postgresql 的 group by 之 grouping sets/rollup/cube

postgresql 从 9.5 开始提供 rollup/cube/grouping sets 分组函数，使用起来更为方便,尤其时用sql直接出报表时，一个sql就把明细和汇总值全部搞定。

https://www.postgresql.org/docs/9.5/static/sql-select.html
https://www.postgresql.org/docs/9.5/static/queries-table-expressions.html#QUERIES-GROUPING-SETS

and grouping_element can be one of:

    ( )
    expression
    ( expression [, ...] )
    ROLLUP ( { expression | ( expression [, ...] ) } [, ...] )
    CUBE ( { expression | ( expression [, ...] ) } [, ...] )
    GROUPING SETS ( grouping_element [, ...] )

测试数据

with tmp_tab as (
  select '20180205' as day_id,'1001' as custno,'a001' as data_type,1 as qty union all
  select '20180205' as day_id,'1001' as custno,'a002' as data_type,2 as qty union all
  select '20180205' as day_id,'1002' as custno,'a001' as data_type,3 as qty union all
  select '20180205' as day_id,'1002' as custno,'a003' as data_type,4 as qty union all
  select '20180206' as day_id,'1001' as custno,'a004' as data_type,5 as qty union all
  select '20180207' as day_id,'1003' as custno,'a001' as data_type,6 as qty 
)
select grouping(t0.day_id) as day_id,
       grouping(t0.data_type) as data_type,
       grouping(t0.custno) as custno,
       t0.day_id,
       t0.data_type,
       t0.custno,
       sum(t0.qty)
from tmp_tab t0
group by t0.day_id,t0.data_type,t0.custno

grouping sets:显示指定的汇总值

1、group by grouping sets ( (t0.day_id,t0.data_type,t0.custno) )
等效于
group by t0.day_id,t0.data_type,t0.custno

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
(6 rows)

2、group by grouping sets ( t0.day_id,t0.data_type,t0.custno )
对 t0.day_id,t0.data_type,t0.custno 这三列分别求合计

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         1 |      1 | 20180205 |           |        |  10
      0 |         1 |      1 | 20180206 |           |        |   5
      0 |         1 |      1 | 20180207 |           |        |   6
      1 |         1 |      0 |          |           | 1001   |   8
      1 |         1 |      0 |          |           | 1002   |   7
      1 |         1 |      0 |          |           | 1003   |   6
      1 |         0 |      1 |          | a001      |        |  10
      1 |         0 |      1 |          | a002      |        |   2
      1 |         0 |      1 |          | a003      |        |   4
      1 |         0 |      1 |          | a004      |        |   5
(10 rows)

3、group by grouping sets ( (t0.day_id,t0.data_type),t0.custno )
(t0.day_id,t0.data_type) 作为一个整体求合计

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      1 |         1 |      0 |          |           | 1001   |   8
      1 |         1 |      0 |          |           | 1002   |   7
      1 |         1 |      0 |          |           | 1003   |   6
      0 |         0 |      1 | 20180205 | a001      |        |   4
      0 |         0 |      1 | 20180205 | a002      |        |   2
      0 |         0 |      1 | 20180205 | a003      |        |   4
      0 |         0 |      1 | 20180206 | a004      |        |   5
      0 |         0 |      1 | 20180207 | a001      |        |   6
(8 rows)

4、group by grouping sets ( (t0.day_id),(t0.data_type),(t0.custno) )

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         1 |      1 | 20180205 |           |        |  10
      0 |         1 |      1 | 20180206 |           |        |   5
      0 |         1 |      1 | 20180207 |           |        |   6
      1 |         1 |      0 |          |           | 1001   |   8
      1 |         1 |      0 |          |           | 1002   |   7
      1 |         1 |      0 |          |           | 1003   |   6
      1 |         0 |      1 |          | a001      |        |  10
      1 |         0 |      1 |          | a002      |        |   2
      1 |         0 |      1 |          | a003      |        |   4
      1 |         0 |      1 |          | a004      |        |   5
(10 rows)

rollup:保留 group by 的基础上,增加汇总行
1、group by rollup ( (t0.day_id,t0.data_type,t0.custno) )
对t0.day_id,t0.data_type,t0.custno 这三列整体求个合计

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      1 |         1 |      1 |          |           |        |  21
(7 rows)

2、group by rollup ( t0.day_id,t0.data_type,t0.custno )
对t0.day_id,t0.data_type,t0.custno 依次求合计

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      1 | 20180205 | a001      |        |   4
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      1 | 20180205 | a002      |        |   2
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      0 |         0 |      1 | 20180205 | a003      |        |   4
      0 |         1 |      1 | 20180205 |           |        |  10
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
      0 |         0 |      1 | 20180206 | a004      |        |   5
      0 |         1 |      1 | 20180206 |           |        |   5
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      0 |         0 |      1 | 20180207 | a001      |        |   6
      0 |         1 |      1 | 20180207 |           |        |   6
      1 |         1 |      1 |          |           |        |  21
(15 rows)

3、group by rollup ( (t0.day_id,t0.data_type),t0.custno )
(t0.day_id,t0.data_type) 作为一个整体求合计

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      1 | 20180205 | a001      |        |   4
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      1 | 20180205 | a002      |        |   2
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      0 |         0 |      1 | 20180205 | a003      |        |   4
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
      0 |         0 |      1 | 20180206 | a004      |        |   5
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      0 |         0 |      1 | 20180207 | a001      |        |   6
      1 |         1 |      1 |          |           |        |  21
(12 rows)

4、group by rollup ( (t0.day_id),(t0.data_type),(t0.custno) )
等效于第二种情况
group by rollup ( t0.day_id,t0.data_type,t0.custno )

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      1 | 20180205 | a001      |        |   4
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      1 | 20180205 | a002      |        |   2
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      0 |         0 |      1 | 20180205 | a003      |        |   4
      0 |         1 |      1 | 20180205 |           |        |  10
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
      0 |         0 |      1 | 20180206 | a004      |        |   5
      0 |         1 |      1 | 20180206 |           |        |   5
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      0 |         0 |      1 | 20180207 | a001      |        |   6
      0 |         1 |      1 | 20180207 |           |        |   6
      1 |         1 |      1 |          |           |        |  21
(15 rows)

cube:保留 group by 的基础上,增加指定列的交叉汇总行
1、group by cube ( (t0.day_id,t0.data_type,t0.custno) )

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      1 |         1 |      1 |          |           |        |  21
(7 rows)

2、group by cube ( t0.day_id,t0.data_type,t0.custno )
t0.day_id,t0.data_type,t0.custno 这三列交叉汇总

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      1 | 20180205 | a001      |        |   4
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      1 | 20180205 | a002      |        |   2
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      0 |         0 |      1 | 20180205 | a003      |        |   4
      0 |         1 |      1 | 20180205 |           |        |  10
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
      0 |         0 |      1 | 20180206 | a004      |        |   5
      0 |         1 |      1 | 20180206 |           |        |   5
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      0 |         0 |      1 | 20180207 | a001      |        |   6
      0 |         1 |      1 | 20180207 |           |        |   6
      1 |         1 |      1 |          |           |        |  21
      0 |         1 |      0 | 20180205 |           | 1001   |   3
      0 |         1 |      0 | 20180206 |           | 1001   |   5
      1 |         1 |      0 |          |           | 1001   |   8
      0 |         1 |      0 | 20180205 |           | 1002   |   7
      1 |         1 |      0 |          |           | 1002   |   7
      0 |         1 |      0 | 20180207 |           | 1003   |   6
      1 |         1 |      0 |          |           | 1003   |   6
      1 |         0 |      0 |          | a001      | 1001   |   1
      1 |         0 |      0 |          | a001      | 1002   |   3
      1 |         0 |      0 |          | a001      | 1003   |   6
      1 |         0 |      1 |          | a001      |        |  10
      1 |         0 |      0 |          | a002      | 1001   |   2
      1 |         0 |      1 |          | a002      |        |   2
      1 |         0 |      0 |          | a003      | 1002   |   4
      1 |         0 |      1 |          | a003      |        |   4
      1 |         0 |      0 |          | a004      | 1001   |   5
      1 |         0 |      1 |          | a004      |        |   5
(32 rows)

3、group by cube ( (t0.day_id,t0.data_type),t0.custno )

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
      1 |         1 |      0 |          |           | 1001   |   8
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      1 |         1 |      0 |          |           | 1002   |   7
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      1 |         1 |      0 |          |           | 1003   |   6
      1 |         1 |      1 |          |           |        |  21
      0 |         0 |      1 | 20180205 | a001      |        |   4
      0 |         0 |      1 | 20180205 | a002      |        |   2
      0 |         0 |      1 | 20180205 | a003      |        |   4
      0 |         0 |      1 | 20180206 | a004      |        |   5
      0 |         0 |      1 | 20180207 | a001      |        |   6
(15 rows)

4、group by cube ( (t0.day_id),(t0.data_type),(t0.custno) )

 day_id | data_type | custno |  day_id  | data_type | custno | sum 
--------+-----------+--------+----------+-----------+--------+-----
      0 |         0 |      0 | 20180205 | a001      | 1001   |   1
      0 |         0 |      0 | 20180205 | a001      | 1002   |   3
      0 |         0 |      1 | 20180205 | a001      |        |   4
      0 |         0 |      0 | 20180205 | a002      | 1001   |   2
      0 |         0 |      1 | 20180205 | a002      |        |   2
      0 |         0 |      0 | 20180205 | a003      | 1002   |   4
      0 |         0 |      1 | 20180205 | a003      |        |   4
      0 |         1 |      1 | 20180205 |           |        |  10
      0 |         0 |      0 | 20180206 | a004      | 1001   |   5
      0 |         0 |      1 | 20180206 | a004      |        |   5
      0 |         1 |      1 | 20180206 |           |        |   5
      0 |         0 |      0 | 20180207 | a001      | 1003   |   6
      0 |         0 |      1 | 20180207 | a001      |        |   6
      0 |         1 |      1 | 20180207 |           |        |   6
      1 |         1 |      1 |          |           |        |  21
      0 |         1 |      0 | 20180205 |           | 1001   |   3
      0 |         1 |      0 | 20180206 |           | 1001   |   5
      1 |         1 |      0 |          |           | 1001   |   8
      0 |         1 |      0 | 20180205 |           | 1002   |   7
      1 |         1 |      0 |          |           | 1002   |   7
      0 |         1 |      0 | 20180207 |           | 1003   |   6
      1 |         1 |      0 |          |           | 1003   |   6
      1 |         0 |      0 |          | a001      | 1001   |   1
      1 |         0 |      0 |          | a001      | 1002   |   3
      1 |         0 |      0 |          | a001      | 1003   |   6
      1 |         0 |      1 |          | a001      |        |  10
      1 |         0 |      0 |          | a002      | 1001   |   2
      1 |         0 |      1 |          | a002      |        |   2
      1 |         0 |      0 |          | a003      | 1002   |   4
      1 |         0 |      1 |          | a003      |        |   4
      1 |         0 |      0 |          | a004      | 1001   |   5
      1 |         0 |      1 |          | a004      |        |   5
(32 rows)

postgresql 的 grouping sets/rollup/cube 与 oracle 的是一样的，使用简单，功能强大。