CODEWITHSUNDEEP
sqlpostgresqlplpgsqlcommon-table-expression
Renato Dinhani
Renato Dinhani

Reputation: 36676

Reuse a complex query result in other queries without redoing the complex query

I have a complex query in PostgreSQL and I want to use the result of it in other operations like UPDATEs and DELETEs, something like:

<COMPLEX QUERY>;
UPDATE WHERE <COMPLEX QUERY RESULT> = ?;
DELETE WHERE <COMPLEX QUERY RESULT> = ?;
UPDATE WHERE <COMPLEX QUERY RESULT> = ?;

I don't want to have to do the complex query one time for each operations. One way to avoid this is store the result in a table and use it for the WHERE and JOINS and after finishing, drop the temporary table.

I want to know if there is another way without storing the results to database, but already using the results in memory.

I already use loops for this, but I think doing only one operation for each thing will be faster than doing the operations per row.

Upvotes: 0

Views: 1487

Answers (2)

Erwin Brandstetter
Erwin Brandstetter

Reputation: 656441

You can loop through the query results like @phatfingers demonstrates (probably with a generic record variable or scalar variables instead of a rowtype, if the result type of the query doesn't match any existing rowtype). This is a good idea for few resulting rows or when sequential processing is necessary.

For big result sets your original approach will perform faster by an order of magnitude. It is much cheaper to do a mass INSERT / UPDATE / DELETE with one SQL command than to write / delete incrementally, one row at a time.

A temporary table is the right thing for reusing such results. It gets dropped automatically at the end of the session. You only have to delete explicitly if you want to get rid of it right away or at the end of a transaction. I quote the manual here:

Temporary tables are automatically dropped at the end of a session, or optionally at the end of the current transaction.

For big temporary tables it might be a good idea to run ANALYZE after they are populated.

Writeable CTE

Here is a demo for what Pavel added in his comment:

CREATE TEMP TABLE t1(id serial, txt text);
INSERT INTO t1(txt)
VALUES ('foo'), ('bar'), ('baz'), ('bax');

CREATE TEMP TABLE t2(id serial, txt text);
INSERT INTO t2(txt)
VALUES ('foo2'),('bar2'),('baz2');

CREATE TEMP TABLE t3 (id serial, txt text);

WITH x AS (
    UPDATE t1
    SET    txt = txt || '2'
    WHERE  txt ~~ 'ba%'
    RETURNING txt
    )
, y AS (
    DELETE FROM t2
    USING  x
    WHERE  t2.txt = x.txt
    RETURNING *
    )
INSERT INTO t3
SELECT *
FROM   y
RETURNING *;

Read more in the chapter Data-Modifying Statements in WITH in the manual.

Upvotes: 2

phatfingers
phatfingers

Reputation: 10250

DECLARE
    r foo%rowtype;
BEGIN
    FOR r IN [COMPLEX QUERY]
    LOOP
    -- process r
    END LOOP;
    RETURN;
END

Upvotes: 1

Related Questions