Postgresql – get an error when filtering subSelect

aliaspostgresql

select "masters".*, (select avg (rating) 
from "reviews" where "master_id" = "masters"."id") as rating 
from "masters" where rating > 3

[42703] ERROR: column "rating" does not exist

Best Answer

You can't use a column alias on the same level where you defined it. You need to wrap the query:

select *
from (
  select masters.*, (select avg(rating) 
                     from reviews 
                     where master_id = masters.id) as rating 
  from masters 
) t
where rating > 3;

A more efficient solution is however, to do the aggregation once for all rows and then join to the result:

select m.*, r.rating
from masters m
  join (
     select master_id, avg(rating)  as rating
     from reviews 
     group by master_id
     having avg(rating) > 3
  ) r on r.master_id = m.id;

Related Solutions

Postgresql – How to find table name for each row result

There must be some kind of misunderstanding. tableoid is a system column that's available for any regular table in any version of Postgres since at least v7.3. Per documentation:

Every table has several system columns that are implicitly defined by the system.

tableoid is one of them. The solution I provided under the questions you refer to is also suggested in the manual here. And it works. I have been using it for years in various Postgres versions.

Either you are not using Postgres, or template is not a table. Is it a VIEW maybe? If you define a VIEW like:

CREATE VIEW template AS
SELECT * FROM some_table;

Then system columns are not included by default.

What do you get for

SELECT relnamespace, relname, relkind
FROM   pg_class
WHERE  relname = 'template';

Is relkind 'r', 'v' or something else?
Or do you get multiple rows? Then consider this:

Why does a PostgreSQL SELECT query return different results when a schema name is specified?

PostgreSQL CASE Expression – Resolving ‘Column Doesn’t Exist’ Error in WHERE Clause

The WHERE clause is evaluated before aliases in the SELECT clause. WHERE is used to locate rows from the base table which are the input to all expressions in the SELECT. Only after locating the rows your CASE can be evaluated with real values and the final_price alias is assigned its value.

There are multiple ways to solve that:

repeat the same expression in the WHERE clause - that way it will be used to locate the rows
wrap your query as a subquery in another query and move the WHERE to the outer query - as you select from a derived table, the alias becomes a "real" column that way
_{(this would work in MySQL, but not in PostgreSQL) use the HAVING clause instead of WHERE - HAVING (in MySQL) works after the SELECT is computed to do a "second wave" of filtering. In standard SQL (and in Postgres), SELECT is logically processed after HAVING, so this won't work.}
create a generated/virtual column from your CASE expression, that way it becomes a part of your base table and you might even be able to index it

Warning: your query will need to do fulltable/index scan unless you can add some "simpler" WHERE conditions or be able to use the indexed virtual column.

Best Answer

Related Solutions

Postgresql – How to find table name for each row result

PostgreSQL CASE Expression – Resolving ‘Column Doesn’t Exist’ Error in WHERE Clause

Related Question