SQL Server 2008：如何更新以前记录的结束日期列

问题描述：

我在下面有一些记录。对于每个客户，只有一条记录可以有END_DATE=12/31/9998。SQL Server 2008：如何更新以前记录的结束日期列

所以我需要结束日期基于EFFECT_DATE列以前的记录。

例如，我需要从 12/31/9998更新ID=2的END_DATE到7/17/2017为CLT_NBR=12375

ID CLT_NBR IS_PRIMARY EFFECT_DATE END_DATE 
----------------------------------------------- 
1 12375  1   8/13/2015 9/30/2015 
2 12375  1   10/1/2015 12/31/9998 
3 12375  1   7/18/2017 12/31/9998 
4 12331  1   2/3/2016 7/8/2016 
5 12331  1   7/9/2016 12/31/9998

为什么不降END_DATE列和使用递归CTE在查询计算它数据？ –

你应该看到我的答案为sql 2008兼容版本 – DhruvJoshi

@SeanLange - 尽可能多地我也推荐这个（或者用'LEAD（）'）的东西，那些查询会消耗大量资源。由于这些数据通常相当静态，所以缓存结束日期通常会更好，尽管可以将其计算为列。也就是说，我绝对建议将它作为**独占**（'

答

您可以使用lead获得下一个生效日期和减去一天来更新当前行的结束日期。

with cte as (select t.*,coalesce(dateadd(day,-1,lead(effect_date) over(partition by clt_nbr order by id),'9998-12-31') as new_end_date 
      from tbl t) 
update cte 
set end_date=new_end_date

对于SQL Server 2008，使用

with rownums as (select t.*,row_number() over(partition by clt_nbr order by id) as rnum from tbl t) 
,cte as (select r1.*,dateadd(day,-1,coalesce(r2.effect_date,'9999-01-01')) as new_end_date 
     from rownums r1 
     left join rownums r2 on r1.clt_nbr=r2.clt_nbr and r1.rnum=r2.rnum-1 
     ) 
update cte 
set end_date=new_end_date

我正在使用sql server 2008 – Ice

答

基础上加入和将相关数据的解决方案。现在我经常使用CTE，但在它们很受欢迎之前，我们不得不考虑使用这样的东西。

select * 
into ##test1 
from 
(
select ID = 1, CLT_NBR = 12375, IS_PRIMARY = 1, EFFECT_DATE = cast('8/13/2015' as date), END_DATE = cast('9/30/2015' as date) 
union all select ID = 2, CLT_NBR = 12375, IS_PRIMARY = 1, EFFECT_DATE = cast('10/1/2015' as date), END_DATE = cast('12/31/9998' as date) 
union all select ID = 3, CLT_NBR = 12375, IS_PRIMARY = 1, EFFECT_DATE = cast('7/18/2017' as date), END_DATE = cast('12/31/9998' as date) 
union all select ID = 4, CLT_NBR = 12331, IS_PRIMARY = 1, EFFECT_DATE = cast('2/3/2016' as date), END_DATE = cast('7/8/2016' as date) 
union all select ID = 5, CLT_NBR = 12331, IS_PRIMARY = 1, EFFECT_DATE = cast('7/9/2016' as date), END_DATE = cast('12/31/9998' as date) 
) x 

select * from ##test1 

select t.ID, t.CLT_NBR, t.IS_PRIMARY, t.EFFECT_DATE, END_DATE = isnull(dateadd(day,-1,min(t_next.EFFECT_DATE)),'9998-12-31') 
from ##test1 t 
left join ##test1 t_next on t_next.CLT_NBR = t.CLT_NBR and t_next.effect_date > t.effect_date 
group by t.ID, t.CLT_NBR, t.IS_PRIMARY, t.EFFECT_DATE 

update t 
set END_DATE = helper.END_DATE 
from ##test1 t 
left join 
(
    select t.ID, t.CLT_NBR, t.IS_PRIMARY, t.EFFECT_DATE, END_DATE = isnull(dateadd(day,-1,min(t_next.EFFECT_DATE)),'9998-12-31') 
    from ##test1 t 
    left join ##test1 t_next on t_next.CLT_NBR = t.CLT_NBR and t_next.effect_date > t.effect_date 
    group by t.ID, t.CLT_NBR, t.IS_PRIMARY, t.EFFECT_DATE 
) helper on helper.id = t.id 

select * from ##test1 

drop table ##test1

答

您可以使用查询类似下面的SQL 2008兼容的答案

CREATE TABLE yourTbl (ID int,CLT_NBR int,IS_PRIMARY int,EFFECT_DATE date, END_DATE date) 
INSERT INTO yourTbl VALUES 
(1 ,12375,1,'8/13/2015','9/30/2015') 
,(2 ,12375,1,'10/1/2015','12/31/9998') 
,(3 ,12375,1,'7/18/2017','12/31/9998') 
,(4 ,12331,1, '2/3/2016','7/8/2016') 
,(5 ,12331,1, '7/9/2016','12/31/9998') 




;with y as 
(
SELECT 
    *, 
    ROW_NUMBER() OVER(PARTITION BY CLT_NBR ORDER BY EFFECT_DATE DESC) R 
FROM yourTbl WHERE END_DATE ='12/31/9998' 
) 


UPDATE y1 
SET y1.END_DATE= y2.EFFECT_DATE 
FROM 
y y1 JOIN y y2 
ON y1.R=y2.R+1 and y1.CLT_NBR=y2.CLT_NBR 


select * from yourTbl

See working demo

SQL Server 2008：如何更新以前记录的结束日期列

相关推荐