Refactor a query with two subselects in WHERE - ruby-on-rails

Question

I've got the following scope on the Project model to find projects to which at least 80 crowdfunding pledges have been done or projects which have reached their goal already:
class Project < ActiveRecord::Base
scope :popular, -> { collecting.where("80 <= (?) OR goal <= (?)",
Pledge.select('COUNT(*)').where("project_id = projects.id").where(paid: true),
Pledge.select('SUM(amount)').where("project_id = projects.id").where(paid: true))
}
(...)
end
This works just fine and produces the following SQL for Postres:
Project.popular
# SELECT "projects".* FROM "projects" WHERE "projects"."state" = 'collecting' AND (80 <= (SELECT COUNT(*) FROM "pledges" WHERE (project_id = projects.id) AND "pledges"."paid" = 't') OR goal <= (SELECT SUM(amount) FROM "pledges" WHERE (project_id = projects.id) AND "pledges"."paid" = 't'))
It's important to get the count as well:
Project.popular.count
# SELECT COUNT(*) FROM "projects" WHERE "projects"."state" = 'collecting' AND (80 <= (SELECT COUNT(*) FROM "pledges" WHERE (project_id = projects.id) AND "pledges"."paid" = 't') OR goal <= (SELECT SUM(amount) FROM "pledges" WHERE (project_id = projects.id) AND "pledges"."paid" = 't'))
Okay, the subselects are nice, but I have this feeling that there's a better and more efficient way to do this. I've tried joins and sum, but the mandatory group to use aggregate functions breaks Project.popular.count.
Any ideas how to refactor this? Maybe just a way to do where("project_id = projects.id") in Hash notation?

Answer 1

I don't know if this is any good for you because I can't express the below query in the ActiveRecord lingo but if you can it will be a big improvement over the poor query currently being generated
select "projects".*, project_count, project_amount
from
"projects"
inner join (
select
id,
count(*) as project_count,
sum(amount) as project_amount
from pledges
group by id
where paid
) pledges using (id)
where
"projects"."state" = 'collecting'
and
(project_count >= 80 or project_amount >= goal)

Answer 2

Thanks, #Clodonaldo! With a few litte touches, your query works like a charm:
select "projects".*, pledges_count, pledges_sum
from
"projects"
inner join (
select
project_id as id,
count(*) as pledges_count,
sum(amount) as pledges_sum
from pledges
where paid
group by project_id
) pledges using (id)
where
"projects"."state" = 'collecting'
and
(pledges_count >= 80 or pledges_sum >= goal)
Not sure thou whether AR or AREL can build this.
UPDATE:
The shortest I can come up with is (I don't need the count and sum values):
Project.collecting.joins(
"INNER JOIN (",
Pledge.paid.group(:project_id).select("
project_id AS id,
COUNT(*) AS count,
SUM(amount) AS sum
").to_sql,
") pledges USING (id)"
).where("count >= 80 OR sum >= goal")
The above produces:
SELECT "projects".*
FROM
"projects"
INNER JOIN (
SELECT
project_id AS id,
COUNT(*) AS count,
SUM(amount) AS sum
FROM "pledges"
WHERE "pledges"."paid" = 't'
GROUP BY project_id
) pledges USING (id)
WHERE
"projects"."state" = 'collecting'
AND
(count >= 80 OR sum >= goal)