如何在限制 oracle json_array 中使用 group by 或 distinct 子句进行过滤

Question

拥有简单的数据库结构

PERSON_ID   FIRST_NAME  LAST_NAME
1             John          Doe
2             John          Doe
3             Peter       Jackson

需要使用 JSON ARRAY 结构构建单行输出，其中包含按 first_name,last_name 条件过滤的唯一数据。

预期结果：

[{
    "firstname": "John",
    "lastname": "Doe"
},
{
    "firstname": "Peter",
    "lastname": "Jackson"
}] 

在数组级别使用 group by 会产生两行

SELECT  json_array(   
                    json_object(  key 'firstname' VALUE t.first_name, 
                                  key 'lastname'  VALUE t.last_name)

                 ) RESPONSEJSON

FROM TESTDATA t
GROUP BY t.first_name, t.last_name



    RESPONSEJSON
1   [{"firstname":"Peter","lastname":"Jackson"}]
2   [{"firstname":"John","lastname":"Doe"}]

Answer 1

使用子查询来创建不同的对象。然后将它们聚合到数组中：

create table t (
  person_id int, first_name varchar2(10), last_name varchar2(10)
);

insert into t values (1, 'John', 'Doe' );
insert into t values (2, 'John', 'Doe' );
insert into t values (3, 'Peter', 'Jackson' );
commit;

with jobjs as (
  select  distinct json_object(  
            key 'firstname' VALUE t.first_name, 
            key 'lastname'  VALUE t.last_name
          ) responsejson
  from    t
)
SELECT  json_arrayagg ( responsejson )
FROM    jobjs;

[
  {"firstname":"John","lastname":"Doe"},
  {"firstname":"Peter","lastname":"Jackson"}
]

---

在 12.2 中存在一个错误，当使用上述 distinct 时会给出错误的结果。你可以改用group by 来解决这个问题：

with jobjs as (
  select  json_object(  
            key 'firstname' VALUE t.first_name, 
            key 'lastname'  VALUE t.last_name
          ) responsejson
  from    t
  group   by t.first_name, t.last_name
)
SELECT  json_arrayagg ( responsejson )
FROM    jobjs;

[
  {"firstname":"Peter","lastname":"Jackson"},
  {"firstname":"John","lastname":"Doe"}
]