Flatten the jsonb nested array in postgresql

Question

I have a data in the table as

id(integer)    | label(text)     | value(jsonb)     |
---------------|-----------------|------------------|
12345          |       Education | [[{"label": "Type", "value": "Under Graduate"},{"label": "Location", "value": "New Delhi"}],[{"label": "Type", "value": "Post Graduate"}]]|

And the required output is :

id    | label               | value          |
------|---------------------|----------------|
12345 | Education_Type_1    | Under Graduate |
12345 | Education_Location_1| New Delhi      |
12345 | Education_Type_2    | Post Graduate  |

Can someone please help me solve this issue that I am facing?

Answer 1

You can use jsonb_array_elements(your_jsonb_column). Tested on Postgres 9.6. You can use json_array_elements(your_json_column) if you are using some other version.

Table:

create table test (id int,label text, value jsonb);

Insert Statement:

insert into test values(12345,'Education','[[{"label": "Type", "value": "Under Graduate"}],[{"label": "Type", "value": "Post Graduate"}]]');
insert into test values(123456,'Education2','[[{"label": "Type2", "value": "Under Graduate2"}],[{"label": "Type2", "value": "Post Graduate2"}]]');

SQL Query:

select id, label,jsonb_array_elements(value)->0->>'value'
from test

Where 0 is used to take first elements from an array. ->> is used to remove quotes from the string.

Output:

id  label   value
12345   Education   Under Graduate
12345   Education   Post Graduate
123456  Education2  Under Graduate2
123456  Education2  Post Graduate2

SQL Fiddle

Answer 2

I found the solution. Thanks @Fahad Anjum. I wrote the solution on top of your soultion.

SELECT 
    'Education_' || (jsonb_array_elements(elem)->>'label')::text || '_' || pos::text AS label, jsonb_array_elements(elem)->>'value'
FROM jsonb_array_elements(
    '{"test": [
        [{"label":"Type", "value": "Under Graduate"},{"label":"Location", "value": "New Delhi"},{"label":"CGPA", "value": "9.07"}],
        [{"label":"Type", "value": "Post Graduate"},{"label":"Location", "value": "Bangalore"}],
        [{"label":"Type", "value": "Some education 1"}]]}'::jsonb->'test'
    ) WITH ordinality arr(elem, pos);

Answer 3

Since we value column is like multi-dimensional array of irregular dimension we will use recursive query to find solution.

Below query result in required output you want

I have populated your sample data in CTE.

with recursive cte(id,label,value,dims) as (
 select
  12345,
  'Education'::text,
  '[
    [
     {"label": "Type", "value": "Under Graduate"},
     {"label": "Location", "value":"New Delhi"}
    ],
    [
     {"label": "Type", "value": "Post Graduate"}
    ]
   ]'::jsonb,
  jsonb_array_length('[[{"label": "Type", "value": "Under Graduate"},{"label": "Location", "value": "New Delhi"}],[{"label": "Type", "value": "Post Graduate"}]]'::jsonb)
), res(id,label,val,dims) as (
    select cte.id,cte.label,l.v,cte.dims-1
    from cte,lateral(
        select jsonb_array_elements(cte.value) as v
    ) l
    union all
    select
      res.id,res.label,l.v,res.dims-1
    from res,lateral(
       select jsonb_array_elements(res.val) as v
     ) l
    where
     res.dims>0
)
select
    res.id,
    res.val->>'value' as value,
    res.label ||
    '_'||
    (res.val->>'label')::text ||
    '_' ||
    row_number() over (partition by id,label,(res.val->>'label')::text) as label
from res
where dims=0