Update row.column1 with value from other_row.column1

Question

I have a problem with a bug recently discovered in one of our systems.
It generates two rows with the same public_order_id (and also same customer_id, order_state, sub_customer_id, order_date and amount_paid).
The "error row" which should be removed is one with:
amount_paid <> 0 AND (amount_paid_main = 0 AND amount_paid_bonus = 0)
The correct row is:
amount_paid <> 0 AND NOT (amount_paid_main = 0 AND amount_paid_bonus = 0)
However it is not as easy to just delete the error once. The customer_class is wrong in the correct row, it is stored in the error row. So I need to update the correct row with the customer_class from the error row. And I have no idea how to do this. I guess I need a PL/SQL to solve this. This error has occurred on 20 000 transactions so I cannot update it manually.
How do I correct this with PL/SQL, or just with SQL? the internal_transaction_id is unique, but it is not certain that the error row is has a higher number than the correct row.
I hope you understand what I want to do.
This is example of the data(seperated by ;):
CUSTOMER_ID;PUBLIC_ORDER_ID;ORDER_STATE;SUB_CUSTOMER_ID;ORDER_DATE;AMOUNT_PAID;CUSOMER_CLASS;INTERNAL_ORDER_ID;AMOUNT_PAID_MAIN;AMOUNT_PAID_BONUS;type of row
SONY_ERICSON;1;FINISHED;45;2009-04-27;200;28;10;200;0;*<- correct row*
SONY_ERICSON;1;FINISHED;45;2009-04-27;200;16;11;0;0;<- error row
NOKIA;2;FINISHED;78;2009-04-16;200;32;12;200;0;*<- correct row*
NOKIA;2;FINISHED;78;2009-04-16;200;5;13;0;0;<- error row
MOTOROLA;3;FINISHED;65;2009-04-20;200;4;14;0;0;<- error row
MOTOROLA;3;FINISHED;65;2009-04-20;200;105;15;200;50;*<- correct row*

Answer 1

Hello, an update should suffice here:
WITH test_table AS (
SELECT 'SONY_ERICSON' CUSTOMER_ID,1 PUBLIC_ORDER_ID,'FINISHED' ORDER_STATE, 45 SUB_CUSTOMER_ID, TO_DATE('2009-04-27','YYYY-MM-DD') ORDER_DATE, 200 AMOUNT_PAID, 28 CUSTOMER_CLASS,10 INTERNAL_ORDER_ID,200 AMOUNT_PAID_MAIN, 0 AMOUNT_PAID_BONUS FROM DUAL UNION ALL
SELECT 'SONY_ERICSON',1,'FINISHED',45, TO_DATE('2009-04-27','YYYY-MM-DD'), 200,16,11,0,0  FROM DUAL UNION ALL
SELECT 'NOKIA',2,'FINISHED',78, TO_DATE('2009-04-16','YYYY-MM-DD'), 200,32,12,200,0 FROM DUAL UNION ALL
SELECT 'NOKIA',2,'FINISHED',78, TO_DATE('2009-04-16','YYYY-MM-DD'), 200,5,13,0,0 FROM DUAL UNION ALL
SELECT 'MOTOROLA',3,'FINISHED',65, TO_DATE('2009-04-20','YYYY-MM-DD'), 200,4,14,0,0 FROM DUAL UNION ALL
SELECT 'MOTOROLA',3,'FINISHED',65, TO_DATE('2009-04-20','YYYY-MM-DD'), 200,105,15,200,50 FROM DUAL)
-- end test data
SELECT *
  FROM test_table
WHERE NVL(amount_paid_main,0) + NVL(amount_paid_bonus,0) = 0;Confirming that we have the error records:
CUSTOMER_ID  PUBLIC_ORDER_ID ORDER_ST SUB_CUSTOMER_ID ORDER_DAT AMOUNT_PAID CUSTOMER_CLASS INTERNAL_ORDER_ID AMOUNT_PAID_MAIN AMOUNT_PAID_BONUS
SONY_ERICSON               1 FINISHED              45 27-APR-09         200             16                11                0                 0
NOKIA                      2 FINISHED              78 16-APR-09         200              5                13                0                 0
MOTOROLA                   3 FINISHED              65 20-APR-09         200              4                14                0                 0Then, something like:
UPDATE test_table tt1
    SET tt1.CUSTOMER_CLASS =
  (SELECT tt2.CUSTOMER_CLASS
      FROM test_table
     WHERE  NVL(amount_paid_main,0) + NVL(amount_paid_bonus,0) = 0
          AND tt1.CUSTOMER_ID = tt2.CUSTOMER_ID
          AND tt1.PUBLIC_ORDER_ID = tt2.PUBLIC_ORDER_ID)
WHERE NVL(tt1.amount_paid_main,0) + NVL(tt1.amount_paid_bonus,0) > 0;