Oracle 行列转换
1、固定列数的行列转换
如
student subject grade
--------- ---------- --------
student1 语文 80
student1 数学 70
student1 英语 60
student2 语文 90
student2 数学 80
student2 英语 100
……
转换为
语文 数学 英语
student1 80 70 60
student2 90 80 100
……
语句如下:select student,
sum(decode(subject,'语文', grade,null)) "语文",
sum(decode(subject,'数学', grade,null)) "数学",
sum(decode(subject,'英语', grade,null)) "英语"
from table
group by student;
2、不定列行列转换
如
c1 c2
--- -----------
1 我
1 是
1 谁
2 知
2 道
3 不
……
转换为
1 我是谁
2 知道
3 不
这一类型的转换可以借助于PL/SQL来完成,这里给一个例子
CREATE OR REPLACE FUNCTION get_c2(tmp_c1 NUMBER)
RETURN VARCHAR2
IS
Col_c2 VARCHAR2(4000);
BEGIN
FOR cur IN (SELECT c2 FROM t WHERE c1=tmp_c1) LOOP
Col_c2 := Col_c2||cur.c2;
END LOOP;
Col_c2 := rtrim(Col_c2,1);
RETURN Col_c2;
END;
select distinct c1 ,get_c2(c1) cc2 from table;
或者不用pl/sql,利用分析函数和 CONNECT_BY 实现:
SELECT c1, SUBSTR (MAX (SYS_CONNECT_BY_PATH (c2, ';')), 2) NAME
FROM (SELECT c1, c2, rn, LEAD (rn) OVER (PARTITION BY c1 ORDER BY rn) rn1
FROM (SELECT c1, c2, ROW_NUMBER () OVER (ORDER BY c2) rn
FROM t))
START WITH rn1 IS NULL
CONNECT BY rn1 = PRIOR rn
GROUP BY c1;
3、列数不固定(交叉表行列转置)
这种是比较麻烦的一种,需要借助pl/sql:
原始数据:
CLASS1 CALLDATE CALLCOUNT
1 2005-08-08 40
1 2005-08-07 6
2 2005-08-08 77
3 2005-08-09 33
3 2005-08-08 9
3 2005-08-07 21
转置后:
CALLDATE CallCount1 CallCount2 CallCount3
------------ ---------- ---------- ----------
2005-08-09 0 0 33
2005-08-08 40 77 9
2005-08-07 6 0 21
试验如下:
1). 建立测试表和数据
CREATE TABLE t(
class1 VARCHAR2(2 BYTE),
calldate DATE,
callcount INTEGER
);
INSERT INTO t(class1, calldate, callcount)
VALUES ('1', TO_DATE ('08/08/2005', 'MM/DD/YYYY'), 40);
INSERT INTO t(class1, calldate, callcount)
VALUES ('1', TO_DATE ('08/07/2005', 'MM/DD/YYYY'), 6);
INSERT INTO t(class1, calldate, callcount)
VALUES ('2', TO_DATE ('08/08/2005', 'MM/DD/YYYY'), 77);
INSERT INTO t(class1, calldate, callcount)
VALUES ('3', TO_DATE ('08/09/2005', 'MM/DD/YYYY'), 33);
INSERT INTO t(class1, calldate, callcount)
VALUES ('3', TO_DATE ('08/08/2005', 'MM/DD/YYYY'), 9);
INSERT INTO t(class1, calldate, callcount)
VALUES ('3', TO_DATE ('08/07/2005', 'MM/DD/YYYY'), 21);
COMMIT ;
2). 建立ref cursor准备输出结果集
CREATE OR REPLACE PACKAGE pkg_getrecord
IS
TYPE myrctype IS REF CURSOR;
END pkg_getrecord;
/
3). 建立动态sql交叉表函数,输出结果集
CREATE OR REPLACE FUNCTION fn_rs
RETURN pkg_getrecord.myrctype
IS
s VARCHAR2 (4000);
CURSOR c1 IS
SELECT ',sum(case when Class1='
|| class1
|| ' then CallCount else 0 end)'
|| ' "CallCount'
|| class1
|| '"' c2
FROM t
GROUP BY class1;
r1 c1%ROWTYPE;
list_cursor pkg_getrecord.myrctype;
BEGIN
s := 'select CallDate ';
OPEN c1;
LOOP
FETCH c1 INTO r1;
EXIT WHEN c1%NOTFOUND;
s := s || r1.c2;
END LOOP;
CLOSE c1;
s := s || ' from T group by CallDate order by CallDate desc ';
OPEN list_cursor FOR s;
RETURN list_cursor;
END fn_rs;
/
4). 测试在sql plus下执行:
var results refcursor;
exec :results := fn_rs;
print results;
CALLDATE CallCount1 CallCount2 CallCount3
--------------- ---------- ---------- ----------
2005-08-09 0 0 33
2005-08-08 40 77 9
2005-08-07 6 0 21
转载自:http://www.cnblogs.com/gkl0818/archive/2009/02/25/1398078.html
===================================================================
有如下表,其中各个i值对应的行数是不定的
I A D
---------- ---------- -------------------
1 b 2008-03-27 10:55:42
1 a 2008-03-27 10:55:46
1 d 2008-03-27 10:55:30
2 z 2008-03-27 10:55:55
2 t 2008-03-27 10:55:59
要获得如下结果,注意字符串需要按照D列的时间排序:
2 z,t
这是一个比较典型的行列转换,有好几种实现方法
1.自定义函数实现
return varchar2
is
type typ_cursor is ref cursor;
v_cursor typ_cursor;
v_temp varchar2(10);
v_result varchar2(4000):= '';
v_sql varchar2(200);
begin
v_sql := 'select a from t where i=' || n ||' order by d';
open v_cursor for v_sql;
loop
fetch v_cursor into v_temp;
exit when v_cursor%notfound;
v_result := v_result ||',' || v_temp;
end loop;
return substr(v_result,2);
end;
SQL> select i,my_concat(i) from t group by i;
I MY_CONCAT(I)
---------- --------------------
1 d,b,a
2 z,t
虽然这种方式可以实现需求,但是如果表t的数据量很大,i的值又很多的情况下,因为针对每个i值都要执行一句select,扫描和排序的次数和i的值成正比,性能会非常差。
2.使用sys_connect_by_path
from
(
select i,a,d,min(d) over(partition by i) d_min,
(row_number() over(order by i,d))+(dense_rank() over (order by i)) numid
from t
)
start with d=d_min connect by numid-1=prior numid
group by i;
从执行计划上来看,这种方式只需要扫描两次表,比自定义函数的方法,效率要高很多,尤其是表中数据量较大的时候:
3.使用wm_sys.wm_concat
这个函数也可以实现类似的行列转换需求,但是似乎没有办法做到直接根据另外一列排序,所以需要先通过子查询或者临时表排好序
I WMSYS.WM_CONCAT(A)
---------- --------------------
1 b,a,d
2 z,t
SQL> select i,wmsys.wm_concat(a)
2 from
3 (select * from t order by i,d)
4 group by i;
I WMSYS.WM_CONCAT(A)
---------- --------------------
1 d,b,a
2 z,t
执行计划上看,只需要做一次表扫描就可以了,但是这个函数是加密过的,执行计划并不能显示函数内部的操作。
不知道大家还有没有更加高效的实现方式,欢迎指教^_^
其他一些方法,可以参考:
http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:2196162600402
http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:15637744429336
转载自:http://www.ningoo.net/html/2008/how_to_do_string_aggregate_on_oracle.html