Experts-please review my function and see if it is "good" enough..!

hi guys,

I have code a function that reads lines from a file into a buffer based on this file format:

H,1,2,3,4,5,6,7,8,9,0
D,1,2,3,4,5
T,3

H,1,2,3,4,5,6,7,8,9,0
D,1,2,3,4,5
T,3

...more records...
where it can also be:
H,1,2,3,4,
5,6,7,8,9,0
D,1,2
,3,4,5
T,3

where the number of fields is FIXED..(ie. H has 10 data fields, D has 5, T has 1)

My function reads one "record" into a buffer:

iErr = DoGetBuffer(szFile,SZ_NOFIELD1, szBuffer1); /* for H */
iErr = DoGetBuffer(szFile,SZ_NOFIELD2, szBuffer2); /* for D */
iErr = DoGetBuffer(szFile,SZ_NOFIELD3, szBuffer3); /* for T */

Note that I use the no of commas as an indicator. This function is very important and there cannot be any bugs in it... can u guys please go through it and comment/improve upon it so that it is "perfect" for production use?

Thanzs!

--------------------------------------
int CMyClass::DoGetBuffer(CFile * myFile, int NoFields, char *buffer)
{
      int iComma;
      char szElement;
      DWORD dwRead;
      int iLine;
      // initialize
      iLine = 0;
      iComma = 0;
      szElement = NULL;

    do
    {
        dwRead =  myFile->Read(&szElement, 1);
            
            // check if end of record
            if ( \
                  ((szElement == '\r')&&(iComma == (NoFields - 1)))|| \
                  ((dwRead  <= 0)&&(iComma == (NoFields - 1))) \
               )      
            {      // should check for record_type in the next line
                        buffer[iLine] = '\0';
                        return 0;
            }

            // ignore newline and cr
            if ((szElement != '\n')&&(szElement != '\r'))
            {
            buffer[iLine] = szElement;
            if (buffer[iLine] == ',')  // one comma add
                  iComma++;
            iLine++;
            };
            

    }
    while (dwRead > 0);
      // dwRead returned <= 0
      return -1;

}
----------------------------------