From this article, I tried to update or delete property of a JSONB column:

CREATE TABLE xxx (id BIGSERIAL, data JSONB);
INSERT INTO xxx(data) VALUES( ‘{"a":1,"b":2}‘ );
SELECT * FROM data;
 id |       data       
----+------------------
  1 | {"a": 1, "b": 2}

create the update function:

CREATE FUNCTION jsonb_merge(JSONB, JSONB) 
RETURNS JSONB AS $$
WITH json_union AS (
    SELECT * FROM JSONB_EACH($1)
    UNION ALL
    SELECT * FROM JSONB_EACH($2)
) SELECT JSON_OBJECT_AGG(key, value)::JSONB FROM json_union;
$$ LANGUAGE SQL;

testing:

-- replace
UPDATE xxx SET data = jsonb_merge(data,‘{"b":3}‘) WHERE id = 1;
SELECT * FROM xxx;
 id |       data       
----+------------------
  1 | {"a": 1, "b": 3}

-- append
UPDATE xxx SET data = jsonb_merge(data,‘{"c":4}‘) WHERE id = 1;
SELECT * FROM xxx;
 id |           data       
----+-------------------------
  1 | {"a": 1, "b": 3, "c": 4}

The question is:

1. is there any drawback of using JSONB_EACH (jsonb_merge) instead of JSONB_EACH_TEXT (from the article) in this case?

2. how to modify the jsonb_merge so if the second parameter property value is null (something like {"b":null}) the value would be erased?

.

-- remove
UPDATE xxx SET data = jsonb_merge(data,‘{"b":null}‘) WHERE id = 1;
SELECT * FROM xxx;
 id |       data       
----+-----------------
  1 | {"a": 1, "c": 4}