在 SQL 中删除重复组

Question

在SQL服务器中，查询后得到如下数据集。我需要按 uid 对这些数据进行分组，然后删除“重复”和 return 与最新列 D 的分组。我还需要 return 2 个最近的重复数据删除组。如果满足以下条件，则组被定义为重复组：

行数相同
A、B、C 列相同
行的顺序相同

Uid	A	B	C	D1	D2
1	6	1	2	2021-02-19	2021-02-19 09:00:00
1	6	2	1	2021-02-19	2021-02-19 10:00:00
1	6	1	2	2021-02-19	2021-02-19 11:00:00
2	6	1	2	2021-01-19	2021-01-19 09:00:00
2	6	2	1	2021-01-19	2021-01-19 10:00:00
3	6	1	2	2020-02-19	2020-02-19 09:00:00
3	6	2	1	2020-02-19	2020-02-19 10:00:00
3	6	1	2	2020-02-19	2020-02-19 11:00:00
4	11	4	5	2000-10-05	2000-10-05 09:00:00

比如上面的数据集中，uid的1和3是重复的，1是最新的。因此上面的数据集应该return

Uid	A	B	C	D1	D2
1	6	1	2	2021-02-19	2021-02-19 09:00:00
1	6	2	1	2021-02-19	2021-02-19 10:00:00
1	6	1	2	2021-02-19	2021-02-19 11:00:00
2	6	1	2	2021-01-19	2021-01-19 09:00:00
2	6	2	1	2021-01-19	2021-01-19 10:00:00

我尝试了以下 window 函数：

FROM (

  SELECT 
    A,
    B,
    C,
    D1,
    D2,
    ROW_NUMBER() over (partition by 
                                    Uid                                    
                                    ORDER BY D2 DESC) as rn
....

where rn = 1

但这不允许我按 uid 分组。我怎样才能做到这一点？

Answer 1

在下面的查询中，子查询 temp 为 A、B、C 创建一个逗号分隔的列。然后我在 A、B、C 上使用分区，并根据子查询 temp2 级别的日期获得排名。外部的最终查询仅提取排名 1 并显示 table.

的输出

select t.Uid, t.A,t.B, t.C , t.D1, t.D2
from (
                    select Uid, A,B,C, D1, rank() over ( partition by A, B,C order by D1 desc) as rank
                    from 
                        (SELECT Uid,  
                    A=STUFF  
                    (  
                         (  
                           SELECT  ', ' + CAST(A AS VARCHAR(MAX))  
                           FROM Table1 t2   
                           WHERE t2.Uid = t1.Uid   
                           FOR XML PATH('')  
                         ),1,1,''  
                    )  ,
                    B=STUFF  
                    (  
                         (              
                           SELECT  ', ' + CAST(B AS VARCHAR(MAX))  
                           FROM Table1 t2   
                           WHERE t2.Uid = t1.Uid   
                           FOR XML PATH('')  
                         ),1,1,''  
                    )  ,
                    C=STUFF  
                    (  
                         (  
                           SELECT  ', ' + CAST(C AS VARCHAR(MAX))  
                           FROM Table1 t2   
                           WHERE t2.Uid = t1.Uid   
                           FOR XML PATH('')  
                         ),1,1,''  
                    )  ,
                    cast (max( [D1] ) as date) D1
                    FROM Table1 t1  
                    GROUP BY Uid  ) as temp 
                    ) as temp2
join Table1 t on temp2.Uid = t.Uid
and temp2. D1= t.D1
where temp2.rank = 1

这是数据库 Fiddle link : https://dbfiddle.uk/?rdbms=sqlserver_2017&fiddle=de2127330c2e60d3733bfc9548504142

在 SQL 中删除重复组

In SQL remove duplicate groups

sql

tsql

sql-server

window-functions