Build multi-tenant RAG with Neon's database-per-user model — no nosy neighbors, max isolation, minimal costs
Docs/Neon Postgres guides/Functions/JSON functions/jsonb_array_elements

Postgres jsonb_array_elements() function

Expands a JSONB array into a set of rows

You can use the jsonb_array_elements function to expand a JSONB array into a set of rows, each containing one element of the array. It is a simpler option compared to complex looping logic. It is also more efficient than executing the same operation on the application side by reducing data transfer and processing overhead.

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

jsonb_array_elements(json)

jsonb_array_elements example

Suppose you have a table with information about developers:

developers

CREATE TABLE developers (
 id INT PRIMARY KEY,
 name TEXT,
 skills JSONB
);

INSERT INTO developers (id, name, skills) VALUES
 (1, 'Alice', '["Java", "Python", "SQL"]'),
 (2, 'Bob', '["C++", "JavaScript"]'),
 (3, 'Charlie', '["HTML", "CSS", "React"]');
| id |  name   |          skills
|----|---------|---------------------------
| 1  | Alice   | ["Java", "Python", "SQL"]
| 2  | Bob     | ["C++", "JavaScript"]
| 3  | Charlie | ["HTML", "CSS", "React"]

Now, let's say you want to extract each individual skill from the skills JSON array. You can use jsonb_array_elements for that:

SELECT id, name, skill
FROM developers,
    jsonb_array_elements(skills) AS skill;

This query returns the following values:

| id |  name   |    skill
|----|---------|--------------
| 1  | Alice   | "Java"
| 1  | Alice   | "Python"
| 1  | Alice   | "SQL"
| 2  | Bob     | "C++"
| 2  | Bob     | "JavaScript"
| 3  | Charlie | "HTML"
| 3  | Charlie | "CSS"
| 3  | Charlie | "React"

Advanced examples

This section shows advanced jsonb_array_elements examples.

Filtering jsonb_array_elements

You can use the jsonb_array_elements function to extract the sizes from the JSON data and then filter the products based on a specific color (or size):

SELECT *
FROM products
WHERE 'Blue' IN (
 SELECT REPLACE(jsonb_array_elements(details->'colors')::text, '"', '')::text
);

This query returns the following values:

| id |   name   |                               details                                  |
|----|----------|------------------------------------------------------------------------|
|  1 | T-Shirt  | {"sizes": ["S", "M", "L", "XL"], "colors": ["Red", "Blue", "Green"]}   |
|  4 | Jeans    | {"sizes": ["28", "30", "32", "34"], "colors": ["Blue", "Black"]}       |

Handling NULL in jsonb_array_elements

This example updates the table to insert another product (Socks) with one of the values in the sizes as null:

products

| id |  name   |                                 details                                 |
|----|---------|-------------------------------------------------------------------------|
|  6 | Socks   | {"sizes": ["S", null, "L", "XL"], "colors": ["White", "Black", "Gray"]} |
INSERT INTO products (id, name, details) VALUES (6, 'Socks', '{"sizes": ["S", null, "L", "XL"], "colors": ["White", "Black", "Gray"]}');

Querying for Socks shows how null values in an array are handled:

SELECT
 id,
 name,
 size
FROM products AS p,
 jsonb_array_elements(p.details -> 'sizes') AS size
WHERE name = 'Socks';

This query returns the following values:

| id | name  | size |
|----|-------|------|
|  6 | Socks | "S"  |
|  6 | Socks | null |
|  6 | Socks | "L"  |
|  6 | Socks | "XL" |

Ordering json_array_elements output using WITH ORDINALITY

Let's consider a scenario where you have a table named workflow with a JSONB column steps representing sequential steps in a workflow:

workflow

CREATE TABLE workflow (
   id SERIAL PRIMARY KEY,
   workflow_name TEXT,
   steps JSONB
);

INSERT INTO workflow (workflow_name, steps) VALUES
   ('Employee Onboarding', '{"tasks": ["Submit Resume", "Interview", "Background Check", "Offer", "Orientation"]}'),
   ('Project Development', '{"tasks": ["Requirement Analysis", "Design", "Implementation", "Testing", "Deployment"]}'),
   ('Order Processing', '{"tasks": ["Order Received", "Payment Verification", "Packing", "Shipment", "Delivery"]}');
| id |    workflow_name    |                                          steps                                          |
|----|---------------------|-----------------------------------------------------------------------------------------|
|  1 | Employee Onboarding | {"tasks": ["Submit Resume", "Interview", "Background Check", "Offer", "Orientation"]}   |
|  2 | Project Development | {"tasks": ["Requirement Analysis", "Design", "Implementation", "Testing", "Deployment"]}|
|  3 | Order Processing    | {"tasks": ["Order Received", "Payment Verification", "Packing", "Shipment", "Delivery"]}|

Each workflow consists of a series of tasks, and you want to extract and display the tasks along with their order in the workflow.

SELECT
   workflow_name,
   task.value AS task_name,
   task.ordinality AS task_order
FROM
   workflow,
   jsonb_array_elements(steps->'tasks') WITH ORDINALITY AS task;

This query returns the following values:

|    workflow_name    |       task_name        | task_order |
|---------------------|------------------------|------------|
| Employee Onboarding | "Submit Resume"        |          1 |
| Employee Onboarding | "Interview"            |          2 |
| Employee Onboarding | "Background Check"     |          3 |
| Employee Onboarding | "Offer"                |          4 |
| Employee Onboarding | "Orientation"          |          5 |
| Project Development | "Requirement Analysis" |          1 |
| Project Development | "Design"               |          2 |
| Project Development | "Implementation"       |          3 |
| Project Development | "Testing"              |          4 |
| Project Development | "Deployment"           |          5 |
| Order Processing    | "Order Received"       |          1 |
| Order Processing    | "Payment Verification" |          2 |
| Order Processing    | "Packing"              |          3 |
| Order Processing    | "Shipment"             |          4 |
| Order Processing    | "Delivery"             |          5 |

Nested arrays in jsonb_array_elements

You can also handle nested arrays with jsonb_array_elements.

Consider a scenario where each product in an electronics_products table has multiple variants, and each variant has an array of sizes and an array of colors.

electronics_products

CREATE TABLE electronics_products (
 id INTEGER PRIMARY KEY,
 name TEXT,
 details JSONB
);


INSERT INTO electronics_products (id, name, details) VALUES
 (1, 'Laptop', '{"variants": [{"model": "A", "sizes": ["13 inch", "15 inch"], "colors": ["Silver", "Black"]}, {"model": "B", "sizes": ["15 inch", "17 inch"], "colors": ["Gray", "White"]}]}'),
 (2, 'Smartphone', '{"variants": [{"model": "X", "sizes": ["5.5 inch", "6 inch"], "colors": ["Black", "Gold"]}, {"model": "Y", "sizes": ["6.2 inch", "6.7 inch"], "colors": ["Blue", "Red"]}]}');
| id |    name    |                                                                                   details
|----|------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
|  1 | Laptop     | {"variants": [{"model": "A", "sizes": ["13 inch", "15 inch"], "colors": ["Silver", "Black"]}, {"model": "B", "sizes": ["15 inch", "17 inch"], "colors": ["Gray", "White"]}]}
|  2 | Smartphone | {"variants": [{"model": "X", "sizes": ["5.5 inch", "6 inch"], "colors": ["Black", "Gold"]}, {"model": "Y", "sizes": ["6.2 inch", "6.7 inch"], "colors": ["Blue", "Red"]}]}

To handle the nested arrays and extract information about each variant, you can run this query using the jsonb_array_elements function:

SELECT
 id,
 name,
 variant->>'model' AS model,
 size,
 color
FROM
 electronics_products,
  jsonb_array_elements(details->'variants') AS variant,
  jsonb_array_elements_text(variant->'sizes') AS t1(size),
  jsonb_array_elements_text(variant->'colors') AS t2(color);

This query returns the following values:

| id |    name    | model |   size   | color  |
|----|------------|-------|----------|--------|
|  1 | Laptop     | A     | 13 inch  | Silver |
|  1 | Laptop     | A     | 13 inch  | Black  |
|  1 | Laptop     | A     | 15 inch  | Silver |
|  1 | Laptop     | A     | 15 inch  | Black  |
|  1 | Laptop     | B     | 15 inch  | Gray   |
|  1 | Laptop     | B     | 15 inch  | White  |
|  1 | Laptop     | B     | 17 inch  | Gray   |
|  1 | Laptop     | B     | 17 inch  | White  |
|  2 | Smartphone | X     | 5.5 inch | Black  |
|  2 | Smartphone | X     | 5.5 inch | Gold   |
|  2 | Smartphone | X     | 6 inch   | Black  |
|  2 | Smartphone | X     | 6 inch   | Gold   |
|  2 | Smartphone | Y     | 6.2 inch | Blue   |
|  2 | Smartphone | Y     | 6.2 inch | Red    |
|  2 | Smartphone | Y     | 6.7 inch | Blue   |
|  2 | Smartphone | Y     | 6.7 inch | Red    |

jsonb_array_elements with joins

Let's assume you want to retrieve a list of users along with their roles in each organization. The data is stored in an organizations table and a users table.

organizations

| id |                           members                            |
|----|--------------------------------------------------------------|
|  1 | [{"id": 23, "role": "admin"}, {"id": 24, "role": "default"}] |
|  2 | [{"id": 23, "role": "user"}]                                 |
|  3 | [{"id": 24, "role": "admin"}, {"id": 25, "role": "default"}] |
|  4 | [{"id": 25, "role": "user"}]                                 |

users

| id  | name  |      email       |
|-----|-------|------------------|
| 23  | Max   | max@gmail.com    |
| 24  | Joe   | joe@gmail.com    |
| 25  | Alice | alice@gmail.com  |
CREATE TABLE organizations (
   id SERIAL PRIMARY KEY,
   members JSONB
);

CREATE TABLE users (
   id INTEGER PRIMARY KEY,
   name TEXT,
   email TEXT
);

INSERT INTO organizations (members) VALUES
   ('[{ "id": 23, "role": "admin" }, { "id": 24, "role": "default" }]'),
   ('[{ "id": 23, "role": "user" }]'),
   ('[{ "id": 24, "role": "admin" }, { "id": 25, "role": "default" }]'),
   ('[{ "id": 25, "role": "user" }]');

INSERT INTO users (id, name, email) VALUES
   (23, 'Max', 'max@gmail.com'),
   (24, 'Joe', 'joe@gmail.com'),
   (25, 'Alice', 'alice@gmail.com');

You can use the jsonb_array_elements function to extract the members from the JSONB array in the organizations table and then join with the users table.

SELECT
   o.id AS organization_id,
   u.id AS user_id,
   u.name AS user_name,
   u.email AS user_email,
   m->>'role' AS member_role
FROM
   organizations o
JOIN jsonb_array_elements(o.members) AS m ON true
JOIN users u ON m->>'id' = u.id::TEXT;

This query returns the following values:

| organization_id | user_id | user_name |   user_email    | member_role |
|-----------------|---------|-----------|-----------------|-------------|
|               2 |      23 | Max       | max@gmail.com   | user        |
|               1 |      23 | Max       | max@gmail.com   | admin       |
|               3 |      24 | Joe       | joe@gmail.com   | admin       |
|               1 |      24 | Joe       | joe@gmail.com   | default     |
|               4 |      25 | Alice     | alice@gmail.com | user        |
|               3 |      25 | Alice     | alice@gmail.com | default     |

Additional considerations

This section outlines additional considerations including alternative functions.

Alternatives to jsonb_array_elements

Use jsonb_array_elements when you need to maintain the JSON structure of the elements for further JSON-related operations or analysis and jsonb_array_elements_text if you need to work with the extracted elements as plain text for string operations, text analysis, or integration with text-based functions.

If you want to create a comma-separated list of all skills for each developer in the developers table, jsonb_array_elements_text can be used along with string_agg.

SELECT name, string_agg(skill, ',') AS skill_list
FROM developers, jsonb_array_elements_text(skills) AS skill
GROUP BY name;

This query returns the following values:

|  name   |   skill_list    |
|---------|-----------------|
|  Alice  | Java,Python,SQL |
|   Bob   | C++,JavaScript  |
| Charlie | HTML,CSS,React  |

Using jsonb_array_elements would result in an error because it returns JSONB values, which cannot be directly concatenated with the string operator.

SELECT name, string_agg(skill, ',') AS skill_list
FROM developers, jsonb_array_elements(skills) AS skill
GROUP BY name;

jsonb_path_query

jsonb_path_query uses JSON Path expressions for flexible navigation and filtering within JSONB structures and returns a JSONB array containing matching elements. It supports filtering within the path expression itself, enabling complex conditions and excels at navigating and extracting elements from nested arrays and objects.

If your query involves navigating through multiple levels of nesting, complex filtering conditions, or updates to JSONB data, jsonb_path_query is often the preferred choice.

Consider a simple example — to extract the first skill of each developer in the developers table:

SELECT jsonb_path_query(skills, '$[0]') AS first_skill
FROM developers;

This query returns the following values:

| first_skill |
|-------------|
|   "Java"    |
|   "C++"     |
|   "HTML"    |

Resources

Last updated on

Was this page helpful?