当“JPA 查询”中列的参数为 NULL 时，如何不 GROUP BY 列？

Question

I want select data from table and GROUP BY with city_id, district_id, ward_id. And if one of these column have param is null then GROUP BY the remaining columns.

Example: If param:ward_id is null, then GROUP BY columns city_id, district_id. And column ward_id not affect to GROUP BY. Thanks.

Edit: I add sum(people) and filter year in query.

@Query(value = "SELECT city_id, district_id, ward_id, sum(people) as sum_people
        FROM table  
        WHERE (:city_id is null OR city_id =:city_id ) 
        AND (:district_id is null OR district_id =:district_id ) 
        AND (:ward_id is null OR ward_id =:ward_id ) 
        AND year =:year
        GROUP BY city_id, district_id, ward_id",
        nativeQuery = true)
List<Object> findAddress(
        @Param("city_id") Integer city_id,
        @Param("district_id") Integer district_id,
        @Param("ward_id") Integer ward_id,
        @Param("year") Integer year
    )

Answer 1

SELECT * FROM table

You are not performing any aggregation.

WHERE (:city_id is null OR city_id =:city_id ) 
AND (:district_id is null OR district_id =:district_id ) 
AND (:ward_id is null OR ward_id =:ward_id ) 

You are also filtering data based on these 3 columns.

So the answer is to remove the group by altogether. If you want to control the ordering of your output, use ORDER BY instead.

---

After the edit, you included sum(people).

Try the following - note that I have not tested.

SELECT SELECT city_id, district_id, ward_id, sum(people) as sum_people
FROM (SELECT city_id, district_id, case :ward_id is null then '' else ward_id end ward_id, people
    FROM table
    WHERE (:city_id is null OR city_id =:city_id ) 
    AND (:district_id is null OR district_id =:district_id ) 
    AND (:ward_id is null OR ward_id =:ward_id )
) tmp
GROUP BY city_id, district_id, ward_id END

Essentially, when you use the null to filter on ward_id, it will replace the results with '' so that the group by will aggregate all ward_ids.