JSON_OBJECT_AGG (Lakehouse v1)

Converts key-value pairs into a JSON object. For each row in the input, it generates a key-value pair where the key is derived from the <key_expression> and the value is derived from the <value_expression>. These key-value pairs are then combined into a single JSON object.

See also: JSON_ARRAY_AGG

Analyze Syntax

func.json_object_agg(<key_expression>, <value_expression>)

Analyze Examples

See SQL Example for details

SQL Syntax

JSON_OBJECT_AGG(<key_expression>, <value_expression>)

Parameter	Description
key_expression	Specifies the key in the JSON object. Only supports string expressions. If the key_expression evaluates to NULL, the key-value pair is skipped.
value_expression	Specifies the value in the JSON object. It can be any supported data type. If the value_expression evaluates to NULL, the key-value pair is skipped.

Return Type

JSON object.

SQL Examples

This example demonstrates how JSON_OBJECT_AGG can be used to aggregate different types of data—such as decimals, integers, JSON variants, and arrays—into JSON objects, with the column b as the key for each JSON object:

CREATE TABLE d (
    a DECIMAL(10, 2),
    b STRING,
    c INT,
    d VARIANT,
    e ARRAY(STRING)
);

INSERT INTO d VALUES
    (20, 'abc', NULL, '{"k":"v"}', ['a','b']),
    (10, 'de', 100, 'null', []),
    (4.23, NULL, 200, '"uvw"', ['x','y']),
    (5.99, 'xyz', 300, '[1,2,3]', ['z']);

SELECT
    json_object_agg(b, a) AS json_a,
    json_object_agg(b, c) AS json_c,
    json_object_agg(b, d) AS json_d,
    json_object_agg(b, e) AS json_e
FROM
    d;

-[ RECORD 1 ]-----------------------------------
json_a: {"abc":20.0,"de":10.0,"xyz":5.99}
json_c: {"de":100,"xyz":300}
json_d: {"abc":{"k":"v"},"de":null,"xyz":[1,2,3]}
json_e: {"abc":["a","b"],"de":[],"xyz":["z"]}