在没有在PostgreSQL中创建新表的情况下选择进入

Question

我有很多像这样的结果在最后结合的查询，所以我可以计算两个不同结果集之间的连接数，然后使用联合查询来生成完整的结果表。这里的问题是使用select into为每个结果创建一个新表，而不是将其存储在内存中。在使用PostgreSQL时，如何将每个结果存储在自己的变量中，而不会在数据库中创建大量假表格，之后我将丢弃它们？

select distinct cust_num into t1 
from all_visits_x_cust 
where date > '06/02/2015' 
and date <= '07/02/2015' 
and cust_num > 9999; 

select distinct cust_num into t1_2 
from cc_calls_x_cust 
where date = '06/02/2015' 
and cust_num > 9999); 

select count(*) as post_call_visitors 
into v1 
from t1_2 join t1 
on t1.cust_num = t1_2.cust_num); 

Answer 1

你可以只使用前两个查询作为子查询您的最终查询：

select count(*) as post_call_visitors 
into temp v1 
from (
     select distinct cust_num 
     from all_visits_x_cust 
     where date > '06/02/2015' 
     and date <= '07/02/2015' 
     and cust_num > 9999 
    ) as t1 
join (
     select distinct cust_num 
     from cc_calls_x_cust 
     where date = '06/02/2015' 
     and cust_num > 9999 
    ) as t1_2 
on t1.cust_num = t1_2.cust_num; 

一定要删除这些子查询的into条款。他们最后在as部分获得他们的名字。

通过在第二行使用temp关键字，temporary表将在特殊模式下创建。它会在会议结束时自动放弃。

作为一个侧面说明，下面较短的查询应返回相同的结果：

select count(distinct cust_num) as post_call_visitors 
from cc_calls_x_cust 
where date = '06/02/2015' 
and cust_num > 9999 
and cust_num in (
     select cust_num 
     from all_visits_x_cust 
     where date > '06/02/2015' 
     and date <= '07/02/2015' 
    ); 

Answer 2

有Postgres里没有全局变量。您可以使用with query作为一个运行这些查询：

with t1 as (
    select distinct cust_num 
    from all_visits_x_cust 
    where date > '06/02/2015' 
    and date <= '07/02/2015' 
    and cust_num > 9999 
    ), 
t1_2 as (
    select distinct cust_num into t1_2 
    from cc_calls_x_cust 
    where date = '06/02/2015' 
    and cust_num > 9999 
    ) 
select count(*) as post_call_visitors 
from t1_2 
join t1 
on t1.cust_num = t1_2.cust_num; 

可以在PL/pgSQL function或DO statement使用SELECT ... INTO variable。