昵称： zjyss

原文地址：https://www.cnblogs.com/zjyss/p/15701439.html

1.可以循环表取出相同字段的第一条去建立临时表或视图

2.使用pg的row_number 函数对相同字段记录分组排序，取出排序分组记录中的第一个。

下例即取出查询结果集合中产品对应date最新的那一条数据集合，相当于根据product_id去重，保留date最大的一条

select s.product_id, s.price
from (
    select *, row_number() over (partition by ttt.product_id order by ttt.date desc) as group_idx
    from (
         select distinct (kpol.product_id),kpol.price, sm.date from
            kthrp_purchase_order_line kpol
            left join stock_move sm on split_part(sm.ref, ',', 1) = 'kthrp.purchase.order.line' and split_part(sm.ref, ',', 2)::INTEGER = kpol.id
            left join product_product pp on pp.id = kpol.product_id
         where sm.date is not null
            order by sm.date desc
             ) ttt
) s
where s.group_idx = 1;

可以简化为：

select s.field_name_s, s.field_name_x
from (
    select *, row_number() over (partition by ttt.field_name_a order by ttt.field_name_b desc) as group_idx
    from 子查询 ttt
) s
where s.group_idx = 1;

1.row_number() 为返回的记录定义各行编号

2.pritition by 分组

3.order by 排序