1

在 SQL Server 2008 R2 中，我在表中添加了两个重复的 ID 和记录。当我尝试删除最后两条记录之一时，我收到以下错误。

更新或删除的行值要么不使行唯一，要么改变多行。

数据是：

7   ABC         6
7   ABC         6
7   ABC         6
8   XYZ         1
8   XYZ         1
8   XYZ         4
7   ABC         6
7   ABC         6

我需要删除最后两条记录：

7   ABC         6
7   ABC         6

我一直在尝试使用“编辑前 200 行”功能删除最后 2 条记录以删除此重复 ID，但出现上述错误。

任何帮助表示赞赏。提前致谢：）

4

2 回答 2

1

由于您不知道表中还有其他列，因此假设您的数据在 3 列 A、B、C 中，您可以使用以下方法删除 2 行：

;with t as (
    select top(2) *
      from tbl
     where A = 7 and B = 'ABC' and C = 6
)
DELETE t;

这将根据条件任意匹配两行，并删除它们。

于 2012-12-07T22:05:44.260 回答
0

这是我用来删除可能有很多副本的表中的副本的代码大纲。

/* I always put the rollback and commit up here in comments until I am sure I have 
   done what I wanted. */
BEGIN tran Jim1 -- rollback tran Jim1 -- Commit tran Jim1; DROP table PnLTest.dbo.What_Jim_Deleted

/* This creates a table to put the deleted rows in just in case I'm really screwed up */
SELECT top 1 *, NULL dupflag 
  INTO jt1.dbo.What_Jim_Deleted --DROP TABLE jt1.dbo.What_Jim_Deleted
  FROM jt1.dbo.tab1;
/* This removes the row without removing the table */
TRUNCATE TABLE jt1.dbo.What_Jim_Deleted;

/* the cte assigns a row number to each unique security for each day, dups will have a
   rownumber > 1.  The fields in the partition by are from the composite key for the 
   table (if one exists.  These are the queries that I ran to show them as dups

SELECT compkey1, compkey2, compkey3, compkey4, COUNT(*) 
  FROM jt1.dbo.tab1
  GROUP BY compkey1, compkey2, compkey3, compkey4
  HAVING COUNT(*) > 1
  ORDER BY 1 DESC


*/
with getthedups as
  (SELECT *,
       ROW_NUMBER() OVER 
         (partition by compkey1,compkey2, compkey3, compkey4 
                          ORDER BY Timestamp desc) dupflag /*This can be anything that gives some order to the rows (even if order doesn't matter) */
     FROM jt1.dbo.tab1)
/* This delete is deleting from the cte which cascades to the underlying table 
   The Where is part of the Delete (even though it comes after the OUTPUT.  The
   OUTPUT takes all of the DELETED row and inserts them into the "oh shit" table,
   just in case.*/
DELETE 
  FROM getthedups 
  OUTPUT DELETED.* INTO jti.dbo.What_Jim_Deleted
  WHERE dupflag > 1

--Check the resulting tables here to ensure that you did what you think you did

/* If all has gone well then commit the tran and drop the "oh shit" table, or let it 
   hang around for a while. */
于 2013-02-11T16:21:35.917 回答