[오라클] COLUMN->ROW, ROW->COLUMN. 데이터 행과 열을 변환

/*---------------------------------------------------
* ROW => COLUMN의 변환
* COLUMN => ROW의 변환
----------------------------------------------------*/
---------------
DEPTNO  EMPNO
---------------
10      7782
10      7839
10      7934
20      7369
20      7566
20      7788
30      7499
30      7521
30      7654

------------------------------
DEPTNO  EMP1  EMP2  EMP3
------------------------------
10      7782  7839  7934
20      7369  7566  7788
30      7499  7521  7654

 

/* COLUMN => ROW 시작 */
SELECT A.DEPTNO,
       DECODE(C.NO, 1, A.EMP1,
                    2, A.EMP2,
                    3, A.EMP3) EMPNO
FROM
     (
     /* ROW => COLUMN 시작 */
     SELECT DEPTNO,
            MAX(DECODE(RID, 1, EMPNO)) EMP1,
            MAX(DECODE(RID, 2, EMPNO)) EMP2,
            MAX(DECODE(RID, 3, EMPNO)) EMP3
     FROM (
          SELECT DEPTNO,
                 ROW_NUMBER() OVER (PARTITION BY DEPTNO ORDER BY EMPNO) RID,
                EMPNO
          FROM EMP
          )
     GROUP BY DEPTNO
     /* ROW => COLUMN 종료 */
     ) A, COPY_T C
WHERE C.NO <= 3
/* COLUMN => ROW 종료 */

 

/*---------------------------------------------------
* CROSSTAB에서 열을 행으로 행을 열로 변환
* ROW => COLUMN, COLUMN => ROW을 한꺼번에 구현
----------------------------------------------------*/
------------------------------
DEPTNO   EMP1  EMP2  EMP3
------------------------------
10       7782  7839  7934
20       7369  7566  7788
30       7499  7521  7654

------------------------------
EMP  DEPT_10 DEPT_20 DEPT_30
------------------------------
EMP1 7782    7369    7499
EMP2 7839    7566    7521
EMP3 7934    7788    7654

SELECT DECODE(C.NO, 1, 'EMP1',
                    2, 'EMP2',
                    3, 'EMP3') EMP,
       MAX(DECODE(A.DEPTNO||C.NO2, '1001', A.EMP1,
                                   '1002', A.EMP2,
                                   '1003', A.EMP3)) DEPT_10,
       MAX(DECODE(A.DEPTNO||C.NO2, '2001', A.EMP1,
                                   '2002', A.EMP2,
                                   '2003', A.EMP3)) DEPT_20,
       MAX(DECODE(A.DEPTNO||C.NO2, '3001', A.EMP1,
                                   '3002', A.EMP2,
                                   '3003', A.EMP3)) DEPT_30
FROM
     (
     /* 원래의 ROW, COLUMN구조 시작 */
     SELECT DEPTNO,
            MAX(DECODE(RID, 1, EMPNO)) EMP1,
            MAX(DECODE(RID, 2, EMPNO)) EMP2,
            MAX(DECODE(RID, 3, EMPNO)) EMP3
     FROM (
          SELECT DEPTNO,
                 ROW_NUMBER() OVER (PARTITION BY DEPTNO ORDER BY EMPNO) RID,
                EMPNO
          FROM EMP
          )
     GROUP BY DEPTNO
     /* 원래의 ROW, COLUMN구조 종료 */
     ) A, COPY_T C
WHERE C.NO <= 3    /* 3은 보여질 COLUMN의 수 */
GROUP BY DECODE(C.NO, 1, 'EMP1',
                      2, 'EMP2',
                      3, 'EMP3')
      
/*---------------------------------------------------
* 참고) COPY_T 의 생성
----------------------------------------------------*/
CREATE TABLE COPY_T
AS
SELECT ROWNUM                  NO
      ,TO_CHAR(ROWNUM, 'FM00') NO2
FROM ALL_OBJECTS
WHERE ROWNUM <= 31

CREATE UNIQUE INDEX COPY_T_IDX1 ON COPY_T(NO)
CREATE UNIQUE INDEX COPY_T_IDX2 ON COPY_T(NO2)

/*---------------------------------------------------
* ROW_NUMBER() 함수의 기능을 구현 => 테이블을 두번 읽기
* ROWNUM이 지원되지 않는 DBMS에서 ROWNUM 구현도 유사
----------------------------------------------------*/
SELECT DEPTNO,
       ROW_NUMBER() OVER (PARTITION BY DEPTNO ORDER BY EMPNO) RID,
       EMPNO
FROM EMP

SELECT A.DEPTNO,
       COUNT(*) RID,
       A.EMPNO
FROM EMP A, EMP B
WHERE A.DEPTNO = B.DEPTNO      /* PARTITION BY 기능 */   
AND A.EMPNO >= B.EMPNO         /* ORDER BY 기능 => 반드시 UNIQUE 해야함 */
GROUP BY A.DEPTNO, A.EMPNO     /* PARTITION BY,ORDER BY 기능 */
ORDER BY A.DEPTNO, A.EMPNO     /* PARTITION BY,ORDER BY 기능 */

/*---------------------------------------------------
* COPY_T 테이블이 없을때 COPY_T 기능 구현방법
----------------------------------------------------*/
SELECT NO, NO2
FROM COPY_T
WHERE NO <= 5

/*------------------------------------------
* 1.USER_OBJECTS 테이블의 이용
*   최대한 가벼운 테이블 이용
*   USER_OBJECTS가 가벼운지는 검증할 문제임
-------------------------------------------*/
SELECT ROWNUM NO,
       TO_CHAR(ROWNUM, 'FM00') NO2
FROM USER_OBJECTS
WHERE ROWNUM <= 5

/*------------------------------------------
* 2.DUAL 테이블의 이용
*   복사갯수가 적을때 이용(2~3개)
-------------------------------------------*/
SELECT NO,
       TO_CHAR(NO, 'FM00') NO2
FROM (
     SELECT 1 NO FROM DUAL
     UNION ALL
     SELECT 2 FROM DUAL
     UNION ALL
     SELECT 3 FROM DUAL
     UNION ALL
     SELECT 4 FROM DUAL
     UNION ALL
     SELECT 5 FROM DUAL
     )

/*---------------------------------------------------
* 참고) DUAL 테이블이 없는 경우의 구현(SQL SERVER)
----------------------------------------------------*/
CREATE VIEW DUAL
AS
SELECT 'X' DUMMY_COL

SELECT GETDATE() FROM DUAL

 

 

 

출처 : http://forkit.egloos.com/314137