如何从与外部查询连接的 COUNT 子查询中设置列值？答案

【问题标题】：How to I set a column value from a COUNT subquery that joins with an outer query?如何从与外部查询连接的 COUNT 子查询中设置列值？
【发布时间】：2019-04-16 17:15:51
【问题描述】：

我有一个查询（最初是为 SQL Server 编写的），它有多个子查询作为计算计数的列。在 SQL Server 中，我可以有一个方程式来将一列设置为计数子查询，例如：

Pass = (select count(*) from report.sub_2018 p where p.ABST = a.ABST and p.RESULT = 'P' and p.STATUS_REASON = 'Pending' and p.MONTH_YEAR = a.MONTH_YEAR)

我正在尝试在 Vertica 中运行查询，但它不允许这种类型的“方程式”。所以我尝试做类似的事情

(select count(*) from report.sub_2018 p where p.ABST = a.ABST and p.RESULT = 'P' and p.STATUS_REASON = 'Pending' and p.MONTH_YEAR = a.MONTH_YEAR) as Pass

但由于它与外部查询相关，我收到错误Correlated subquery with aggregate function COUNT is not supported

这是我的查询：

select UserId = u.USER_ID,
Name = u.LNAME + ', ' + u.FNAME,
a.Month_Year,
(select count(*) from report.sub_2018 p where p.ABST = a.ABST and p.RESULT = 'P' and p.STATUS_REASON = 'Pending' and p.MONTH_YEAR = a.MONTH_YEAR) as Pass,
(select count(*) from report.sub_2018 p where p.ABST = a.ABST and p.RESULT = 'F' and p.STATUS_REASON = 'Pending' and p.MONTH_YEAR = a.MONTH_YEAR) as Fail,
(select count(*) from report.sub_2018 p where p.ABST = a.ABST and p.STATUS_REASON = 'Pending' and p.MONTH_YEAR = a.MONTH_YEAR) as Total
from report.sub_2018 a inner join pd_user_info u on a.ABST = u.USER_ID
where MONTH_YEAR like '2018-%' and u.USER_ID like 'MMN%'
group by u.LNAME, u.FNAME, a.MONTH_YEAR, a.ABST, u.USER_ID
order by u.LNAME, u.FNAME, a.MONTH_YEAR

我不太确定如何重新排列查询以使其与表 report.sub_2018 a 的外部查询一起使用

感谢任何帮助！

【问题讨论】：

标签： sql vertica

【解决方案1】：

改为编写此查询

select 
  u.USER_ID as UserId,
  u.LNAME || ', ' || u.FNAME as Name,
  a.Month_Year,
  count(case when a.RESULT = 'P' and a.STATUS_REASON = 'Pending' then 1 end) as Pass,
  count(case when a.RESULT = 'F' and a.STATUS_REASON = 'Pending' then 1 end) as Fail,
  count(case when                    a.STATUS_REASON = 'Pending' then 1 end) as Total
from report.sub_2018 a 
inner join pd_user_info u on a.ABST = u.USER_ID
where MONTH_YEAR like '2018-%' and u.USER_ID like 'MMN%'
group by u.LNAME, u.FNAME, a.MONTH_YEAR, a.ABST, u.USER_ID
order by u.LNAME, u.FNAME, a.MONTH_YEAR

这样，您只需访问sub_2018 表一次。 I have blogged about this technique recently.

另请注意，在除 SQL Server 之外的大多数其他数据库中，使用此语法对列进行别名处理

u.USER_ID as UserId

...不是这个 SQL Server 特定的：

UserId = u.USER_ID

【讨论】：