处理DBNull.Value

Question

I frequently have to deal with DataTables connected to grid controls, custom updating always seems to produce a lot of code related to DBNull.Value. I saw a similar question here but think there must be a better answer:

What is the best way to deal with DBNull's

The thing I find is I tend to encapsulate my database updates in methods so I end up with code like below where I move the DBNull.value to a nullable type and then back for the update:

private void UpdateRowEventHandler(object sender, EventArgs e)
{
    Boolean? requiresSupport = null;
    if (grdMainLevel1.GetFocusedRowCellValue(colASRequiresSupport) != DBNull.Value)
        requiresSupport = (bool)grdMainLevel1.GetFocusedRowCellValue(colASRequiresSupport);

    AdditionalSupport.UpdateASRecord(year, studentID, requiresSupport)
}

internal static void UpdateASRecord(
        string year,
        string studentID,            
        bool? requiresSupport)
    {
        List<SqlParameter> parameters = new List<SqlParameter>();

        parameters.Add(new SqlParameter("@year", SqlDbType.Char, 4) { Value = year });
        parameters.Add(new SqlParameter("@student_id", SqlDbType.Char, 11) { Value = studentID });

        if (requiresSupport == null)
            parameters.Add(new SqlParameter("@requires_support", SqlDbType.Bit) { Value = DBNull.Value });
        else
            parameters.Add(new SqlParameter("@requires_support", SqlDbType.Bit) { Value = requiresSupport });

        //execute sql query here to do update
    }

That was just an example of the flow and not working code. I realize I could do things like pass objects or swallow potential casting problems using "as type" to get DBUll straight to null but both of these to me appear to hide potential errors, I like the type safety of the method with nullable types.

Is there a cleaner method to do this while maintaining type safety?

Answer 1

A couple of (very) simple generic helper methods might at least concentrate the test into one piece of code:

static T FromDB<T>(object value)
{
    return value == DBNull.Value ? default(T) : (T)value;
}

static object ToDB<T>(T value)
{
    return value == null ? (object) DBNull.Value : value;
}

These methods can then be used where appropriate:

private void UpdateRowEventHandler(object sender, EventArgs e)
{
    AdditionalSupport.UpdateASRecord(year, studentID, 
        FromDB<Boolean?>(grdMainLevel1.GetFocusedRowCellValue(colASRequiresSupport)));
}

internal static void UpdateASRecord(
        string year,
        string studentID,
        bool? requiresSupport)
{
    List<SqlParameter> parameters = new List<SqlParameter>();

    parameters.Add(new SqlParameter("@year", SqlDbType.Char, 4) { Value = year });
    parameters.Add(new SqlParameter("@student_id", SqlDbType.Char, 11) { Value = studentID });
    parameters.Add(new SqlParameter("@requires_support", SqlDbType.Bit) { Value = ToDB(requiresSupport) });

    //execute sql query here to do update
}

Answer 2

I don't see what's wrong with as -casting and null coalescing.

as -casting is used for reading:

bool? requiresSupport =
  grdMainLevel1.GetFocusedRowCellValue(colASRequiresSupport) as bool?;
AdditionalSupport.UpdateASRecord(year, studentID, requiresSupport);

null coalescing is used for writing:

parameters.Add(new SqlParameter("@student_id", SqlDbType.Char, 11)
  { Value = studentID });
parameters.Add(new SqlParameter("@requires_support", SqlDbType.Bit)
  { Value = (object)requiresSupport ?? DBNull.Value });

Both of these are completely typesafe and do not "hide" errors.

If you really want, you can wrap these into static methods, so you end up with this for reading:

//bool? requiresSupport =
//  grdMainLevel1.GetFocusedRowCellValue(colASRequiresSupport) as bool?;
bool? requiresSupport = FromDBValue<bool?>(
  grdMainLevel1.GetFocusedRowCellValue(colASRequiresSupport));

and this for writing:

//parameters.Add(new SqlParameter("@requires_support", SqlDbType.Bit)
//  { Value = (object)requiresSupport ?? DBNull.Value });
parameters.Add(new SqlParameter("@requires_support", SqlDbType.Bit)
  { Value = ToDBValue(requiresSupport) });

The static method code is slightly cleaner in the writing case, but the intent is less clear (especially in the reading case).

Answer 3

public static object DbNullable<T>(T? value) where T : struct
{
    if (value.HasValue)
    {
        return value.Value;
    }
    return DBNull.Value;
}

public static object ToDbNullable<T>(this T? value) where T : struct
{
    return DbNullable(value);
}

This is my implementation of DBNULL helper. The usage is simple:

new SqlParameter("Option1", option1.ToDbNullable())

Answer 4

parameters.Add("@requires_support", SqlDbType.Bit).Value = (object)requiresSupport ?? DBNull.Value;

which means the same as

parameters.Add("@requires_support", SqlDbType.Bit).Value = (requiresSupport != null) ? (object)requiresSupport : DBNull.Value;

or

if (requiresSupport != null)
    parameters.Add("@requires_support", SqlDbType.Bit).Value = requiresSupport 
else
    parameters.Add("@requires_support", SqlDbType.Bit).Value = DBNull.Value;

(additional cast to object is required to remove the type ambiguity)