You can find an index of a searched element using jsonb_array_elements() with ordinality
(note, ordinality
starts from 1 while the first index of json array is 0):
select
pos- 1 as elem_index
from
samples,
jsonb_array_elements(sample->'result') with ordinality arr(elem, pos)
where
id = 26 and
elem->>'8410' = 'FERR_R';
elem_index
------------
2
(1 row)
Use the above query to update the element based on its index (note that the second argument of jsonb_set()
is a text array):
update
samples
set
sample =
jsonb_set(
sample,
array['result', elem_index::text, 'ratingtext'],
'"some individual text"'::jsonb,
true)
from (
select
pos- 1 as elem_index
from
samples,
jsonb_array_elements(sample->'result') with ordinality arr(elem, pos)
where
id = 26 and
elem->>'8410' = 'FERR_R'
) sub
where
id = 26;
Result:
select id, jsonb_pretty(sample)
from samples;
id | jsonb_pretty
----+--------------------------------------------------
26 | { +
| "result": [ +
| { +
| "8410": "ABNDAT", +
| "8411": "Abnahmedatum" +
| }, +
| { +
| "8410": "ABNZIT", +
| "8411": "Abnahmezeit" +
| }, +
| { +
| "8410": "FERR_R", +
| "8411": "Ferritin", +
| "ratingtext": "Some individual text"+
| } +
| ] +
| }
(1 row)
The last argument in jsonb_set()
should be true
to force adding a new value if its key does not exist yet. It may be skipped however as its default value is true
.
Though concurrency issues seem to be unlikely (due to the restrictive WHERE condition and a potentially small number of affected rows) you may be also interested in Atomic UPDATE .. SELECT in Postgres.