postgresql 总列总和

Question

选择

SELECT pp.id, TO_CHAR(pp.created_dt::date, 'dd.mm.yyyy') AS "Date", CAST(pp.created_dt AS time(0)) AS "Time",
       au.username AS "User", ss.name AS "Service", pp.amount, REPLACE(pp.status, 'SUCCESS', ' ') AS "Status",
       pp.account AS "Props", pp.external_id AS "External", COALESCE(pp.external_status, null, 'indefined') AS "External status"
  FROM payment AS pp
 INNER JOIN auth_user AS au ON au.id = pp.creator_id
 INNER JOIN services_service AS ss ON ss.id = pp.service_id
 WHERE pp.created_dt::date = (CURRENT_DATE - INTERVAL '1' day)::date
 AND ss.name = 'Some Name' AND pp.status = 'SUCCESS'


id    |    Date   |    Time   |   Service  |amount | Status |                 
------+-----------+-----------+------------+-------+--------+---
    9 | 2021.11.1 |  12:20:01 |  some serv | 100   | stat   |
   10 | 2021.12.1 |  12:20:01 |  some serv | 89    | stat   |
------+-----------+-----------+------------+-------+--------+-----        
Total |           |           |            | 189   |        |

我有一个这样的选择。我需要得到类似上面显示的东西。也就是说，我需要得到一列的总数。我已经尝试了很多东西，但对我来说没有任何效果。

Answer 1

如果我理解正确，您想要一个结果，即在原始查询结果之后附加带有聚合值的额外行。您可以通过多种方式实现它：

1. （推荐）最简单的方法可能是将您的原始查询与辅助查询联合起来：

with t(id,other_column1,other_column2,amount) as (values
  (9,'some serv','stat',100),
  (10,'some serv','stat',89)
)
select t.id::text, t.other_column1, t.other_column2, t.amount from t
union all
select 'Total', null, null, sum(amount) from t

2.您也可以使用group by rollup 子句，其目的正是如此。由于您的查询包含许多不涉及聚合的列，因此您的情况变得更加困难。因此，最好先计算聚合，然后再加入不重要的数据：

with t(id,other_column1,other_column2,amount) as (values
  (9,'some serv','stat',100),
  (10,'some serv','stat',89)
)
select case when t.id is null then 'Total' else t.id::text end as id
     , t.other_column1
     , t.other_column2
     , case when t.id is null then ext.sum else t.amount end as amount
from (
  select t.id, sum(amount) as sum
  from t
  group by rollup(t.id)
) ext
left join t on ext.id = t.id
order by ext.id

3. 为了完整起见，我只是向您展示了应该做些什么来避免加入。在这种情况下，group by 子句必须使用除amount 之外的所有列（以保留原始行）加上聚合（以获取总和行），因此具有 2 个集合的grouping sets 子句很方便。 （rollup 子句毕竟是grouping sets 的特例。）明显的缺点是对未参与聚合的每一列重复case grouping... 表达式。

with t(id,other_column1,other_column2,amount) as (values
  (9,'some serv','stat',100),
  (10,'some serv2','stat',89)
)
select case grouping(t.id) when 0 then t.id::text else 'Total' end as id
     , case grouping(t.id) when 0 then t.other_column1 end as other_column1
     , case grouping(t.id) when 0 then t.other_column2 end as other_column2
     , sum(t.amount) as amount
from t
group by grouping sets((t.id, t.other_column1, t.other_column2), ())
order by t.id

查看示例 (db fiddle)：

（坦率地说，除了简单的报告之外，我几乎无法想象任何其他目的，其中列将number 类型的ID 与text 类型的标签Total 混合在一起。）