How to store comma-separated values row by row in postgresql

Question

Input:

('{"user":{"status":1,"loginid":1,"userids":{"userid":"5,6"}}}')  

I want to insert into my table like this:

userid    loginid    status  
---------------------------
5            1         1      
6            1         1

Answer 1

Use regexp_split_to_table(). Assuming that the columns are integers:

with input_data(data) as (
values
    ('{"user":{"status":1,"loginid":1,"userids":{"userid":"5,6"}}}'::json)
)

-- insert into my_table(userid, loginid, status)
select 
    regexp_split_to_table(data->'user'->'userids'->>'userid', ',')::int as userid,
    (data->'user'->>'loginid')::int as loginid,
    (data->'user'->>'status')::int as status
from input_data

 userid | loginid | status 
--------+---------+--------
      5 |       1 |      1
      6 |       1 |      1
(2 rows)    

Answer 2

Would be simpler with an array (JSON array) to begin with. The you can use json_array_elements_text(json) . See:

• How to turn json array into postgres array?

Convert the list you have to an array with string_to_array() . Then unnest() .

SELECT unnest(string_to_array(js#>>'{user,userids,userid}', ',')) AS userid
     , (js#>>'{user,loginid}')::int AS loginid
     , (js#>>'{user,status}')::int  AS status
FROM  (
   SELECT json '{"user":{"status":1,"loginid":1,"userids":{"userid":"5,6"}}}'
   ) i(js);

db<>fiddle here

I advise Postgres 10 or later for the simple form with unnest() in the SELECT list. See:

• What is the expected behaviour for multiple set-returning functions in select clause?

I avoid regexp functions for simple tasks. Those are powerful, but substantially more expensive.