BUG: GetFieldValue Returns Empty String for SQL_LONGVARCHAR

The information in this article applies to:

• The Microsoft Foundation Classes (MFC), included with: Microsoft Visual C++, 32-bit Edition, version 4.2

SYMPTOMS

CRecordset::GetFieldValue() incorrectly returns an empty string for SQL_LONGVARCHAR data that is 1 character long.

CAUSE

GetFieldValue() allocates a buffer and then calls CRecordset::GetData() to retrieve data from the field into this buffer. For long binary and character data, the length of the data is not known in advance, so GetFieldValue() does not know what size of buffer to allocate.

Therefore, for SQL_LONGVARCHAR and SQL_LONGVARBINARY data, GetFieldValue() calls GetData() with a buffer length of 1. GetData() tries to retrieve the data and will likely fail because the buffer is too small. Whether GetData() fails or not, it returns the actual length of the data in the field excluding the NULL termination byte. GetFieldValue() then calls the function CRecordset::GetLongCharDataAndCleanup() to reallocate the buffer and retrieve the rest of the data, if necessary.

GetLongCharDataAndCleanup() must determine whether GetData() was able to retrieve the data by checking whether the actual data length is larger than the buffer length:

   // If long data, may need to call SQLGetData again
   if (nLen < nActualSize &&
       (nSQLType == SQL_LONGVARCHAR || nSQLType == SQL_LONGVARBINARY))

GetLongCharDataAndCleanup() fails to consider the space required for the NULL termination byte. If the data length is 1, then both nLen and nActualSize are 1, which means that GetLongCharDataAndCleanup() will not attempt to get the additional data.

RESOLUTION

Override the GetLongCharDataAndCleanup() function and change the length comparison above from "<" to "<=". See the MORE INFORMATION section below about how to implement this workaround.

STATUS

Microsoft has confirmed this to be a bug in the Microsoft products listed at the beginning of this article. We are researching this bug and will post new information here in the Microsoft Knowledge Base as it becomes available.

MORE INFORMATION

The following steps detail one method of making the needed changes to GetFieldValue():

1. If you do not currently have a class derived from CRecordset, create one. If you have multiple classes derived from CRecordset, you may want to create a new base class derived from CRecordset and modify your existing recordset classes to be derived from this new base class so that you do not have to make the changes to every recordset class.

   For example, create GetSet.H:

   class CGetSet : public CRecordset { public:

          CGetSet(CDatabase* pDatabase = NULL);
   

   };

   and GetSet.CPP:

      #include "stdafx.h"
      #include "GetSet.h"
   
      CGetSet::CGetSet(CDatabase* pdb)
          : CRecordset(pdb)
      {
      }
   
   

2. Copy the prototypes of the CRecordset GetFieldValue() and GetLongCharDataAndCleanup() functions from MSDEV\MFC\SRC\AFXDB.H to your recordset .H file as public member functions. Note that all four of the GetFieldValue() prototypes must be copied.

3. Copy the implementations of the CRecordset::GetFieldValue() and CRecordset::GetLongCharDataAndCleanup() functions from MSDEV\MFC\SRC\DBCORE.CPP to your recordset .CPP file. Change the "CRecordset::" prefix to your new class name. Note that all four of the GetFieldValue() implementations must be copied.

4. In your implementation of GetLongCharDataAndCleanup(), replace "<" with "<=".

   Change:

      // If long data, may need to call SQLGetData again
      if (nLen < nActualSize &&
          (nSQLType == SQL_LONGVARCHAR || nSQLType == SQL_LONGVARBINARY))
   
      to:
   
      // If long data, may need to call SQLGetData again
      if (nLen <= nActualSize &&
          (nSQLType == SQL_LONGVARCHAR || nSQLType == SQL_LONGVARBINARY))
   
   

5. If you were using the CRecordset class directly, you will need to change your application to use the new class instead.