Question

So here is the code:

return new DistrictInfo { 
    rid = Convert.ToUInt32(((OracleNumber)o.GetOracleValue("rid")).Value), 
    doc = Convert.ToUInt16(((OracleNumber)o.GetOracleValue("doctor")).Value), 
    secdoc = Convert.ToUInt16(((OracleNumber)o.GetOracleValue("secdoctor")).Value), 
    num = Convert.ToUInt16(((OracleNumber)o.GetOracleValue("num")).Value), 
    docname = o.GetOracleValue("doctorname") as string, 
    slpuname = o.GetOracleValue("lpuname") as string, 
    reason = o.GetOracleValue("reason") as string, 
    secdocname = o.GetOracleValue("secdocname") as string 
};

Now I need to rewrite this code to do a checking whether object property is existing or not. This should be like this piece of code:

DistrictInfo di;
if (!(o["rid"].Equals(DBNull.Value)) && !(o.GetOracleValue("rid").Equals(DBNull.Value)) && (((OracleNumber)o.GetOracleValue("rid")).Value != null))
{
    di.rid = Convert.ToUInt32(((OracleNumber)o.GetOracleValue("rid")).Value);
}

But I found this code some kind of awkward and not elegant. I did a many of checking ’cause I want to escape exceptions.
So the question is how can we refactor this code? Tell me your thoughts. I think there’s no need in such many checkings. Another thing we need to specify a variety of object property names to do one block of code for all variety members. I think there’s ability to use LINQ for that. Also in first piece of code we see different casting so we need to mention it in our new code.
Thanks in advance guys!

P.S. Library for working with database is dotConnect for Oracle from devArt.

Answer 1

I would suggest you to use UInt32.TryParse() which provides a safe method of converting raw string value to UInt32, there are two cases – value can be converted to UInt32 or can not. So basically check return value of TryParse() method to see whether value was converted successfully.

So

string rawValue = o["rid"].ToString();
UInt32 parsedValue;
if (UInt32.TryParse(rawValue, out parsedValue))
{
    // was converted successfully
}
else
{
    // was not converted
}

Regarding automation of this process to handle multiple properties by their names you also need to consider property type, so I can not see a nice way to leverage LINQ for this purpose.

EDIT: Added proposal regarding automation of a field transformation

You can leverage Extension Methods feature of .NET to decorate OracleObject type by a set of helpful methods.

public static class OracleObjectExtensions
{
    public static UInt32 GetUInt32Value(this OracleObject oracleObject, string fieldName)
    {
        UInt32 returnValue = default(UInt32);

        if (oracleObject[fieldName] != null)
        {
            string rawValue = oracleObject[fieldName].ToString();
            UInt32.TryParse(rawValue, out returnValue);                
        }

        return returnValue;
    }

    public static UInt16 GetUInt16Value(this OracleObject oracleObject, string fieldName)
    {
        UInt16 returnValue = default(UInt16);

        if (oracleObject[fieldName] != null)
        {
            string rawValue = oracleObject[fieldName].ToString();
            UInt16.TryParse(rawValue, out returnValue);
        }

        return returnValue;
    }
}

EDIT2: Extension Methods description

Extension methods enable you to “add” methods to existing types
without creating a new derived type, recompiling, or otherwise
modifying the original type. Extension methods are a special kind of
static method, but they are called as if they were instance methods on
the extended type.