T-Sql 按 MIN() 选择和分组答案

【问题标题】：T-Sql select and group by MIN()T-Sql 按 MIN() 选择和分组
【发布时间】：2016-07-31 18:27:34
【问题描述】：

我有 3 个表格，例如：

ProductCategory [1 - m] Product [1-m] ProductPrice

这样的简单脚本：

select pc.CategoryId ,pp.LanguageId , pp.ProductId ,pp.Price
from ProductCategory as pc
    inner join Product as p on pc.ProductId = p.Id
    inner join ProductPrice as pp on p.Id = pp.ProductId
order by CategoryId , LanguageId , ProductId

显示这些表格数据：

CategoryId  LanguageId  ProductId   Price
----------- ----------- ----------- ---------------------------------------
1           1           1           55.00
1           1           2           55.00
1           2           1           66.00
1           2           2           42.00
2           1           3           76.00
2           1           4           32.00
2           2           3           89.00
2           2           4           65.00
4           1           4           32.00
4           1           5           77.00
4           2           4           65.00
4           2           5           85.00

现在我需要的是：对于每个类别，按原样获取整行，但仅限于具有最低价格的产品。

我刚刚写了一个简单的查询，如下所示：

with dbData as
(
select pc.CategoryId ,pp.LanguageId , pp.ProductId ,pp.Price
from ProductCategory as pc
    inner join Product as p on pc.ProductId = p.Id
    inner join ProductPrice as pp on p.Id = pp.ProductId
)
select distinct db1.*
from dbData as db1
inner join dbData as db2 on db1.CategoryId = db2.CategoryId
where db1.LanguageId = db2.LanguageId
    and db1.Price = (select Min(Price) 
                        from dbData 
                        where CategoryId = db2.CategoryId
                                and LanguageId = db2.LanguageId)

其结果是正确的：

CategoryId  LanguageId  ProductId   Price
----------- ----------- ----------- ---------------------------------------
1           1           1           55.00
1           1           2           55.00
1           2           2           42.00
2           1           4           32.00
2           2           4           65.00
4           1           4           32.00
4           2           4           65.00

有更酷的方法吗？

注意：查询必须符合Sql-Server 2008 R2+

【问题讨论】：

标签： sql sql-server sql-server-2008 tsql greatest-n-per-group

【解决方案1】：

你可以使用像RANK()这样的窗口函数：

WITH cte AS 
(
  select pc.CategoryId, pp.LanguageId, pp.ProductId, pp.Price,
    rnk = RANK() OVER(PARTITION BY pc.CategoryId ,pp.LanguageId ORDER BY pp.Price) 
  from ProductCategory as pc
  join Product as p on pc.ProductId = p.Id
  join ProductPrice as pp on p.Id = pp.ProductId
)
SELECT CategoryId, LanguageId, ProductId, Price
FROM cte
WHERE rnk = 1;

LiveDemo

【讨论】：

【解决方案2】：

如果您需要每个 categoryid 和 languageid 的产品价格，您可以将 languageid 添加到分区中

select top 1 with ties pc.CategoryId ,pp.LanguageId , pp.ProductId ,pp.Price
    from ProductCategory as pc
    inner join Product as p on pc.ProductId = p.Id
    inner join ProductPrice as pp on p.Id = pp.ProductId
        order by row_number() over  (partition by pc.categoryid order by price)

【讨论】：

【解决方案3】：

您没有在查询中使用Product 表，因此似乎没有必要。我会这样写：

select ppc.*
from (select pc.CategoryId, pp.LanguageId , pp.ProductId, pp.Price,
             row_number() over (partition by pc.CategoryId order by pp.Price) as seqnum
      from ProductCategory pc inner join
           ProductPrice pp
           on pc.ProductId = pp.ProductId
     ) ppc
where seqnum = 1
order by CategoryId, LanguageId, ProductId;

【讨论】：