Question

以下是我的表格（仅包括相关栏目）

Table: carts
address_id - integer

Table: addresses
name - varchar
phone - varchar

Table: orders
order_number - integer (this is the foreign key for cart table)

我想获取仅订购一次的客户的电话号码，因此我构建了以下查询

select addresses.phone 
from orders 
   inner join carts on orders.order_number = carts.id 
   inner join address on carts.address_id = addresses.id 
group by addresses.phone 
having count(orders.*) = 1;

这很棒！但我还需要选择客户名称和订单号，我将select语句更新为

select addresses.phone, addresses.name, orders.order_number ...

现在，postgres敦促我在GROUP BY子句中包含这些列，但这不会给我带来预期的结果。

我尝试使用子查询如下，这似乎得到了我想要的结果

select addresses.phone, (select ad.name from addresses ad where ad.phone = addresses.phone) ...

但是使用子查询是解决这个问题的唯一方法吗？或者有更简单/最佳的方式吗？

Answer 1

您可以使用window function来实现此目的，不需要对所有内容进行分组：

select *
from (
  select addresses.phone, addresses.name, orders.order_number, 
         count(orders.order_number) over (partition by addresses.phone) as cnt
  from orders 
     inner join carts on orders.order_number = carts.id 
     inner join address on carts.address_id = addresses.id 
) t 
where cnt = 1;

不要在group by中包含选择列

1 个答案: