Question

我有下表

表A

ASSIGNMENT_DT    START_DATE    END_DATE    CLT_ID  C_ID First_Service_Date FIRST_SERVICE_DATE2
    2/13/15       1/26/18         1/29/18     1      1                     1/30/18
    3/14/16       1/26/18         1/29/18     1      1                     1/30/18
    1/30/18       1/26/18         1/29/18     1      1                       NULL

查询：

,CASE WHEN FIRST_SERVICE_DATE IS NOT NULL THEN FIRST_SERVICE_DATE
  WHEN FIRST_SERVICE_DATE IS NULL AND ASSIGNMENT_DT > START_DATE THEN NULL
  WHEN FIRST_SERVICE_DATE IS NULL AND ASSIGNMENT_DT < START_DATE THEN LAST_VALUE(TRUNC(ASSIGNMENT_DT)) OVER (PARTITION BY CLT_ID
   ORDER BY CLIENTID ,ASSIGNMENT_DT RANGE BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING)

ELSE TO_DATE('9999/01/01', 'yyyy/mm/dd')

END FIRST_SERVICE_DATE2

我需要什么： First_Service_Date2列每个Clt_id应该有一个Agg值。因此，对于该示例，当End_Date为Not null时，请使用小于开始日期的先前assignment_dt

当end_date不为null时，表示在start_Date之后的LEAD赋值日期。

应该是

First_Service_Date
   3/14/16

Answer 1

我想我会将查询拆开并执行UNION。

select a.*,
  FIRST_SERVICE_DATE as FIRST_SERVICE_DATE2
from TableA a
where FIRST_SERVICE_DATE IS NOT NULL
    OR ASSIGNMENT_DT > START_DATE 
union all
select a.*,
  LAST_VALUE(TRUNC(ASSIGNMENT_DT)) OVER (PARTITION BY CLT_ID) as FIRST_SERVICE_DATE2
from TableA a
where FIRST_SERVICE_DATE IS NULL 
    AND ASSIGNMENT_DT < START_DATE
;

（我猜这里，因为你没有分享你的整个查询，但我认为这可以解决这个问题。）

两个日期之间的`LAST_VALUE`

1 个答案: