SQL子查询在分组时返回多行

Question

我有两张表，都有列：tstamp、weight、destination 我正在尝试编写一个查询以将表 a 和 b 中的以下列放入一个表中： Count() from a as Acounts, Count() from b as Bcounts, Sum(Acounts, Bcounts) 其中目的地=x 其中 datepart(hh,tstamp) =y

我已经尝试了几个小时使用联合、联接和子查询，但似乎找不到正确的解决方案。 Bcounts 列未作为列显示在我的表中。相反，结果显示为帐户下的单独行。

我不是 SQL 专家，因为我主要使用 PLC。明天我会发布我的代码。

Answer 1

这是一个 PostgreSQL 示例。

假设这些是您的表格

create table a (tstamp timestamp, weight int, destination text);
create table b (tstamp timestamp, weight int, destination text);

insert into a values
('2020-01-01 10:00:00', 40, 'germany'), ('2019-01-01 10:00:00', 50, 'germany'), 
('2020-04-01 10:00:00', 10, 'germany'), ('2019-04-01 10:00:00', 20, 'germany'), 
('2020-01-01 11:00:00', 40, 'congo'), ('2019-01-01 11:00:00', 50, 'congo'), 
('2020-04-01 11:00:00', 10, 'congo'), ('2019-04-01 12:00:00', 20, 'congo');

insert into b values
('2020-01-01 10:00:00', 40, 'germany'), ('2019-01-01 10:00:00', 50, 'germany'), 
('2020-04-01 11:00:00', 10, 'congo'), ('2019-04-01 11:00:00', 20, 'congo');

查询

我们从表 A 中获取计数并放置一个关于源的标记。我们将它与表 b 中的相同类型的信息结合起来。然后，我们使用case语句选择性地分别拉取a和b的计数，也可以组合拉取。

with combined as (
  select count(*) as counter, 'a' as source from a
  where destination = 'germany' and date_part('hour', tstamp) = 10

  union all

  select count(*) as counter, 'b' as source from b
  where destination = 'germany' and date_part('hour', tstamp) = 10
)
select
  sum(case when source = 'a' then counter else 0 end) as count_a,
  sum(case when source = 'b' then counter else 0 end) as count_b,
  sum(counter) as counter_a_b
from combined

结果

count_a | count_b | counter_a_b ------: | ------: | ----------: 4 | 2 | 6

示例：https://dbfiddle.uk/?rdbms=postgres_12&fiddle=100b4d5d4c94789d2f2c05d79c464166

SQL Server 等效项

with combined as (
  select count(*) as counter, 'a' as source from a
  where destination = 'germany' and datepart(hh, tstamp) = 10

  union all

  select count(*) as counter, 'b' as source from b
  where destination = 'germany' and datepart(hh, tstamp) = 10
)
select
  sum(case when source = 'a' then counter else 0 end) as count_a,
  sum(case when source = 'b' then counter else 0 end) as count_b,
  sum(counter) as counter_a_b
from combined

示例：https://dbfiddle.uk/?rdbms=sqlserver_2017&fiddle=ff9dd4f864792437cb27d143106db186

MySQL 等效项

select
  sum(case when source = 'a' then counter else 0 end) as count_a,
  sum(case when source = 'b' then counter else 0 end) as count_b,
  sum(counter) as counter_a_b
from (
  select count(*) as counter, 'a' as source from a
  where destination = 'germany' and hour(tstamp) = 10

  union all

  select count(*) as counter, 'b' as source from b
  where destination = 'germany' and hour(tstamp) = 10
) t

例如：https://dbfiddle.uk/?rdbms=mysql_5.5&fiddle=660b0c84c0c1b04141d19cc8c6b6af6d