WM_CONCAT(colName)为Oracle 10g中行数据转列数据的内置函数,执行效果如下
表T
A B
------------
1 2
1 3
1 4
2 1
2 3
执行语句 :
SELECT A, WM_CONCAT(B) B FROM T GROUP BY A
执行结果:
A B
---------------
1 2,3,4
2 1,3
注意:
当在包体内使用行转列时,WM_CONCAT(colName)可以使用,但WM_CONCAT(DISTINCT colName)会在包编译时会报错,
例如,在某一包的某一存储过程中有如下语句:
1 UPDATE D_LOWAREA L 2 SET L.BYQXH = ( SELECT /* +index(b) */ 3 WM_CONCAT( DISTINCT XH.XHMC) 4 FROM OLIVE.OL$_ASSOCIATIONS A, 5 PMSBI.F_SB_SBXX_BASIC B, 6 PMSBI.D_XH XH 7 WHERE A.A IN (L.PDBYQ1, L.PDBYQ2, L.PDBYQ3) 8 AND A.B = B.SBBH 9 AND B.XHDM != ' wz ' 10 AND B.XHDM = XH.XHDM); 11 COMMIT ;
编译时会报如下错误:
PL/SQL:ORA-30482:DISTINCT option not allowed for this function.
解决方案:
使用动态SQL(字符串中有单引号由两个单引号表示):
1 EXECUTE IMMEDIATE 2 ' UPDATE D_LOWAREA L 3 SET L.BYQXH = (SELECT /*+index(b)*/ 4 WMSYS.WM_CONCAT(DISTINCT XH.XHMC) 5 FROM OLIVE.OL$_ASSOCIATIONS A, 6 PMSBI.F_SB_SBXX_BASIC B, 7 PMSBI.D_XH XH 8 WHERE A.A IN (L.PDBYQ1, L.PDBYQ2, L.PDBYQ3) 9 AND A.B = B.SBBH 10 AND B.XHDM != '' wz '' 11 AND B.XHDM = XH.XHDM) ' ; 12 COMMIT ;
PS:Oracle 11g以上版本可以用LISTAGG实现行转列,写法如下:
SELECT A, LISTAGG(B, ' , ' ) WITHIN GROUP ( ORDER BY B) B FROM T GROUP BY A