sql-server – 可以根据多个列删除数据库副本吗？

我 asked this question a while back删除基于列的重复记录.答案很棒：

delete from tbl
where id NOT in
(
select  min(id)
from tbl
group by sourceid
)

我现在有一个模拟的情况,但重复记录的定义是基于多个列.如何更改此上述SQL以识别重复记录,其中唯一记录定义为从Col1 Col2 Col3连接.我会做这样的事情吗？

delete from tbl
where id NOT in
(
select  min(id)
from tbl
group by col1,col2,col3
)

解决方法

这显示您要保留的行：

;WITH x AS 
(
  SELECT col1,col3,rn = ROW_NUMBER() OVER 
      (PARTITION BY col1,col3 ORDER BY id)
  FROM dbo.tbl
)
SELECT col1,col3 FROM x WHERE rn = 1;

这显示您要删除的行：

;WITH x AS 
(
  SELECT col1,col3 FROM x WHERE rn > 1;

一旦你高兴,上述两套是正确的,以下将实际删除它们：

;WITH x AS 
(
  SELECT col1,col3 ORDER BY id)
  FROM dbo.tbl
)
DELETE x WHERE rn > 1;

请注意,在所有三个查询中,前6行都相同,只有CTE之后的后续查询已更改.