group_owner

group_id  |   user_id
--------- | ----------
  51      |     1
  51      |     2
  51      |     3
  52      |     2
  53      |     4

用户

user_id   |   username
--------- | -----------
   1      |    John
   2      |    John
   3      |    John
   4      |    Peter

user_id的{{1}}与group_owner的{{1}}具有外键关系。

我正从users表中删除重复项。如果有重复项，我需要让用户保持最低ID，并删除其余部分。我创建了一个临时表，如下所示。

user_id

我需要将users替换为SELECT t2.id AS Old_id, t1.MinID AS New_id INTO #Translation FROM (SELECT MIN(id) AS MinID, username FROM users GROUP BY username) t1 INNER JOIN users t2 ON t1.username = t2.username表中的Old_ids。 New_ids对group_owner和group_owner有约束力。

group_id

如果该组合已经存在，它将会爆炸。因此，执行以下操作将无效。

user_id

如果该组合不存在，有人可以帮我更新group_owner吗？如果该组合存在，可以删除该组合吗？

以下是group_owner和users表在处理后应该如下：

group_owner

ALTER TABLE [dbo].[group_owner] ADD  CONSTRAINT [pk_group_owner] PRIMARY KEY CLUSTERED 
(
    [group_id] ASC,
    [user_id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, IGNORE_DUP_KEY = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON)
GO

用户

UPDATE group_owner 
  SET group_owner.[user_id] = New_id
  FROM group_owner 
    INNER JOIN #Translation T
      ON group_owner.[user_id] = T.Old_id
  WHERE T.Old_id <> T.New_id

参考：http://weblogs.sqlteam.com/jeffs/archive/2004/10/07/2190.aspx

Answer 1

尝试以下步骤。

从group_owner表中删除不需要的记录。

正如你所说的group_id和user_id组合是唯一的，我们只需要一个group_id = 51和user_id = 1的条目，所以我们可以从group_owner中删除（51,2）和（51,3）组合。

脚本：

;with cte_1
    as
    (SELECT Username,[user_id],COUNT([user_id])OVER(PARTITION BY Username Order by [Username]) as CNT
    FROM users  )
    ,cte_2
    as
    (SELECT Username,a.[user_id],b.group_id  , ROW_NUMBER()OVER(PARTITION BY b.group_id Order by [user_id]) as Rno
    FROM cte_1 a
    JOIN group_owner b on a.[user_id]=b.[user_id]
    WHERE CNT>1)
    DELETE c3
    FROM cte_2 c2
      JOIN group_owner c3 on c2.group_id=c3.group_id and c2.[user_id]=c3.[user_id]
    WHERE Rno>1

下一步是使用user table中的Minimum user_id更新group_owner表中的[user_id]。

脚本：

;with cte_1
as
(SELECT Username,[user_id]
 ,COUNT([user_id])OVER(PARTITION BY Username Order by [Username]) as CNT
 ,MIN([user_id])OVER(PARTITION BY Username Order by [user_id]) as [MINUserID]
FROM users  )
UPDATE b
  SET b.user_id=a.[MINUserID]
FROM cte_1 a
 JOIN group_owner b on a.[user_id]=b.[user_id]
 WHERE CNT>1

第三步也是最后一步是通过保持用户ID的最小值来从User表中删除重复的条目。

脚本：

;with cte_1
    as
    (SELECT Username,[user_id],ROW_NUMBER()OVER(PARTITION BY Username Order by [user_id]) as RNO
    FROM users  )
    DELETE 
    FROM cte_1
    WHERE RNO>1

如果不存在2列的组合则更新

group_owner

用户

group_owner

用户

1 个答案: