Change table name at runtime

Question

Lets suppose that I have a db table with name Employee and a respective EF 6.0 db-first model.

Getting all rows of table Employee is done through q

Answer 1

I don't know if you should do that, but I think you can. You will have to dig into Entity Framework metadata structures, like MetadataWorkspace, which you can get from the underlying ObjectContext. See an example here: http://weblogs.asp.net/ricardoperes/entity-framework-metadata.

Answer 2

I know it's been been a while since the original post, but I'll add my answer to help someone else. I had generic SQL queue tables with different table names. I.e. the schema is exactly the same for both tables. I created a framework so that you can dynamically poll the table of your choice by providing the name and that's why I needed to update the table name at run-time. Basically, you can create an interceptor to intercept the raw SQL queries from entity framework and update the table name from there.

public class MyInterceptor : IDbCommandInterceptor
{
    private const string TableReplaceString = "[TheTableNameToReplace]";

    private void ReplaceTableName(DbCommand command, IEnumerable<DbContext> contexts)
    {
        var myContext = contexts?.FirstOrDefault(x => x is MyContext) as MyContext;
        if (myContext != null && command != null && command.CommandText.Contains(TableReplaceString))
        {
            command.CommandText = command.CommandText.Replace(TableReplaceString, $"[{myContext.NewTableName}]");
        }
    }

    public void NonQueryExecuting(DbCommand command, DbCommandInterceptionContext<int> interceptionContext)
    {
        ReplaceTableName(command, interceptionContext.DbContexts);
    }

    public void NonQueryExecuted(DbCommand command, DbCommandInterceptionContext<int> interceptionContext)
    {
        ReplaceTableName(command, interceptionContext.DbContexts);
    }

    public void ReaderExecuting(DbCommand command, DbCommandInterceptionContext<DbDataReader> interceptionContext)
    {
        ReplaceTableName(command, interceptionContext.DbContexts);
    }

    public void ReaderExecuted(DbCommand command, DbCommandInterceptionContext<DbDataReader> interceptionContext)
    {
        ReplaceTableName(command, interceptionContext.DbContexts);
    }

    public void ScalarExecuting(DbCommand command, DbCommandInterceptionContext<object> interceptionContext)
    {
        ReplaceTableName(command, interceptionContext.DbContexts);
    }

    public void ScalarExecuted(DbCommand command, DbCommandInterceptionContext<object> interceptionContext)
    {
        ReplaceTableName(command, interceptionContext.DbContexts);
    }
}

Of course, you have to get the new table name from somewhere. Either from the constructor or from a stored field in your custom DBContext which you can grab from interceptionContext.DbContexts.

Then you just have to register the interceptor for your context.

public class MyContext : DBContext
{
    public readonly string NewTableName;

    public MyContext(string connectionString, string newTableName)
        : base(connectionString)
    {
        NewTableName = newTableName;
        // Set interceptor
        DbInterception.Add(new MyInterceptor());
    }
}

UPDATE: I found that if you add the interceptor in the constructor above will cause memory leaks. DotMemory doesn't tell you about this. Make sure you add the interceptor in a static constructor instead.

public class MyContext : DBContext
{
    public readonly string NewTableName;

    static MyContext()
    {
        // Set interceptor only in static constructor
        DbInterception.Add(new MyInterceptor());
    }

    public MyContext(string connectionString, string newTableName)
        : base(connectionString)
    {
        NewTableName = newTableName;
    }
}

Answer 3

You must create a new dbcontext that inherits from the db-first model context and treat it like code-first in ef. Check the link, please. Same problem as yours.

https://www.codeproject.com/Articles/421643/How-to-Use-MVC-Net-on-the-Dynamics-NAV-Database-St#_articleTop

So when mapping, you can get the table name dynamically.

Answer 4

Old question, but based on the problem I would suggest that you look at partioning your table based on a "current" bit or a datetime field. Partitioning is based on a column value & is supported by most modern DBMS. It would avoid issue at the ORM level.

Answer 5

Why not use some good old fashioned polymorphism?

partial class Employee : IEmployee { }
partial class HistoricalEmployee : IEmployee { }

interface IEmployee {
    public string Name { get; set; }
}

void PrintEmployeeName(IEmployee employee)
{
    Debug.WriteLine(employee.Name);
}

PrintEmployeeName(context.Employees.First());
PrintEmployeeName(context.HistoricalEmployees.First());

Answer 6

Thanks for the answers.

I think that my case is a real-world scenario that is, typically neglected in all "getting-started" typical scenarios of EF tutorials and examples.

Based on the fact that I use db-first approach and the switch should be at the application level, I think that I will create a Context instance, based on different SSDL with the new table name that the user will use on demand