Automatically create an Enum based on values in a database lookup table?

How do I automatically create an enum and subsequently use its values in C# based on values in a database lookup table (using enterprise library data layer)?

For example, If I add a new lookup value in the database, I don't want to have to manually add the extra static enum value declaration in code - I'd like to keep the enum in sync with the database.

Is there such a thing as this?


I don't want to create a code generated static enum (as per The Code Project article Enum Code Generator - Generating enum code automatically from database look up tables) and would prefer it to be completely automatic.

163966 次浏览

Does it have to be an actual enum? How about using a Dictionary<string,int> instead?

for example

Dictionary<string, int> MyEnum = new Dictionary()\{\{"One", 1}, {"Two", 2}};
Console.WriteLine(MyEnum["One"]);

Enums must be specified at compile time, you can't dynamically add enums during run-time - and why would you, there would be no use/reference to them in the code?

From Professional C# 2008:

The real power of enums in C# is that behind the scenes they are instantiated as structs derived from the base class, System.Enum . This means it is possible to call methods against them to perform some useful tasks. Note that because of the way the .NET Framework is implemented there is no performance loss associated with treating the enums syntactically as structs. In practice, once your code is compiled, enums will exist as primitive types, just like int and float .

So, I'm not sure you can use Enums the way you want to.

You want System.Web.Compilation.BuildProvider

I also doubt the wisdom of doing this, but then there maybe a good use case that I can't think of.

What you're looking for are Build Providers i.e. System.Web.Compilation.BuildProvider

They're used very effectively by SubSonic, you can download the source and see how they use them, you won't need anything half as intricate as what they're doing.

Hope this helps.

Let's say you have the following in your DB:

table enums
-----------------
| id | name     |
-----------------
| 0  | MyEnum   |
| 1  | YourEnum |
-----------------


table enum_values
----------------------------------
| id | enums_id | value | key    |
----------------------------------
| 0  | 0        | 0     | Apple  |
| 1  | 0        | 1     | Banana |
| 2  | 0        | 2     | Pear   |
| 3  | 0        | 3     | Cherry |
| 4  | 1        | 0     | Red    |
| 5  | 1        | 1     | Green  |
| 6  | 1        | 2     | Yellow |
----------------------------------

Construct a select to get the values you need:

select * from enums e inner join enum_values ev on ev.enums_id=e.id where e.id=0

Construct the source code for the enum and you'll get something like:

String enumSourceCode = "enum " + enumName + "{" + enumKey1 + "=" enumValue1 + "," + enumKey2 + ... + "}";

(obviously this is constructed in a loop of some kind.)

Then comes the fun part, Compiling your enum and using it:

CodeDomProvider provider = CodeDomProvider.CreateProvider("CSharp");
CompilerParameters cs = new CompilerParameters();
cp.GenerateInMemory = True;


CompilerResult result = provider.CompileAssemblyFromSource(cp, enumSourceCode);


Type enumType = result.CompiledAssembly.GetType(enumName);

Now you have the type compiled and ready for use.
To get a enum value stored in the DB you can use:

[Enum].Parse(enumType, value);

where value can be either the integer value (0, 1, etc.) or the enum text/key (Apple, Banana, etc.)

I'm doing this exact thing, but you need to do some kind of code generation for this to work.

In my solution, I added a project "EnumeratedTypes". This is a console application which gets all of the values from the database and constructs the enums from them. Then it saves all of the enums to an assembly.

The enum generation code is like this:

// Get the current application domain for the current thread
AppDomain currentDomain = AppDomain.CurrentDomain;


// Create a dynamic assembly in the current application domain,
// and allow it to be executed and saved to disk.
AssemblyName name = new AssemblyName("MyEnums");
AssemblyBuilder assemblyBuilder = currentDomain.DefineDynamicAssembly(name,
AssemblyBuilderAccess.RunAndSave);


// Define a dynamic module in "MyEnums" assembly.
// For a single-module assembly, the module has the same name as the assembly.
ModuleBuilder moduleBuilder = assemblyBuilder.DefineDynamicModule(name.Name,
name.Name + ".dll");


// Define a public enumeration with the name "MyEnum" and an underlying type of Integer.
EnumBuilder myEnum = moduleBuilder.DefineEnum("EnumeratedTypes.MyEnum",
TypeAttributes.Public, typeof(int));


// Get data from database
MyDataAdapter someAdapter = new MyDataAdapter();
MyDataSet.MyDataTable myData = myDataAdapter.GetMyData();


foreach (MyDataSet.MyDataRow row in myData.Rows)
{
myEnum.DefineLiteral(row.Name, row.Key);
}


// Create the enum
myEnum.CreateType();


// Finally, save the assembly
assemblyBuilder.Save(name.Name + ".dll");

My other projects in the solution reference this generated assembly. As a result, I can then use the dynamic enums in code, complete with intellisense.

Then, I added a post-build event so that after this "EnumeratedTypes" project is built, it runs itself and generates the "MyEnums.dll" file.

By the way, it helps to change the build order of your project so that "EnumeratedTypes" is built first. Otherwise, once you start using your dynamically generated .dll, you won't be able to do a build if the .dll ever gets deleted. (Chicken and egg kind of problem -- your other projects in the solution need this .dll to build properly, and you can't create the .dll until you build your solution...)

I got most of the above code from this msdn article.

Hope this helps!

Just showing the answer of Pandincus with "of the shelf" code and some explanation: You need two solutions for this example ( I know it could be done via one also ; ), let the advanced students present it ...

So here is the DDL SQL for the table :

USE [ocms_dev]
GO


CREATE TABLE [dbo].[Role](
[RoleId] [int] IDENTITY(1,1) NOT NULL,
[RoleName] [varchar](50) NULL
) ON [PRIMARY]

So here is the console program producing the dll:

using System;
using System.Collections.Generic;
using System.Text;
using System.Reflection;
using System.Reflection.Emit;
using System.Data.Common;
using System.Data;
using System.Data.SqlClient;


namespace DynamicEnums
{
class EnumCreator
{
// after running for first time rename this method to Main1
static void Main ()
{
string strAssemblyName = "MyEnums";
bool flagFileExists = System.IO.File.Exists (
AppDomain.CurrentDomain.SetupInformation.ApplicationBase +
strAssemblyName + ".dll"
);


// Get the current application domain for the current thread
AppDomain currentDomain = AppDomain.CurrentDomain;


// Create a dynamic assembly in the current application domain,
// and allow it to be executed and saved to disk.
AssemblyName name = new AssemblyName ( strAssemblyName );
AssemblyBuilder assemblyBuilder =
currentDomain.DefineDynamicAssembly ( name,
AssemblyBuilderAccess.RunAndSave );


// Define a dynamic module in "MyEnums" assembly.
// For a single-module assembly, the module has the same name as
// the assembly.
ModuleBuilder moduleBuilder = assemblyBuilder.DefineDynamicModule (
name.Name, name.Name + ".dll" );


// Define a public enumeration with the name "MyEnum" and
// an underlying type of Integer.
EnumBuilder myEnum = moduleBuilder.DefineEnum (
"EnumeratedTypes.MyEnum",
TypeAttributes.Public,
typeof ( int )
);


#region GetTheDataFromTheDatabase
DataTable tableData = new DataTable ( "enumSourceDataTable" );


string connectionString = "Integrated Security=SSPI;Persist " +
"Security Info=False;Initial Catalog=ocms_dev;Data " +
"Source=ysg";


using (SqlConnection connection =
new SqlConnection ( connectionString ))
{


SqlCommand command = connection.CreateCommand ();
command.CommandText = string.Format ( "SELECT [RoleId], " +
"[RoleName] FROM [ocms_dev].[dbo].[Role]" );


Console.WriteLine ( "command.CommandText is " +
command.CommandText );


connection.Open ();
tableData.Load ( command.ExecuteReader (
CommandBehavior.CloseConnection
) );
} //eof using


foreach (DataRow dr in tableData.Rows)
{
myEnum.DefineLiteral ( dr[1].ToString (),
Convert.ToInt32 ( dr[0].ToString () ) );
}
#endregion GetTheDataFromTheDatabase


// Create the enum
myEnum.CreateType ();


// Finally, save the assembly
assemblyBuilder.Save ( name.Name + ".dll" );
} //eof Main
} //eof Program
} //eof namespace

Here is the Console programming printing the output ( remember that it has to reference the dll ). Let the advance students present the solution for combining everything in one solution with dynamic loading and checking if there is already build dll.

// add the reference to the newly generated dll
use MyEnums ;


class Program
{
static void Main ()
{
Array values = Enum.GetValues ( typeof ( EnumeratedTypes.MyEnum ) );


foreach (EnumeratedTypes.MyEnum val in values)
{
Console.WriteLine ( String.Format ( "{0}: {1}",
Enum.GetName ( typeof ( EnumeratedTypes.MyEnum ), val ),
val ) );
}


Console.WriteLine ( "Hit enter to exit " );
Console.ReadLine ();
} //eof Main
} //eof Program

You could use CodeSmith to generate something like this:

http://www.csharping.com/PermaLink,guid,cef1b637-7d37-4691-8e49-138cbf1d51e9.aspx

I don't think there is a good way of doing what you want. And if you think about it I don't think this is what you really want.

If you would have a dynamic enum, it also means you have to feed it with a dynamic value when you reference it. Maybe with a lot of magic you could achieve some sort of IntelliSense that would take care of this and generate an enum for you in a DLL file. But consider the amount of work it would take, how uneffective it would be to access the database to fetch IntelliSense information as well as the nightmare of version controlling the generated DLL file.

If you really don't want to manually add the enum values (you'll have to add them to the database anyway) use a code generation tool instead, for example T4 templates. Right click+run and you got your enum statically defined in code and you get all the benefits of using enums.

Using dynamic enums is bad no matter which way. You will have to go through the trouble of "duplicating" the data to ensure clear and easy code easy to maintain in the future.

If you start introducing automatic generated libraries, you are for sure causing more confusion to future developers having to upgrade your code than simply making your enum coded within the appropriate class object.

The other examples given sound nice and exciting, but think about the overhead on code maintenance versus what you get from it. Also, are those values going to change that frequently?

I've done this with a T4 template. It is fairly trivial to drop a .tt file into your project, and set up Visual Studio to run the T4 template as a pre-build step.

The T4 generates a .cs file, which means you can have it just query the database and build an enum in a .cs file from the result. Wired up as a pre-build task, it would re-create your enum on every build, or you can run the T4 manually as needed instead.

I always like to write my own "custom enum". Than I have one class that is a little bit more complex, but I can reuse it:

public abstract class CustomEnum
{
private readonly string _name;
private readonly object _id;


protected CustomEnum( string name, object id )
{
_name = name;
_id = id;
}


public string Name
{
get { return _name; }
}


public object Id
{
get { return _id; }
}


public override string ToString()
{
return _name;
}
}


public abstract class CustomEnum<TEnumType, TIdType> : CustomEnum
where TEnumType : CustomEnum<TEnumType, TIdType>
{
protected CustomEnum( string name, TIdType id )
: base( name, id )
{ }


public new TIdType Id
{
get { return (TIdType)base.Id; }
}


public static TEnumType FromName( string name )
{
try
{
return FromDelegate( entry => entry.Name.Equals( name ) );
}
catch (ArgumentException ae)
{
throw new ArgumentException( "Illegal name for custom enum '" + typeof( TEnumType ).Name + "'", ae );
}
}


public static TEnumType FromId( TIdType id )
{
try
{
return FromDelegate( entry => entry.Id.Equals( id ) );
}
catch (ArgumentException ae)
{
throw new ArgumentException( "Illegal id for custom enum '" + typeof( TEnumType ).Name + "'", ae );
}
}


public static IEnumerable<TEnumType> GetAll()
{
var elements = new Collection<TEnumType>();
var infoArray = typeof( TEnumType ).GetFields( BindingFlags.Public | BindingFlags.Static );


foreach (var info in infoArray)
{
var type = info.GetValue( null ) as TEnumType;
elements.Add( type );
}


return elements;
}


protected static TEnumType FromDelegate( Predicate<TEnumType> predicate )
{
if(predicate == null)
throw new ArgumentNullException( "predicate" );


foreach (var entry in GetAll())
{
if (predicate( entry ))
return entry;
}


throw new ArgumentException( "Element not found while using predicate" );
}
}

Now I just need to create my enum I want to use:

 public sealed class SampleEnum : CustomEnum<SampleEnum, int>
{
public static readonly SampleEnum Element1 = new SampleEnum( "Element1", 1, "foo" );
public static readonly SampleEnum Element2 = new SampleEnum( "Element2", 2, "bar" );


private SampleEnum( string name, int id, string additionalText )
: base( name, id )
{
AdditionalText = additionalText;
}


public string AdditionalText { get; private set; }
}

At last I can use it like I want:

 static void Main( string[] args )
{
foreach (var element in SampleEnum.GetAll())
{
Console.WriteLine( "{0}: {1}", element, element.AdditionalText );
Console.WriteLine( "Is 'Element2': {0}", element == SampleEnum.Element2 );
Console.WriteLine();
}


Console.ReadKey();
}

And my output would be:

Element1: foo
Is 'Element2': False


Element2: bar
Is 'Element2': True

One way to keep the Enums and to create a Dynamic list of values at the same time is to use the Enums that you currently have with a Dynamically created Dictionary.

Since most Enums are used in the context that they are defined to be used, and the "dynamic enums" will be supported by dynamic processes, you can distinguish the 2.

The first step is to create a table/collection that houses the IDs and References for the Dynamic Entries. In the table you will autoincrement much larger than your largest Enum value.

Now comes the part for your dynamic Enums, I am assuming that you will be using the Enums to create a set of conditions that apply a set of rules, some are dynamically generated.

Get integer from database
If Integer is in Enum -> create Enum -> then run Enum parts
If Integer is not a Enum -> create Dictionary from Table -> then run Dictionary parts.

Aren't we coming to this from the wrong direction?

If the data is likely to change at all during the lifetime of the deployed release then an enum is just not appropriate, and you need to use a dictionary, hash or other dynamic collection.

If you know the set of possible values is fixed for the life of the deployed release, then an enum is preferable.

If you must have something in your database that replicates the enumerated set, then why not add a deployment step to clear and repopulate the database table with the definitive set of enum values?

enum builder class

public class XEnum
{
private EnumBuilder enumBuilder;
private int index;
private AssemblyBuilder _ab;
private AssemblyName _name;
public XEnum(string enumname)
{
AppDomain currentDomain = AppDomain.CurrentDomain;
_name = new AssemblyName("MyAssembly");
_ab = currentDomain.DefineDynamicAssembly(
_name, AssemblyBuilderAccess.RunAndSave);


ModuleBuilder mb = _ab.DefineDynamicModule("MyModule");


enumBuilder = mb.DefineEnum(enumname, TypeAttributes.Public, typeof(int));




}
/// <summary>
/// adding one string to enum
/// </summary>
/// <param name="s"></param>
/// <returns></returns>
public FieldBuilder add(string s)
{
FieldBuilder f = enumBuilder.DefineLiteral(s, index);
index++;
return f;
}
/// <summary>
/// adding array to enum
/// </summary>
/// <param name="s"></param>
public void addRange(string[] s)
{
for (int i = 0; i < s.Length; i++)
{
enumBuilder.DefineLiteral(s[i], i);
}
}
/// <summary>
/// getting index 0
/// </summary>
/// <returns></returns>
public object getEnum()
{
Type finished = enumBuilder.CreateType();
_ab.Save(_name.Name + ".dll");
Object o1 = Enum.Parse(finished, "0");
return o1;
}
/// <summary>
/// getting with index
/// </summary>
/// <param name="i"></param>
/// <returns></returns>
public object getEnum(int i)
{
Type finished = enumBuilder.CreateType();
_ab.Save(_name.Name + ".dll");
Object o1 = Enum.Parse(finished, i.ToString());
return o1;
}
}

create an object

string[] types = { "String", "Boolean", "Int32", "Enum", "Point", "Thickness", "long", "float" };
XEnum xe = new XEnum("Enum");
xe.addRange(types);
return xe.getEnum();

Word up, I as well got tired of writing out enumerations based on Id / Name db table columns, copying and pasting stuff from queries in SSMS.

Below is a super dirty stored procedure that takes as input a table name, the column name you want to use for the c# enumeration name, and the column name that you want to use for the c# enumeration value.

Most of theses table names I work with a) end with "s" b) have a [TABLENAME]Id column and c) have a [TABLENAME]Name column, so there are a couple if statements that will assume that structure, in which case, the column name parameters are not required.

A little context for these examples - "Stonk" here doesn't really mean "stock" but kinda, the way I'm using "stonk" it means "a thing that has some numbers associated to it for a time period" But that's not important, it's just an example of table with this Id / Name schema. It looks like this:

CREATE TABLE StonkTypes (
StonkTypeId TINYINT IDENTITY(1,1) PRIMARY KEY NOT NULL,
StonkTypeName VARCHAR(200) NOT NULL CONSTRAINT UQ_StonkTypes_StonkTypeName UNIQUE (StonkTypeName)
)

After I create the proc, this statement:

EXEC CreateCSharpEnum 'StonkTypes'

Selects this string:

public enum StonkTypes { Stonk = 1, Bond = 2, Index = 3, Fund = 4, Commodity = 5,
PutCallRatio = 6, }

Which I can copy and paste into a C# file.

I have a Stonks table and it has StonkId and StonkName columns so this exec:

EXEC CreateCSharpEnum 'Stonks'

Spits out:

public enum Stonks { SP500 = 1, DowJonesIndustrialAverage = 2, ..... }

But for that enum I want to use the "Symbol" column for the enum name values so this:

EXEC CreateCSharpEnum 'Stonks', 'Symbol'

Does the trick and renders:

public enum Stonks { SPY = 1, DIA = 2, ..... }

Without further ado, here is this dirty piece of craziness. Yeah, very dirty, but I'm kind of pleased with myself - it's SQL code that constructs SQL code that constructs C# code. Couple layers involved.


CREATE OR ALTER PROCEDURE CreateCSharpEnum
@TableName VARCHAR(MAX),
@EnumNameColumnName VARCHAR(MAX) = NULL,
@EnumValueColumnName VARCHAR(MAX) = NULL
AS


DECLARE @LastCharOfTableName VARCHAR(1)
SELECT @LastCharOfTableName = RIGHT(@TableName, 1)


PRINT 'Last char = [' + @LastCharOfTableName + ']'


DECLARE @TableNameWithoutS VARCHAR(MAX)
IF UPPER(@LastCharOfTableName) = 'S'
SET @TableNameWithoutS = LEFT(@TableName, LEN(@TableName) - 1)
ELSE
SET @TableNameWithoutS = @TableName


PRINT 'Table name without trailing s = [' + @TableNameWithoutS + ']'


IF @EnumNameColumnName IS NULL
BEGIN
SET @EnumNameColumnName = @TableNameWithoutS + 'Name'
END


PRINT 'name col name = [' + @EnumNameColumnName + ']'


IF @EnumValueColumnName IS NULL
SET @EnumValueColumnName = @TableNameWithoutS + 'Id'


PRINT 'value col name = [' + @EnumValueColumnName + ']'


-- replace spaces and punctuation
SET @EnumNameColumnName  = 'REPLACE(' + @EnumNameColumnName + ', '' '', '''')'
SET @EnumNameColumnName  = 'REPLACE(' + @EnumNameColumnName + ', ''&'', '''')'
SET @EnumNameColumnName  = 'REPLACE(' + @EnumNameColumnName + ', ''.'', '''')'
SET @EnumNameColumnName  = 'REPLACE(' + @EnumNameColumnName + ', ''('', '''')'
SET @EnumNameColumnName  = 'REPLACE(' + @EnumNameColumnName + ', '')'', '''')'


PRINT 'name col name with replace sql = [' + @EnumNameColumnName + ']'


DECLARE @SqlStr VARCHAR(MAX) = 'SELECT ' + @EnumNameColumnName
+ ' + '' = '''
+ ' + LTRIM(RTRIM(STR(' + @EnumValueColumnName + '))) + '','' FROM ' + @TableName + ' ORDER BY ' + @EnumValueColumnName


PRINT 'sql that gets rows for enum body = [' + @SqlStr + ']'


CREATE TABLE #EnumRowsTemp (s VARCHAR(MAX))


INSERT
INTO #EnumRowsTemp
EXEC(@SqlStr)


--SELECT * FROM #EnumRowsTemp


DECLARE @csharpenumbody VARCHAR(MAX)
SELECT @csharpenumbody = COALESCE(@csharpenumbody + ' ', '') + s FROM #EnumRowsTemp


--PRINT @csharpenumbody


DECLARE @csharpenum VARCHAR(MAX) = 'public enum ' + @TableName + ' { ' + @csharpenumbody + ' }'


PRINT @csharpenum


SELECT @csharpenum


DROP TABLE #EnumRowsTemp

Please, be critical. One funky thing I didn't understand, how come I have to create and drop this #EnumRowsTemp table and not just "SELECT INTO #EnumRowsTemp" to create the temp table on the fly? I don't know the answer, I tried that and it didn't work. That's probably the least of the problems of this code...

As dirty as it may be... I hope this saves some of you fellow dorks a little bit of time.