How can I select only the rows from a table with ids that don't exist in an inner join in PostgreSQL? - sql

Question

I have the following table
postgres=# select * from joins_example;
user_id | price | id | email
---------+--------+----+--------------------------
1 | $30.00 | |
5 | $50.00 | |
7 | $20.00 | |
| | 1 | hadil#example.com
| | 5 | saiid#example.com
| | 2 | fahir#example.com
6 | $60.00 | 6 | oma#example.com
8 | $40.00 | 8 | nasim#example.com
| | 8 | nasim.hassan#example.com
9 | $40.00 | 9 | farah#example.com
9 | $70.00 | |
10 | $80.00 | | majid#example.com
| | 10 | majid.seif#example.com
(13 rows)
A self inner join between user_id and id produces
postgres=# select * from joins_example as x inner join joins_example as y on x.user_id = y.id;
user_id | price | id | email | user_id | price | id | email
---------+--------+----+-------------------+---------+--------+----+--------------------------
1 | $30.00 | | | | | 1 | hadil#example.com
5 | $50.00 | | | | | 5 | saiid#example.com
6 | $60.00 | 6 | oma#example.com | 6 | $60.00 | 6 | oma#example.com
8 | $40.00 | 8 | nasim#example.com | | | 8 | nasim.hassan#example.com
8 | $40.00 | 8 | nasim#example.com | 8 | $40.00 | 8 | nasim#example.com
9 | $40.00 | 9 | farah#example.com | 9 | $40.00 | 9 | farah#example.com
9 | $70.00 | | | 9 | $40.00 | 9 | farah#example.com
10 | $80.00 | | majid#example.com | | | 10 | majid.seif#example.com
(8 rows)
What I want is either:
user_id | price | id | email | user_id | price | id | email
---------+--------+----+-------------------+---------+--------+----+--------------------------
7 | $50.00 | | | | | |
| | | | | | 2 | fahir#example.com
or:
user_id | price | id | email | user_id | price | id | email
---------+--------+----+-------------------+---------+--------+----+--------------------------
| | | | 7 | $50.00 | |
| | 2 | fahir#example.com | | | |
Even
user_id | price | id | email
---------+--------+----+--------------------------
5 | $50.00 | |
| | 2 | fahir#example.com
would be a good start.
Specifically I want to know how to select only the rows from joins_example with user_ids or ids that don't exist in the inner join.

Answer 1

You could consider an approach that uses correlated subqueries with NOT EXISTS conditions:
select *
from joins_example as x
where
(
x.user_id is not null
and not exists (
select 1 from joins_example y where x.user_id = y.id
)
)
or (
x.id is not null
and not exists (
select 1 from joins_example y where x.id = y.user_id
)
)
Demo on DB Fiddle:
| user_id | price | id | email |
| ------- | ----- | --- | ----------------- |
| 7 | 20.00 | | |
| | | 2 | fahir#example.com |

Answer 2

SELECT *
FROM joins_example j
WHERE (j.user_id IS NULL AND j.id IS NULL)
OR (j.user_id IS NOT NULL AND NOT EXISTS(SELECT 1 FROM joins_example j2 WHERE j2.id = j.user_id))
OR (j.id IS NOT NULL AND NOT EXISTS(SELECT 1 FROM joins_example j2 WHERE j2.user_id = j.id));

Answer 3

SELECT *
FROM joins_example AS w
LEFT JOIN (
select x.user_id
from joins_example as x
inner join joins_example as y on x.user_id = y.id
) AS z ON z.user_id = w.user_id or z.user_id = w.id
WHERE z.user_id IS NULL;
Is a good enough start i.e.
user_id | price | id | email | user_id
---------+--------+----+-------------------+---------
7 | $20.00 | | |
| | 2 | fahir#example.com |