PostgreSQL : LEFT JOIN with tempory table adding extra lines - postgresql

Question

I have a view storing water pipes references together with details on operations on the water system. I need to extract from that view the water pipes on which more than one operation were registered over the past 12 months. Here is how I proceed :
Here is the view structure and a sample of data:
CREATE TABLE schema.pipe (
id INTEGER,
code VARCHAR,
ope_date DATE,
ope_type VARCHAR(2),
system VARCHAR(2));
INSERT INTO schema.pipe (code, ope_date, ope_type, system) VALUES
('0001', '2014-11-11', '01', 'EU'),
('0001', '2014-11-11', '03', 'EU'),
('0002', '2014-12-03', '03', 'EP'),
('0002', '2014-01-03', '03', 'EP'),
('0003', '2014-08-11', '01', 'EP'),
('0003', '2014-03-03', '03', 'EP'),
('0003', '2012-02-27', '03', 'EP'),
('0004', '2014-08-11', '01', 'UN'),
('0004', '2013-12-30', '03', 'UN'),
('0004', '2013-06-01', '03', 'UN'),
('0004', '2012-07-31', '03', 'UN'),
('0005', '2013-10-01', '03', 'EU'),
('0005', '2012-11-01', '03', 'EU'),
('0006', '2014-04-01', '01', 'UN'),
('0006', '2014-05-15', '01', 'UN');
code is the pipe reference
ope_date is the operation date
ope_type is the operation type
system is the system type
Here is the query I'm using:
SELECT code, ope_date FROM schema.pipe
WHERE (NOW()::DATE - ope_date) < 365
GROUP BY code, ope_date
HAVING count(*) = 1 ;
I get this:
code | ope_date
---------+--------------
0002 | 2014-12-03
0002 | 2014-01-03
0003 | 2014-08-11
0003 | 2014-03-03
0004 | 2013-12-30
0004 | 2014-08-11
0006 | 2014-04-01
0006 | 2014-05-15
Now, I need to bring back the other columns with this selection. So I use:
WITH temptable AS (
SELECT code, ope_date FROM schema.pipe WHERE (NOW()::DATE - ope_date) < 365 GROUP BY code, ope_date HAVING count(*) = 1)
SELECT DISTINCT a.code, a.ope_date, b.ope_type, b.system FROM temptable a LEFT JOIN schema.pipe b on a.code = b.code ;
I get this, which is too many lines (I need 8 lines and I get 12):
code | ope_date | ope_type | system
-------+--------------+------------+---------
0002 | 2014-01-03 | 03 | EP
0002 | 2014-12-03 | 03 | EP
0003 | 2014-03-03 | 01 | EP
0003 | 2014-03-03 | 03 | EP
0003 | 2014-08-11 | 01 | EP
0003 | 2014-08-11 | 03 | EP
0004 | 2013-12-30 | 01 | UN
0004 | 2013-12-30 | 03 | UN
0004 | 2014-08-11 | 01 | UN
0004 | 2014-08-11 | 03 | UN
0006 | 2014-04-01 | 01 | UN
0006 | 2014-05-15 | 01 | UN
So here comes my question: how can I get just the lines matching my selection?
Many thanks in advance.
EDIT :
What I need is:
code | ope_date | ope_type | system
---------+-------------+-----------+---------
0002 | 2014-12-03 | 03 | EP
0002 | 2014-01-03 | 03 | EP
0003 | 2014-08-11 | 01 | EP
0003 | 2014-03-03 | 03 | EP
0004 | 2013-12-30 | 03 | UN
0004 | 2014-08-11 | 01 | UN
0006 | 2014-04-01 | 01 | UN
0006 | 2014-05-15 | 01 | UN

Answer 1

I've found myself a solution that consists in using both code and ope_date columns to join the tables (not just code):
WITH temptable AS (
SELECT code, ope_date FROM schema.pipe WHERE (NOW()::DATE - ope_date) < 365 GROUP BY code, ope_date HAVING count(*) = 1)
SELECT DISTINCT a.code, a.ope_date, b.ope_type, b.system FROM temptable a, schema.pipe b WHERE a.code = b.code AND a.ope_date = b.ope_date;
Any comment on this solution?