Create json with column values as object keys - json

Question

I have a table defined like this:
CREATE TABLE data_table AS (
id bigserial,
"name" text NOT NULL,
"value" text NOT NULL,
CONSTRAINT data_table_pk PRIMARY KEY (id)
);
INSERT INTO data_table ("name", "value") VALUES
('key_1', 'value_1'),
('key_2', 'value_2');
I would like to get a JSON object from this table content, which will look like this:
{
"key_1":"value_1",
"key_2":"value_2"
}
Now I'm using the client application to parse the result set into JSON format. Is it possible to accomplish this by a postgresl query?

Answer 1

If you're on 9.4 you can do the following:
$ select json_object_agg("name", "value") from data_table;
json_object_agg
----------------------------------------------
{ "key_1" : "value_1", "key_2" : "value_2" }

Answer 2

select
format(
'{%s}',
string_agg(format(
'%s:%s',
to_json("name"),
to_json("value")
), ',')
)::json as json_object
from data_table;
json_object
---------------------------------------
{"key_1":"value_1","key_2":"value_2"}

Answer 3

In a generic scenario you can nest more than one json_object_agg functions on top of a subquery. The inner subqueries should always have at least one column that will be used by outer subquery as keys for the json_object_agg function.
In the example, in the subquery C the values of the column action are used as keys in the subquery A. In A the values of column role are used as keys in query A.
-- query A
select json_object_agg(q1.role, q1.actions) from (
-- subquery B
select q2.role, json_object_agg(q2.action, q2.permissions) as actions from (
-- subquery C
select r.name as role, a.name as action, json_build_object (
'enabled', coalesce(a.bit & bit_and(p.actionids) <> 0, false),
'guestUnsupported', r.name = 'guest' and a."guestUnsupported"
) as permissions
from role r
left join action a on a.entity = 'route'
left join permission p on p.roleid = r.id
and a.entity = p.entityname
and (p.entityid = 1 or p.entityid is null)
where
1 = 1
and r.enabled
and r.deleted is null
group by r.name, a.id
) as q2 group by q2.role
) as q1
The result is a single row/single column with the following content:
{
"Role 1": {
"APIPUT": {
"enabled": false,
"guestUnsupported": false
},
"APIDELETE": {
"enabled": false,
"guestUnsupported": false
},
"APIGET": {
"enabled": true,
"guestUnsupported": false
},
"APIPOST": {
"enabled": true,
"guestUnsupported": false
}
},
"Role 2": {
"APIPUT": {
"enabled": false,
"guestUnsupported": false
},
"APIDELETE": {
"enabled": false,
"guestUnsupported": false
},
"APIGET": {
"enabled": true,
"guestUnsupported": false
},
"APIPOST": {
"enabled": false,
"guestUnsupported": false
}
}
}