Sql 关于固定手动列值的条件列的总和

这是我的源数据，其中TIMESTAMP_WID表示YYYYMMDD

我试图实现的输出结果如下所示

(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-2),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)
UNION
(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-3),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)
(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-4),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)

逻辑如下:

(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-2),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)
UNION
(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-3),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)
(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-4),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)

引用CUR_时间戳

到目前为止，我试过的代码如下：

(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-2),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)
UNION
(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-3),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)
(SELECT 
OUTLET, 
SUM(SALES_VALUE), 
TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE,'MON'),-4),'YYYYMMDD')) AS CUR_TIMESTAMP
FROM SOME_TABLE)

在上面的sql查询中，我可以填充与目标相同的内容，但是，在计算SUMSALLES_值时有一个非常重要的逻辑

当

当

当

---

任何人请引导我这听起来像是一个累积的总和：

select t.*
from (select t.*,
             sum(sales_value) over (partition by outlet order by timestamp_wid) as running_sales_value
      from some_table st
     ) t

然后可以添加where子句来筛选所需的行。我认为：

select t.*
from (select t.*,
             sum(sales_value) over (partition by outlet order by timestamp_wid) as running_sales_value
      from some_table st
     ) t
where timestamp_wid >= TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE, 'MON'), -4), 'YYYYMMDD')) and
      timestamp_wid < TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE, 'MON'), -2), 'YYYYMMDD')) and

select t.*
from (select t.*,
             sum(sales_value) over (partition by outlet order by timestamp_wid) as running_sales_value
      from some_table st
     ) t

select t.*
from (select t.*,
             sum(sales_value) over (partition by outlet order by timestamp_wid) as running_sales_value
      from some_table st
     ) t
where timestamp_wid >= TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE, 'MON'), -4), 'YYYYMMDD')) and
      timestamp_wid < TO_NUMBER(TO_CHAR(ADD_MONTHS(TRUNC(SYSDATE, 'MON'), -2), 'YYYYMMDD')) and