假设我有一张表:
SELECT * FROM settings;
| id | name | strategies |
| -- | --- | --- |
| 1 | default | [{name: xyz, enabled: true}, {name: bot2, enabled: true}] |
| 2 | new1 | [{name: bot2, enabled: true}, {name: xyz, enabled: false}] |
我希望在bot2
之前添加一个新的对象{name: bot1, enabled: true}
。
我正在尝试使用这个已回答的问题中提供的解决方案:
WITH bot2_index AS (SELECT
pos- 1 AS bot2_index
FROM
settings,
jsonb_array_elements(strategies) WITH ordinality arr(elem, pos)
WHERE
NAME = 'default'
AND elem->>'name' = 'bot2')
UPDATE settings
SET strategies = jsonb_set(strategies, '{bot2_index}', '{
"name": "bot1",
"enabled": false
}', TRUE);
但是我却得到了
ERROR: path element at position 1 is not an integer: "bot2_index"
bot2_index
是bigint
类型,为什么这个语法不起作用?我还尝试了其他变体,例如只使用bot2_index
、bot2_index::int
、bot2_index::string
,甚至将其作为两个单独的查询运行(就像接受的答案中所示),但它也不起作用。编辑#1 这个语法可以工作,但它似乎替换了该索引处的元素,而不是在给定索引处的元素之前或之后附加元素。如何使其像JS的
splice()
函数一样工作?UPDATE settings
SET strategies = jsonb_set(strategies, concat('{',(SELECT
pos- 1 AS bot2_index
FROM
settings,
jsonb_array_elements(strategies) WITH ordinality arr(elem, pos)
WHERE
NAME = 'default'
AND elem->>'name' = 'js:bot2'),'}')::text[], '{
"name": "bot1",
"enabled": false
}', TRUE);