需要使用row_number()过滤3列

Question

我的数据库中有这样的数据

---------------------------------------------------------------
shopid | shopname | keyword | date     | filename
---------------------------------------------------------------
1      | shop1    | A_key       |2020-04-24| A     
---------------------------------------------------------------
1      | shop1    | A_key       |2020-04-24| B
---------------------------------------------------------------
1      | shop1    | B_key       |2020-04-25| B     
---------------------------------------------------------------
1      | shop1    | B_key       |2020-04-25| C     
---------------------------------------------------------------
2      | shop2    | C_key       |2020-04-24| A     
---------------------------------------------------------------
2      | shop2    | C_key       |2020-04-24| B
---------------------------------------------------------------
3      | shop3    | D_key       |2020-04-24| A    
---------------------------------------------------------------
3      | shop3    | D_key       |2020-04-24| B    
---------------------------------------------------------------
3      | shop3    | E_key       |2020-04-24| A 
---------------------------------------------------------------
3      | shop3    | E_key       |2020-04-25| B
---------------------------------------------------------------

我想要的结果是来自最新文件名的最新数据

---------------------------------------------------------------
shopid | shopname | keyword | date     | filename
---------------------------------------------------------------
1      | shop1    | A_key       |2020-04-24| B
---------------------------------------------------------------
1      | shop1    | B_key       |2020-04-25| C     
---------------------------------------------------------------
2      | shop2    | C_key       |2020-04-24| B
---------------------------------------------------------------
3      | shop3    | D_key       |2020-04-24| B    
---------------------------------------------------------------
3      | shop3    | E_key       |2020-04-24| A 
---------------------------------------------------------------
3      | shop3    | E_key       |2020-04-25| B
---------------------------------------------------------------

我想知道如何使用sql我的sql是

select *,rank() over (partition by date,keyword order by filename desc) rank
from table
 order by date asc;

Answer 1

我想要的结果是最新文件名中的最新数据

我不完全理解这一点。下面返回每个商店的最新日期行：

select t.*
from (select t.*,
             rank() over (partition by shopid order by date desc) as seqnum
      from table t
     ) t
where seqnum = 1;

Answer 2

看起来您需要为每个商店/关键字/日期占一行，如果有多个行满足此要求，则使用文件名来决定哪个是最新的(C晚于B等)。

你可以使用分析器来分析它：

SELECT 
  x.*
FROM 
  (
    SELECT
      *,
      row_number() over (partition by shopid,keyword,date order by filename desc) as rn
    FROM 
      table 
  ) x
WHERE 
  x.rn = 1

但我敢说，您也可以使用标准分组：

SELECT
  shopid, shopname, keyword, date, MAX(filename) as filename
FROM 
  table 
GROUP BY 
  shopid, shopname, keyword, date

如果您想要开始在中添加其他detial列，此方法将失败，但据我从您的Q中可以看出，没有任何