Learn how Neon compares to Aurora Serverless v2 - TL;DR: faster cold starts, responsive autoscaling, 80% lower costs

Postgres json_each() function

Expands JSON into a record per key-value pair

The json_each function in Postgres is used to expand a JSON object into a set of key-value pairs.

It is useful when you need to iterate over a JSON object's keys and values, such as when you're working with dynamic JSON structures where the schema is not fixed. Another important use case is performing data transformations and analytics.

Try it on Neon!

Neon is Serverless Postgres built for the cloud. Explore Postgres features and functions in our user-friendly SQL editor. Sign up for a free account to get started.

Sign Up

Function signature

json_each(json JSON) -> SETOF record(key text, value json)

The function returns a set of rows, each containing a key and the corresponding value for each field in the input JSON object. The key is of type text, while the value is of type json.

Example usage

Consider a JSON object representing a user's profile information. The JSON data will have multiple attributes and might look like this:

{
  "username": "johndoe",
  "age": 30,
  "email": "johndoe@example.com"
}

We can go over all the fields in the profile JSON object using json_each, and produce a row for each key-value pair.

SELECT key, value
FROM json_each('{"username": "johndoe", "age": 30, "email": "johndoe@example.com"}');

This query returns the following results:

| key      | value                 |
|----------|-----------------------|
| username | "johndoe"             |
| age      | 30                    |
| email    | "johndoe@example.com" |

Advanced examples

json_each custom column names

You can use AS to specify custom column names for the key and value columns.

SELECT attr_name, attr_value
FROM json_each('{"username": "johndoe", "age": 30, "email": "johndoe@example.com"}')
AS user_data(attr_name, attr_value);

This query returns the following results:

| attr_name | attr_value            |
|-----------|-----------------------|
| username  | "johndoe"             |
| age       | 30                    |
| email     | "johndoe@example.com" |

Use json_each as a table or row source

Since json_each returns a set of rows, you can use it as a table source in a FROM clause. This lets us join the expanded JSON data in the output with other tables.

Here, we're joining each row in the user_data table with the output of json_each:

CREATE TABLE user_data (
    id INT,
    profile JSON
);
INSERT INTO user_data (id, profile)
VALUES
    (123, '{"username": "johndoe", "age": 30, "email": "johndoe@example.com"}'),
    (140, '{"username": "mikesmith", "age": 40, "email": "mikesmith@example.com"}');

SELECT id, key, value
FROM user_data, json_each(user_data.profile);

This query returns the following results:

| id  | key      | value                   |
|-----|----------|-------------------------|
| 123 | username | "johndoe"               |
| 123 | age      | 30                      |
| 123 | email    | "johndoe@example.com"   |
| 140 | username | "mikesmith"             |
| 140 | age      | 40                      |
| 140 | email    | "mikesmith@example.com" |

Additional considerations

Performance implications

When working with large JSON objects, json_each may lead to performance overhead, as it expands each key-value pair into a separate row.

Alternative functions

  • json_each_text - Similar functionality to json_each but returns the value as a text type instead of JSON.
  • json_object_keys - It returns only the set of keys in the JSON object, without the values.
  • jsonb_each - It provides the same functionality as json_each, but accepts JSONB input instead of JSON.

Resources

Last updated on

Was this page helpful?