SQL 累积总和直到一个标志值并重置总和

Question

我仍在学习 SQL，并试图找出一个我无法解决的问题。所以我的问题是我正在尝试选择一个按日期排序的表（比如说费用），并且在表中我有一个名为 Charged 的​​列，我想将费用添加为累积的（这部分我想通了）。但是，在那之后，我有另一列将充当名为 PayOut 的标志。当 PayOut 值为 1 时，我希望 Charged(SumValue) 的总和重置为零。我该怎么做？这是我尝试过的，我得到的当前输出和我想要的输出。注意：我看到一些使用 CTE 的帖子，但不是相同的场景并且更复杂。

select ex.date, 
       ex.Charged,
       (case when(ex.PayOut=1) then 0 
             else sum(ex.Charged) over (order by ex.date)end) as SumValue, 
      ex.PayOut
from Expense ex 
order by ex.date asc

数据是这样的

Date       Charged   PayOut
01/10/2018   10        0
01/20/2018   5         0
01/30/2018   3         0
02/01/2018   0         1
02/11/2018   12        0
02/21/2018   15        0

我得到的输出

Date       Charged   PayOut  SumValue
01/10/2018   10        0        10
01/20/2018   5         0        15
01/30/2018   3         0        18
02/01/2018   0         1        0
02/11/2018   12        0        30
02/21/2018   15        0        45

想要的输出

Date       Charged   PayOut  SumValue
01/10/2018   10        0        10
01/20/2018   5         0        15
01/30/2018   3         0        18
02/01/2018   0         1        0
02/11/2018   12        0        12
02/21/2018   15        0        27

Answer 1

只需从您的 PayOut 列创建组并将其用作 OVER 中的分区

WITH Expense AS (
              SELECT CAST('01/10/2018' AS DATE) AS Date, 10 AS Charged, 0 AS PayOut
    UNION ALL SELECT CAST('01/20/2018' AS DATE), 5, 0
    UNION ALL SELECT CAST('01/30/2018' AS DATE), 3, 0
    UNION ALL SELECT CAST('02/01/2018' AS DATE), 0, 1
    UNION ALL SELECT CAST('02/11/2018' AS DATE), 12, 0
    UNION ALL SELECT CAST('02/21/2018' AS DATE), 15, 0
)
SELECT
     dat.date
    ,dat.Charged
    ,dat.PayOut
    ,dat.PayOutGroup
    ,SUM(dat.Charged) OVER (PARTITION BY dat.PayOutGroup ORDER BY dat.date) as SumValue
FROM (
    SELECT 
         e.date
        ,e.Charged
        ,e.PayOut
        ,SUM(e.PayOut) OVER (ORDER BY e.date) AS PayOutGroup
    FROM Expense e
) dat