Develop Reference

Inject a string property in a class using Ninject

One of my interfaces has a string property that will depend on where the interface is being used. I want to avoid hardcoding the property every time the object is created. I can set the property in constructor, but the object is injected using a factory.
The interface as follows:
public interface IObjectStore
{
string StorageTableName { get; set;}
void UpdateObjectStore(string key, string value);
string ReadObjectStore(string key);
}
Which is used in a service
public class CategoryService<T> : ICategoryService<T> where T : Company
{
private readonly IObjectStore objectStore;
public CategoryService(IObjectStore objStore)
{
this.objectStore = objStore;
objectStore.StorageTableName = "CategoryTable"; // I want to avoid this hard coding
}
...
}
The service is created using service factory (Ninject.Extensions.Factory)
public interface IServiceFactory
{
ICategoryService<T> CreateCategoryService<T>() where T : class;
}
Which is then injected using Ninject at the controller level. Here are my bindings
bool storeInNoSql = true;
kernel.Bind<IServiceFactory>().ToFactory().InSingletonScope();
kernel.Bind<ICategoryService<Article>>().To<CategoryService<Article>>();
kernel.Bind<IObjectStore>().ToMethod(ctx => storeInNoSql ? ctx.Kernel.Get<ObjectStore>() : null);
So the question is: how do i tell Ninject to set the property StorageTableName to "CategoryTable" everytime the object is injected into CategoryService and to "ArticleTable" everytime it is inserted into ArticleService?

I think this is what you are looking for.
It's just a very small sample project I just did, but this should solve your problem.
public class Ninject_34091099
{
public static void Run()
{
using (IKernel kernel = new StandardKernel())
{
kernel.Bind<IInterface<Generic1>>()
.To<Class<Generic1>>()
.WithConstructorArgument("name", "STRING ONE");
kernel.Bind<IInterface<Generic2>>()
.To<Class<Generic2>>()
.WithConstructorArgument("name", "The other string");
kernel.Bind<IServiceFactory>().ToFactory().InSingletonScope();
var factory = kernel.Get<IServiceFactory>();
var c1 = factory.CreateInterface<Generic1>();
var c2 = factory.CreateInterface<Generic2>();
Console.WriteLine(c1.Name);
Console.WriteLine(c2.Name);
}
Console.WriteLine("Done");
Console.ReadLine();
}
}
public interface IInterface<T> where T : class
{
string Name { get; set; }
}
public class Generic1
{
}
public class Generic2
{
}
public class Class<T> : IInterface<T> where T : class
{
public string Name { get; set; }
public Class(string name)
{
Name = name;
}
}
public interface IServiceFactory
{
IInterface<T> CreateInterface<T>() where T : class;
}
Sorry that the names mean nothing :D
Hope it helps

C# Repository with multiple constraints and inheritance

i wasn't able to find a similar issue but feel free to redirect me if i just missed it.
I am trying to get familiar with the Repository pattern.
I'll give you an example of the code i'm trying to get to work unsuccessfully.
These are the classes and interfaces that represent the entity i'm using.
public class AbsObj
{
public string Code { get; set; }
}
public interface IAbsObj
{
bool Save();
}
public class User : AbsObj
{
public string Language{get; set;}
}
public class DbUser : User, IAbsObj
{
public bool Save()
{
return true;
}
}
Then to the repository Interface
public interface IRepository<T>
{
void Add(T value);
void Update(T value);
void Delete(T value);
}
The generic Repository
public class Repository<T> : IRepository<T> where T : AbsObj, IAbsObj
{
protected List<T> _lst;
public Repository()
{
_lst = new List<T>();
}
public void Add(T value)
{
}
public void Update(T value)
{
}
public void Delete(T value)
{
}
public bool Save()
{
for (int i = 0; i < _lst.Count; i++)
{
_lst[i].Save();
}
return true;
}
}
Then a more specific repository, which should handle the loading of the users from the db:
public class UserRepository<T> : Repository<T> where T : AbsObj, IAbsObj
{
public void Load()
{
DbUser us = new DbUser();
us.Code = "Cod";
us.Language = "IT";
_lst.Add(us);
}
}
I created the DBUser class just to have the freedom to create an XMLUser in the future which would handle a different type of saving.
It inherits from User which in turn inherits from AbsObj.
It implements IAbsObj.
Nonetheless i got a compile time error when i try to add to the list the DbUser object created, stating that it's impossible to convert from DBUser to T.
Given the constraints i tought it was possible: what am i missing here?
Thanks in advance for any help!

Your UserRepository definition could be:
public class UserRepository : Repository<DbUser>
{
....
}
But since you want to make it generic for XMLUser as well:
public class UserRepository<T> : Repository<T> where T: User, new()
{
public void Load()
{
User us = new T() as User;
us.Code = "Cod";
us.Language = "IT";
_lst.Add(us);
}
}
To use:
new UserRepostitory<DbUser>();
new UserRepostitory<XmlUser>();

Setting the connection string of a DBContext in my repository class using Ninject

I have an MVC 5 application that uses EF 6 and implements Repository pattern with dependency injection using the DI container Ninject. The connection string for the dbcontext is stored in the Web.config file which the EF Context properly finds. Everything works fine. Lately, I have a requirement that the connection to my DBContext need to be determined at runtime and connect to different databases (but with exactly the same structure). So, I need to change the sql connectionstring part from the entity connectionstring at run-time before the repository is instantiated. I would really appreciate some help in doing it. I am not a DI guru; know just enough Ninject to get my things going.
Here is my Repository Base Interface:
public interface IRepositoryBase<T> where T : class
{
void Add(T entity, string userGuid = "");
void Delete(T entity);
// ... removed other method signatures for brevity
}
My Repository base implementation:
public abstract class RepositoryBase<D, T> : IRepositoryBase<T>, IDisposable
where T : class
where D : DbContext, new()
{
private Guid? currUserGuid = null;
private D dataContext;
protected D DataContext
{
get
{
if (dataContext == null)
dataContext = new D();
return dataContext;
}
set { dataContext = value; }
}
public IQueryable<T> FindBy(Expression<Func<T, bool>> predicate)
{
return DataContext.Set<T>().Where(predicate);
}
public virtual IQueryable<T> GetAll()
{
IQueryable<T> query = DataContext.Set<T>();
return query;
}
public virtual void Delete(T entity)
{
OperationStatus stat = TryDelete(entity);
}
// .... removed rest for brevity
}
Interface and implementation for concrete class:
public interface ICustomerRepository : IRepositoryBase<Customer>
{
Customer GetCustomerAndStatus( Guid custGuid );
}
public class CustomerRepository : RepositoryBase<PCDataEFContext, Customer>, ICustomerRepository
{
public Customer GetCustomerAndStatus( Guid custGuid )
{
return DataContext.Customers.Include( x => x.CustStatusType )
.SingleOrDefault( x => x.PKGuid == custGuid );
}
}
My Ninject dependency resolver:
public class NinjectDependencyResolver : IDependencyResolver
{
private IKernel kernel;
public NinjectDependencyResolver()
{
kernel = new StandardKernel();
AddBindings();
}
public IKernel Kernel { get { return kernel; } }
private void AddBindings()
{
kernel.Bind<ICustomerRepository>().To<CustomerRepository>();
// ... other bindings are omitted for brevity
}
}
and finally, here is my Entity Framework generated DBContext:
public partial class PCDataEFContext : DbContext
{
public PCDataEFContext()
: base("name=PCDataEFContext")
{
}
protected override void OnModelCreating(DbModelBuilder modelBuilder)
{
throw new UnintentionalCodeFirstException();
}
public virtual DbSet<Customer> Customers { get; set; }
}
All the above code works great! But as I said in the beginning, I don't know how to inject the connection string into my Repositorybase class at runtime so that I don't have to modify any of my inherited repositories (I have plenty of them). Someone please help.
Babu.

Could you do it like this?
public partial class PCDataEFContext : DbContext
{
public PCDataEFContext()
: base(Util.GetTheConnectionString())
{ }
}
public class MyDerivedContext : PCDataEFContext
{
public MyDerivedContext()
: base()
{ }
}
class Util
{
public static string GetTheConnectionString()
{
// return the correct name based on some logic...
return "name=PCDataEFContext";
}
}
Another way of doing it, could be in the RepositorBase class you defined, by altering the connectionstring after the creation of the dbcontext:
protected D DataContext
{
get
{
if (dataContext == null)
{
dataContext = new D();
dataContext.Database.Connection.ConnectionString = "the new connectionstring";
}
return dataContext;
}
set { dataContext = value; }
}

How to call a method in another class from a generic method?

Why doesn't the line marked with //Dont work in the bottom of the code compile?
I want to reuse the WriteMessage method with different Classes, I try to use generics, but I'm not sure how to use it.
class ClassOne
{
public string MethodOne()
{
return ("ClassOne");
}
public string MethodTwo()
{
return ("ClassOne -MethodTwo ");
}
}
class ClassTwo
{
public string MethodOne()
{
return ("ClassTwo");
}
public string MethodTwo()
{
return ("ClassOne -MethodTwo ");
}
}
class Program
{
private static void Main()
{
var objectOne = new ClassOne();
WriteMessage(objectOne);
var objectTwo = new ClassTwo();
WriteMessage(objectTwo);
Console.ReadKey();
}
public static void WriteMessage<T>(T objectA)
{
var text = objectA.MethodTwo(); //Dont Work
Console.WriteLine("Text:{0}", text);
}
}

Try implementing a interface :
Example :
public interface IHasTwoMethods
{
string MethodOne()
string MethodTwo()
}
Implement this inteface on your classes :
class ClassOne : IHasTwoMethods
class ClassTwo : IHasTwoMethods
Then in your generic method do like this :
public static void WriteMessage<T>(T objectA) where T : IHasTwoMethods
{
var text = objectA.MethodTwo(); //Will work
Console.WriteLine("Text:{0}", text);
}
You can read more about interfaces here : http://msdn.microsoft.com/en-us/library/87d83y5b.aspx

This doesn't compile because as far as the compiler is concerned objectA is just an Object.
To get this to work, you need to use a generic type constraint:
public interface MyInterface
{
string MethodTwo();
}
public class A : MyInterface
{
...
}
public class B : MyInterface
{
...
}
public static void WriteMessage<T>(T objectA) where T: MyInterface
{
var text = objectA.MethodTwo(); //Will Work!
Console.WriteLine("Text:{0}", text);
}
MSDN : Constraints on Type Parameters

Since you're passing in a generically-typed object with T, the compiler doesn't know what class you're using--for all it knows, it could be an int or an Application or anything.
What you probably want is to have ClassOne and ClassTwo inherit from another class that has an abstract MethodTwo class that both implement. Something like...
abstract class SuperClass
{
public abstract string MethodOne();
}
class ClassOne : SuperClass
{
public override string MethodOne()
{
return ("ClassOne");
}
}
then in Main:
public static void WriteMessage<T>(T objectA) where T : SuperClass
{
var text = objectA.MethodOne();
Console.WriteLine("Text:{0}", text);
}
Read up on C# inheritance here: http://msdn.microsoft.com/en-us/library/ms173149.aspx

Dynamically selecting the type of a component given the type of its generic type parameter

I frequently seems to come up to a situation where I have an abstract type which needs to be processed differently depending on which concrete implementation it has.
As an example, an abstract class Payment could be subclassed as class CreditCard or class StoredCredit. To actually process the payment, we want to use an implementation of
interface IPaymentTaker {
PaymentProcessingResult Process(Payment payment); }
i.e. either
class CreditCardPaymentTaker : IPaymentTaker { ... }
or
class StoredCreditPaymentTaker : IPaymentTaker { ... }
In the past I have injected an IDictionary into the parent component and then done
_paymentTakers[payment.GetType()].Process(payment);
The downside of this is that the IPaymentTaker implementations are not strongly typed enough, so the first bit of the Process method has to be:
Process(Payment payment)
{
var creditCardPayment = payment as CreditCardPayment;
if (creditCardPayment == null)
throw new Exception("Payment must be of type CreditCard");
}
I'm sure there must be a name for the pattern I'm trying to implement but I don't know what it is!
Ideally I would
(a) be able to instantiate the PaymentProcessor based just on the type of the Payment, without creating the dictionary;
(b) be able to have strongly typed PaymentProcessors that only accept the subclass they can use.
Does anyone have a neat way of solving this problem?

You can solve this with a visitor:
interface IPaymentVisitor {
void Visit(CreditCard payment);
void Visit(StoredCredit payment);
}
abstract class Payment {
public abstract void Accept(IPaymentVisitor visitor);
}
class CreditCard : Payment {
public override void Accept(IPaymentVisitor visitor) {
visitor.Visit(this);
}
}
class StoredCredit : Payment {
public override void Accept(IPaymentVisitor visitor) {
visitor.Visit(this);
}
}
class PaymentTaker : IPaymentVisitor, IPaymentTaker {
public void Visit(CreditCard payment) {
// ...
}
public void Visit(StoredCredit payment) {
// ...
}
public PaymentProcessingResult Process(Payment payment) {
payment.Accept(this);
// ...
}
}
If you still want to separate the different payment takers, or if your hierarchy jitters, you can use an acyclic visitor (pdf):
interface IPaymentVisitor {
}
interface IPaymentVisitor<TPayment> : IPaymentVisitor where TPayment : Payment {
void Visit(TPayment payment);
}
abstract class Payment {
public abstract void Accept(IPaymentVisitor visitor);
}
class CreditCard : Payment {
public override void Accept(IPaymentVisitor visitor) {
if (visitor is IPaymentVisitor<CreditCard>) {
((IPaymentVisitor<CreditCard>)visitor).Visit(this);
}
}
}
class StoredCredit : Payment {
public override void Accept(IPaymentVisitor visitor) {
if (visitor is IPaymentVisitor<StoredCredit>) {
((IPaymentVisitor<StoredCredit>)visitor).Visit(this);
}
}
}
class CreditCardPaymentTaker : IPaymentVisitor<CreditCard>, IPaymentTaker {
public void Visit(CreditCard payment) {
// ...
}
public PaymentProcessingResult Process(Payment payment) {
payment.Accept(this);
// ...
}
}
class StoredCreditPaymentTaker : IPaymentVisitor<StoredCredit>, IPaymentTaker {
public void Visit(StoredCredit payment) {
// ...
}
public PaymentProcessingResult Process(Payment payment) {
payment.Accept(this);
// ...
}
}

interface IPayment
{
IPaymentTaker Taker {get;}
}
class CreditCardPayment : IPayment
{
IPaymentTaker Taker{ get {return new CreditCardPaymentTaker();}}
}
payment.Taker.Process(payment);

Even though James' method is ideal, using an IoC container could be difficult. Here's my Reflection or dynamics based approach. Doing the following will allow you to still use an IoC to setup the mapping between the PaymentTaker and Payment.
public class Payment
{
}
public class CreditCardPayment : Payment
{
}
public class StoreCreditPayment : Payment
{
}
public interface IPaymentTaker
{
}
public interface IPaymentTaker<T> : IPaymentTaker
{
void Process(T payment);
}
public static class PaymentTaker
{
public static void Process(Payment payment)
{
var paymentType = payment.GetType();
// You would have these already setup and loaded via your IOC container...
var paymentTakers = new Dictionary<Type, IPaymentTaker>();
paymentTakers.Add(typeof(CreditCardPayment), new CreditCardPaymentTaker());
paymentTakers.Add(typeof(StoreCreditPayment), new StoreCreditPaymentTaker());
// Get the payment taker for the specific payment type.
var paymentTaker = paymentTakers[paymentType];
// Execute the 'Process' method.
paymentTaker.GetType().GetMethod("Process").Invoke(paymentTaker, new object[]{ payment });
// If .NET 4.0 - dynamics can be used.
// dynamic paymentTaker = paymentTakers[paymentType];
// paymentTaker.Process((dynamic)payment);
}
}
public class CreditCardPaymentTaker : IPaymentTaker<CreditCardPayment>
{
public void Process(CreditCardPayment payment)
{
Console.WriteLine("Process Credit Card Payment...");
}
}
public class StoreCreditPaymentTaker : IPaymentTaker<StoreCreditPayment>
{
public void Process(StoreCreditPayment payment)
{
Console.WriteLine("Process Credit Card Payment...");
}
}
And then you can use it like this:
var cc = new CreditCardPayment();
PaymentTaker.Process(cc);

If you can ensure the names of the Payment and PaymentTaker match you can use something like this:
Process(Payment payment)
{
String typeName = "YourPathToPaymentTakers." + payment.GetType().Name + "Taker";
Type type = typeof(IPaymentTaker).Assembly.GetType(typeName);
IPaymentTaker taker = (IPaymentTaker)Activator.CreateInstance(type);;
}
I have used this approach in the past, but if you do not have 100% control of the names of the classes this could be a problem.