如何在列中划分句点

Posted 2023-02-16

【中文标题】如何在列中划分句点

【英文标题】：How to divide a period in columns

【发布时间】：2021-12-03 06:25:03

【问题描述】：

我正在尝试创建一个查询，其中第一列显示公司列表，其他 3 列显示他们每月的收入。我就是这样做的：

WITH time_frame AS
  (SELECT date_trunc('month',NOW())-interval '0 week'),
  
  
  time_frame1 AS
  (SELECT date_trunc('month',NOW())-interval '1 month'),
  
  time_frame2 AS
  (SELECT date_trunc('month',NOW())-interval '2 month')


select table1.company_name,

            (CASE
               WHEN  table2.date_of_transaction = (SELECT * FROM time_frame2) THEN sum(table2.amount)
               ELSE NULL
           END) AS "current week - 2",
           
              (CASE
               WHEN  table2.date_of_transaction = (SELECT * FROM time_frame1) THEN sum(table2.amount)
               ELSE NULL
           END) AS "current week - 1",
           
           
       (CASE
               WHEN  table2.date_of_transaction = (SELECT * FROM time_frame2) THEN 
 sum(table2.amount)
               ELSE NULL
           END) AS "current week - 2"




from table1
join table2 on table2.table1_id = table.id
where table1.company_joined >= '04-20-2019'

group by 1

当我执行表时，会出现：运行查询时出错：列“table2.date_of_transaction”必须出现在 GROUP BY 子句中或用于聚合函数第 15 行：WHEN table2.date_of_transaction = (SELECT * FROM time_frame)嗯……^

您对如何解决它有任何想法吗？谢谢。

company name	month1	month2
name 1	£233	£343
name 2	£243	£34
name 3	£133	£43

【问题讨论】：

尝试删除group by 1

【参考方案1】：

您可以使用filter() 运算符来简化语句

select t1.company_name,
       sum(t2.amount) filter (where t2.date_of_transaction = date_trunc('month',NOW())-interval '2 month'),
       sum(t2.amount) filter (where t2.date_of_transaction = date_trunc('month',NOW())-interval '1 month'),
       sum(t2.amount) filter (where t2.date_of_transaction = date_trunc('month',NOW()))
from table1 t1
  join table2 t2 on t2.table1_id = t1.id
where t1.company_joined >= date '2019-04-20'
group by t1.company_name;

如果您真的想将日期范围放入 CTE，您只需要一个：

with dates (r1, r2, r3) as (
  values 
    (date_trunc('month',NOW())-interval '2 month',
     date_trunc('month',NOW())-interval '1 month',
     date_trunc('month',NOW()))
)
select t1.company_name,
       sum(t2.amount) filter (where t2.date_of_transaction = d.r1),
       sum(t2.amount) filter (where t2.date_of_transaction = d.r2),
       sum(t2.amount) filter (where t2.date_of_transaction = d.r3)
from table1 t1
  cross join dates d
  join table2 t2 on t2.table1_id = t1.id
where t1.company_joined >= date '2019-04-20'
group by t1.company_name
;

CTE dates 返回包含三列的单行，因此交叉连接不会更改结果行数。

【讨论】：

感谢您的回复。 “过滤器”是什么意思？它在 Postgresql 中不被识别为函数

@RMolok：那么你使用的是过时的 Postgres 版本

好的，谢谢，我想我用你的答案和案例解决了每个月作为过滤器的问题