How to use a sum (case) into second table as condition - sql

Question

This is my query:
select matricule,nom_ligne,count(num_ticket) as nbre_ticket, sum(prix) as moant,receveur, nom_gie,
(SELECT SUM(CASE WHEN depense.id_bus = 1 THEN depense.montant ELSE 0 END) AS NbrARRNP from depense) as tot,
receveur, nom_gie from journal_recettes
group by matricule,nom_ligne,receveur, nom_gie
And I obtain this results:
Results
But what I need is to use this condition :
CASE WHEN depense.id_bus = journal_recettes.id_bus THEN depense.montant ELSE 0 END
But When I try, i obtain this error :
Error

Answer 1

Consider modifying your query like below by joining both tables on id_bus column
select j.matricule,
j.nom_ligne,
count(j.num_ticket) as nbre_ticket,
sum(j.prix) as moant,
j.receveur,
j.nom_gie,
SUM(d.id_bus) AS NbrARRNP
from journal_recettes j
join depense d on d.id_bus = j.id_bus
group by j.matricule,
j.nom_ligne,
j.receveur,
j.nom_gie;

Answer 2

You are looking for a correlated subquery. This is usually handled in the where clause:
select matricule, nom_ligne, count(num_ticket) as nbre_ticket,
sum(prix) as moant, receveur, nom_gie,
(select SUM(d.montant) AS NbrARRNP
from depense d
where d.id_bus = jr.id_bus
) as tot,
receveur, nom_gie
from journal_recettes jr
group by matricule, nom_ligne, receveur, nom_gie;
In this case, you may have an issue because the correlation condition is not a group by key. I'm not sure if Postgres supports this construct. If this is a problem, the simplest solution is to include id_bus in the group by, but without sample data, I'm not sure if that works in this case.

Answer 3

Assuming you are doing the join on id_bus, you can use Common Table Expressions (http://www.postgresql.org/docs/9.5/static/queries-with.html)
with sums as (SELECT id_bus, sum(CASE
WHEN depense.id_bus = 1 THEN depense.montant
ELSE 0
END) AS NbrARRNP
FROM depense
group by id_bus)
SELECT matricule,
nom_ligne,
Count(num_ticket) AS nbre_ticket,
Sum(prix) AS moant,
receveur,
nom_gie,
d.NbrARRNP AS tot,
receveur,
nom_gie
FROM journal_recettes j left outer join sums d on j.id_bus = d.id_bus
GROUP BY matricule,
nom_ligne,
receveur,
nom_gie,
j.id_bus

Answer 4

Many thanks to all.
I post my final solution to contribute :
select p1.matricule,p1.nom_ligne,
Count(p1.num_ticket) as nbre_ticket, Sum(p1.prix) as montant,
(select sum(montant) as tot from depense p2
where date_trunc('day',p2.date_depense) = date_trunc('day',p1.date_vendu)
group by date_depense) som_depense ,
(Sum(p1.prix) -((select sum(montant) as tot from depense p3
where date_trunc('day',p3.date_depense) = date_trunc('day',p1.date_vendu)
group by date_depense) ) ) net_reverse ,to_char(p1.date_vendu,'YYYYMMDD'),
p1.receveur, p1.nom_gie, id_line
from journal_recettes p1
group by
matricule,nom_ligne, date_vendu,receveur,nom_gie, id_line
And it's work as well.