Table of Contents
In a usual MFC database project, we have a lot of CRecordset
derived classes generated by ClassWizard
. The class presented here, CODBCRecordset
is a very easy to use replacement of all these CRecordset
classes. CODBCRecordset
is derived from CRecordset
but it does not have hardcoded inside the number and the type of the database fields.
I have seen some implementations trying to solve this problem. But CODBCRecordset
has at least two advantages:
CODBCRecordset
can be used not only to get values from the database, but to write values into the database using the MFC way. - The other implementations cannot simultaneously open more than one recordset through one database connection when the database is MS SQL Server. In some cases, this could be a very big problem because opening a new connection is expensive in terms of time and resources.
Because CODBCRecordset
is derived from CRecordset
and uses its data exchange mechanism, it is fully compatible with MFC ODBC class CDatabase
.
Each field value is stored in a CDBField
object. The CDBField
class is inherited from CDBVariant
and has some added functionality so it's simple to use.
CODBCRecordset
and CDBField
classes support all database data types. CDBField
makes implicit type conversion where it is appropriate to supply the data in the requested format.
Here is a list of CODBCRecordset
and CDBField
methods:
CODBCRecordset class |
Constructor | The same as for CRecordset accepting CDatabase * |
BOOL Open( LPCTSTR lpszSQL, UINT nOpenType, DWORD dwOptions ); | Open the recordset
lpszSQL is a SQL statement that returns recordset
e.g. SELECT * FROM tablename
nOpenType is CRecordset open type, see CRecordset::Open() dwOptions is CRecordset options, see CRecordset::Open()
Note that lpszSQL and nOpenType have exchanged their positions compared to CRecordset::Open()
|
short GetODBCFieldCount() | Returns number of the fields (columns) in the recordset . This method is defined in CRecordset . |
int GetFieldID( LPCTSTR ) | Returns a field index by given field name. It is case insensitive. CRecordset::GetFieldIndexByName() works, but is case sensitive. |
CString GetFieldName( int ) | Returns a field name by given field index |
CDBField& Field( LPCTSTR )<br />
CDBField& Field( int ) | Through this method, you can get a reference to the internal CDBField object corresponding to the specified column (See the CDBField class table for details about CDBField ).
There are two forms of the method - with argument of type LPCTSTR szName - specifying the name of the column, and int nID - specifying the index of the coulmn in the recordset.
These methods can be used as lvalue in expressions. Thus, you can write values to the database.
|
CDBField& operator( LPCTSTR )<br />
CDBField& operator( int ) | The function operator is defined for ease of use of the class and just calls the corresponding Field() method
There are two forms of the function operator - with argument of type LPCTSTR szName - specifying the name of the column, and int nID - specifying the index of the column in the recordset.
These can be used as lvalue in expressions. Thus you can write values to the database.
|
bool | GetBool() | unsigned char | GetChar() | short | GetShort() | int | GetInt() | long | GetLong() | float | Getfloat() | double | GetDouble() | COleDateTime | GetDate() | CString | GetString() | CLongBinary* | GetBinary() |
| All of these methods do the appropriate type conversions depending on their return value and the type of the underlying data in the database.
These methods just call Field().AsXXX()
(See the CDBField class table for details about CDBField ).
There are two forms of these methods - with argument of type LPCTSTR szName - specifying the name of the column, and int nID - specifying the index of the coulmn in the recordset .
These cannot be used as lvalue in expressions.
|
CDBField class |
Constructors | No public constructors. CDBField cannot be instantiated except by CODBCRecordset to be used in internal structures to support data exchange. These objects are accessible through CODBCRecordset methods. |
bool | AsBool() | unsigned char | AsChar() | short | AsShort() | int | AsInt() | long | AsLong() | float | AsFloat() | double | AsDoble() | COleDateTime | AsDate() | CString | AsString() | CLongBinary* | AsBinary() |
| All of these methods do the appropriate type conversions depending on their return value and the type of the underlying data in the database (See the AsXXX methods table for data conversion rules).
There is no data type Int but AsInt() is equal to AsLong()
|
assignment operators | There are defined assignment operators accepting boo l, unsigned char , short , int , long , COleDateTime and CString . So CDBField objects can be lvalues . There is no assignment operator accepting CLongBinary because MFC does not support writing CLongBinary values into database. These assignment operators do appropriate conversions to the underlying database column data type except CLongBinary (See the assignment operators table for data conversion rules). |
const CString& GetName() | Returns the field name this object corresponds to. |
bool | IsNull() | bool | IsBool() | bool | IsChar() | bool | IsShort() | bool | IsInt() | bool | IsLong() | bool | IsFloat() | bool | IsDouble() | bool | IsNumber() | bool | IsDate() | bool | IsString() | bool | IsBinary() |
| Each of these return true if the field contains a value of the corresponding data type.
There is no data type Number but IsNumber() returns true if IsShort() || IsLong() || IsFloat() || IsDouble() .
There is no data type Int but IsInt() returns true if IsLong() returns true .
|
Conversions made by AsXXX methods
|
| Values in the database
|
| NULL | BOOL | UCHAR | SHORT | LONG | SINGLE | DOUBLE | DATE | STRING | BINARY |
AsBool | false | * | * | * | * | * | * | | * | |
AsChar | 0x32 | * | * | * | * | * | * | | * | |
AsShort | 0 | * | * | * | * | * | * | | * | |
AsInt | 0 | * | * | * | * | * | * | | * | |
AsLong | 0 | * | * | * | * | * | * | | * | |
AsFloat | 0.0 | * | * | * | * | * | * | | * | |
AsDouble | 0.0 | * | * | * | * | * | * | | * | |
AsDate | null | invalid | invalid | * | * | * | * | * | * | |
AsString | empty | * | * | * | * | * | * | * | * | * |
AsLongBinary | NULL | | | | | | | | | * |
Empty cells indicate the conversion is not available, thus code asserts.
Cells marked with * indicate conversion is available (See the Conversion algorithms table for data conversion rules). |
Conversions made by assignment operators
|
Database Field Type | Argument of the assignment operator
|
| bool | unsigned char | short | int | long | float | double | COleDateTime | String |
NULL | | | | | | | | | |
BOOL | * | * | * | * | * | * | * | | * |
UCHAR | * | * | * | * | * | * | * | | * |
SHORT | * | * | * | * | * | * | * | | * |
LONG | * | * | * | * | * | * | * | | * |
SINGLE | * | * | * | * | * | * | * | | * |
DOUBLE | * | * | * | * | * | * | * | | * |
DATE | | | | | | | | * | * |
STRING | * | * | * | * | * | * | * | * | * |
BINARY | | | | | | | | | |
Empty cells indicate the conversion is not available, thus code asserts.
Cells marked with * indicate conversion is available (See the Conversion algorithms table for data conversion rules). |
Conversion algorithms |
String to Bool | comparing the first character of the string with 'T ' |
Char to Bool | comparing the character with 'T ' |
Bool to String | String = (bVal) ? 'T' : 'F' |
Bool to Char | Char = (bVal) ? 'T' : 'F' |
String to Number | appropriate atoX() function |
Number to String | CString::Format() method using the appropriate format specifier string |
String to Date | COleDateTime::ParseDateTime() method |
Date to String | COleDateTime::Format() method |
You should include the files ODBCRecordset.h and ODBCRecordset.cpp in your project.
I usually include this line in my StdAfx.h file.
#include "ODBCRecordset.h"
Here is a simple code showing how CODBCRecordset
can be used.
/////////////////////////////////////////////////////////////////////////////
CDatabase db;
// This connect string will pop up the ODBC connect dialog
CString cConnect = "ODBC;";
db.Open( NULL, // DSN
FALSE, // Exclusive
FALSE, // ReadOnly
cConnect, // ODBC Connect string
TRUE // Use cursor lib
);
COleDateTime dOrderDate;
CODBCRecordset rs( &db );
rs.Open( "SELECT * FROM Orders \
WHERE ORDER_DATE > 'jan 1 2000' \
ORDER BY ORDER_DATE" );
for( ; ! rs.IsEOF(); rs.MoveNext() )
{
// The choice is yours. You may choose whatever way
// you want to get the values
//
// These return COleDateTime value
dOrderDate = rs.GetDate( "ORDER_DATE" );
dOrderDate = rs.Field("ORDER_DATE").AsDate();
// These make implicit call to AsDate()
dOrderDate = rs("ORDER_DATE");
dOrderDate = rs.Field("ORDER_DATE");
// Now edit the fields in the recordset
rs.Edit();
rs("ORDER_DATE") = "jan 1 1999"; // Implicit conversion
rs.Field("ORDER_DATE") = "jan 1 1999"; // Implicit conversion
rs.Update();
} // for(....
//////////////////////////////////////////////////////////////////////////////
If ORDER_DATE
is stored in the database as datetime
or compatible data type, the value will be got directly.
If ORDER_DATE
is stored in the database as string
or compatible data type (char
, varchar
) the value will be converted via COleDateTime::ParseDateTime()
method. If conversion fails, dOrderDate
will be set to COleDatetime::invalid
.
When opening a resultset
generated by join
statements, it is possible to get 2 or more columns that have the same name. CODBCRecordset
leaves the name of the first column intact but other repeated columns are renamed with adding the number this columns repeats the name. Not repeated column names are left intact. E.g.
SELECT * FROM Orders, Customers WHERE Orders.CUST_ID = Customers.ID
If the table Orders
have a column with name ID
and Customers
have a column with name ID
, CODBCRecordset
will rename ID
from Customers
to ID2
and all other not repeating column names will be intact.
Well, here is a tip: Rename columns manually to be sure what name they have, e.g.:
SELECT Orders.*, Customers.ID as CUSTOMERS_ID
FROM Orders, Customers
WHERE Orders.CUST_ID = Customers.ID
CODBCRecordset
allocates storage for all fields in the resultset
and uses MFC Record Field eXchange mechanism like it has been inherited from CRecordset
using ClassWizard
.