From: | Andreas Joseph Krogh <andreak(at)officenet(dot)no> |
---|---|
To: | pgsql-sql(at)postgresql(dot)org |
Subject: | Problem with UNION-queries |
Date: | 2007-11-12 20:27:14 |
Message-ID: | 200711122127.14860.andreak@officenet.no |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-sql |
Hi all
I have a somewhat furry solution to a problem for which I think there might be
a better way to do it.
The table looks like this (simplified for the sake of this example):
drop table if exists test1;
create table test1(
id serial primary key,
key varchar,
username varchar,
value varchar
);
-- Insert test-data
insert into test1(username, key, value) values('andreak', 'A', 'value1');
insert into test1(username, key, value) values('andreak', 'A', 'value2');
insert into test1(username, key, value) values('andreak', 'A', 'value3');
insert into test1(username, key, value) values('andreak', 'B', 'value2');
insert into test1(username, key, value) values('andreak', 'B', 'value3');
insert into test1(username, key, value) values('andreak', 'B', 'value4');
insert into test1(username, key, value) values('andreak', null, 'value3');
insert into test1(username, key, value) values('andreak', null, 'value4');
insert into test1(username, key, value) values('andreak', null, 'value5');
For the sake of this example the test-case is greatly simplified, so I have
the following query to give me all rows for value='A' and value='B'
select t.username, t.key, t.value from test1 t where t.key = 'A'
UNION
select t.username, t.key, t.value from test1 t where t.key = 'B'
username | key | value
----------+-----+--------
andreak | A | value1
andreak | A | value2
andreak | A | value3
andreak | B | value2
andreak | B | value3
andreak | B | value4
(6 rows)
Again, I know there are other, better, ways to accomplish that with this
simple schema, but again it's for the sake of the example. The important
thing here is that it's 2 UNION-queries providing the result.
Now - what I'm trying to accomplish is getting the following result-set:
username | key | value
----------+-----+--------
andreak | A | value1
andreak | A | value2
andreak | A | value3
andreak | B | value2
andreak | B | value3
andreak | B | value4
andreak | | value5
(7 rows)
That is - I want all rows with username='andreak' AND (key IS NULL) where
the "value" is not in the previous result (not part of the other
UNION-queries). The hard, and important, part is that the resulting
rows' "value" must not exist in the "value"-column for any previous rows.
Here is one way I figured out how to do it:
select t.username, t.key, t.value from test1 t where t.key = 'A'
UNION
select t.username, t.key, t.value from test1 t where t.key = 'B'
UNION
select t.username, t.key, t.value from test1 t where t.value NOT IN (
select value from (
select t.username, t.key, t.value from test1 t where t.key = 'A'
UNION
select t.username, t.key, t.value from test1 t where t.key = 'B'
) tmp1
)
;
Given that my real schema is way more complex I'm looking for a solution which
doesn't involve issuing the NOT IN (original UNION-query) as it is a rather
heavy query.
Does anybody have a better approach to this problem?
--
Andreas Joseph Krogh <andreak(at)officenet(dot)no>
Senior Software Developer / Manager
------------------------+---------------------------------------------+
OfficeNet AS | The most difficult thing in the world is to |
Karenslyst Allé 11 | know how to do a thing and to watch |
PO. Box 529 Skøyen | somebody else doing it wrong, without |
0214 Oslo | comment. |
NORWAY | |
Tlf: +47 24 15 38 90 | |
Fax: +47 24 15 38 91 | |
Mobile: +47 909 56 963 | |
------------------------+---------------------------------------------+
From | Date | Subject | |
---|---|---|---|
Next Message | Fernando Hevia | 2007-11-12 21:48:19 | Re: design of tables for sparse data |
Previous Message | Sabin Coanda | 2007-11-12 17:12:36 | Re: show value of backslashes in string array argument |