Unable to cast object of type 'System.DBNull' to type 'System.String`

With a simple generic function you can make this very easy. Just do this:

return ConvertFromDBVal<string>(accountNumber);

using the function:

public static T ConvertFromDBVal<T>(object obj)
{
    if (obj == null || obj == DBNull.Value)
    {
        return default(T); // returns the default value for the type
    }
    else
    {
        return (T)obj;
    }
}

A shorter form can be used:

return (accountNumber == DBNull.Value) ? string.Empty : accountNumber.ToString()

EDIT: Haven't paid attention to ExecuteScalar. It does really return null if the field is absent in the return result. So use instead:

return (accountNumber == null) ? string.Empty : accountNumber.ToString() 

ExecuteScalar will return

• null if there is no result set
• otherwise the first column of the first row of the resultset, which may be DBNull.

If you know that the first column of the resultset is a string, then to cover all bases you need to check for both null and DBNull. Something like:

object accountNumber = ...ExecuteScalar(...);
return (accountNumber == null) ? String.Empty : accountNumber.ToString();

The above code relies on the fact that DBNull.ToString returns an empty string.

If accountNumber was another type (say integer), then you'd need to be more explicit:

object accountNumber = ...ExecuteScalar(...);
return (accountNumber == null || Convert.IsDBNull(accountNumber) ?     
         (int) accountNumber : 0;

If you know for sure that your resultset will always have at least one row (e.g. SELECT COUNT(*)...), then you can skip the check for null.

In your case the error message "Unable to cast object of type ‘System.DBNull’ to type ‘System.String`" indicates that the first column of your result set is a DBNUll value. This is from the cast to string on the first line:

string accountNumber = (string) ... ExecuteScalar(...);

Marc_s's comment that you don't need to check for DBNull.Value is wrong.

You can use C#'s null coalescing operator

return accountNumber ?? string.Empty;

I suppose you can do it like this:

string accountNumber = DBSqlHelperFactory.ExecuteScalar(...) as string;

If accountNumber is null it means it was DBNull not string :)

Since I got an instance which isn't null and if I compared to DBNULL I got Operator '==' cannot be applied to operands of type 'string' and 'system.dbnull' exeption, and if I tried to change to compare to NULL, it simply didn't work ( since DBNull is an object) even that's the accepted answer.

I decided to simply use the 'is' keyword. So the result is very readable:

data = (item is DBNull) ? String.Empty : item

based on answer from @rein

public static class DbDataReaderExtensions
{
    public static TObjProp Get<TObj, TObjProp>(
        this DbDataReader reader,
        Expression<Func<TObj, TObjProp>> expression)
    {
        MemberExpression member = expression.Body as MemberExpression;
        string propertyName = member.Member.Name;

        //PropertyInfo propInfo = member.Member as PropertyInfo;

        var recordOrdinal = reader.GetOrdinal(propertyName);
        var obj = reader.GetValue(recordOrdinal);

        if (obj == null || obj == DBNull.Value)
        {
            return default(TObjProp);
        }
        else
        {
            return (TObjProp)obj;
        }
    }
}

Given:

public class MyClass
{
    public bool? IsCheckPassed { get; set; }
}

Use as:

var test = reader.Get<MyClass, bool?>(o => o.IsCheckPassed);

or, if you hardcode class type in exception method:

var test = reader.Get(o => o.IsCheckPassed);

p.s. I haven't figured yet how to make generics implicit without sacrificing code length.. fee free to comment and suggest improvements

Full example:

public async Task<MyClass> Test(string connectionString) {
    var result = new MyClass();
    
    await using var con = new SQLiteConnection(connectionString);
    con.Open();

    await using var cmd = con.CreateCommand();
    cmd.CommandText = @$"SELECT Id, IsCheckPassed FROM mytable";
    
    var reader = await cmd.ExecuteReaderAsync();
    while (reader.Read()) {
        // old, not working! Throws exception!
        //bool? isCheckPassed1 = reader.GetBoolean(reader.GetOrdinal("IsCheckPassed"));
        
        // old, working, but too long (also if you have like 20 properties then all the more reasons to refactor..)
        bool? isCheckPassed2 = null;
        bool? isCheckPassed2Temp = reader.GetValue(reader.GetOrdinal("IsCheckPassed"));
        if (isCheckPassed2Temp != null && isCheckPassed2Temp != DBNull.Value)
            isCheckPassed2 = (bool?)isCheckPassed2Temp;
        
        // new
        var isCheckPassed3 = reader.Get<MyClass, bool?>(o => o.IsCheckPassed);
        // repeat for 20 more properties :)
        
        result.IsCheckPassed = isCheckPassed3;
    }
    
    return result;
}

Solution will work for as long as table column names match property names of the class. And might not be production-grade performance wise, so use or modify at your own risk :)