CODEWITHSUNDEEP
sqlpostgresqljoincount
Ggolo
Ggolo

Reputation: 405

How to sort by count with postgresql?

I have two tables:

Companies: (id, name, city)
Workers: (id, name)

I would like to get all companies and sort them by numbers of employes.

The result should give:

count | company id | company name | city
------------------------------------------
90         6           foo corp      NY
45         9           bar corp      LA
0          3         foobar corp     HO

I tried:

select 
    c.*, 
    count(w.id) as c 
from 
    companies c 
left join 
    workers w 
on 
    c.id = w.company_id 
group by 
    c.id 
order by 
    c desc;

But that's not working as it tells me to group by g.name too :/

Any ideas?

Upvotes: 34

Views: 33695

Answers (3)

aardbol
aardbol

Reputation: 2295

If you don't want the count result to be returned (because of an ORM framework or so), you could apply it directly in the order by clause:

select 
    c.*
from 
    companies c 
left join 
    workers w 
on 
    c.id = w.company_id 
group by 
    c.id 
order by 
    count(w.id) desc;

Tested with postgreSQL 11

Upvotes: 8

David Andres
David Andres

Reputation: 31801

Try this as a subquery:

SELECT C.*
FROM 
(
  SELECT C.Id, C.Company_Name, C.City, COUNT(W.Id) AS CNT
  FROM Companies C
  LEFT JOIN Workers W ON W.Company_Id = C.Id
  GROUP BY C.Id, C.Company_Name, C.City
) T
ORDER BY T.CNT

Upvotes: 1

Eric
Eric

Reputation: 95173

You've aliased the table and column as the same thing, so don't do that. It's not invalid, just tough to follow.

Anyway, include all columns that you're selecting that aren't aggregates in your group by:

select 
    count(w.id) as mycount,
    w.company_id,
    c.company_name,
    c.city
from 
    companies c 
    left join workers w on 
        c.id=w.company_id 
group by 
    w.company_id,
    c.company_name,
    c.city
order by mycount desc;

Upvotes: 42

Related Questions