基于父表的Sql总结

如何解决基于父表的Sql总结

我的sql看起来像：

create table ad(
  ad_id int,ad_name varchar(10)
);


insert into ad(ad_id,ad_name) values
  (1,'ad1'),(2,'ad2'),(3,'ad3');
 
 
 create table ad_insight(
   id int,ad_id int,date date,clicks int
 );


insert into ad_insight(id,ad_id,date,clicks) values
    (1,1,'2021-04-25',1),'2021-04-24',4),'2021-04-23',2),(4,2,6),(5,'2021-03-03',7);


create table product(
    product_id int,product_name varchar(10)
);

insert into product(product_id,product_name) values
    (1,'prod1'),'prod2'),'prod3'),'prod4');
    (1,3,'prod1');
    
create table product_insight(
    id int,product_id int,sale int,date date
);
insert into product_insight(id,product_id,sale,date) values
    (1,12,'2021-04-25'),11,'2021-04-24'),13,'2021-04-23'),14,'2021-04-22'),17,'2021-04-21'),(6,15,'2021-04-20'),(7,'2021-04-19'),(8,19,'2021-04-25');

这里有fiddle

模式的快速解释： 我有广告：

1. 每个广告都有洞察力，可以告诉我们某个广告何时处于活动状态。
2. 每个广告都有产品。每个产品都有 product_insight，它告诉我们该产品在某一天产生了多少销售额。

现在我想得到以下表格：

1. 将汇总 ad_insight 表中的点击次数，并汇总 product_insight 至 2021-04-23 中 2021-04-25 的 product_sale。

+----------+--------+--------------+--------------+
| ad_name  | clicks | product_sale |   products   |
+----------+--------+--------------+--------------+
| ad1      |      7 |           55 | prod1,prod2 |
| ad2      |      6 |            0 | prod3,prod4 |
| ad3      |      0 |           36 | prod1        |
+----------+--------+--------------+--------------+

2. 汇总行将汇总上表中的所有内容：

+------------+--------------+--------------------+----------------------------+
| total_ads  | total_clicks | total_product_sale |    unique_all_products     |
+------------+--------------+--------------------+----------------------------+
|          3 |            13|                 91 | prod1,prod2,prod3,prod4 |
+------------+--------------+--------------------+----------------------------+

我尝试了什么？

# 1) table
select ad_name,SUM(ad_insight.clicks) as clicks 
from ad 
left join ad_insight on ad.ad_id = ad_insight.ad_id 
where ad_insight.date >= '2021-04-23' and ad_insight.date <= '2021-04-25'
group by ad.ad_id;

# 2) table
select count(distinct ad_insight.ad_id) as total,SUM(ad_insight.clicks) as clicks
from ad_insight
left join ad on ad.ad_id = ad_insight.ad_id
where ad_insight.date >= '2021-04-23' and ad_insight.date <= '2021-04-25'

但是不知道怎么select product_sale 表和用逗号分隔的产品！

解决方法

如果我理解正确的话，您希望为每个广告汇总两个不同的维度（点击次数和销售额）。加入前汇总：

select ad.ad_name,ai.clicks,p.sales,p.products
from ad left join
     (select ai.ad_id,sum(ai.clicks) as clicks
      from ad_insight ai
      where ai.date >= '2021-04-23' and ai.date <= '2021-04-25'
      group by ai.ad_id
     ) ai
     on ad.ad_id = ai.ad_id left join
     (select p.ad_id,sum(pi.sale) as sales,group_concat(distinct p.product_name) as products
      from product p join
           product_insight pi
           on pi.product_id = p.product_id 
      where pi.date >= '2021-04-23' and pi.date <= '2021-04-25'
      group by p.ad_id
     ) p
     on p.ad_id = ad.ad_id ;

第二个查询只是再次聚合。