Postgres：根据相关表中的条件获取总相关计数答案

【问题标题】：Postgres: Getting a total related count based on a condition from a related tablePostgres：根据相关表中的条件获取总相关计数
【发布时间】：2016-07-11 15:46:39
【问题描述】：

我的 sql-fu 并不强大，我确信我在尝试使其正常工作时遗漏了一些简单的东西。我有一组相当标准的表格：

users
-----
id
name


carts
-----
id
user_id
purchased_at


line_items
----------
id
cart_id
product_id


products
--------
id
permalink

如果该用户购买了特定产品，我想获取每个用户购买的购物车的总数。也就是说：如果他们购买的购物车中至少有一个具有特定永久链接的产品，我想要计算购买的购物车的总数，无论其内容如何。

购买的购物车的定义是当 carts.purchased_at 不为空时。

select
  u.id,
  count(c2.*) as purchased_carts

from users u

inner join carts c on u.id = c.user_id
inner join line_items li on c.id = li.cart_id
inner join products p on p.id = li.product_id 
left join carts c2 on u.id = c2.user_id

where 
  c.purchased_at is not NULL 
  and
  c2.purchased_at is not NULL 
  and
  p.permalink = 'product-name'
group by 1
order by 2 desc

purchase_carts 的数字异常高，可能与订单项总数乘以购物车数有关？也许？我对结果感到很困惑。任何帮助将不胜感激。

【问题讨论】：

标签： postgresql count aggregate

【解决方案1】：

这应该会有所帮助：

select u.id,
       count(*)
from   users u join
       carts c on c.user_id = u.id
where  c.purchased_at is not NULL and
       exists (
         select null
         from   carts      c2
         join   line_items l on l.cart_id = c2.id
         join   products   p on p.id      = l.product_id
         where  c2.user_id      = u.id and
                c2.purchased_at is not NULL 
                p.permalink     = 'product-name')
group by u.id
order by count(*) desc;

exists 谓词是一个半连接。

【讨论】：

【解决方案2】：

bool_or 是你所需要的

select
    u.id,
    count(distinct c.id) as purchased_carts
from
    users u
    inner join
    carts c on u.id = c.user_id
    inner join
    line_items li on c.id = li.cart_id
    inner join
    products p on p.id = li.product_id 
where c.purchased_at is not NULL 
group by u.id
having bool_or (p.permalink = 'product-name')
order by 2 desc

【讨论】：