sql - ORA-00937: 不是单组组函数

Posted 2023-05-09

【中文标题】sql - ORA-00937: 不是单组组函数

【英文标题】：sql - ORA-00937: not a single-group group function

【发布时间】：2019-01-12 12:04:15

【问题描述】：

 select
    location,
    home_team_name,
    count(case when extract(year from match_date)='2018' and extract(month from match_date)=1 then 1 end) january_2018,
    count(case when extract(year from match_date)='2018' and extract(month from match_date)=2 then 1 end) february_2018,
    count(case when extract(year from match_date)='2018' and extract(month from match_date)=3 then 1 end) march_2018,
    count(case when extract(year from match_date)='2018' then 1 end) Total
from match_results

union all

select 
    'total' as location,
    'total' as home_team_name,
    count(case when extract(month from match_date)=1 then 1 end) january_2018,
    count(case when extract(month from match_date)=2 then 1 end) february_2018,
    count(case when extract(month from match_date)=3 then 1 end) march_2018,
    count(case when extract(year from match_date)='2018' then 1 end) Total
from match_results
group by location,home_team_name;

错误消息：ORA-00937：不是单组组函数。目前在 oracle live SQL 上运行。

关于如何解决这个问题的任何想法？

【问题讨论】：

将group by 移动到第一个选择。并查看grouping sets 以在一次选择中获得两者。

【参考方案1】：

我推荐grouping sets：

select coalesce(location, 'Total') as location,
       coalesce(home_team_name, 'Total') as home_team_name,
       sum(case when match_date >= date '2018-01-01' and 
                     match_date < date '2018-02-01'
                then 1 else 0
           end) as january_2018,
       sum(case when match_date >= date '2018-02-01' and 
                     match_date < date '2018-03-01'
                then 1 else 0
           end) as february_2018,
       sum(case when match_date >= date '2018-03-01' and 
                     match_date < date '2018-04-01'
                then 1 else 0
           end) as march_2018,
       sum(case when match_date >= date '2018-01-01' and 
                     match_date < date '2019-01-01'
               then 1 else 0
           end) as total_2018
from match_results
group by grouping sets ( (location, home_team_name), () );

也就是说，不需要重复查询。我还更改了日期比较以使用实际日期。我发现这比提取日期部分更具可读性和可维护性。

【讨论】：

@sammywemmy 。 . .我发现您更喜欢更详细的解决方案，这让我感到惊讶。

我从这两个答案中学到了很多，对sql的理解也更多了。不是一个比一个，我只能勾选一个答案

【参考方案2】：

运行聚合查询时，每个非聚合列都必须出现在GROUP BY 子句中。您有两个 UNIONed 子查询：并且只有第一个具有非聚合列（即 location 和 home_team），但是您碰巧将 GROUP BY 子句放在第二个子查询中，因为它似乎计算总数，可能不需要一个。您可以将GROUP BY 子句放在第一个查询中而不是第二个：

select
    location,
    home_team_name,
    count(case when extract(year from match_date)='2018' and extract(month from match_date)=1 then 1 end) january_2018,
    count(case when extract(year from match_date)='2018' and extract(month from match_date)=2 then 1 end) february_2018,
    count(case when extract(year from match_date)='2018' and extract(month from match_date)=3 then 1 end) march_2018,
    count(case when extract(year from match_date)='2018' then 1 end) Total
from match_results
group by location,home_team_name

union all

select 
    'total' as location,
    'total' as home_team_name,
    count(case when extract(month from match_date)=1 then 1 end) january_2018,
    count(case when extract(month from match_date)=2 then 1 end) february_2018,
    count(case when extract(month from match_date)=3 then 1 end) march_2018,
    count(case when extract(year from match_date)='2018' then 1 end) Total
from match_results
;

【讨论】：

谢谢。我真的很感激