Find duplication values from multiple columns in a big table

Question

Hi All,
I am working on a 11gR2 database in linux. i want to display record that have duplicate values in 2 columns.
1. Table Structure :-
CREATE TABLE A
ID NUMBER(10),
F_NAME VARCHAR2(100 BYTE),
L_NAME VARCHAR2(100 BYTE)
2. Sample Data:-
Insert into A
(ID, F_NAME, L_NAME) Values (1,'TONY' ,'SUMIT');
Insert into A
(ID, F_NAME, L_NAME) Values (2,'SUMIT' ,'KEITH');
Insert into A
(ID, F_NAME, L_NAME) Values (3,'NORA','SMITH');
Insert into A
(ID, F_NAME, L_NAME) Values (4,'APRIL','TONY');
Insert into A
(ID, F_NAME, L_NAME) Values (5,'ROSS','TAM');
ID F_NAME L_NAME
1 TONY SUMIT
2 SUMIT KEITH
3 NORA SMITH
4 APRIL TONY
5 ROSS TAM
4. My requirement is i need display IDs that it's F_NAME or L_NAME has duplication in F_NAME or L_NAME columns.
The result should be
ID
1 reason: F_NAME (TONY) equals to L_NAME of record 4, L_NAME (SUMIT) equals to F_NAME of record 2
2 reason: F_NAME (SUMIT) equals to L_NAME of record 1
4 reason: L_NAME (TONY) equals to F_NAME of record 1
record 3, 5 aren't in the result because there is no duplication in F_NAME or L_NAME columns for NORA,SMITH, ROSS, TAM
The table contains 10 million records, i really need to consider the performance.
kindly suggest me the solution

Answer 1

Note: Forum members please suggest better approach to this -- below.. convert into SQL :)
I know I will be opposed by many people in this forum for posting such in-efficient solution.
But trying to learn along with you.. its an interesting problem which must deal with all rows vs all rows to get all combinations.
But I am still thinking how to write it in SQL, probably will learn from this post after we receive some good SQL solution for the code what I am currently doing now.
step 1: created a table B similar to table A and added a column reason
CREATE TABLE B
  ID      NUMBER(10),
  F_NAME  VARCHAR2(100 BYTE),
  L_NAME  VARCHAR2(100 BYTE),
  REASON  VARCHAR2(1000 BYTE)  --- ADDED THIS
)Definetely inefficient :(
BEGIN
   FOR rec_outer IN (SELECT * FROM A) LOOP
      FOR rec_inner IN (SELECT * FROM A) LOOP
         IF (rec_outer.f_name = rec_inner.l_name) THEN
            UPDATE B
               SET reason =
                         rec_outer.id
                      || ' reason: F_NAME ('
                      || rec_outer.f_name
                      || ') equals to L_NAME of record '
                      || rec_inner.id
             WHERE b.id = rec_outer.id;
         END IF;
      END LOOP;
      FOR rec_inner IN (SELECT * FROM A) LOOP
         IF (rec_outer.l_name = rec_inner.f_name) THEN
            UPDATE B
               SET reason =
                      reason
                      || CASE
                            WHEN reason IS NULL THEN
                               rec_outer.id || ' reason: '
                            ELSE
                         END
                      || 'L_NAME ('
                      || rec_inner.f_name
                      || ') equals to F_NAME of record '
                      || rec_inner.id
             WHERE b.id = rec_outer.id;
         END IF;
      END LOOP;
   END LOOP;
   COMMIT;
EXCEPTION
   WHEN OTHERS THEN
      rollback;
      RAISE;
END;OUTPUT:
ID     F_NAME     L_NAME     REASON
1     TONY     SUMIT     1 reason: F_NAME (TONY) equals to L_NAME of record 4,L_NAME (SUMIT) equals to F_NAME of record 2
2     SUMIT     KEITH     2 reason: F_NAME (SUMIT) equals to L_NAME of record 1
3     NORA     SMITH     
4     APRIL     TONY     4 reason: L_NAME (TONY) equals to F_NAME of record 1
5     ROSS     TAM     Cheers,
Manik.
Edited : Added rollback