ORA-00937: 不是单组函数变通方法

Question

select
        o.customer_id,
        c.cust_first_name,
        c.cust_last_name,
        c.income_level,
        to_char(o.order_date, 'DD-MON-YY HH12:MI') as order_date,
        ROW_NUMBER() over (partition by o.customer_id order by o.order_date) as order#,
        o.order_total,
        lag(o.order_total, 1, 0) over (partition by o.customer_id order by o.customer_id) 
        + lag(o.order_total, 2, 0) over (partition by o.customer_id order by o.customer_id) as last_two_orders_sum,
        min(o.order_date) keep (dense_rank last order by o.customer_id) as first_order_total
from orders o, customers c
where o.customer_id = c.customer_id

这是我产生此错误的查询：

ORA-00937: not a single-group group function

最后一列的预期结果 - 每个客户的第一个订单：

我知道我必须使用 GROUP BY 子句来处理这个错误，但按所有其他列分组似乎不太合适。我知道这是 Oracle 11g 中的一个限制，这个问题已在12c，但我仍然需要某种解决方法。

尝试过：

CTE --> 在 CTE

中有一个 select 有点奇怪

子查询 --> 无法找出正确的语法，结果一团糟

感谢任何帮助！

Answer 1

而不是 MIN(...) KEEP (...) 聚合函数，您想要使用 MIN(...) KEEP (...) OVER (...) 分析函数，或者，可能没有 KEEP，使用 MIN(...) OVER (PARTITION BY ...):

select
        o.customer_id,
        c.cust_first_name,
        c.cust_last_name,
        c.income_level,
        to_char(o.order_date, 'DD-MON-YY HH12:MI') as order_date,
        ROW_NUMBER() over (partition by o.customer_id order by o.order_date)
          as order#,
        o.order_total,
        lag(o.order_total, 1, 0)
          over (partition by o.customer_id order by o.customer_id) 
        + lag(o.order_total, 2, 0)
          over (partition by o.customer_id order by o.customer_id)
          as last_two_orders_sum,
        MIN(o.order_date) OVER (PARTITION BY o.customer_id) as first_order_date,
        MIN(o.order_total) KEEP (DENSE_RANK FIRST ORDER BY o.order_date)
                           OVER (PARTITION BY o.customer_id) as first_order_total
from    orders o
        INNER JOIN customers c
        ON o.customer_id = c.customer_id

注意：另外，使用 ANSI 连接语法而不是旧的 comma-join 语法。

I know that this is a limitation in Oracle 11g and this issue is fixed in 12c, but I still need some kind of workaround.

不，这不是您可以通过升级到更高版本来解决的问题。您的查询的一个基本问题是您混合使用了聚合列和 non-aggregated 列，并且正如错误消息所暗示的那样，您没有包含所有 [=33] 的 GROUP BY 子句=]列。

解决方案是不使用任何聚合函数，改为仅使用分析函数。

db<>fiddle here

ORA-00937: 不是单组函数变通方法

ORA-00937: not a single-group group function Workaround

sql

oracle

group-by

oracle11g

ORA-00937: 不是单组函数 变通方法

ORA-00937: not a single-group group function Workaround

sql

oracle

group-by

oracle11g

ORA-00937: 不是单组函数变通方法