new List<Type>();
 
            foreach (var type in types)
            {
                //http://www.hanselman.com/blog/DoesATypeImplementAnInterface.aspx
                if (typeof(IAttributeScanTask).IsAssignableFrom(type)
                    && type.FullName != typeof(IAttributeScanTask).FullName)
                    result.Add(type);
            }
 
            return result;
        }
 
        private static void AddRange(ICollection<Type> src, IEnumerable<Type> des)
        {
            foreach (var type in des)
            {
                src.Add(type);
            }
        }
    }

Now we need to have some attribute for decorating on the top of method for Audit, Logging, CatchException. And details of it as below:

    [AttributeUsage(AttributeTargets.Method, AllowMultiple = true)]
    public class AuditMethodAttribute : Attribute
    {
    }

    [AttributeUsage(AttributeTargets.Method, AllowMultiple = true)]
    public class CatchMethodExceptionAttribute : Attribute
    {
    }

    [AttributeUsage(AttributeTargets.Method, AllowMultiple = true)]
    public class LockObjectAttribute : Attribute
    {
    }

    [AttributeUsage(AttributeTargets.Method, AllowMultiple = true)]
    public class LoggingAttribute : Attribute
    {
    }

And for each attributes we shall have the scan task for it as:

    public class AuditMethodScanTask : IAttributeScanTask
    {
        private readonly ILog _logger;
 
        public AuditMethodScanTask(ILog logger)
        {
            _logger = logger;
        }
 
        public void Run(IInvocation invocation, bool isFirstCall)
        {
            if (AttributeRecognize(invocation.Method) && isFirstCall)
            {
                CatchExceptionHelper.TryCatchAction(
                    () => AuditMethod(invocation.Proceed, invocation.Method.Name, isFirstCall),
                    _logger);
            }
        }
 
        public bool AttributeRecognize(ICustomAttributeProvider methodInfo)
        {
            var attribute = methodInfo.GetCustomAttributes(typeof(AuditMethodAttribute), true);
 
            if (attribute == null)
                return false;
 
            return attribute is AuditMethodAttribute[];
        }
 
        private static void AuditMethod(Action method, string methodName, bool isFirstCall)
        {
            //connect to database for audit this method call
            if (isFirstCall)
                method();
        }
    }

    public class LoggingScanTask : IAttributeScanTask
    {
        private readonly ILog _logger;
 
        public LoggingScanTask()
        {
        }
 
        public LoggingScanTask(ILog logger)
        {
            _logger = logger;
        }
 
        public void Run(IInvocation invocation, bool isFirstCall)
        {
            if (AttributeRecognize(invocation.Method))
            {
                CatchExceptionHelper.TryCatchAction(
                    () => LogInformation(invocation.Proceed, invocation.Method.Name, isFirstCall),
                    _logger);
            }
        }
 
        public bool AttributeRecognize(ICustomAttributeProvider methodInfo)
        {
            var attribute = methodInfo.GetCustomAttributes(typeof(LoggingAttribute), true);
 
            if (attribute == null)
                return false;
 
            return attribute is LoggingAttribute[];
        }
 
        private void LogInformation(Action method, string methodName, bool isFirstCall)
        {
            var startTime = DateTime.Now;
 
            if (isFirstCall)
                method();
 
            var endTime = DateTime.Now;
            var duration = endTime.Subtract(startTime).TotalSeconds.ToString("N3");
 
            var msg = string.Format("Duration of raw {0}(): {1}s",
                                    methodName, duration);
 
            _logger.Info(msg);
        }
    }

    public class LockObjectScanTask : IAttributeScanTask
    {
        private readonly ILog _logger;
        private static readonly object Locker = new object();
 
        public LockObjectScanTask()
        {
        }
 
        public LockObjectScanTask(ILog logger)
        {
            _logger = logger;
        }
 
        public void Run(IInvocation invocation, bool isFirstCall)
        {
            if (AttributeRecognize(invocation.Method) && isFirstCall)
            {
                CatchExceptionHelper.TryCatchAction(
                    () => LockMethod(invocation.Proceed, invocation.Method.Name, isFirstCall),
                    _logger);
            }
        }
 
        public bool AttributeRecognize(ICustomAttributeProvider methodInfo)
        {
            var attribute = methodInfo.GetCustomAttributes(typeof(LockObjectAttribute), true);
 
            if (attribute == null)
                return false;
 
            return attribute is LockObjectAttribute[];
        }
 
        private static void LockMethod(Action method, string methodName, bool isFirstCall)
        {
            lock (Locker)
            {
                if (isFirstCall)
                    method();
            }
        }
    }

Next step we introduce the customize interceptor for StructureMap that extend from Dynamic Proxy

    public class MyExInterceptor : IInterceptor
    {
        private readonly ILog _logger;
        private readonly IAttributeScanEngine _attributeEngine;
        private static string _attrNamespace;
 
        public MyExInterceptor(string attrNamespace)
            : this(MyObjectFactory.GetInstance<ILog>(),
                   MyObjectFactory.GetInstance<IAttributeScanEngine>(), attrNamespace)
        {
        }
 
        public MyExInterceptor(ILog logger, IAttributeScanEngine attributeEngine, string attrNamespace)
        {
            _logger = logger;
            _attributeEngine = attributeEngine;
            _attrNamespace = attrNamespace;
        }
 
        public void Intercept(IInvocation invocation)
        {
            var attributes = _attributeEngine.AutoFindAttribute("");
 
            foreach (var attribute in attributes)
            {
                var nameTemp = GetAttributeName(GetScanTaskName(attribute.FullName));
 
                var type = Type.GetType(nameTemp);
 
                if (type != null)
                {
                    var existedMethod = Attribute.GetCustomAttribute(invocation.GetConcreteMethodInvocationTarget(), type, false);
 
                    if (existedMethod != null)
                    {
                        var localAttribute = attribute;
                        CatchExceptionHelper.TryCatchAction(() => _attributeEngine.Run(invocation, localAttribute), _logger);
                    }
                }
            }
        }
 
        private static string GetScanTaskName(string fullName)
        {
            var result = fullName.Substring(fullName.LastIndexOf(".") + 1, fullName.Length - fullName.LastIndexOf(".") - 1);
            result = result.Substring(0, result.IndexOf("ScanTask"));
 
            return result;
        }
 
        private static string GetAttributeName(string scanTaskName)
        {
            return _attrNamespace + "." + scanTaskName + "Attribute";
        }
    }

and also have a Dynamic Proxy helper for registering this custom interceptor with Dynamic Castle:

    public class DynamicProxyHelper
    {
        public static object CreateInterfaceProxyWithTargetInterface(Type interfaceType, object concreteObject)
        {
            var dynamicProxy = new ProxyGenerator();
 
            var result = dynamicProxy.CreateInterfaceProxyWithTargetInterface(
                interfaceType,
                concreteObject,
                new[] { new MyExInterceptor("ConfORMSample.Core.Interceptor") });
 
            return result;
        }
    }

As you see we will make a object proxy based on the interface, so if you don't have a object that implemented from one interface, you will not be able to creating a proxy for it.

In this post I make a hard code for namespace ConfORMSample.Core,Interceptor. But I will improve in the future by pass the list of namespaces and scan all ScanTasks inside this namespaces.

We must to register it to StructureMap, in all class extend the Registry class. We have one here:

    public class TestingRegistry : Registry
    {
        public TestingRegistry()
        {
            For<ITenantContext>().Use<SimpleTenantContext>();
 
            For<INewsRepository>().Use<NewsRepository>()
                .EnrichWith(ex => DynamicProxyHelper.CreateInterfaceProxyWithTargetInterface(typeof(INewsRepository), ex));
        }
    }

As here I use the EnrichWith method fron StructureMap and registering this intercept with my MyExInterceptor. And I also have some unit testing for this

    [TestClass]
    public class AutoAttributeScanEngineTesting
    {
        private IAttributeScanEngine _engine;
        private IDirectoryTask _directoryTask;
        private ILog _logger;
 
        [TestInitialize]
        public void InitTestCase()
        {
            _directoryTask = new DirectoryWrapper();
            _logger = MockRepository.GenerateMock<ILog>();
            _engine = new AutoAttributeScanEngine(_directoryTask, _logger);
        }
 
        [TestMethod]
        public void Can_Init_Scan_Engine()
        {
            Assert.IsNotNull(_engine);
        }
 
        [TestMethod]
        public void Can_Scan_All_Instance()
        {
            var allTypes = _engine.AutoFindAttribute("");
            Assert.IsNotNull(allTypes);
            Assert.IsTrue(allTypes.Count() >= 0);
        }
 
        public void Can_Scan_And_Run_All_Instance()
        {
        }
 
        [TestCleanup]
        public void CleanUpTestCase()
        {
        }
    }

    [TestClass]
    public class NewsRepositoryIntereptorTesting
    {
        private IList<string> _assemblyNames;
        private IConfigurator _configurator;
        private INewsRepository _repository;
 
        [TestInitialize]
        public void InitTestCase()
        {
            _assemblyNames = new List<string>
                              {
                                  "ConfORMSample.Core",
                                  "ConfORMSample.NewsMgt.Entities",
                                  "ConfORMSample.Repository",
                                  "ConfORMSample.EventHandlers",
                                  "ConfORMTesting"
                              };
 
            _configurator = new Configurator();
            new MyBootStrapper(_assemblyNames, _configurator);
 
            _repository = MyObjectFactory.GetInstance<INewsRepository>();
        }
 
        [TestMethod]
        public void Can_Interceptor_Method()
        {
            _repository.TestInterceptor();
        }
 
        [TestCleanup]
        public void CleanUpTestCase()
        {
            GC.SuppressFinalize(_configurator);
            GC.SuppressFinalize(_repository);
        }
    }

Finally we shall decorate is as below:

    public class NewsRepository : RepositoryWithTypedId<News, Guid>, INewsRepository
    {
        private readonly ITenantContext _tenantContext;
 
        public NewsRepository()
            : this(MyObjectFactory.GetInstance<ITenantContext>())
        {
        }
 
        public NewsRepository(ITenantContext tenantContext)
        {
            _tenantContext = tenantContext;
        }
 
        [Logging]
        public override IDbContext GetDbContext(string factoryKey = "")
        {
            Contract.Requires(_tenantContext != null, "Tenant Context is null");
 
            return !string.IsNullOrEmpty(factoryKey)
                ? base.GetDbContext(factoryKey)
                : base.GetDbContext(_tenantContext.GetFactory("NewsMgtKey"));
        }
 
        [Logging]
        protected override ISession GetSession(string factoryKey = "")
        {
            Contract.Requires(_tenantContext != null, "Tenant Context is null");
 
            return !string.IsNullOrEmpty(factoryKey)
                ? base.GetSession(factoryKey)
                : base.GetSession(_tenantContext.GetFactory("NewsMgtKey"));
        }
 
        [AuditMethod]
        [LockObject]
        [CatchMethodException]
        [Logging]
        public virtual void TestInterceptor()
        {
        }
    }

If you set a break point on GetDbContext, GetSession, and TestInterceptor methods and debug this class, you will see a dynamic proxy will be created for this class and all the converns will run first after that it will call your concrete method, and it seen to be the proxy will control your process. Carefull with this approach, it can be cause some security hole in your code.

Conclusion, AOP is make your concern can expand on many modules in your application. It's really good because you don't need to modify code when you make change something in your app. so you will not violate the Open-Closed Principle and one concern is only have one responsibility (follow Single Responsibility Principle), all your class must be implemented the interface (follow Interface-Segregation Principle). If you have some concerns about this post, leaving comment here, and I will try to make you happy. Thanks for your read.

A first look at ConfORM - Part 2

2011-01-21T16:41:00Z

You can find my part 1 here and all source codes in my git.

Today I will continue writing about ConfORM. I've said in the previous parts of ConfORM and the advantages of it. This part I will only concentrate on how to build a multiple session factory for ConfORM (also known as multiple database). I get my ideas mainly from Sharp Architecture. First of all, why we need multiple Session Factory? If anyone knows about NHibernate knows that build a Session Factory is very expensive. As it will be a lot of work as a validation schema, entity, preparing data for mapping, build mapping files, prepared caching, preparing relational object model, ... As we know, on a request, the NHibernate Session Factory will build it again, so would be uneconomic and effort-consuming (maybe it would be better in second times, but it is also painful). What if we just build a SessionFactory, and spend a lot of times. That is a part of the day this post. And another part is in the business applications, people often connect to multiple databases simultaneously. And the question is how can we build an engine to store multiple connections to the database? The simple things but it is often very difficult to execute. I've been doing it for ADO.NET and not really very happy. Sharp Architecture worked so well on multiple database. So why do we need to "reinventing the wheel", just use it. I joined the community of Sharp Architecture and felt that this was a really good community. Everyone here works pretty positive and have a lot of feedback every day. I do not take advantage of Sharp Architecture that only a small part of it, so I copied it and changed slightly to suit ConfORM. This is shameful.
And now it's time I start my work.
To realize the multiple databases, I will build a model for the system entity. This time I concentrated on the football model. I will not repeat how to build it, please refer to part 1 of this. I only speak rudimentary knowledge of its domain.
As a management team should be related to the club, players, match. A player will play in one or more positions. A match will be held at a location and will take place between the two clubs, and of course we will have 22 players participating in. That's it. We will have the Domain Diagram as follows:

After that we would have ConfORm database schema and we will not care about it anymore.

We'll need one to take care about transaction IDbContext, it is really a Unit of Work:

    public interface IDbContext
    {
        void CommitChanges();
 
        IDisposable BeginTransaction();
 
        void CommitTransaction();
 
        void RollbackTransaction();
    }

and a ISessionStorage to store and keep the Session in memory:

    public interface ISessionStorage
    {
        ISession GetSessionForKey(string factoryKey);
 
        void SetSessionForKey(string factoryKey, ISession session);
 
        IEnumerable<ISession> GetAllSessions();
    }

and then we will also need a factory ITenantContext to get the key:

    public interface ITenantContext
    {
        string GetFactory(string key);
    }

and finally we will need to take care about persistence IRepository for us:

    public interface IRepository<T> : IRepositoryWithTypedId<T, int> { }
 
    public interface IRepositoryWithTypedId<T, in TId>
    {
        T Get(TId id);
 
        IList<T> GetAll();
 
        IList<T> FindAll(IDictionary<string, object> propertyValuePairs);
 
        T FindOne(IDictionary<string, object> propertyValuePairs);
 
        T SaveOrUpdate(T entity);
 
        void Delete(T entity);
 
        IDbContext GetDbContext(string factoryKey = "");
    }

After completing the contract for the system, we will take steps to build it in more detail at the following:

+ DbContext:

    public class DbContext : RootObject, IDbContext
    {
        public DbContext(string factoryKey)
        {
            Contract.Requires(!string.IsNullOrEmpty(factoryKey), "factoryKey may not be null or empty");
 
            FactoryKey = factoryKey;
        }
 
        private ISession Session
        {
            get
            {
                return NHibernateSession.CurrentFor(FactoryKey);
            }
        }
 
        public void CommitChanges()
        {
            Session.Flush();
        }
 
        public IDisposable BeginTransaction()
        {
            return Session.BeginTransaction();
        }
 
        public void CommitTransaction()
        {
            Session.Transaction.Commit();
        }
 
        public void RollbackTransaction()
        {
            Session.Transaction.Rollback();
        }
 
        public string FactoryKey { get; set; }
    }

+ SimpleSessionStorage, it is just a simple storage that contain a Dictionary for store a pairs of key and value, in the future if you use the ASP.NET MVC, Silverlight or WPF, you can wrapping this class and adapt with your requirement:

    public class SimpleSessionStorage : RootObject, ISessionStorage
    {
        public SimpleSessionStorage() { }
 
        public ISession GetSessionForKey(string factoryKey)
        {
            ISession session;
 
            return !_storage.TryGetValue(factoryKey, out session) ? null : session;
        }
 
        public void SetSessionForKey(string factoryKey, ISession session)
        {
            _storage[factoryKey] = session;
        }
 
        public IEnumerable<ISession> GetAllSessions()
        {
            return _storage.Values;
        }
 
        private readonly Dictionary<string, ISession> _storage = new Dictionary<string, ISession>();
    }

+ SimpleTenantContext, yeah it is also a simple tenant context class, you can wrap this in the future for your need, now I just hard code this class (violated the Open-Close :D ), just keep it simple in this sample:

    public class SimpleTenantContext : ITenantContext
    {
        private readonly ConcurrentDictionary<string, string> _tenants = new ConcurrentDictionary<string, string>();
 
        public SimpleTenantContext()
        {
            InitTenants();
        }
 
        public string GetFactory(string key)
        {
            string value;
 
            _tenants.TryGetValue(key, out value);
 
            return value;
        }
 
        private void InitTenants()
        {
            _tenants.TryAdd("NewsMgtKey", "NewsMgtFactoryKey");
            _tenants.TryAdd("FootballMgtKey", "FootballMgtFactoryKey");
        }
    }

+ NHibernateSession, just keep it from Sharp Architecture and improved some thing else:

    public static class NHibernateSession
    {
        #region private variables
 
        private static IInterceptor _registeredInterceptor;
 
        private static readonly ConcurrentDictionary<string, ISessionFactory> SessionFactories = new ConcurrentDictionary<string, ISessionFactory>();
 
        private static ISessionFactory _slug; // for the out slug
 
        private static IConfigBuilder _configBuilder;
 
        #endregion private variables
 
        #region Init() overloads
 
        [CLSCompliant(false)]
        public static Configuration Init(
                IConfigBuilder configBuilder,
                ISessionStorage storage,
                string connectionString,
                string factoryKey
            )
        {
            _configBuilder = configBuilder;
 
            InitStorage(storage);
 
            try
            {
                if (!string.IsNullOrEmpty(factoryKey))
                {
                    var config = AddConfiguration(factoryKey, connectionString);
 
                    if (config != null)
                        DefaultFactoryKey = factoryKey;
 
                    return config;
                }
 
                return AddConfiguration(DefaultFactoryKey, connectionString);
            }
            catch
            {
                Storage = null;
                throw;
            }
        }
 
        #endregion Init() overloads
 
        public static void InitStorage(ISessionStorage storage)
        {
            Contract.Requires(storage != null, "storage mechanism was null but must be provided");
            Contract.Requires(Storage == null, "A storage mechanism has already been configured for this application");
 
            Storage = storage;
        }
 
        [CLSCompliant(false)]
        public static Configuration AddConfiguration(
                string factoryKey,
                string connectionString
            )
        {
            var config = AddConfiguration(
                factoryKey,
                _configBuilder.BuildConfiguration(connectionString, factoryKey)
                );
 
            return config;
        }
 
        [CLSCompliant(false)]
        public static Configuration AddConfiguration(
            string factoryKey,
            Configuration cfg)
        {
            var sessionFactory = CreateSessionFactoryFor(cfg);
 
            return AddConfiguration(factoryKey, sessionFactory, cfg);
        }
 
        [CLSCompliant(false)]
        public static Configuration AddConfiguration(
            string factoryKey,
            ISessionFactory sessionFactory,
            Configuration cfg)
        {
            Contract.Requires(!SessionFactories.ContainsKey(factoryKey),
                "A session factory has already been configured with the key of " + factoryKey);
 
            SessionFactories.TryAdd(factoryKey, sessionFactory);
 
            return cfg;
        }
 
        public static ISession Current
        {
            get
            {
                Contract.Requires(!IsConfiguredForMultipleDatabases(),
                    "The NHibernateSession.Current property may " +
                    "only be invoked if you only have one NHibernate session factory; i.e., you're " +
                    "only communicating with one database.  Since you're configured communications " +
                    "with multiple databases, you should instead call CurrentFor(string factoryKey)");
 
                return CurrentFor(DefaultFactoryKey);
            }
        }
 
        public static void RegisterInterceptor(IInterceptor interceptor)
        {
            Contract.Requires(interceptor != null, "interceptor may not be null");
 
            _registeredInterceptor = interceptor;
        }
 
        public static bool IsConfiguredForMultipleDatabases()
        {
            return SessionFactories.Count > 1;
        }
 
        public static ISession CurrentFor(string factoryKey)
        {
            Contract.Requires(!string.IsNullOrEmpty(factoryKey), "factoryKey may not be null or empty");
            Contract.Requires(Storage != null, "An ISessionStorage has not been configured");
            Contract.Requires(SessionFactories.ContainsKey(factoryKey), "An ISessionFactory does not exist with a factory key of " + factoryKey);
 
            var session = Storage.GetSessionForKey(factoryKey);
 
            if (session == null)
            {
                if (_registeredInterceptor != null)
                {
                    session = SessionFactories[factoryKey].OpenSession(_registeredInterceptor);
                }
                else
                {
                    session = SessionFactories[factoryKey].OpenSession();
                }
 
                Storage.SetSessionForKey(factoryKey, session);
            }
 
            return session;
        }
 
        public static void CloseAllSessions()
        {
            if (Storage != null)
                foreach (var session in Storage.GetAllSessions().Where(session => session.IsOpen))
                {
                    session.Close();
                }
        }
 
        public static void Reset()
        {
            if (Storage != null)
            {
                foreach (var session in Storage.GetAllSessions())
                {
                    session.Dispose();
                }
            }
 
            SessionFactories.Clear();
 
            Storage = null;
            _registeredInterceptor = null;
        }
 
        public static ISessionFactory GetSessionFactoryFor(string factoryKey)
        {
            if (!SessionFactories.ContainsKey(factoryKey))
                return null;
 
            return SessionFactories[factoryKey];
        }
 
        public static void RemoveSessionFactoryFor(string factoryKey)
        {
            if (GetSessionFactoryFor(factoryKey) != null)
            {
                SessionFactories.TryRemove(factoryKey, out _slug);
            }
        }
 
        public static ISessionFactory GetDefaultSessionFactory()
        {
            return GetSessionFactoryFor(DefaultFactoryKey);
        }
 
        public static string DefaultFactoryKey = PersistanceConstants.DEFAULT_FACTORY_SESSION_KEY;
 
        public static ISessionStorage Storage { get; set; }
 
        private static ISessionFactory CreateSessionFactoryFor(Configuration cfg)
        {
            return cfg.BuildSessionFactory();
        }
    }

+ RepositoryBase class, contain something common for alot of Repository to use (Abstract is good, I think so, but should be fine if we can use composition):

    public abstract class RepositoryBase<T> : RepositoryWithTypedId<T, int>, IRepository<T> { }
 
    public abstract class RepositoryWithTypedId<T, TId> : RootObject, IRepositoryWithTypedId<T, TId>
    {
        private IDbContext _dbContext;
 
        protected virtual ISession GetSession(string factoryKey = "")
        {
            if (string.IsNullOrEmpty(factoryKey))
            {
                return NHibernateSession.CurrentFor(PersistanceConstants.DEFAULT_FACTORY_SESSION_KEY);
            }
 
            return NHibernateSession.CurrentFor(factoryKey);
        }
 
        public virtual IDbContext GetDbContext(string factoryKey = "")
        {
            if (string.IsNullOrEmpty(factoryKey))
            {
                return new DbContext(PersistanceConstants.DEFAULT_FACTORY_SESSION_KEY);
            }
 
            return _dbContext ?? (_dbContext = new DbContext(factoryKey));
        }
 
        public virtual T Get(TId id)
        {
            return GetSession().Get<T>(id);
        }
 
        public virtual IList<T> GetAll()
        {
            var criteria = GetSession().CreateCriteria(typeof(T));
            return criteria.List<T>();
        }
 
        public virtual IList<T> FindAll(IDictionary<string, object> propertyValuePairs)
        {
            Contract.Requires(propertyValuePairs != null && propertyValuePairs.Count > 0,
                "propertyValuePairs was null or empty; " +
                "it has to have at least one property/value pair in it");
 
            var criteria = GetSession().CreateCriteria(typeof(T));
 
            foreach (var key in propertyValuePairs.Keys)
            {
                if (propertyValuePairs[key] != null)
                {
                    criteria.Add(Restrictions.Eq(key, propertyValuePairs[key]));
                }
                else
                {
                    criteria.Add(Restrictions.IsNull(key));
                }
            }
 
            return criteria.List<T>();
        }
 
        public virtual T FindOne(IDictionary<string, object> propertyValuePairs)
        {
            var foundList = FindAll(propertyValuePairs);
 
            if (foundList.Count > 1)
            {
                throw new NonUniqueResultException(foundList.Count);
            }
 
            return foundList.Count == 1 ? foundList[0] : default(T);
        }
 
        public virtual void Delete(T entity)
        {
            GetSession().Delete(entity);
        }
 
        public virtual T SaveOrUpdate(T entity)
        {
            Contract.Requires(!(entity is IEntity<TId>),
                "For better clarity and reliability, Entities with an assigned Id must call Save or Update");
 
            GetSession().SaveOrUpdate(entity);
            return entity;
        }
    }

+ And we will have 2 repositories for News and Football. In this case, we will connect to 2 database for get data:

    public class NewsRepository : RepositoryWithTypedId<News, Guid>, INewsRepository
    {
        private readonly ITenantContext _tenantContext;
 
        public NewsRepository(ITenantContext tenantContext)
        {
            _tenantContext = tenantContext;
        }
 
        public override IDbContext GetDbContext(string factoryKey = "")
        {
            Contract.Requires(_tenantContext != null, "Tenant Context is null");
 
            return !string.IsNullOrEmpty(factoryKey)
                ? base.GetDbContext(factoryKey)
                : base.GetDbContext(_tenantContext.GetFactory("NewsMgtKey"));
        }
 
        protected override ISession GetSession(string factoryKey = "")
        {
            Contract.Requires(_tenantContext != null, "Tenant Context is null");
 
            return !string.IsNullOrEmpty(factoryKey)
                ? base.GetSession(factoryKey)
                : base.GetSession(_tenantContext.GetFactory("NewsMgtKey"));
        }
    }

and,

    public class FootballResultRepository : RepositoryWithTypedId<Game, Guid>, IFootballResultRepository
    {
        public static readonly string FootballFactoryKey = "FootBallMsgtKey";
 
        protected override ISession GetSession(string factoryKey = "")
        {
            return base.GetSession(FootballFactoryKey);
        }
    }

+ Finally, I have some Unit testing for this:

    [TestClass]
    public class ConfigurationTesting
    {
        private ConfORMConfigBuilder _confOrmConfigBuilder;
        private string _connectionString;
        private string _sessionFactoryName;
        private IEnumerable<string> _assemblies;
 
        [TestInitialize]
        public void Init()
        {
            _connectionString = "Server=localhost;database=ConfORMSample;Integrated Security=SSPI;";
            _sessionFactoryName = "SessionFactory1";
 
            _assemblies = new List<string>
                              {
                                  "ConfORMSample.NewsMgt.Entities.dll"
                              };
 
            _confOrmConfigBuilder = new SqlServerConfORMConfigBuilder(_assemblies, "ConfORMSample.dbo");
 
            NHibernateSession.Init(_confOrmConfigBuilder, new SimpleSessionStorage(), _connectionString, string.Empty);
        }
 
        [TestMethod]
        public void Can_Init_Config()
        {
            var config = _confOrmConfigBuilder.BuildConfiguration(_connectionString, _sessionFactoryName);
 
            Assert.IsNotNull(config);
        }
 
        ...
    }

and,

    [TestClass]
    public class FootBallConfigurationTesting
    {
        private ConfORMConfigBuilder _confOrmConfigBuilder;
        private string _connectionString;
        private IEnumerable<string> _assemblies;
 
        [TestInitialize]
        public void InitTest()
        {
            _connectionString = "Server=localhost;database=ConfORMSample_FootballMsgt;Integrated Security=SSPI;";
 
            _assemblies = new List<string>
                              {
                                  "ConfORMSample.Football.Entities.dll"
                              };
 
            _confOrmConfigBuilder = new Football.SqlServerConfORMConfigBuilder(_assemblies, "ConfORMSample_FootballMsgt.dbo");
 
            NHibernateSession.Init(_confOrmConfigBuilder, new SimpleSessionStorage(), _connectionString, "Football");
        }
 
        [TestMethod]
        public void Can_Init_Football_Config()
        {
            Assert.IsNotNull(_confOrmConfigBuilder);
        }
 
        [TestMethod]
        public void Can_Get_Configuration_From_NHibernate_Session()
        {
            var session = NHibernateSession.CurrentFor("Football");
            Assert.IsNotNull(session);
        }
 
        [TestCleanup]
        public void CleanUp()
        {
            GC.SuppressFinalize(_confOrmConfigBuilder);
        }
    }

and,

    [TestClass]
    public class MultipleDatabaseTesting : TestingBase
    {
        private ConfORMConfigBuilder _newsMgtConfigBuilder;
        private string _newsMgtConnectionString;
        private IEnumerable<string> _newsMgtAssemblies;
 
        private ConfORMConfigBuilder _footballConfigBuilder;
        private string _footballConnectionString;
        private IEnumerable<string> _footballAssemblies;
 
        private ISessionStorage _sessionStorage;
 
        [TestInitialize]
        public override void Init()
        {
            base.Init();
 
            _sessionStorage = new SimpleSessionStorage();
 
            // news management config area
            _newsMgtConnectionString = "Server=localhost;database=ConfORMSample;Integrated Security=SSPI;";
 
            _newsMgtAssemblies = new List<string>
                              {
                                  "ConfORMSample.NewsMgt.Entities.dll"
                              };
 
            _newsMgtConfigBuilder = new Football.SqlServerConfORMConfigBuilder(_newsMgtAssemblies, "ConfORMSample.dbo");
 
            NHibernateSession.Init(_newsMgtConfigBuilder, _sessionStorage, _newsMgtConnectionString, "NewsMgt");
 
            // football management config area
            _footballConnectionString = "Server=localhost;database=ConfORMSample_FootballMsgt;Integrated Security=SSPI;";
 
            _footballAssemblies = new List<string>
                              {
                                  "ConfORMSample.Football.Entities.dll"
                              };
 
            _footballConfigBuilder = new Football.SqlServerConfORMConfigBuilder(_footballAssemblies, "ConfORMSample_FootballMsgt.dbo");
 
            NHibernateSession.Init(_footballConfigBuilder, _sessionStorage, _footballConnectionString, "FootballMgt");
        }
 
        [TestMethod]
        public void Can_Init_All_Config()
        {
            Assert.IsNotNull(_newsMgtConfigBuilder);
            Assert.IsNotNull(_footballConfigBuilder);
        }
 
        [TestMethod]
        public void Can_Get_NewsMgt_Session()
        {
            Assert.IsNotNull(NHibernateSession.CurrentFor("NewsMgt"));
        }
 
        [TestMethod]
        public void Can_Get_FootballMgt_Session()
        {
            Assert.IsNotNull(NHibernateSession.CurrentFor("FootballMgt"));
        }
 
        [TestCleanup]
        public override void CleanUp()
        {
            base.Init();
            GC.SuppressFinalize(_newsMgtConfigBuilder);
            GC.SuppressFinalize(_footballConfigBuilder);
            GC.SuppressFinalize(_sessionStorage);
        }
    }

Anyway my approach is maybe not a good, but I just want to write it as here and find some feedbacks from .NET community.

So please leave your thinking about my post.

Happy coding!

A first look at ConfORM - Part 1

2011-01-08T20:12:00Z

All source codes for this post can be found at here.

Have you ever heard of ConfORM is not? I have read it three months ago when I wrote an post about NHibernate and Autofac. At that time, this project really has just started and still in beta version, so I still do not really care much. But recently when reading a book by Jason Dentler NHibernate 3.0 Cookbook, I started to pay attention to it. Author have mentioned quite a lot of OSS in his book. And now again I have reviewed ConfORM once again. I have been involved in ConfORM development group on google and read some articles about it. Fabio Maulo spent a lot of work for the OSS, and I hope it will adapt a great way for NHibernate (because he contributed to NHibernate that).

So what is ConfORM? It is stand for Configuration ORM, and it was trying to use a lot of heuristic model for identifying entities from C# code. Today, it's mostly Model First Driven development, so the first thing is to build the entity model. This is really important and we can see it is the heart of business software. Then we have to tell DB about the entity of this model. We often will use Inversion Engineering here, Database Schema is will create based on recently Entity Model. From now we will absolutely not interested in the DB again, only focus on the Entity Model.

Fluent NHibenate really good, I liked this OSS. Sharp Architecture and has done so well in Fluent NHibernate integration with applications. A Multiple Database technical in Sharp Architecture is truly awesome. It can receive configuration, a connection string and a dll containing entity model, which would then create a SessionFactory, finally caching inside the computer memory. As the number of SessionFactory can be very large and will full of the memory, it has also devised a way of caching SessionFactory in the file. This post I hope this will not completely explain about and building a model of multiple databases. I just tried to mount a number of posts from the community and apply some of my knowledge to build a management model Session for ConfORM.

As well as Fluent NHibernate, ConfORM also supported on the interface mapping, see this to understand it. So the first thing we will build the Entity Model for it, and here is what I will use the model for this article. A simple model for managing news and polls, it will be too easy for a number of people, but I hope not to bring complexity to this post.

I will then have some code to build super type for the Entity Model.

    public interface IEntity<TId>
    {
        TId Id { get; set; }
    }

    public abstract class EntityBase<TId> : IEntity<TId>
    {
        public virtual TId Id { get; set; }
 
        public override bool Equals(object obj)
        {
            return Equals(obj as EntityBase<TId>);
        }
 
        private static bool IsTransient(EntityBase<TId> obj)
        {
            return obj != null &&
            Equals(obj.Id, default(TId));
        }
 
        private Type GetUnproxiedType()
        {
            return GetType();
        }
 
        public virtual bool Equals(EntityBase<TId> other)
        {
            if (other == null)
                return false;
            if (ReferenceEquals(this, other))
                return true;
            if (!IsTransient(this) &&
            !IsTransient(other) &&
            Equals(Id, other.Id))
            {
                var otherType = other.GetUnproxiedType();
                var thisType = GetUnproxiedType();
                return thisType.IsAssignableFrom(otherType) ||
                otherType.IsAssignableFrom(thisType);
            }
            return false;
        }
 
        public override int GetHashCode()
        {
            if (Equals(Id, default(TId)))
                return base.GetHashCode();
            return Id.GetHashCode();
        }
    }

Database schema will be created as:

The next step is to build the ConORM builder to create a NHibernate Configuration. Patrick have a excellent article about it at here. Contract of it below:

    public interface IConfigBuilder
    {
        Configuration BuildConfiguration(string connectionString, string sessionFactoryName);
    }

The idea here is that I will pass in a connection string and a set of the DLL containing the Entity Model and it makes me a NHibernate Configuration (shame that I stole this ideas of Sharp Architecture). And here is its code:

    public abstract class ConfORMConfigBuilder : RootObject, IConfigBuilder
    {
        private static IConfigurator _configurator;
 
        protected IEnumerable<Type> DomainTypes;
 
        private readonly IEnumerable<string> _assemblies;
 
        protected ConfORMConfigBuilder(IEnumerable<string> assemblies)
            : this(new Configurator(), assemblies)
        {
            _assemblies = assemblies;
        }
 
        protected ConfORMConfigBuilder(IConfigurator configurator, IEnumerable<string> assemblies)
        {
            _configurator = configurator;
            _assemblies = assemblies;
        }
 
        public abstract void GetDatabaseIntegration(IDbIntegrationConfigurationProperties dBIntegration, string connectionString);
 
        protected abstract HbmMapping GetMapping();
 
        public Configuration BuildConfiguration(string connectionString, string sessionFactoryName)
        {
            Contract.Requires(!string.IsNullOrEmpty(connectionString), "ConnectionString is null or empty");
            Contract.Requires(!string.IsNullOrEmpty(sessionFactoryName), "SessionFactory name is null or empty");
            Contract.Requires(_configurator != null, "Configurator is null");
 
            return CatchExceptionHelper.TryCatchFunction(
                () =>
                {
                    DomainTypes = GetTypeOfEntities(_assemblies);
 
                    if (DomainTypes == null)
                        throw new Exception("Type of domains is null");
 
                    var configure = new Configuration();
                    configure.SessionFactoryName(sessionFactoryName);
 
                    configure.Proxy(p => p.ProxyFactoryFactory<ProxyFactoryFactory>());
                    configure.DataBaseIntegration(db => GetDatabaseIntegration(db, connectionString));
 
                    if (_configurator.GetAppSettingString("IsCreateNewDatabase").ConvertToBoolean())
                    {
                        configure.SetProperty("hbm2ddl.auto", "create-drop");
                    }
 
                    configure.Properties.Add("default_schema", _configurator.GetAppSettingString("DefaultSchema"));
                    configure.AddDeserializedMapping(GetMapping(),
                                                     _configurator.GetAppSettingString("DocumentFileName"));
 
                    SchemaMetadataUpdater.QuoteTableAndColumns(configure);
 
                    return configure;
                }, Logger);
        }
 
        protected IEnumerable<Type> GetTypeOfEntities(IEnumerable<string> assemblies)
        {
            var type = typeof(EntityBase<Guid>);
            var domainTypes = new List<Type>();
 
            foreach (var assembly in assemblies)
            {
                var realAssembly = Assembly.LoadFrom(assembly);
 
                if (realAssembly == null)
                    throw new NullReferenceException();
 
                domainTypes.AddRange(realAssembly.GetTypes().Where(
                    t =>
                    {
                        if (t.BaseType != null)
                            return string.Compare(t.BaseType.FullName,
                                          type.FullName) == 0;
                        return false;
                    }));
            }
 
            return domainTypes;
        }
    }

I do not want to dependency on any RDBMS, so I made a builder as an abstract class, and so I will create a concrete instance for SQL Server 2008 as follows:

    public class SqlServerConfORMConfigBuilder : ConfORMConfigBuilder
    {
        public SqlServerConfORMConfigBuilder(IEnumerable<string> assemblies)
            : base(assemblies)
        {
        }
 
        public override void GetDatabaseIntegration(IDbIntegrationConfigurationProperties dBIntegration, string connectionString)
        {
            dBIntegration.Dialect<MsSql2008Dialect>();
            dBIntegration.Driver<SqlClientDriver>();
            dBIntegration.KeywordsAutoImport = Hbm2DDLKeyWords.AutoQuote;
            dBIntegration.IsolationLevel = IsolationLevel.ReadCommitted;
            dBIntegration.ConnectionString = connectionString;
            dBIntegration.LogSqlInConsole = true;
            dBIntegration.Timeout = 10;
            dBIntegration.LogFormatedSql = true;
            dBIntegration.HqlToSqlSubstitutions = "true 1, false 0, yes 'Y', no 'N'";
        }
 
        protected override HbmMapping GetMapping()
        {
            var orm = new ObjectRelationalMapper();
 
            orm.Patterns.PoidStrategies.Add(new GuidPoidPattern());
 
            var patternsAppliers = new CoolPatternsAppliersHolder(orm);
            //patternsAppliers.Merge(new DatePropertyByNameApplier()).Merge(new MsSQL2008DateTimeApplier());
            patternsAppliers.Merge(new ManyToOneColumnNamingApplier());
            patternsAppliers.Merge(new OneToManyKeyColumnNamingApplier(orm));
 
            var mapper = new Mapper(orm, patternsAppliers);
 
            var entities = new List<Type>();
 
            DomainDefinition(orm);
            Customize(mapper);
 
            entities.AddRange(DomainTypes);
 
            return mapper.CompileMappingFor(entities);
        }
 
        private void DomainDefinition(IObjectRelationalMapper orm)
        {
            orm.TablePerClassHierarchy(new[] { typeof(EntityBase<Guid>) });
            orm.TablePerClass(DomainTypes);
 
            orm.OneToOne<News, Poll>();
            orm.ManyToOne<Category, News>();
 
            orm.Cascade<Category, News>(Cascade.All);
            orm.Cascade<News, Poll>(Cascade.All);
            orm.Cascade<User, Poll>(Cascade.All);
        }
 
        private static void Customize(Mapper mapper)
        {
            CustomizeRelations(mapper);
            CustomizeTables(mapper);
            CustomizeColumns(mapper);
        }
 
        private static void CustomizeRelations(Mapper mapper)
        {
        }
 
        private static void CustomizeTables(Mapper mapper)
        {
        }
 
        private static void CustomizeColumns(Mapper mapper)
        {
            mapper.Class<Category>(
                cm =>
                {
                    cm.Property(x => x.Name, m => m.NotNullable(true));
                    cm.Property(x => x.CreatedDate, m => m.NotNullable(true));
                });
 
            mapper.Class<News>(
                cm =>
                {
                    cm.Property(x => x.Title, m => m.NotNullable(true));
                    cm.Property(x => x.ShortDescription, m => m.NotNullable(true));
                    cm.Property(x => x.Content, m => m.NotNullable(true));
                });
 
            mapper.Class<Poll>(
                cm =>
                {
                    cm.Property(x => x.Value, m => m.NotNullable(true));
                    cm.Property(x => x.VoteDate, m => m.NotNullable(true));
                    cm.Property(x => x.WhoVote, m => m.NotNullable(true));
                });
 
            mapper.Class<User>(
                cm =>
                {
                    cm.Property(x => x.UserName, m => m.NotNullable(true));
                    cm.Property(x => x.Password, m => m.NotNullable(true));
                });
        }
    }

As you can see that we can do so many things in this class, such as custom entity relationships, custom binding on the columns, custom table name, ... Here I only made two so-Appliers for OneToMany and ManyToOne relationships, you can refer to it here

    public class ManyToOneColumnNamingApplier : IPatternApplier<PropertyPath, IManyToOneMapper>
    {
        #region IPatternApplier<PropertyPath,IManyToOneMapper> Members
 
        public void Apply(PropertyPath subject, IManyToOneMapper applyTo)
        {
            applyTo.Column(subject.ToColumnName() + "Id");
        }
 
        #endregion
 
        #region IPattern<PropertyPath> Members
 
        public bool Match(PropertyPath subject)
        {
            return subject != null;
        }
 
        #endregion
    }

    public class OneToManyKeyColumnNamingApplier : OneToManyPattern, IPatternApplier<PropertyPath, ICollectionPropertiesMapper>
    {
        public OneToManyKeyColumnNamingApplier(IDomainInspector domainInspector) : base(domainInspector) { }
 
        #region Implementation of IPattern<PropertyPath>
 
        public bool Match(PropertyPath subject)
        {
            return Match(subject.LocalMember);
        }
 
        #endregion Implementation of IPattern<PropertyPath>
 
        #region Implementation of IPatternApplier<PropertyPath,ICollectionPropertiesMapper>
 
        public void Apply(PropertyPath subject, ICollectionPropertiesMapper applyTo)
        {
            applyTo.Key(km => km.Column(GetKeyColumnName(subject)));
        }
 
        #endregion Implementation of IPatternApplier<PropertyPath,ICollectionPropertiesMapper>
 
        protected virtual string GetKeyColumnName(PropertyPath subject)
        {
            Type propertyType = subject.LocalMember.GetPropertyOrFieldType();
            Type childType = propertyType.DetermineCollectionElementType();
            var entity = subject.GetContainerEntity(DomainInspector);
            var parentPropertyInChild = childType.GetFirstPropertyOfType(entity);
            var baseName = parentPropertyInChild == null ? subject.PreviousPath == null ? entity.Name : entity.Name + subject.PreviousPath : parentPropertyInChild.Name;
            return GetKeyColumnName(baseName);
        }
 
        protected virtual string GetKeyColumnName(string baseName)
        {
            return string.Format("{0}Id", baseName);
        }
    }

Everyone also can download the ConfORM source at google code and see example inside it. Next part I will write about multiple database factory. Hope you enjoy about it. happy coding and see you next part.

Generating Report for NUnit

2010-12-17T13:59:00Z

All source codes for this post can be found at my github.

Time ago, I received a request that people ask me how they can generate reports of the results of testing using NUnit? In fact, I may never do this. In the little world of my programming, I only care about the test results, red-green-refactoring, and that was it. When I got that question quite a bit unexpected, I knew that I could use NCover to generate reports, but reports of NCover too simple, it did not give us more details on the number of test cases, test methods, ...

And I began to see about creating interesting report for NUnit.I was lucky to find an open source here. Its authors call it NUnit2Report, but one disadvantage is it only running on .NET 1.0. Indeed too old compared to the current version 4.0. And I try to download the preview, but I could not run. I had to open its source code and found that it uses XSLT to convert the output of NUnit results from XML to HTML. Nothing really special, because I also knew that after NUnit run output file extension is XML is created. Author only use this file to convert to HTML using XSLT. And I decided to convert it to. NET 4.0, because I will not have to code from scratch. Conversion work made me take some time, but was lucky that I finally have what I want. Thanks Gilles for the this OSS. I will send a mail to thank him for his efforts but put this out for the OSS. Now I will show people how to do it. I used the auto built NAnt and NUnit for running TestCase, and I use Selenium testing framework. After writing three TestCase using Selenium, I ran NUnit, and got the following results:

There are 1 fail and 2s success. In the bin directory of this project will have the NUnit output file as shown below:

Then I create a build file, and a bat file for easy running (can use PowerShell is here also.) Double click in the bat file to create a report like this:

Finally open the index.html file in the folder to view report.

As everyone can see, it is the TestCase and divide very clearly, that I meet the requirements. This is really good.

Once again I really thank NUnit2Report from Gilles. People can contact him via the mail address gilles.bayon@laposte.net or website http://nunit2report.sourceforge.net. It really is useful to those who promised to QA. Hopefully this post will help anyone really interested in doing reports for NUnit.

DRY with Lambda Expression (errors management)

2010-11-19T14:59:00Z

In my current project, I have made a miracle with Lambda Expression. Have you ever implemented something like this:

    public abstract class GenericRepository<TEntity> : IRepository<TEntity> where TEntity : class, IEntity
    {
        private readonly ILog _logger;
 
        protected GenericRepository()
            : this(IoC.GetInstance<IUnitOfWork>(), IoC.GetInstance<ILog>()
            )
        {
        }
 
        protected GenericRepository(IUnitOfWork unitOfWork, ILog logger)
        {
            UnitOfWork = unitOfWork;
            _logger = logger;
        }
 
        public IUnitOfWork UnitOfWork { get; set; }
 
        public IEnumerable<TEntity> All()
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
 
            try
            {
                return UnitOfWork.Session.QueryOver<TEntity>().List();
            }
            catch (Exception ex)
            {
                _logger.Error(ex.Message);
                throw;
            }
        }
 
        public int Add(TEntity entity)
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            try
            {
                return (int)UnitOfWork.Session.Save(entity);
            }
            catch (Exception ex)
            {
                _logger.Error(ex.Message);
                throw;
            }
        }
 
        public void Remove(TEntity entity)
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            try
            {
                UnitOfWork.Session.Delete(entity);
            }
            catch (Exception ex)
            {
                _logger.Error(ex.Message);
                throw;
            }
        }
 
        public void Update(TEntity entity)
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            try
            {
                UnitOfWork.Session.Update(entity);
            }
            catch (Exception ex)
            {
                _logger.Error(ex.Message);
                throw;
            }
        }
 
        public IEnumerable<TEntity> GetBy(Expression<Func<TEntity, bool>> condition)
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
            CodeContract.Contract.Assert(condition != null, "Expression is null");
 
            try
            {
                return UnitOfWork.Session.QueryOver<TEntity>()
                        .Where(condition)
                        .List();
            }
            catch (Exception ex)
            {
                _logger.Error(ex.Message);
                throw;
            }
        }
    }

Yeah. Many try...catch... you use for error management. And I think it is very stupid, and violate the DRY (Don't repeat yourself) principle. I give a example, one day, I will change the error management, so I must to go from the begin this class to end this class for modify it. Once again, it is very stupid and take a lot of time to manage. So I always think one solution for resolve it some weeks ago. And now in new project, I apply one technical that use the lambda expression for inverting code control.
It is very simply. I think after I present it at here, anyone also know about it.
I just make a helper class as here:

    public static class CatchExceptionHelper
    {
        /// <summary>
        /// Tries the catch action.
        /// for example, void Foo() is a action
        /// </summary>
        /// <param name="action">The action.</param>
        /// <param name="logger">The logger.</param>
        public static void TryCatchAction(Action action, ILog logger)
        {
            try
            {
                action();
            }
            catch (Exception ex)
            {
                logger.Error(ex.Message);
                throw;
            }
        }
 
        /// <summary>
        /// Tries the catch function.
        /// for example, int Foo() is a function
        /// </summary>
        /// <typeparam name="T"></typeparam>
        /// <param name="function">The function.</param>
        /// <param name="logger">The logger.</param>
        /// <returns></returns>
        public static T TryCatchFunction<T>(Func<T> function, ILog logger)
        {
            try
            {
                return function();
            }
            catch (Exception ex)
            {
                logger.Error(ex.Message);
                throw;
            }
        }
    }

After that just do that as here:

    public abstract class GenericRepository<TEntity> : IRepository<TEntity> where TEntity : class, IEntity
    {
        private readonly ILog _logger;
 
        protected GenericRepository()
            : this(IoC.GetInstance<IUnitOfWork>(), IoC.GetInstance<ILog>()
            )
        {
        }
 
        protected GenericRepository(IUnitOfWork unitOfWork, ILog logger)
        {
            UnitOfWork = unitOfWork;
            _logger = logger;
        }
 
        public IUnitOfWork UnitOfWork { get; set; }
 
        public IEnumerable<TEntity> All()
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
 
            return CatchExceptionHelper.TryCatchFunction(() =>
            {
                return UnitOfWork.Session.QueryOver<TEntity>().List();
            }, _logger);
        }
 
        public int Add(TEntity entity)
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            return CatchExceptionHelper.TryCatchFunction(() =>
            {
                return (int)UnitOfWork.Session.Save(entity);
            }, _logger);
        }
 
        public void Remove(TEntity entity)
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            CatchExceptionHelper.TryCatchAction(() => UnitOfWork.Session.Delete(entity), _logger);
        }
 
        public void Update(TEntity entity)
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            CatchExceptionHelper.TryCatchAction(() => UnitOfWork.Session.Update(entity), _logger);
        }
 
        public IEnumerable<TEntity> GetBy(Expression<Func<TEntity, bool>> condition)
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
            CodeContract.Contract.Assert(condition != null, "Expression is null");
 
            return CatchExceptionHelper.TryCatchFunction(() =>
            {
                return UnitOfWork.Session.QueryOver<TEntity>()
                        .Where(condition)
                        .List();
            }, _logger);
        }
    }

Now we shall be easy to change error management any time that not touch to this class, just modify helper class. But as you see we still need to check the Code Contract at the begin of each method. I also think it is also not good. So I will re factoring something like that:

    public abstract class GenericRepository<TEntity> : IRepository<TEntity> where TEntity : class, IEntity
    {
        private readonly ILog _logger;
 
        protected GenericRepository()
            : this(IoC.GetInstance<IUnitOfWork>(), IoC.GetInstance<ILog>()
            )
        {
        }
 
        protected GenericRepository(IUnitOfWork unitOfWork, ILog logger)
        {
            UnitOfWork = unitOfWork;
            _logger = logger;
        }
 
        public IUnitOfWork UnitOfWork { get; set; }
 
        public IEnumerable<TEntity> All()
        {
            return CheckContractViolated(() =>
            {
                return CatchExceptionHelper.TryCatchFunction(() =>
                {
                    return UnitOfWork.Session.QueryOver<TEntity>().List();
                }, _logger);
            });
        }
 
        public int Add(TEntity entity)
        {
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            return CheckContractViolated(() =>
            {
                return CatchExceptionHelper.TryCatchFunction(() =>
                {
                    return (int)UnitOfWork.Session.Save(entity);
                }, _logger);
            });
        }
 
        public void Remove(TEntity entity)
        {
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            CheckContractViolated(() =>
            {
                CatchExceptionHelper.TryCatchAction(() => UnitOfWork.Session.Delete(entity), _logger);
            });
        }
 
        public void Update(TEntity entity)
        {
            CodeContract.Contract.Assert(entity != null, "Entity is null");
 
            CheckContractViolated(() =>
            {
                CatchExceptionHelper.TryCatchAction(() => UnitOfWork.Session.Update(entity), _logger);
            });
        }
 
        public IEnumerable<TEntity> GetBy(Expression<Func<TEntity, bool>> condition)
        {
            CodeContract.Contract.Assert(condition != null, "Expression is null");
            return CheckContractViolated(() =>
            {
                return CatchExceptionHelper.TryCatchFunction(() =>
                {
                    return UnitOfWork.Session.QueryOver<TEntity>()
                            .Where(condition)
                            .List();
                }, _logger);
            });
        }
 
        private void CheckContractViolated(Action action)
        {
            CheckContractViolated();
 
            action();
        }
 
        private TReturn CheckContractViolated<TReturn>(Func<TReturn> function)
        {
            CheckContractViolated();
 
            return function();
        }
 
        private void CheckContractViolated()
        {
            CodeContract.Contract.Assert(UnitOfWork != null, "Unit of work is null");
            CodeContract.Contract.Assert(UnitOfWork.Session != null, "Session is null");
        }
    }

As a code above, I add three functions with named is CheckContractViolated and just delegate the responsibility to it.If anyone don't like delegated the LambdaExpression for manage the contract, you only need the third CheckContractViolated method at end of this class. But I like Lambda Expression :D. Very simple, but with me it is very cool! Happy coding!

DateTime formatting extension method

2010-11-01T11:17:00Z

Notes: I also updated the pieces of code at http://www.extensionmethod.net/Details.aspx?ID=393

In current project, I get some troubles in DateTime class, about format the date time object with the pattern specific string and the current culture, so I decide to code some extension methods in DataTime class in .NET library. And I think it’s it very useful if somebody can use it as right way. I take some idea from this link and coding the enum class for its. After that I also used the Lambda Expression for pass the parameter into the DateTime formatting method. It’s really easy for read code. And below is my implementation for that:

+ DateTimeFormat enum class:

    public enum DateTimeFormat
    {
        [StringValue("unknown")]
        UnknownSpecific = 0,
        [StringValue("d")]
        ShortDate = 1,
        [StringValue("D")]
        LongDate = 2,
        [StringValue("t")]
        ShortTime = 3,
        [StringValue("T")]
        LongTime = 4,
        [StringValue("f")]
        FullShortDateTime = 5,
        [StringValue("F")]
        FullLongDateTime = 6,
        [StringValue("g")]
        GeneralShortDateTime = 7,
        [StringValue("G")]
        GeneralLongDateTime = 8,
        [StringValue("m")]
        MonthDayLowerCase = 9,
        [StringValue("M")]
        MonthDayUpperCase = 10,
        [StringValue("r")]
        Rfc1123LowerCase = 11,
        [StringValue("R")]
        Rfc1123UpperCase = 12,
        [StringValue("s")]
        SortableDateTimeIso8601 = 13,
        [StringValue("U")]
        UniversalSortableDateTime = 14,
        [StringValue("y")]
        YearMonthLowerCase = 15,
        [StringValue("Y")]
        YearMonthUpperCase = 16
    }

+ DateTime extension class:

    public static class DateTimeExtension
    {
        /// <summary>
        /// Toes the string format.
        /// </summary>
        /// <param name="source">The source.</param>
        /// <param name="dateTimeFormat">The date time format.</param>
        /// <returns></returns>
        public static string ToStringFormat(this DateTime source, Expression<Func<DateTimeFormat>> dateTimeFormat)
        {
            var dateTimeFormatCompiled = dateTimeFormat.Compile().Invoke();
 
            var dateTimeStringFormat = StringEnum.GetStringValue(dateTimeFormatCompiled);
 
            var currentCulture = Thread.CurrentThread.CurrentCulture;
 
            return source.ToString(dateTimeStringFormat, currentCulture);
        }
    }

+ Use it:

var shortDateResult = DateTime.Now.ToStringFormat(() => DateTimeFormat.ShortDate);
var fullLongDateTimeResult = DateTime.Now.ToStringFormat(() => DateTimeFormat.FullLongDateTime);
var rfc1123Result = DateTime.Now.ToStringFormat(() => DateTimeFormat.Rfc1123LowerCase);

+ Some unit testing:

    [TestClass]
    public class DateTimeFormatTesting
    {
        [TestMethod]
        public void Can_Get_String_Of_Date_Time_Format()
        {
            var shortDate = StringEnum.GetStringValue(DateTimeFormat.ShortDate);
            Assert.AreEqual(shortDate, "d");
            var longDate = StringEnum.GetStringValue(DateTimeFormat.LongDate);
            Assert.AreEqual(longDate, "D");
            var shortTime = StringEnum.GetStringValue(DateTimeFormat.ShortTime);
            Assert.AreEqual(shortTime, "t");
            var longTime = StringEnum.GetStringValue(DateTimeFormat.LongTime);
            Assert.AreEqual(longTime, "T");
            var monthDayLowerCase = StringEnum.GetStringValue(DateTimeFormat.MonthDayLowerCase);
            Assert.AreEqual(monthDayLowerCase, "m");
            var monthDayUpperCase = StringEnum.GetStringValue(DateTimeFormat.MonthDayUpperCase);
            Assert.AreEqual(monthDayUpperCase, "M");
            var rfc1123UpperCase = StringEnum.GetStringValue(DateTimeFormat.Rfc1123UpperCase);
            Assert.AreEqual(rfc1123UpperCase, "R");
 
            // and more testing for this enum at here
            // ...
        }
    }

    [TestClass]
    public class DateTimeStringFormatTesting
    {
        private DateTime _originalDateTime;
 
        [TestInitialize]
        public void InitTestCase()
        {
            _originalDateTime = new DateTime(1984, 3, 11, 2, 30, 58);
        }
 
        [TestMethod]
        public void Can_Get_Short_Date_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.ShortDate);
            Assert.AreEqual(_originalDateTime.ToShortDateString(), result);
        }
 
        [TestMethod]
        public void Can_Get_Long_Date_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.LongDate);
            Assert.AreEqual("11 March 1984", result);
        }
 
        [TestMethod]
        public void Can_Get_Short_Time_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.ShortTime);
            Assert.AreEqual("02:30", result);
        }
 
        [TestMethod]
        public void Can_Get_Long_Time_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.LongTime);
            Assert.AreEqual("02:30:58", result);
        }
 
        [TestMethod]
        public void Can_Get_Full_Short_Date_Time_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.FullShortDateTime);
            Assert.AreEqual("11 March 1984 02:30", result);
        }
 
        [TestMethod]
        public void Can_Get_Full_Long_Date_Time_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.FullLongDateTime);
            Assert.AreEqual("11 March 1984 02:30:58", result);
        }
 
        [TestMethod]
        public void Can_Get_General_Short_Date_Time_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.GeneralShortDateTime);
            Assert.AreEqual("11/03/1984 02:30", result);
        }
 
        [TestMethod]
        public void Can_Get_General_Long_Date_Time_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.GeneralLongDateTime);
            Assert.AreEqual("11/03/1984 02:30:58", result);
        }
 
        [TestMethod]
        public void Can_Get_Month_Day_Lower_Case_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.MonthDayLowerCase);
            Assert.AreEqual("11 March", result);
        }
 
        [TestMethod]
        public void Can_Get_Month_Day_Upper_Case_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.MonthDayUpperCase);
            Assert.AreEqual("11 March", result);
        }
 
        [TestMethod]
        public void Can_Get_Rfc1123_Lower_Case_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.Rfc1123LowerCase);
            Assert.AreEqual("Sun, 11 Mar 1984 02:30:58 GMT", result);
        }
 
        [TestMethod]
        public void Can_Get_Rfc1123_Upper_Case_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.Rfc1123UpperCase);
            Assert.AreEqual("Sun, 11 Mar 1984 02:30:58 GMT", result);
        }
 
        [TestMethod]
        public void Can_Get_Sortable_DateTime_Iso8601_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.SortableDateTimeIso8601);
            Assert.AreEqual("1984-03-11T02:30:58", result);
        }
 
        [TestMethod]
        public void Can_Get_Universal_Sortable_DateTime_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.UniversalSortableDateTime);
            Assert.AreEqual("11 March 1984 02:30:58", result);
        }
 
        [TestMethod]
        public void Can_Get_Year_Month_Lower_Case_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.YearMonthLowerCase);
            Assert.AreEqual("March 1984", result);
        }
 
        [TestMethod]
        public void Can_Get_Year_Month_Upper_Case_String()
        {
            var result = _originalDateTime.ToStringFormat(() => DateTimeFormat.YearMonthUpperCase);
            Assert.AreEqual("March 1984", result);
        }
 
        [TestCleanup]
        public void TeardownTestCase()
        {
            if (_originalDateTime != null)
            {
                GC.SuppressFinalize(_originalDateTime);
            }
        }
    }

That is it. Just 2 classes for this extension and if you have some concern about that, don't hesitate to contact with me. Code for this extension is attached in this post as below. Happy coding, mate!

NHibernate 3 and Autofac, can go together? (part 2)

2010-10-06T15:29:00Z

In part 1, I introduced about NHibernate and domain model for News Management System. And today, I will continue to show you about Autofac and how to make Autofac can understand all core instances of NHIbernate. If you do not read part 1, please read it at here. The time went very quickly, so we will start jumping into Autofac now.

All source code are placed in my github at here

Introduce about Autofac

Autofac is an IoC container for Microsoft .NET. It manages the dependencies between classes so that applications stay easy to change as they grow in size and complexity. This is achieved by treating regular .NET classes as components. I don’t want to talk more. Just go to here for more information

Inversion of Control (IoC)

Don’t ambiguity between Dependency Injection and Inversion of Control? Please make clearly about it before you can continue. If you have the IoC container you will don’t care about creation object task, object lifetime management, auto-wiring up… Please google for more information about tasks that IoC container can do for you.

AOP (Interceptor)

Cross-cutting concerns, Advice, Pointcut, Aspect are four features that you should care about. In this example, I only focus on implementation the Exception aspect on some Repository class using AutoFac (in Autofac they are interceptor).

Module (separated of concern: SoC): Advantages of Structuring with Modules:

Decreased Configuration Complexity
Configuration Parameters are Explicit
Abstraction from the Internal Application Architecture
Better Type Safety
Dynamic Configuration

In my example, I used 2 modules for object configuration: One for registering all components for NHibernate, and one for registering all core service, interceptor for my example.

Intergrated NHibernate into Autofac

To NHibernate can run well, it need us config something for it. Such as: Dialect, cache, auto generated schema, show sql...; and in Fluent NHibernate, we need adding some mapping class for configuration, adding some conventions for customize generated object. So we need Configuration object for store all things above.
After we have the configuration, we will base on it for creating the SessionFactory. Because SessionFactory is very expensive, so we usually only create only one SessionFactory on one AppDomain of your application (web form or window form). And then we can suggest SessionFactory for get to you one Session. Because Session is only run in short time, we call it is Session per Request (Session-per-request is an ORM pattern that means a new session is created for each request and terminated at the end of the request). So we can have many instances of it in AppDomain. Up to you! I usually set the Unit of work above ISession of NHibernate. It make all persistence actions is become clearly. I think my explain about Configuration, SessionFactory and Session is enough to you. Below is a code the implemented for it.

    public class NHibernateComponentModule : Module
    {
        public string ConnectionString { get; set; }
 
        public Assembly AssemblyMapper { get; set; }
 
        protected override void Load(ContainerBuilder builder)
        {
            Contract.Assert(builder != null, "Builder container is null");
            Contract.Assert(!string.IsNullOrEmpty(ConnectionString), "Cannot find connection string in config file");
            Contract.Assert(AssemblyMapper != null, "AssemblyMapper is null");
 
            var cfg = BuildConfiguration();
 
            var persistenceModel = BuildPersistenceModel();
            persistenceModel.Configure(cfg);
 
            var sessionFactory = BuildSessionFactory(cfg);
 
            RegisterConponents(builder, cfg, sessionFactory);
        }
 
        public Configuration BuildConfiguration()
        {
            var config = Fluently.Configure()
                .Database(MsSqlConfiguration.MsSql2005.ConnectionString(ConnectionString))
                .ExposeConfiguration(c => c.SetProperty(Environment.ReleaseConnections, "on_close"))
                .ExposeConfiguration(c => c.SetProperty(Environment.ProxyFactoryFactoryClass, typeof(NHibernate.ByteCode.Castle.ProxyFactoryFactory).AssemblyQualifiedName))
                .ExposeConfiguration(c => c.SetProperty(Environment.Hbm2ddlAuto, "create"))
                .ExposeConfiguration(c => c.SetProperty(Environment.ShowSql, "true"))
                .ExposeConfiguration(BuildSchema)
                .BuildConfiguration();
 
            if (config == null)
                throw new Exception("Cannot build NHibernate configuration");
 
            return config;
        }
 
        public AutoPersistenceModel BuildPersistenceModel()
        {
            var persistenceModel = new AutoPersistenceModel();
 
            persistenceModel.Conventions.Setup(c =>
            {
                c.Add(typeof(ForeignKeyNameConvention));
                c.Add(typeof(ReferenceConvention));
                c.Add(typeof(PrimaryKeyNameConvention));
                c.Add(typeof(TableNameConvention));
            });
 
            persistenceModel.AddMappingsFromAssembly(AssemblyMapper);
 
            persistenceModel.WriteMappingsTo(@"./");
 
            return persistenceModel;
        }
 
        public ISessionFactory BuildSessionFactory(Configuration config)
        {
            var sessionFactory = config.BuildSessionFactory();
 
            if (sessionFactory == null)
                throw new Exception("Cannot build NHibernate Session Factory");
 
            return sessionFactory;
        }
 
        public void RegisterConponents(ContainerBuilder builder, Configuration config, ISessionFactory sessionFactory)
        {
            builder.RegisterInstance(config).As<Configuration>().SingleInstance();
            builder.RegisterInstance(sessionFactory).As<ISessionFactory>().SingleInstance();
            builder.Register(x => x.Resolve<ISessionFactory>().OpenSession()).As<ISession>().InstancePerLifetimeScope();
            builder.Register(x => new UnitOfWork(x.Resolve<ISessionFactory>())).As<IUnitOfWork>();
            builder.Register(x => new UnitOfWorkFactory()).As<IUnitOfWorkFactory>().SingleInstance();
        }
 
        private static void BuildSchema(Configuration config)
        {
            new SchemaExport(config).SetOutputFile(@"./Schema.sql").Create(true, true);
        }
    }

As you see we registered SessionFactory is a InstancePerLifeTimeScope.

So we only have one SessionFactory in AppDomain.

For more information about object life cycle, please go to Autofac website.

I think I will not explain deeply about Repository, because it has already been rather popular. You can go back some my post in the past for have knowledge about it. I just show some code example for registering Repository and Interceptor to Autofac as here:

    public class ServiceComponentModule : Module
    {
        protected override void Load(ContainerBuilder builder)
        {
            Contract.Assert(builder != null, "Builder container is null");
 
            builder.RegisterType<ExceptionInterceptor>()
                .As<ExceptionInterceptor>();
 
            builder.RegisterType<CategoryRepository>()
                .As<ICategoryRepository>()
                .EnableInterfaceInterceptors()
                .InterceptedBy(typeof(ExceptionInterceptor));
 
            builder.RegisterType<NewsRepository>()
                .As<INewsRepository>()
                .EnableInterfaceInterceptors()
                .InterceptedBy(typeof(ExceptionInterceptor));
 
            builder.RegisterType<PollRepository>()
                .As<IPollRepository>()
                .EnableInterfaceInterceptors()
                .InterceptedBy(typeof(ExceptionInterceptor));
 
            builder.RegisterType<UserRepository>()
                .As<IUserRepository>()
                .EnableInterfaceInterceptors()
                .InterceptedBy(typeof(ExceptionInterceptor));
        }
    }

and the ExceptionInterceptor is:

    public class ExceptionInterceptor : IInterceptor
    {
        private readonly ILog _logger;
 
        public ExceptionInterceptor()
            : this(Log4NetWrapper.GetLogger(MethodBase.GetCurrentMethod().DeclaringType))
        {
        }
 
        public ExceptionInterceptor(ILog logger)
        {
            _logger = logger;
        }
 
        public void Intercept(IInvocation invocation)
        {
            try
            {
                _logger.Info("Before call " + invocation.Method.Name + " method");
                invocation.Proceed();
                _logger.Info("After call " + invocation.Method.Name + " method");
            }
            catch (Exception ex)
            {
                _logger.Error(ex.Message);
                throw;
            }
        }
    }

It is cool or not, man???

Intergrated CommonServiceLocator into Autofac

Why do you need know many things about register object, get objects, resolve objects? I am very lazy, so I usually delegate all tasks above for CommonServiceLocator doing. And here is the implementing of it:

    public class BootStrapper
    {
        private static IContainer _container;
 
        public BootStrapper()
        {
            _container = GetContainer();
 
            ServiceLocator.SetLocatorProvider(() => new AutofacServiceLocator(_container));
        }
 
        public static IContainer GetContainer()
        {
            var builder = new ContainerBuilder();
 
            builder.RegisterModule(new NHibernateComponentModule()
            {
                // TODO: should wrap the ConfigurationManager for unit testing
                ConnectionString = ConfigurationManager.ConnectionStrings["NHibernate3Sample"].ConnectionString,
                AssemblyMapper = typeof(CategoryMapping).Assembly
            });
 
            builder.RegisterModule(new ServiceComponentModule());
 
            return builder.Build();
        }
    }

As you see, I use a BootStrapper for boot my application. This way make the application more elegant, clearly.

Wrap CommonServiceLocator with IoC wrapper (Law of Demeter at here and here)

After you had read 2 links above about Law of Demeter (LoD), I thought you understood why did I wrap CommonServiceLocator :D.

public static class IoC
    {
        static IoC()
        {
            new BootStrapper();
        }
 
        public static TService GetInstance<TService>()
        {
            return ServiceLocator.Current.GetInstance<TService>();
        }
 
        public static object GetInstance(Type serviceType)
        {
            return ServiceLocator.Current.GetInstance(serviceType);
        }
 
        ...
    }

```
Boot your application
```

Just call new BootStrapper(); at the place that you want to boot your application.

In this example is in IoC class, but you can also call in Global.asax class in web form or App.xaml in WPF and Silverlight :)

That is all things I want to talking in this post. Thanks for your read and happy coding. Just dance!!!