SQL JOINing a Table to Itself

Getting two sets of information from one table in a select statement often leads people to write subselects, but it really doesn't matter that this is the same table twice, we can just give it a new alias and treat it as if it were a different table. This is one of those techniques where, once you've seen it, it's really obvious, but until that point it can be very confusing. I explained this to someone else recently, so I thought I'd capture it here in case it's helpful to anyone else.

Consider that tried-and-tested example: employees and managers. Here's the staff table from the database (today's imaginary data isn't particularly imaginative, sorry):

mysql> select * from staff;
+----+------------+-----------+------------+
| id | first_name | last_name | manager_id |
+----+------------+-----------+------------+
|  1 | Hattie     | Hopkins   |          4 |
|  2 | Henry      | Hopkins   |          4 |
|  3 | Harry      | Hopkins   |          5 |
|  4 | Helen      | Hopkins   |       NULL |
|  5 | Heidi      | Hopkins   |          4 |
|  6 | Hazel      | Hopkins   |          1 |
+----+------------+-----------+------------+
6 rows in set (0.00 sec)

In order to see who reports to whom, we need to query the table to get the employees, and again to get the managers' names. MySQL will get horribly confused if we use the same table twice, so we alias it each time to indicate which one we mean. I do this by drawing the picture of which tables I need and labelling them - here I've used "e" for employees and "m" for managers:

mysql> select e.first_name as employee, m.first_name as reports_to
    -> from staff e
    -> inner join staff m on e.manager_id = m.id;
+----------+------------+
| employee | reports_to |
+----------+------------+
| Hattie   | Helen      |
| Henry    | Helen      |
| Harry    | Heidi      |
| Heidi    | Helen      |
| Hazel    | Hattie     |
+----------+------------+
5 rows in set (0.02 sec)

Did you spot that there's one row less in this output? That's because we did an inner join, but Helen has no manager since she owns the company, so she doesn't appear in the second set of results. We could easily have left joined instead to include her if we wanted to.

You can use this technique regardless of how many other tables are needed for the query and whether you need to join the table to itself or onto other places. You might have a table of users, and a table of user_friends, linking one user to another. You can start from the users table, join to user_friends, then join to user again with a different alias to pull back the information you need - the possiblities are almost endless.

Lorna is an independent web development consultant, writer and trainer, open source project lead and community evangelist. This post was originally published at LornaJane