Make Column value as key using JSON function in Postgres - sql

Question

I'm using Postgres database and have a table as below
Table Name: Test
id firstname lastname
1 Sam Crews
2 John Dave
I'm trying to get result set back in below JSON format but with no luck
Expected:
[{"1": {"firstname": "Sam", "lastname": "Crews"}},
{"2": {"firstname": "John", "lastname": "Dave"}}
]
I tried using row_to_json, json_build_object functions but the output is bit different (as below).
Actual:
[{"id": "1", "firstname": "Sam", "lastname": "Crews"},
{"id": "2", "firstname": "John", "lastname": "Dave"}
]
Any pointers on how to achieve expected result would be greatly appreciated.
Thank you!

Answer 1

row_to_json() will always use all column names to create the keys in the JSON document. If you only want two columns, you need to explicitly specify them.
You also want a nested JSON object, so you need to nest the methods to create one:
select json_agg(json_build_object(
id,
json_build_object('firstname', firstname, 'lastname', lastname))
)
from the_table;
Klin made me realize that with Postgres 9.6 you can simplify the creation of the firstname/lastname object:
select json_agg(json_build_object(id, to_jsonb(the_table) - 'id'))
from the_table;

Answer 2

with my_table(id, firstname, lastname) as (
values
(1, 'Sam', 'Crews'),
(2, 'John', 'Dave')
)
select jsonb_agg(obj)
from my_table,
jsonb_build_object(id, to_jsonb(my_table)- 'id') obj
jsonb_agg
------------------------------------------------------------------------------------------------------
[{"1": {"lastname": "Crews", "firstname": "Sam"}}, {"2": {"lastname": "Dave", "firstname": "John"}}]
(1 row)