ROLLUP AND CUBE OPERATORS IN ORACLE 8I

Question

제품 : PL/SQL
작성날짜 : 2000-06-29
========================================
ROLLUP AND CUBE OPERATORS IN ORACLE 8I
========================================
PURPOSE
ROLLUP 과 CUBE Operator에 대해 설명하고자 한다.
Explanation
ROLLUP operator는 SELECT문의 GROUP BY절에 사용된다.
SELECT절에 ROLLUP 을 사용함으로써 'regular rows'(보통의 select된 data)와
'super-aggregate rows'(총계)을 구할 수 있다. 기존에는 select ... union select
를 이용해 구사해야 했었던 것이다. 'super-aggregate rows'는 'sub-total'
(중간 Total, 즉 소계)을 포함한다.
CUBE operator는 Cross-tab에 대한 Summary를 추출하는데 사용된다. 모든 가능한
dimension에 대한 total을 나타낸다. 즉 ROLLUP에 의해 나타내어지는 item total값과
column total값을 나타낸다.
NULL값은 모든 값에 대한 super-aggregate 을 나타낸다. GROUPING() function은
모든 값에 대한 set을 나타내는 null값과 column의 null값과 구별하는데 쓰여진다.
GROUPING() function은 GROUP BY절에서 반드시 표현되어야 한다. GROUPING()은 모든
값의 set을 표현합에 있어서 null이면 1을 아니면 0을 return한다.
ROLLUP과 CUBE는 CREATE MATERIALIZED VIEW에서 사용되어 질수 있다.
Example
아래와 같이 테스트에 쓰여질 table과 data을 만든다.
create table test_roll
(YEAR NUMBER(4),
REGION CHAR(7),
DEPT CHAR(2),
PROFIT NUMBER );
insert into test_roll values (1995 ,'West' , 'A1' , 100);
insert into test_roll values (1995 ,'West' , 'A2' , 100);
insert into test_roll values (1996 ,'West' , 'A1' , 100);
insert into test_roll values (1996 ,'West' , 'A2' , 100);
insert into test_roll values (1995 ,'Central' ,'A1' , 100);
insert into test_roll values (1995 ,'East' , 'A1' , 100);
insert into test_roll values (1995 ,'East' , 'A2' , 100);
SQL> select * from test_roll;
YEAR REGION DE PROFIT
1995 West A1 100
1995 West A2 100
1996 West A1 100
1996 West A2 100
1995 Central A1 100
1995 East A1 100
1995 East A2 100
7 rows selected.
예제 1: ROLLUP
SQL> select year, region, sum(profit), count(*)
from test_roll
group by rollup(year, region);
YEAR REGION SUM(PROFIT) COUNT(*)
1995 Central 100 1
1995 East 200 2
1995 West 200 2
1995 500 5
1996 West 200 2
1996 200 2
700 7
7 rows selected.
위의 내용을 tabular로 나타내어 보면 쉽게 알 수 있다.
Year Central(A1+A2) East(A1+A2) West(A1+A2)
1995 (100+NULL) (100+100) (100+100) 500
1996 (NULL+NULL) (NULL+NULL) (100+100) 200
700
예제 2: ROLLUP and GROUPING()
SQL> select year, region, sum(profit),
grouping(year) "Y", grouping(region) "R"
from test_roll
group by rollup (year, region);
YEAR REGION SUM(PROFIT) Y R
1995 Central 100 0 0
1995 East 200 0 0
1995 West 200 0 0
1995 500 0 1
1996 West 200 0 0
1996 200 0 1
700 1 1
7 rows selected.
참고) null값이 모든 값의 set에 대한 표현으로 나타내어지면 GROUPING function은
super-aggregate row에 대해 1을 return한다.
예제 3: CUBE
SQL> select year, region, sum(profit), count(*)
from test_roll
group by cube(year, region);
YEAR REGION SUM(PROFIT) COUNT(*)
1995 Central 100 1
1995 East 200 2
1995 West 200 2
1995 500 5
1996 West 200 2
1996 200 2
Central 100 1
East 200 2
West 400 4
700 7
위의 내용을 tabular로 나타내어 보면 쉽게 알 수 있다.
Year Central(A1+A2) East(A1+A2) West(A1+A2)
1995 (100+NULL) (100+100) (100+100) 500
1996 (NULL+NULL) (NULL+NULL) (100+100) 200
100 200 400 700
예제 4: CUBE and GROUPING()
SQL> select year, region, sum(profit),
grouping(year) "Y", grouping(region) "R"
from test_roll
group by cube (year, region);
YEAR REGION SUM(PROFIT) Y R
1995 Central 100 0 0
1995 East 200 0 0
1995 West 200 0 0
1995 500 0 1
1996 West 200 0 0
1996 200 0 1
Central 100 1 0
East 200 1 0
West 400 1 0
700 1 1
10 rows selected.
===============================================
HOW TO USE ROLLUP AND CUBE OPERATORS IN PL/SQL
===============================================
Release 8.1.5 PL/SQL에서는 CUBE, ROLLUP 이 지원되지 않는다. 8i에서는 DBMS_SQL
package을 이용하여 dynamic SQL로 구현하는 방법이 workaround로 제시된다.
Ordacle8i에서는 PL/SQL block에 SQL절을 직접적으로 위치시키는 Native Dynamic SQL
(참고 bul#11721)을 지원한다.
Native Dynamic SQL을 사용하기 위해서는 COMPATIBLE 이 8.1.0 또는 그 보다 높아야
한다.
SVRMGR> show parameter compatible
NAME TYPE VALUE
compatible string 8.1.0
예제 1-1: ROLLUP -> 위의 예제 1과 비교한다.
SQL> create or replace procedure test_rollup as
my_year test_roll.year%type;
my_region test_roll.region%type;
my_sum int;
my_count int;
begin
select year, region, sum(profit), count(*)
into my_year, my_region, my_sum, my_count
from test_roll
group by rollup(year, region);
end;
Warning: Procedure created with compilation errors.
SQL> show errors
Errors for PROCEDURE TEST_ROLLUP:
LINE/COL ERROR
10/8 PL/SQL: SQL Statement ignored
13/18 PLS-00201: identifier 'ROLLUP' must be declared
SQL> create or replace procedure test_rollup as
type curTyp is ref cursor;
sql_stmt varchar2(200);
tab_cv curTyp;
my_year test_roll.year%type;
my_region test_roll.region%type;
my_sum int;
my_count int;
begin
sql_stmt := 'select year, region, sum(profit), count(*) ' ||
'from test_roll ' ||
'group by rollup(year, region)';
open tab_cv for sql_stmt;
loop
fetch tab_cv into my_year, my_region, my_sum, my_count;
exit when tab_cv%NOTFOUND;
dbms_output.put_line (my_year || ' '||
nvl(my_region,' ') ||
' ' || my_sum || ' ' || my_count);
end loop;
close tab_cv;
end;
SQL> set serveroutput on
SQL> exec test_rollup
1995 Central 100 1
1995 East 200 2
1995 West 200 2
1995 500 5
1996 West 200 2
1996 200 2
700 7
PL/SQL procedure successfully completed.
예제 2-1: ROLLUP and GROUPING() -> 위의 예제 2와 비교한다.
SQL> create or replace procedure test_rollupg as
my_year test_roll.year%type;
my_region test_roll.region%type;
my_sum int;
my_count int;
my_g_region int;
my_g_year int;
begin
select year, region, sum(profit),
grouping(year), grouping(region)
into my_year, my_region, my_sum, my_count,
my_g_year, my_g_region
from test_roll
group by rollup(year, region);
end;
Warning: Procedure created with compilation errors.
SQL> show error
Errors for PROCEDURE TEST_ROLLUPG:
LINE/COL ERROR
12/4 PL/SQL: SQL Statement ignored
17/13 PLS-00201: identifier 'ROLLUP' must be declared
SQL> create or replace procedure test_rollupg as
type curTyp is ref cursor;
sql_stmt varchar2(200);
tab_cv curTyp;
my_year test_roll.year%type;
my_region test_roll.region%type;
my_sum int;
my_count int;
my_g_region int;
my_g_year int;
begin
sql_stmt := 'select year, region, sum(profit), count(*), ' ||
'grouping(year), grouping(region) ' ||
'from test_roll ' ||
'group by rollup(year, region)';
open tab_cv for sql_stmt;
loop
fetch tab_cv into my_year, my_region, my_sum, my_count,
my_g_year, my_g_region;
exit when tab_cv%NOTFOUND;
dbms_output.put_line (my_year || ' '||my_region ||
' ' || my_sum || ' ' || my_count ||
' ' || my_g_year || ' ' || my_g_region);
end loop;
close tab_cv;
end;
Procedure created.
SQL> set serveroutput on
SQL> exec test_rollupg
1995 Central 100 1 0 0
1995 East 200 2 0 0
1995 West 200 2 0 0
1995 500 5 0 1
1996 West 200 2 0 0
1996 200 2 0 1
700 7 1 1
PL/SQL procedure successfully completed.
예제 3-1: CUBE -> 위의 예제 3과 비교한다.
SQL> create or replace procedure test_cube as
my_year test_roll.year%type;
my_region test_roll.region%type;
my_sum int;
my_count int;
begin
select year, region, sum(profit), count(*)
into my_year, my_region, my_sum, my_count
from test_roll
group by cube(year, region);
end;
Warning: Procedure created with compilation errors.
SQL> show error
Errors for PROCEDURE TEST_CUBE:
LINE/COL ERROR
10/4 PL/SQL: SQL Statement ignored
13/13 PLS-00201: identifier 'CUBE' must be declared
SQL> create or replace procedure test_cube as
type curTyp is ref cursor;
sql_stmt varchar2(200);
tab_cv curTyp;
my_year test_roll.year%type;
my_region test_roll.region%type;
my_sum int;
my_count int;
begin
sql_stmt := 'select year, region, sum(profit), count(*) ' ||
'from test_roll ' ||
'group by cube(year, region)';
open tab_cv for sql_stmt;
loop
fetch tab_cv into my_year, my_region, my_sum, my_count;
exit when tab_cv%NOTFOUND;
dbms_output.put_line (my_year || ' '||
nvl(my_region,' ') ||
' ' || my_sum || ' ' || my_count);
end loop;
close tab_cv;
end;
Procedure created.
SQL> set serveroutput on
SQL> exec test_cube
1995 Central 100 1
1995 East 200 2
1995 West 200 2
1995 500 5
1996 West 200 2
1996 200 2
Central 100 1
East 200 2
West 400 4
700 7
PL/SQL procedure successfully completed.
예제 4-1: CUBE and GROUPING() -> 위의 예제 4와 비교한다.
SQL> create or replace procedure test_cubeg as
my_year test_roll.year%type;
my_region test_roll.region%type;
my_sum int;
my_count int;
my_g_region int;
my_g_year int;
begin
select year, region, sum(profit),
grouping(year), grouping(region)
into my_year, my_region, my_sum, my_count,
my_g_year, my_g_region
from test_roll
group by cube(year, region);
end;
Warning: Procedure created with compilation errors.
SQL> show error
Errors for PROCEDURE TEST_CUBEG:
LINE/COL ERROR
12/4 PL/SQL: SQL Statement ignored
17/13 PLS-00201: identifier 'CUBE' must be declared
SQL> create or replace procedure test_cubeg as
type curTyp is ref cursor;
sql_stmt varchar2(200);
tab_cv curTyp;
my_year test_roll.year%type;
my_region test_roll.region%type;
my_sum int;
my_count int;
my_g_region int;
my_g_year int;
begin
sql_stmt := 'select year, region, sum(profit), count(*), ' ||
'grouping(year), grouping(region) ' ||
'from test_roll ' ||
'group by cube(year, region)';
open tab_cv for sql_stmt;
loop
fetch tab_cv into my_year, my_region, my_sum, my_count,
my_g_year, my_g_region;
exit when tab_cv%NOTFOUND;
dbms_output.put_line (my_year || ' '||my_region ||
' ' || my_sum || ' ' || my_count ||
' ' || my_g_year || ' ' || my_g_region);
end loop;
close tab_cv;
end;
Procedure created.
SQL> set serveroutput on
SQL> exec test_cubeg
1995 Central 100 1 0 0
1995 East 200 2 0 0
1995 West 200 2 0 0
1995 500 5 0 1
1996 West 200 2 0 0
1996 200 2 0 1
Central 100 1 1 0
East 200 2 1 0
West 400 4 1 0
700 7 1 1
PL/SQL procedure successfully completed.
Reference Ducumment
Note:67988.1

Answer 1

Hello,
As previously posted you should use export and import utilities.
To execute exp or imp statements you have just to open a command line interface, for instance,
a DOS box on Windows.
So you don't have to use SQL*Plus or TOAD.
About export/import you may care on the mode, to export a single or a list of Tables the Table mode
is enough.
Please, find here an example to begin:
http://wiki.oracle.com/page/Oracle+export+and+import+
Hope this help.
Best regards,
Jean-Valentin