如何对同一个查询 SQL 进行 2 个不同的计数？答案

【问题标题】：How to do 2 different counts on the same query SQL?如何对同一个查询 SQL 进行 2 个不同的计数？
【发布时间】：2021-09-24 07:47:47
【问题描述】：

我已经加入了 2 张桌子

SELECT ccc.supplier_id as supplier_id, c1.name as supplier, ccc.dealer_id as dealer_id, c2.name as dealer
FROM company_companyconnection ccc 
left JOIN companies_company c1 ON c1.id = ccc.supplier_id
left JOIN companies_company c2 on c2.id = ccc.dealer_id
ORDER BY supplier_id ASC

这样我得到的结果如下：

supplier_id	supplier	dealer_id	dealer
1	company1	3	company3
1	company1	2	company2
2	company2	3	company3

现在我想对同一个查询进行 2 次不同的计数，以便我可以看到计数是双向的，例如

company_id	company	as_supplier_count	as_dealer_count
1	company1	2	0
2	company2	1	1
3	company3	0	2

到目前为止我已经做到了

SELECT ccc.supplier_id as company_id, c1.name as company, count(ccc.dealer_id) as as_supplier_connections
FROM company_companyconnection ccc 
left JOIN companies_company c1 ON c1.id = ccc.supplier_id
JOIN companies_company c2 on c2.id = ccc.dealer_id
GROUP BY ccc.supplier_id, c1.name
ORDER BY supplier_id ASC

company_id	company	as_supplier_count
1	company1	2
2	company2	1
3	company3	0

但我在最后阶段遇到了麻烦，任何帮助都会非常感谢您！

【问题讨论】：

标签： sql postgresql join count

【解决方案1】：

您可以取消透视和聚合：

SELECT cc.*,
       count(*) filter (where v.which = 'supplier') as num_supplier,
       count(*) filter (where v.which = 'dealer') as num_dealer
FROM companies_company cc left join
     (company_companyconnection ccc cross join lateral
      (values (supplier_id, 'supplier'),
              (dealer_id, 'dealer')
      ) v(company_id, which)
     )
     on v.company_id = cc.id
group by cc.id;

注意：group by cc.id 假定companies_company(id) 是表的主键（或至少声明为unique）。否则，您需要同时列出select 和group by 中的列。

【讨论】：