How exactly do aliases work in Oracle databases? - sql

Question

I've been dabbling around in sql code and recently was reading up on aliases. I am kind of confused why the following statement does not work:
select id, data from table1 a
inner join
(
select id, data from table2 b,
(
select id, data from table3 e
where b.id = e.id
) c
where b.id = a.id
)d on a.id = d.id
What I want is something like this to work:
select id, data from table1 a
inner join
(
select id, data from
(
select id, data from table3 e
where a.id = e.id
) c
)d on a.id = d.id
Currently my solution doesn't have the WHERE class at the end, meaning all of the table gets fetched.
...
where a.id = e.id
...
My point here, would be to use an ID present in table A in the table E. I'm open to suggestions as to changing the structure, but unfortunately I think the structure will have to stay the same since the actual query is much more complex. This is just an exert from the full query.
EDIT:
I'll try to elaborate as to why I have the current structure.
I have table 1 which contains ID's and text and other columns.
| id | data |
| -------- | ------ |
| table1_1 | text |
| table1_2 | text |
...
The second table contains multiple entries for an ID of table1.
| id | refid | data |
| -------- | -------- | ------ |
| table2_1 | table1_1 | proj1 |
| table2_1 | table1_1 | proj2 |
| table2_2 | table1_1 | proj1 |
| table2_3 | table1_2 | proj5 |
| table2_3 | table1_2 | proj1 |
What I now do is join the entries from table2 to a list of entries with:
LISTAGG(table2.refid, ',') WITHIN GROUP( ORDER BY table2.refid) list_of_projects,
To use this, I need to use group by
My problem was, that I couldn't use the table1.ID in table2.refid.

Answer 1

For better understanding of how sub-queries work, just imagine that database processes them separate from each other.
It means the sub-query
select id, data
from table3 e
where b.id = e.id
will be executed first. There is no alias b in this context -> an error
The next sub-query has the same problem
select id, data from table2 b,
(
select id, data from table3 e
where b.id = e.id
) c
where b.id = a.id
There is no data source called "a" -> another error
And to be honest, using sub-queries in this case is a bad idea. Join is what you need here
I believe something like this will help you out.
select a.id, a.data
from table1 a
inner join table3 e
on a.id = e.id;

Answer 2

From Oracle 12c, you can use CROSS APPLY or a LATERAL join to pass the outer scope into the inner sub-query:
SELECT id, data
FROM table1 a
CROSS APPLY
(
SELECT data
FROM table2 b
CROSS APPLY (
SELECT data
FROM table3 e
WHERE a.id = e.id
) c
WHERE b.id = a.id
) d
or:
SELECT a.id, data
FROM table1 a
INNER JOIN LATERAL
(
SELECT b.id,
c.data
FROM table2 b
INNER JOIN LATERAL (
SELECT e.id,
e.data
FROM table3 e
) c
ON ( a.id = c.id )
) d
ON ( d.id = a.id )
Which, for the sample data:
CREATE TABLE table1 ( id ) AS
SELECT 1 FROM DUAL;
CREATE TABLE table2 ( id ) AS
SELECT 1 FROM DUAL;
CREATE TABLE table3 ( id, data ) AS
SELECT 1, 'A' FROM DUAL;
Both output:
ID | DATA
-: | :---
1 | A
db<>fiddle here