请查看我的第一个问题:SQL: partition over two columns
我有以下表格:
----------------------------------
| No1 | No2 | Amount| Timestamp
----------------------------------
| A | B | 10 | 01.01.2018
| C | D | 20 | 02.01.2018
| B | A | 30 | 03.01.2018
| D | C | 40 | 04.01.2018
----------------------------------
目前我的结果如下:
-----------------------------------------------------
| No1 | No2 | Sum(Amount) over partition | Timestamp
-----------------------------------------------------
| A | B | 40 | 01.01.2018
| C | D | 60 | 02.01.2018
| B | A | 40 | 03.01.2018
| D | C | 60 | 04.01.2018
-----------------------------------------------------
使用SQL(来自Vamsi Prabhala的第一个问题'答案):
select no1,no2,sum(amount) over(partition by least(no1,no2),greatest(no1,no2)) as total, timestamp
from tbl
现在我的问题是如何在结果中添加行,如:
----------------------------------------------------
| No1 | No2 | Sum(Amount) over partition | Timestamp
----------------------------------------------------
| A | B | 40 (optional) | 01.01.2018
| B | A | 40 (optional) | 02.01.2018
| AB |(NULL)| 40 |
| C | D | 60 (optional) | 03.01.2018
| D | C | 60 (optional) | 04.01.2018
| CD |(NULL)| 60 |
----------------------------------------------------
请注意,可以有多行,例如值(No1 = A,No2 = B)
更新:添加时间戳列以更具体地实现我想要实现的目标
答案 0 :(得分:2)
SELECT
LEAST(No1, No2) || ':' || GREATEST(No1, No2) AS set_label,
No1,
No2,
SUM(Amount) AS Amount,
Stamp
FROM
tbl
GROUP BY
GROUPING SETS (
(LEAST(No1, No2), GREATEST(No1, No2), No1, No2, Stamp),
(LEAST(No1, No2), GREATEST(No1, No2))
)
http://sqlfiddle.com/#!4/9afd5/18
如果每一行都有唯一的标识符会更好......
答案 1 :(得分:0)
一种方法是UNION ALL
:
select no1, no2,
sum(amount) over (partition by least(no1, no2), greatest(no1, no2)) as total
from tbl
union all
select least(no1, no2) || greatest(no1, no2), NULL, sum(amount)
from tbl
group by least(no1, no2), greatest(no1, no2);