leetcode 1082典型题 ，开窗函数

 

示例零： 

SELECT seller_id , RANK() OVER( ORDER BY SUM(price) DESC) AS r
,SUM(price) 
FROM Sales 

结果： 返回一条数据 ， 没有partition 关键字，等于是全表做order by ， 基于全表 做sum 排序 ， 
其实这个sql 有问题. 

| seller_id | rk | SUM(price) |
| --------- | -- | ---------- |
| 1         | 1  | 6400       |

　　

示例 一： 

 SELECT seller_id, RANK() OVER( partition by seller_id    ORDER BY SUM(price) DESC) rk 
 ,SUM(price)
    FROM Sales

结果也是返回一条数据

 | seller_id | rk | SUM(price) |
| --------- | -- | ---------- |
| 1         | 1  | 6400       |

 

 

示例二： 

 SELECT seller_id, RANK() OVER(    ORDER BY SUM(price) DESC) rk 
 ,SUM(price)
    FROM Sales
GROUP BY seller_id
结果： 如上sql会返回多条数据, 基于  seller_id 做分组了。 



| seller_id | rk | SUM(price) |
| --------- | -- | ---------- |
| 1         | 1  | 2800       |
| 3         | 1  | 2800       |
| 2         | 3  | 800        |

示例三：

 SELECT seller_id, RANK() OVER( partition by seller_id    ORDER BY SUM(price) DESC) rk 
 ,SUM(price)
    FROM Sales
GROUP BY seller_id

| seller_id | rk | SUM(price) |
| --------- | -- | ---------- |
| 1         | 1  | 2800       |
| 2         | 1  | 800        |
| 3         | 1  | 2800       |

 

 

 

 

 

产品表：Product

+--------------+---------+
| Column Name  | Type    |
+--------------+---------+
| product_id   | int     |
| product_name | varchar |
| unit_price   | int     |
+--------------+---------+
product_id 是这个表的主键(具有唯一值的列)。
该表的每一行显示每个产品的名称和价格。
销售表：Sales

+-------------+---------+
| Column Name | Type    |
+-------------+---------+
| seller_id   | int     |
| product_id  | int     |
| buyer_id    | int     |
| sale_date   | date    |
| quantity    | int     |
| price       | int     |
+------ ------+---------+
这个表它可以有重复的行。 
product_id 是 Product 表的外键(reference 列)。
该表的每一行包含关于一个销售的一些信息。
 

编写解决方案，找出总销售额最高的销售者，如果有并列的，就都展示出来。

以 任意顺序 返回结果表。

返回结果格式如下所示。

 

示例 1:

输入：
Product 表：
+------------+--------------+------------+
| product_id | product_name | unit_price |
+------------+--------------+------------+
| 1          | S8           | 1000       |
| 2          | G4           | 800        |
| 3          | iPhone       | 1400       |
+------------+--------------+------------+
Sales 表：
+-----------+------------+----------+------------+----------+-------+
| seller_id | product_id | buyer_id | sale_date  | quantity | price |
+-----------+------------+----------+------------+----------+-------+
| 1         | 1          | 1        | 2019-01-21 | 2        | 2000  |
| 1         | 2          | 2        | 2019-02-17 | 1        | 800   |
| 2         | 2          | 3        | 2019-06-02 | 1        | 800   |
| 3         | 3          | 4        | 2019-05-13 | 2        | 2800  |
+-----------+------------+----------+------------+----------+-------+
输出：
+-------------+
| seller_id   |
+-------------+
| 1           |
| 3           |
+-------------+
解释：Id 为 1 和 3 的销售者，销售总金额都为最高的 2800。

 

如下是新案例.

案例1 .如下sql含义 : count(1) ,基于groupby 进行汇总求和, 然后排序 ,另外是全局排序.

select project_id,  count(1) , dense_rank() over (order by count(*) desc) as rk from Project group by project_id 





Project =
| project_id | employee_id |
| ---------- | ----------- |
| 1          | 1           |
| 1          | 2           |
| 1          | 3           |
| 2          | 1           |
| 2          | 4           |
 
输出
| project_id | count(1) | rk |
| ---------- | -------- | -- |
| 1          | 3        | 1  |
| 2          | 2        | 2  |

案例2 .
select project_id,  count(1) , dense_rank() over ( partition by project_id order by count(*) desc) as rk from Project group by project_id 

这个结果是区内排序, rk 那地方会有区别， 不加partition by 是全局排序.

案例3.
select project_id,  count(1) , dense_rank() over (  order by count(*) desc) as rk from Project  
不加 group by 等于是全局 count(1) ,只会返回一条结果， 没有意义的.