Question

我有一个包含许多记录的表，其中有一个nic列，它应具有唯一值。但它包含一些污垢，意味着那里有一些重复的值。

与

用户一样

id = 567 has nic = '786-786'

以及

id = 897 has nic = '786-786'.

我想编写一个存储过程，应该删除那些副本并在最后添加1，如

id = 567 , nic= '786-786'
id = 788 , nic = '786-786-1'
id = 2344, nic = '786-786-2'

有可能吗？

Answer 1

使用窗口函数/解析函数，例如

Row_number() over (partition by Nic, order by ID)

SQL Fiddle 有点像...

UPDATE B
SET B.Nic = C.NewNic
FROM Foo B
INNER JOIN (SELECT ID, NIC + '-'  + cast(Row_number() 
over (partition by NIC order by ID ) as varchar(20)) as newNic
FROM FOo) C
 on C.ID = B.ID

这会给你

id = 567 , nic= '786-786-1'
id = 788 , nic = '786-786-2'
id = 2344, nic = '786-786-3'

你总是可以做一个case语句，当row为1然后不显示，如果必须离开第一个没有-1，则减去1

SQL Fiddle getting rid of the -1 for first record...

UPDATE B
SET B.Nic = Case WHEN C.RN = 1 then B.Nic Else B.Nic+ '-'+cast((RN-1) as varchar(20)) end
FROM Foo B
INNER JOIN (SELECT ID, NIC, cast(Row_number() 
over (partition by NIC order by ID ) as varchar(20)) as RN
FROM FOo) C
 on C.ID = B.ID

导致：

ID  NIC
567 786-786
897 786-786-1

和SQL Fiddle显示多个网站

ID  NIC
567 786-786
600 786-786-1
897 786-786-2
10  786-787
12  786-787-1
900 786-787-2

Answer 2

for tsql

您可以通过简单的更新来完成此操作，请尝试以下查询：

select id, nic, rank() over (partition by nic order by id)
from your_table as t1
where exists(select 1 from your_table where nic = t1.nic and id <> t1.id)

您可以构建更新查询，或将更新字符串作为此选择的字段返回。阅读rank()

更正SQL Server数据库中的重复记录

2 个答案: