SQL - 仅当另一列匹配时才删除重复列[重复]答案

【问题标题】：SQL - Deleting duplicate columns only if another column matches [duplicate]SQL - 仅当另一列匹配时才删除重复列[重复]
【发布时间】：2019-01-19 22:32:11
【问题描述】：

我的下表 (TBL_VIDEO) 在“TIMESTAMP”中有重复的列条目，我想仅在“CAMERA”编号匹配时删除它们。

之前：

ANALYSIS_ID | TIMESTAMP | EMOTION | CAMERA
-------------------------------------------    
 1          | 5         | HAPPY   | 1
 2          | 10        | SAD     | 1
 3          | 10        | SAD     | 1
 4          | 5         | HAPPY   | 2
 5          | 15        | ANGRY   | 2
 6          | 15        | HAPPY   | 2

之后：

ANALYSIS_ID | TIMESTAMP | EMOTION | CAMERA
-------------------------------------------    
 1          | 5         | HAPPY   | 1
 2          | 10        | SAD     | 1
 4          | 5         | HAPPY   | 2
 5          | 15        | ANGRY   | 2

我已尝试此语句，但列不会相应地删除。我感谢所有帮助生成正确的 SQL 语句。提前致谢！

delete y
from TBL_VIDEO y 
where exists (select 1 from TBL_VIDEO y2 where y.TIMESTAMP = y2.TIMESTAMP and y2.ANALYSIS_ID < y.ANALYSIS_ID, y.CAMERA = y.CAMERA, y2.CAMERA = y2.CAMERA);

【问题讨论】：

标签： sql sql-server

【解决方案1】：

试试这个：

delete f2 from (
select row_number() over(partition by TIMESTAMP, CAMERA order by ANALYSIS_ID) rang
from yourtable f1
) f2 where f2.rang>1

【讨论】：

+1 这行得通。但只是一个小评论。如果您想将 delete f2 替换为 select * 以测试将删除的内容，则在子查询中添加字段非常方便。但这些实际上并不是删除所必需的。仅rang 就足够了。
没错，我删除了它；）

【解决方案2】：

其他解决方案：

delete f1 from yourtable f1
where exists 
(
  select * from yourtable f2
  where f2.TIMESTAMP=f1.TIMESTAMP and f2.CAMERA=f1.CAMERA and f1.ANALYSIS_ID>f2.ANALYSIS_ID
)

【讨论】：

这行得通。谢谢！
不客气 :)

【解决方案3】：

使用row_number 找到重复的并删除它们

delete from
(select *,row_number() over(partition by TIMESTAMP,CAMERA order by ANALYSIS_ID) as rn from TBL_VIDEO
) t1 where rn>1

【讨论】：

【解决方案4】：

;WITH cte
AS
(
    select ANALYSIS_ID,
    ROW_NUMBER() over(partition by TIMESTAMP, CAMERA order by ANALYSIS_ID) rnk
)

DELETE FROM cte WHERE cte.rnk > 1

【讨论】：

【解决方案5】：

你可以使用subquery：

select v.*
from tbl_video v
where analysis_id = (select min(v1.analysis_id)
                     from tbl_video v1
                     where v1.timestamp = v.timestamp and
                           v1.camera = v.camera 
                    );

不过，带有top (1) with ties子句的解析函数也很有用：

select top (1) with ties v.*
from tbl_video v
order by row_number() over (partition by v.timestamp, v.camera order by v.analysis_id);

所以，您的delete 版本将是：

delete v    
from tbl_video v
where analysis_id = (select min(v1.analysis_id)
                     from tbl_video v1
                     where v1.timestamp = v.timestamp and
                           v1.camera = v.camera 
                    );

【讨论】：