SQL Trigger to allow only customers who are old enough to buy a book? - sql

Question

I have 3 tables, person, audiobook and audiobook_purchases. My database is running MariaDB.
person has fields: id, date_of_birth;
audiobook has fields: ISBN, age_rating, title;
audiobook_purchases has fields: ISBN, customer_id, date_of_purchase;
I'm trying to write a trigger to make sure that when a customer tried to purchases an audiobook, they are old enough to do so according to the age_rating in audiobook.
For example, If Harry Potter and the Philosipher's Stone had age rating 16 and customer Jennifer (ID 1) with date of birth 2010-01-01 tried to purchase this book, this would not be allowed, but Dominick(ID 2) with date_of_birth 1978-01-01 would be allowed.
Please could someone show me a way to run this trigger?

Answer 1

I don't know MariaDB in particular, so my answer may need some adjustments.
You want to create an insert trigger on audiobook_purchase so that a new order will be inserted only if the person who wants to place the order is old enough according to audiobook.age_rating.
First you need to figure out a way of extracting the year from person.date_of_birth. Something like the YEAR() scalar function will probably be available. MariaDB may also provide a NOW() function, which gives the current date. So the person age right now will be: YEAR(NOW()) - YEAR(person.date_of_birth).
Then you have to write the insert trigger. The tricky part is to query the person table to get the person date_of_birth from his id, then to compare it to audiobook.age_rating.
Let's set out an example. First we declare the tables schemas:
CREATE TABLE person(id, name, date_of_birth);
CREATE TABLE audiobook(isbn, age_rating, title);
CREATE TABLE audiobook_purchases(isbn, customer_id, date_of_purchase);
Then we put in some data:
INSERT INTO person VALUES (10, "jennifer", '2010-01-01');
INSERT INTO person VALUES (20, "dominick", '1978-01-01');
INSERT INTO audiobook VALUES (1234, 16, "harry potter");
Then we create the trigger:
CREATE TRIGGER check_purchases
AFTER INSERT ON audiobook_purchases
FOR EACH ROW
WHEN (
SELECT strftime('%Y', 'now') - strftime('%Y', date_of_birth) AS age
FROM person
WHERE new.customer_id=person.id) < (
SELECT audiobook.age_rating
FROM audiobook
WHERE audiobook.isbn=new.isbn)
BEGIN
DELETE FROM audiobook_purchases
WHERE isbn=new.isbn AND
customer_id=new.customer_id AND
date_of_purchase=new.date_of_purchase;
END;
I'll broke down the trigger into smaller steps:
AFTER INSERT ON audiobook_purchases creates a trigger on table audiobook_purchases which will be triggered after the insertion of a new record.
FOR EACH ROW applies the trigger to each new record inserted.
The WHEN clause limits triggering only to those records who satisfy its condition. On the left side of the < sign of the condition there is a query which selects the age of the customer. On the right side there is a query which selects the age rating of the book. Notice the reference to a new table. This table stores the record which triggers the event (see the two examples below). strftime is a scalar function which formats datetime stamps in SQLite. You can read:
strftime('%Y', 'now') as YEAR(NOW()) and
strftime('%Y', date_of_birth) as YEAR(date_of_birth).
Finally between BEGIN and END there are instructions that will be executed on triggering. In this case there is a single instruction which removes the record just inserted. MariaDb may provide a ROLLBACK statement, which can be more efficient than the DELETE statement.
So, for example:
INSERT INTO audiobook_purchases VALUES (1234, 10, '2018-11-25');
will activate the trigger, because the customer with id=10 ('jennifer') is 8 years old and the book with isbn=1234 requires the customer to be at least 16 years old, while:
INSERT INTO audiobook_purchases VALUES (1234, 20, '2018-11-25');
will not activate the trigger, because this customer is 40 years old.
You must be aware that this solution silently ignore the invalid order. I don't know if this is your desired behaviour.
I tested this trigger on SQLite 3.11.0, so it may not be compatible with your SQL interpreter.