Oracle PIVOT子句中的用户定义聚合函数

Question

Unable to use user-defined aggregate functions in Oracle PIVOT clause.

I created a user-defined aggregate function called string_agg.
I am able to use it in a simple statement such as...

select id, string_agg(value) from
(
    select 'user1' as id, 'BMW' as value,   'CAR'      as type from dual    union
    select 'user1' as id, 'Audi' as value,  'CAR'      as type from dual    union
    select 'user2' as id, 'Honda' as value, 'CAR'      as type from dual    union
    select 'user1' as id, 'Dell' as value,  'COMPUTER' as type from dual    union
    select 'user1' as id, 'Sony' as value,  'COMPUTER' as type from dual    union
    select 'user2' as id, 'HP' as value,    'COMPUTER' as type from dual 
)
group by id, type

Results in:

  user1 CAR Audi,BMW user1 COMPUTER Dell,Sony user2 CAR Honda user2 COMPUTER HP 

However when I try to use the same function in a pivot clause

 select * from ( select id, type, string_agg(value) as value from ( select 'user1' as id, 'BMW' as value, 'CAR' as type from dual union select 'user1' as id, 'Audi' as value, 'CAR' as type from dual union select 'user2' as id, 'Honda' as value, 'CAR' as type from dual union select 'user1' as id, 'Dell' as value, 'COMPUTER' as type from dual union select 'user1' as id, 'Sony' as value, 'COMPUTER' as type from dual union select 'user2' as id, 'HP' as value, 'COMPUTER' as type from dual ) group by id, type ) PIVOT (string_agg(value) FOR id IN ('user1' user1, 'user2' user2) ); 

I get the following error...

 ORA-56902: expect aggregate function inside pivot operation 
 

result is...

  COMPUTER Dell,Sony HP CAR Audi,BMW Honda 

Answer 1

Pivot doesn't have to be over the same aggregate function:

select * from
( 
    select id, type, LISTAGG(value) WITHIN GROUP (ORDER BY 1) as value from
    (
        select 'user1' as id, 'BMW' as value,   'CAR'      as type from dual    union
        select 'user1' as id, 'Audi' as value,  'CAR'      as type from dual    union
        select 'user2' as id, 'Honda' as value, 'CAR'      as type from dual    union
        select 'user1' as id, 'Dell' as value,  'COMPUTER' as type from dual    union
        select 'user1' as id, 'Sony' as value,  'COMPUTER' as type from dual    union
        select 'user2' as id, 'HP' as value,    'COMPUTER' as type from dual 
    )
    group by id, type
)
PIVOT (max(value) FOR id IN ('user1' user1, 'user2' user2) );

Answer 2

那么尝试使用wmsys.wm_concat而不是用户定义的聚合呢？