Question

I need to compare 2 consecutive rows in the same table. If the data is missing in the second row, I should update it with the first row’s data.

For Eg:

Row   EmpID  DATE        PosID        EmpStatus  EmpDept  EmpVP
----------------------------------------------------------------------
1     21     2010-12-31  NULL         TC         NULL     40
2     21     2010-01-25  90156840101  NULL       407      NULL
3     21     2003-11-25  NULL         AC         NULL     NULL

First Iteration: Since Row1 EmpStatus = TC, I want to Update the EmpStatus on Row2 to TC (Since its NULL) and EmpVP to 40 on Row2, as shown below:

Row   EmpID  DATE        PosID        EmpStatus  EmpDept  EmpVP
----------------------------------------------------------------------
1     21     2010-12-31  NULL         TC          NULL    40
2     21     2010-01-25  90156840101  TC          407     40 
3     21     2003-11-25  NULL         AC          NULL    NULL

Second Iteration: Since PositionID is NULL on Row3, I want to update Row3 with the PositionID of Row2. Since Row2 now has EmpStatus = TC, I want to compare the row2 and row3 data. Since Row3 has a new value, I want to retain the new value “AC”. But at the same time I want to update the value of EmpDept of Row3 = 40 since it’s NULL. The desired results are shown below:

Row   EmpID  DATE        PosID        EmpStatus  EmpDept  EmpVP
----------------------------------------------------------------------
1     21     2010-12-31  NULL         TC          NULL    40
2     21     2010-01-25  90156840101  TC          407     40 
3     21     2003-11-25  90156840101  AC          407     40

I am working on historical data load and I have to build records going backwards in terms of Dates.

Can anyone please tell me how to code this?
I want to know if we can do these updates, preferably without using cursors, as I have a lot of employees in this table.

Thanks a lot!

Answer 1

This is an iterative process which depends not only on each row visited, but also on data changes made during the processing. I can’t even begin to imagine how you’d do it without cursors.

I think once you resolve yourself to using cursors, then it’s just a matter of determining the rules (in a more general fashion than you’ve presented them) and then implementing those rules using cursors.

This appears to be a one-time process, so I’m not sure I understand why you’d be concerned about writing a routine that uses cursors to process the data, and running it during a maintenance period or just during a slow-usage period.

Here’s some pseudocode for how you might approach this problem (you’ll need to get the actual syntax details from a manual and you’ll need to make sure your business rules are being applied in the right order and the right manner):

BEGIN

DECLARE @row NUMBER, @posID VARCHAR(20), @empStatus VARCHAR(2), @empDept NUMBER, @empVP NUMBER;
DECLARE @prevPosID VARCHAR(20), @prevEmpStatus VARCHAR(2), @prevEmpDept NUMBER, @prevEmpVP NUMBER;
DECLARE @isFirst NUMBER;
SET @isFirst = 1;

DECLARE  mycurs CURSOR FOR
SELECT   Row, PosID, EmpStatus, EmpDept
FROM     Empl
ORDER BY Date DESC
FOR UPDATE OF Empl;

OPEN mycurs;

FETCH NEXT mycurs INTO @row, @posID, @empStatus, @empDept, @empVP;

WHILE (@@FETCH_STATUS != NOTFOUND)
BEGIN
    IF (@isFirst == 1) THEN
    BEGIN
        @isFirst = 0;
    END
    ELSE
    BEGIN
        IF (@empStatus IS NULL AND @prevEmpStatus IS NOT NULL) THEN
        BEGIN
            SET @empStatus = @prevEmpStatus;
            UPDATE Empl SET EmpStatus = @empStatus WHERE Row = @row;
        END
        IF (@posID IS NULL AND @prevPosID IS NOT NULL) THEN
        BEGIN
            SET @posID = @prevPosID ;
            UPDATE Empl SET PosID = @posID WHERE Row = @row;
        END
        IF (@empDept IS NULL AND @prevEmpDept  IS NOT NULL) THEN
        BEGIN
            SET @empDept = @prevEmpDept;
            UPDATE Empl SET EmpDept= @empDept WHERE Row = @row;
        END
    END

    SET @prevPosID = @posID;
    SET @prevEmpStatus = @empStatus;
    SET @prevEmpDept = @empDept;
    SET @prevEmpVP = @empVP;

    FETCH NEXT mycurs INTO @row, @posID, @empStatus, @empDept, @empVP;
END

DEALLOCATE mycurs;
CLOSE mycurs;

END