PostgreSQL与SUM 2/4表的左连接

我试图用4个表中的2个表进行左连接，但我被卡住了

在下面底部的预期结果中，有一个名为begin_stock和end_stockresult of daily_stock.qty的新字段，该字段按日期_stat进行过滤，但不需要使用总和，因为这些值是最终值

您可以在这里尝试：

以下是表格：

产品：

用法：需要使用sum

id  product_id  used    date_out
1   1           10      2020-12-18
2   1           20      2020-12-19
3   2           20      2020-12-18

id  product_id  restock_amount  date_in
1   1           10              2020-12-18
2   1           20              2020-12-19
3   4           10              2020-12-19

产品重新进货：需要使用sum

id  product_id  used    date_out
1   1           10      2020-12-18
2   1           20      2020-12-19
3   2           20      2020-12-18

id  product_id  restock_amount  date_in
1   1           10              2020-12-18
2   1           20              2020-12-19
3   4           10              2020-12-19

每日库存

2018年12月2020日至2019年12月2020日的预期结果：

---

可以使用聚合和横向联接：

select p.*, ds.*, pr.*, u.*
from product p
cross join lateral (
    select 
        max(case when ds.date_stat = date '2020-12-18' then qty end) as begin_stock,
        max(case when ds.date_stat = date '2020-12-19' then qty end) as end_stock
    from daily_stock ds
    where 
        ds.product_id = p.id
        and ds.date_stat in (date '2020-12-18', date '2020-12-19')
) ds
cross join lateral (
    select coalesce(sum(u.used), 0) as used
    from usage u
    where 
        u.product_id = p.id 
        and u.date_out >= date '2020-12-18'
        and u.date_out <= date '2020-12-19'
) u
cross join lateral (
    select coalesce(sum(pr.restock_amount), 0) as restock
    from product_restock pr
    where 
        pr.product_id = p.id 
        and pr.date_in >= date '2020-12-18'
        and pr.date_in <= date '2020-12-19'     
) pr

:

---

您能显示您迄今为止尝试过的查询吗？请在加入之前聚合每个表。。。从p LEFT JOIN SELECT id，SUMx AS x FROM y GROUP BY id AS z ON p.id=z.idhello，感谢您的回答，但仍然发现了一个错误：@Diand:I修复了查询-并测试了id.hmm，是否有办法将结束库存放在表的末尾？看起来begin_stock和end_stock是不能分开的。@Diand：这只是对外部select子句的一个小改动：select p.*，ds.begin_stock，pr.*，u.*，ds.end_stock from…嗯，无论如何，谢谢你，伙计！祝您有个美好的一天！您可以使用windws函数。因此，您可以获得begin\u stock和end\u stock

select p.*, ds.*, pr.*, u.*
from product p
cross join lateral (
    select 
        max(case when ds.date_stat = date '2020-12-18' then qty end) as begin_stock,
        max(case when ds.date_stat = date '2020-12-19' then qty end) as end_stock
    from daily_stock ds
    where 
        ds.product_id = p.id
        and ds.date_stat in (date '2020-12-18', date '2020-12-19')
) ds
cross join lateral (
    select coalesce(sum(u.used), 0) as used
    from usage u
    where 
        u.product_id = p.id 
        and u.date_out >= date '2020-12-18'
        and u.date_out <= date '2020-12-19'
) u
cross join lateral (
    select coalesce(sum(pr.restock_amount), 0) as restock
    from product_restock pr
    where 
        pr.product_id = p.id 
        and pr.date_in >= date '2020-12-18'
        and pr.date_in <= date '2020-12-19'     
) pr

id | product_name | begin_stock | end_stock | restock | used -: | :----------- | ----------: | --------: | ------: | ---: 1 | abc | 10 | 10 | 30 | 30 2 | aaa | 10 | -10 | 0 | 20 3 | bbb | 10 | 10 | 0 | 0 4 | ddd | 10 | 20 | 10 | 0

    with start_end as
(
    select ds.product_id,ds.date_stat,ds.qty
    ,row_number() over(partition by ds.product_id order by ds.date_stat asc) start_dt
    ,row_number() over(partition by ds.product_id order by ds.date_stat desc) end_dt
    from daily_stock ds
)
,sum_of_restock_amount as
(
    select 
    pr.product_id ,sum(pr.restock_amount) restock
    from product_restock pr
    group by pr.product_id  
)
,sum_of_usage as
(
    select 
    u.product_id ,sum(u.used) used
    from usage u
    group by u.product_id
)
select pn.*,st.qty,res.restock,us.used,en.qty
from product  pn
left join start_end st
on pn.id = st.product_id
    and st.start_dt = 1
    and st.date_stat between '2020-12-18' and '2020-12-19'
left join sum_of_restock_amount res
on pn.id = res.product_id
left join sum_of_usage us
on pn.id = us.product_id
left join start_end en
on pn.id = en.product_id
    and en.end_dt = 1
    and en.date_stat between '2020-12-18' and '2020-12-19'