PostgreSQL和嵌套json对象

我需要在这样的结构中构建json对象：

{
    "id":1, 
    "name": "Jessica", //other top-level key-values
    "add_info": {
        "first_object":{"date":"2017-04-17","id":1},
        "second_object":{"date":"2017-04-17","id":1} //etc.
    }
}

不幸的是，我对大量伟大但未举例的PostgreSQL json函数感到困惑，无法找到更好的实现方法。 我试过这个：

SELECT row_to_json(t, TRUE)
FROM (
       SELECT
         id, name
         , (
             (
               SELECT row_to_json(b) AS first_object
               FROM (
                      SELECT *
                      FROM table1
                      WHERE client_id = 1
                    ) b
             ),
             (
               SELECT row_to_json(b) AS second_object
               FROM (
                      SELECT *
                      FROM table2
                      WHERE client_id = 1
                    ) b
             )
           ) AS add_info

       FROM main_table
       WHERE id = 1
     ) t;

但我明白了：

{
    "id":1, 
    "name": "Jessica", //other top-level key-values
    "add_info": {
        "f1":{"date":"2017-04-17","id":1},
        "f2":{"date":"2017-04-17","id":1} //etc.
    }
}

f1、f2！！什么

看来我可以给他们起个名字：

SELECT row_to_json(t, TRUE)
FROM (
       SELECT
         id, name
         , (
             (SELECT row_to_json(first_row) first_row
              FROM
                (
                  SELECT row_to_json(b) AS first_object
                  FROM (
                         SELECT *
                         FROM first_table
                         WHERE client_id = 1
                       ) b
                ) AS first_row
             ),
             (SELECT row_to_json(second_row) second_row
              FROM
                (
                  SELECT row_to_json(b) AS second_object
                  FROM (
                         SELECT *
                         FROM second_table
                         WHERE client_id = 1
                       ) b
                ) AS second_row
             )
           ) AS add_info

       FROM main_table
       WHERE id = 1
     ) t;

但这会产生一个额外的对象：

{
    "id":1,
    "name":"CRED", 
    "add_info":{
        "f1":{"first_object":{"date":"2017-04-17","id":1}},
        "f2":{"second_object":{"date":"2017-04-17","id":1}}
    }
}

---

我可以解决这个问题吗？

您可以使用一个公共表表达式：

WITH add_info(first_object, second_object) AS
     (SELECT
        (SELECT row_to_json(table1)
                FROM table1
                WHERE id = 1
        ),
        (SELECT row_to_json(table2)
                FROM table2
                WHERE id = 1
        )
     )
SELECT row_to_json(t, TRUE)
FROM (
       SELECT
         id, name, add_info
       FROM main_table CROSS JOIN add_info
       WHERE id = 1
     ) t;

---

您可以使用公共表表达式：

WITH add_info(first_object, second_object) AS
     (SELECT
        (SELECT row_to_json(table1)
                FROM table1
                WHERE id = 1
        ),
        (SELECT row_to_json(table2)
                FROM table2
                WHERE id = 1
        )
     )
SELECT row_to_json(t, TRUE)
FROM (
       SELECT
         id, name, add_info
       FROM main_table CROSS JOIN add_info
       WHERE id = 1
     ) t;

---

看来这正是我需要的。试图避免多个

选择
，但现在似乎无法实现…似乎这正是我需要的。试图避免多个
选择
，但现在似乎无法实现。。。