Re: Update Duplicate Rows

Question

Oracle 11.2.0.2
Windows 2008
I have this table (case_file) with duplicate rows (case_number).
I will like to make the case_number unique through out the table.
For example, if I have two rows with the same case_number(225), I will like to change one of the numbers and record that case_number to the privious_case_number column.
I am looking for a way to write up a script to automate the process.
Any help will be appreciated.
CREATE sequence case_id_seq
start with 25;
create table case_file
( case_id number not null, case_number varchar2(25), privious_case_number varchar(25), modified_by varchar2(50));
INSERT into case_file (case_id, case_number, privious_case_number, modified_by)
VALUES (case_id_seq.nextval, '220', '2002', 'DBA');
INSERT into case_file (case_id, case_number, privious_case_number, modified_by)
VALUES (case_id_seq.nextval, '225', null, 'DBA');
INSERT into case_file (case_id, case_number, privious_case_number, modified_by)
VALUES (case_id_seq.nextval, '224', null, 'PM');
INSERT into case_file (case_id, case_number, privious_case_number, modified_by)
VALUES (case_id_seq.nextval, '205', null, 'Supervisor');
INSERT into case_file (case_id, case_number, privious_case_number, modified_by)
VALUES (case_id_seq.nextval, '225', null, 'DBA');When I run the below query, I get this result, with 225 showing up twice:
select * from case_file;
26     225     2002     DBA
27     220     2002     DBA
28     225          DBA
29     224          PM
30     205          Supervisor
31     225          DBAWhen I run the below script:
SELECT case_number, COUNT(case_number) as Number_Of_Case_Number
FROM case_file
GROUP BY case_number
HAVING COUNT(case_number) > 1I get this result
CASE_NUMBER NUMBER_OF_CASE_NUMBER
225 3
SQL>Edited by: 868332 on Feb 6, 2012 1:23 PM

Answer 1

Hi,
Do you really need two unique numbers for each row?
To make case_numebr unique, first, see what the highest existing case_number is, and create a new sequence that starts after that point:
CREATE SEQUENCE     case_number_seq
START WITH     250;Always use that sequence when assigning case_numbers in the future.
To re-number the existing duplicates :
UPDATE     case_file
SET     case_number          = case_number_seq.NEXTVAL
,     previous_case_number     = case_number
WHERE     case_id         IN (
                     WITH  got_r_num     AS
                      SELECT  case_id
                      ,       ROW_NUMBER () OVER ( PARTITION BY  case_number
                                                          ORDER BY         case_id
                                     )  AS r_num
                      FROM    case_file
                  SELECT  case_id
                  FROM        got_r_num
                  WHERE   r_num     > 1
;The sub-query got_r_num assigns numbers 1, 2, 3, ... to all rows, in order by case_id (lowest case_id gets r_num=1) with a separate set of numebrs for each case_number. When there are duplicates, one of the rows with that case_number will keep that case_number, and all the others will be new case_numebrs from case_numebr_seq.
Edited by: Frank Kulash on Feb 6, 2012 4:37 PM
Originally missed the part about previous_case_number. (Be careful how to spell "previous".)