SQL UNION总和

Question

我有以下SQL：

select distinct cat, sum(cnt) 
from 
(
select distinct prcIncidentManagement.CallCategory cat, count(eFolder.eFolderName) cnt 
from prcIncidentManagement 
inner join eFolder 
on prcIncidentManagement.EFOLDERID=eFolder.eFolderID 
group by prcIncidentManagement.CallCategory 

union all 

select distinct prcRequestFulfilment.CallCategory cat, count(eFolder.eFolderName) cnt 
from prcRequestFulfilment 
inner join eFolder 
on prcRequestFulfilment.EFOLDERID=eFolder.eFolderID 
group by prcRequestFulfilment.CallCategory 
) results 
order by 1 

我发现很难总结总的和组它靠在呼叫类别作为错误信息如下：

列“results.cat”在选择列表中无效，因为包含在聚合函数或GROUP BY子句是不

反正这个呢？

Answer 1

删除distinct并在外部查询中添加GROUP BY：

select cat, sum(cnt) 
from 
(
    select prcIncidentManagement.CallCategory cat, count(eFolder.eFolderName) cnt 
    from prcIncidentManagement 
    inner join eFolder 
    on prcIncidentManagement.EFOLDERID=eFolder.eFolderID 
    group by prcIncidentManagement.CallCategory 
    union all 
    select prcRequestFulfilment.CallCategory cat, count(eFolder.eFolderName) cnt 
    from prcRequestFulfilment 
    inner join eFolder 
    on prcRequestFulfilment.EFOLDERID=eFolder.eFolderID 
    group by prcRequestFulfilment.CallCategory 
) results 
GROUP BY cat 
order by 1; 

当您添加和聚合功能，您可以选择额外的列是不是group by子句，也不在聚合函数。

也不需要DISTINCT在子查询中具有聚合函数，因为它没用，因为group by将为您提供具有聚合函数的不同列，以聚合每个组的其他列，因此您不需要需要它。

Answer 2

如果我认为eFolderName从未NULL和eFolderId有适当的外键关系（这两者都是非常合理的假设），那么你可以通过删除子查询的基本连接简化查询：

select cat, sum(cnt) 
from ((select im.CallCategory as cat, count(im.eFolderID) as cnt 
     from prcIncidentManagement im 
     group by im.CallCategory 
    ) union all 
     (select f.CallCategory as cat, count(im.eFolderID) as cnt 
     from prcRequestFulfilment f 
     group by f.CallCategory 
    ) 
    ) results 
group by cat 
order by cat;