Build multi-tenant RAG with Neon's database-per-user model — no nosy neighbors, max isolation, minimal costs

PostgreSQL WITH CHECK OPTION

Summary: in this tutorial, you will learn how to create an updatable view using the WITH CHECK OPTION clause to ensure that the changes to the base tables through the view satisfy the view-defining condition.

Introduction to the PostgreSQL WITH CHECK OPTION clause

In PostgreSQL, a view is a named query stored in the PostgreSQL database server. A simple view can be updatable.

To ensure that any data modification made through a view adheres to certain conditions in the view’s definition, you use the WITH CHECK OPTION clause.

Typically, you specify the WITH CHECK OPTION when creating a view using the CREATE VIEW statement:

CREATE VIEW view_name AS
query
WITH CHECK OPTION;

When you create a view WITH CHECK OPTION, PostgreSQL will ensure that you can only modify data of the view that satisfies the condition in the view’s defining query (query).

Scope of check

In PostgreSQL, you can specify a scope of check:

  • LOCAL
  • CASCADED

The LOCAL scope restricts the check option enforcement to the current view only. It does not enforce the check to the views that the current view is based on.

Here’s the syntax for creating a view with the WITH LOCAL CHECK OPTION:

CREATE VIEW view_name AS
query
WITH LOCAL CHECK OPTION;

The CASCADED scope extends the check option enforcement to all underlying views of the current view. Here’s the syntax for creating a view with the WITH CASCADED CHECK OPTION.

CREATE VIEW view_name AS
query
WITH CASCADED CHECK OPTION;

To change the scope of check for an existing view, you can use the ALTER VIEW statement.

PostgreSQL WITH CHECK OPTION examples

Let’s take some examples of using the WITH CHECK OPTION.

Setting up a sample table

The following statements create a new table called employees and insert data into it:

CREATE TABLE employees (
    id SERIAL PRIMARY KEY,
    first_name VARCHAR(50) NOT NULL,
    last_name VARCHAR(50) NOT NULL,
    department_id INT,
    employee_type VARCHAR(20)
       CHECK (employee_type IN ('FTE', 'Contractor'))
);

INSERT INTO employees (first_name, last_name, department_id, employee_type)
VALUES
    ('John', 'Doe', 1, 'FTE'),
    ('Jane', 'Smith', 2, 'FTE'),
    ('Bob', 'Johnson', 1, 'Contractor'),
    ('Alice', 'Williams', 3, 'FTE'),
    ('Charlie', 'Brown', 2, 'Contractor'),
    ('Eva', 'Jones', 1, 'FTE'),
    ('Frank', 'Miller', 3, 'FTE'),
    ('Grace', 'Davis', 2, 'Contractor'),
    ('Henry', 'Clark', 1, 'FTE'),
    ('Ivy', 'Moore', 3, 'Contractor');

1) Basic PostgreSQL WITH CHECK OPTION example

First, create a view called fte that retrieves the employees with the type FTE from the employees table:

CREATE OR REPLACE VIEW fte AS
SELECT
  id,
  first_name,
  last_name,
  department_id,
  employee_type
FROM
  employees
WHERE
  employee_type = 'FTE';

Second, retrieve data from the fte view:

SELECT * FROM fte;

Output:

id | first_name | last_name | department_id
----+------------+-----------+---------------
  1 | John       | Doe       |             1
  2 | Jane       | Smith     |             2
  4 | Alice      | Williams  |             3
  6 | Eva        | Jones     |             1
  7 | Frank      | Miller    |             3
  9 | Henry      | Clark     |             1
(6 rows)

Third, insert a new row into the employees table via the fte view:

INSERT INTO fte(first_name, last_name, department_id, employee_type)
VALUES ('John', 'Smith', 1, 'Contractor');

It succeeds.

The issue is that we can insert an employee with the type of Contractor into the employee table via the view that exposes the employee to the type of FTE.

To ensure that we can insert only employees with the type FTE into the employees table via the fte view, you can use the WITH CHECK OPTION:

Fourth, replace the fte view and add the WITH CHECK OPTION:

CREATE OR REPLACE VIEW fte AS
SELECT
  id,
  first_name,
  last_name,
  department_id,
  employee_type
FROM
  employees
WHERE
  employee_type = 'FTE'
WITH CHECK OPTION;

After adding the WITH CHECK OPTION, you perform insert, update, and delete on employees table that satisfies the WHERE clause in the defining query of the view.

For example, the following INSERT statement will fail with an error:

INSERT INTO fte(first_name, last_name, department_id, employee_type)
VALUES ('John', 'Snow', 1, 'Contractor');

Error:

ERROR:  new row violates check option for view "fte"
DETAIL:  Failing row contains (12, John, Snow, 1, Contractor).

The reason is that the employee_type Contractor does not satisfy the condition defined in the defining query of the view:

employee_type = 'FTE';

But if you modify the row with the employee type FTE, it’ll be fine.

Fifth, change the last name of the employee id 2 to 'Doe':

UPDATE fte
SET last_name = 'Doe'
WHERE id = 2;

It works as expected.

2) Using WITH LOCAL CHECK OPTION example

First, recreate the fte view without using the WITH CHECK OPTION:

CREATE OR REPLACE VIEW fte AS
SELECT
  id,
  first_name,
  last_name,
  department_id,
  employee_type
FROM
  employees
WHERE
  employee_type = 'FTE';

Second, create a new view fte_1 based on the fte view that returns the employees of department 1, with the WITH LOCAL CHECK OPTION:

CREATE OR REPLACE VIEW fte_1
AS
SELECT
  id,
  first_name,
  last_name,
  department_id,
  employee_type
FROM
  fte
WHERE
  department_id = 1
WITH LOCAL CHECK OPTION;

Third, retrieve the data from the fte_1 view:

SELECT * FROM fte_1;

Output:

id | first_name | last_name | department_id | employee_type
----+------------+-----------+---------------+---------------
  1 | John       | Doe       |             1 | FTE
  6 | Eva        | Jones     |             1 | FTE
  9 | Henry      | Clark     |             1 | FTE
(3 rows)

Since we use the WITH LOCAL CHECK OPTION, PostgreSQL checks only the fte_1 view when we modify the data in the employees table via the fte_1 view.

Fourth, insert a new row into the employees table via the fte_1 view:

INSERT INTO fte_1(first_name, last_name, department_id, employee_type)
VALUES ('Miller', 'Jackson', 1, 'Contractor');

It succeeded. The reason is that the INSERT statement inserts a row with department 1 that satisfies the condition in the fte_1 view:

department_id = 1

Fifth, query data from the employees table:

SELECT
  *
FROM
  employees
WHERE
  first_name = 'Miller'
  and last_name = 'Jackson';

Output:

id | first_name | last_name | department_id | employee_type
----+------------+-----------+---------------+---------------
 12 | Miller     | Jackson   |             1 | Contractor
(1 row)

3) Using WITH CASCADED CHECK OPTION example

First, recreate the view fte_1 with the WITH CASCADED CHECK OPTION:

CREATE OR REPLACE VIEW fte_1
AS
SELECT
  id,
  first_name,
  last_name,
  department_id,
  employee_type
FROM
  fte
WHERE
  department_id = 1
WITH CASCADED CHECK OPTION;

Second, insert a new row into employee table via the fte_1 view:

INSERT INTO fte_1(first_name, last_name, department_id, employee_type)
VALUES ('Peter', 'Taylor', 1, 'Contractor');

Error:

ERROR:  new row violates check option for view "fte"
DETAIL:  Failing row contains (24, Peter, Taylor, 1, Contractor).

The WITH CASCADED CHECK OPTION instructs PostgreSQL to check the constraint on the fte_1 view and also its base view which is the fte view.

That’s why the INSERT statement fails the condition of both fte_1 and fte views.

Summary

  • Use the WITH CHECK OPTION clause to enforce constraints on data modifications through views and ensure that only valid data can be changed.

Last updated on

Was this page helpful?