您好,登录后才能下订单哦!
ROLLUP()函数是对于GROUP BY 分组统计的功能扩展,可以实现分组统计求和的效果。
下面我们开始准备我们的试验的环境
--创建新表employee_salary,存储数据来自用户hr.employees
SQL>CREATE TABLE employee_salary ASSELECT E.FIRST_NAME,E.JOB_ID,E.MANAGER_ID,E.SALARY FROM HR.EMPLOYEES E WHERE E.JOB_ID='IT_PROG';
--查看新创建的表
SQL>SELECT * FROM employee_salary;
--显示效果如下
FIRST_NAME JOB_ID MANAGER_ID SALARY
-------------------- ---------- --------------------------------
Alexander IT_PROG 102 9000.00
Bruce IT_PROG 103 6000.00
David IT_PROG 103 4800.00
Valli IT_PROG 103 4800.00
Diana IT_PROG 103 4200.00
--首先按照JOB_ID进行分组,查看salary和
SQL>SELECT sa.job_id,SUM(sa.salary) FROM employee_salary sa GROUP BY sa.job_id;
--显示效果如下
JOB_ID SUM(SA.SALARY)
---------- -----------------------
IT_PROG 28800
--按照MANAGER_ID进行分组,查看salary和
SELECT sa.manager_id,SUM(sa.salary) FROM employee_salary sa GROUP BY sa.manager_id;
--显示效果如下
MANAGER_ID SUM(SA.SALARY)
---------- ----------------------------
102 9000
103 19800
--我们使用ROLLUP函数看看是什么效果
SELECT sa.job_id,sa.manager_id,SUM(sa.salary) FROM employee_salary sa GROUP BY ROLLUP(sa.job_id,sa.manager_id);
--显示效果如下
JOB_ID MANAGER_ID SUM(SA.SALARY)
---------- ---------- --------------------------------
IT_PROG 102 9000
IT_PROG 103 19800
IT_PROG 28800
28800
说明:ROLLUP解析过程,以ROLLUP(a,b)为例
ROLLUP(a,b)== GROUP(a,b) UNION ALL GROUP(a) UNIONALL GROUP()
即:解析顺序是从右至左,显示按照a,b分组,接下来是按照a分组,最后是对全表分组;
上面ROLLUP (sa.job_id,sa.manager_id)等同下面UNION ALL的集合操作
SELECT sa.job_id,sa.manager_id,SUM(sa.salary) FROM employee_salary sa
GROUP BY sa.job_id,sa.manager_id
UNION ALL
SELECT sa.job_id,NULL,SUM(sa.salary) FROM employee_salary sa
GROUP BY sa.job_id
UNION ALL
SELECT NULL,NULL,SUM(sa.salary) FROM employee_salary sa GROUP BY()
ORDER BY 1,2;
--显示效果如下
JOB_ID MANAGER_ID SUM(SA.SALARY)
---------- ---------- --------------------------------
IT_PROG 102 9000
IT_PROG 103 19800
IT_PROG 28800
28800
说明:虽然最后展示的效果是相同的,但是ROLLUP()函数的执行效率要比UNION ALL的效率要高、要快。
ROLLUP(A,B,C)是在执行组合操作,无顺序,组合公式是(n+1),当n=3时,组合结果就是有4个。
ROLLUP()中的参数位置不同,得出的结果可能不一样!
在理解ROLLUP的基础上再来理解CUBE()就比较容易,ROLLUP()在执行组合操作,CUBE()就是在执行排序动作,从左至右,排序公式是2N次方。
CUBE(A,B,C)==GROUP BY (A,B,C) UNION ALL GROUP BY (A,B) UNION ALL GROUP BY (A,C) UNION ALL GROUPBY (A) UNION ALL GROUP BY (B) UNION ALL GROUP BY (C) UNION ALL GROUP BY ()
例如:执行下列语句
SQL> SELECT sa.job_id,sa.manager_id,SUM(sa.salary) FROM employee_salary sa GROUP BY CUBE(sa.job_id,sa.manager_id);
--显示效果如下
JOB_ID MANAGER_ID SUM(SA.SALARY)
---------- ---------- -----------------------------
28800
102 9000
103 19800
IT_PROG 28800
IT_PROG 102 9000
IT_PROG 103 19800
6 rows selected
上面的语句等同执行下列语句
SQL>SELECT sa.job_id,sa.manager_id,SUM(sa.salary) FROM employee_salary sa GROUP BY (sa.job_id,sa.manager_id)
UNION ALL
SELECT sa.job_id,NULL,SUM(sa.salary) FROM employee_salary sa
GROUP BY (sa.job_id)
UNION ALL
SELECT NULL,sa.manager_id,SUM(sa.salary) FROM employee_salary sa
GROUP BY (sa.manager_id)
UNION ALL
SELECT NULL,NULL,SUM(sa.salary) FROM employee_salary sa
GROUPBY ();
关于GROUPING
GROUPING(A)用于判断对于分组后的列是否是空值NULL,返回值有0和1两个值,1表示,该列为空——NULL,这个NULL值是因为分组时产生,否则则为0;
免责声明:本站发布的内容(图片、视频和文字)以原创、转载和分享为主,文章观点不代表本网站立场,如果涉及侵权请联系站长邮箱:is@yisu.com进行举报,并提供相关证据,一经查实,将立刻删除涉嫌侵权内容。