Correctly join multiple many-to-many tables - MySQL query - mysql

Question

a seemingly generic SQL query really left me clueless.
Here's the case.
I have 3 generic tables (simplified versions here):
Movie
id | title
-----------------------
1 | Evil Dead
-----------------------
2 | Bohemian Rhapsody
....
Genre
id | title
-----------------------
1 | Horror
-----------------------
2 | Comedy
....
Rating
id | title
-----------------------
1 | PG-13
-----------------------
2 | R
....
And 2 many-to-many tables to connect them:
Movie_Genre
movie_id | genre_id
Movie_Rating
movie_id | rating_id
The initial challenge was to write a query which allows me to fetch movies that belong to multiple genres (e.g. horror comedies or sci-fi action).
Thankfully, I was able to find this solution here
MySQL: Select records where joined table matches ALL values
However, what would be the correct option to fetch records that belong to multiple many-to-many tables? E.g. rated R horror comedies. Is there any way to do so without subquery (or a single one only)?

Answer 1

One method uses correlated subqueries:
select m.*
from movies m
where (select count(*)
from movie_genre mg
where mg.movie_id = m.id
) > 1 and
(select count(*)
from movie_rating mr
where mr.movie_id = m.id
) > 1 ;
With indexes on movie_genre(movie_id) and movie_rating(movie_id) this probably has quite reasonable performance.
The above is possibly the most efficient method. However, if you wanted to avoid subqueries, one method would be:
select mg.movie_id
from movie_genres mg join
movie_ratings mr
on mg.movie_id = mr.movie_id
group by mg.movie_id
having count(distinct mg.genre_id) > 0 and
count(distinct mr.genre_id) > 0;
More efficient than the above is aggregating before the join:
select mg.movie_id
from (select movie_id
from mg_genres
group by movie_id
having count(*) >= 2
) mg join
(select movie_id
from mg_ratings
group by movie_id
having count(*) >= 2
) mr
on mg.movie_id = mr.movie_id;
Although you state that you want to avoid subqueries, the irony is that the version with no subqueries probably has the worst performance of these three options.

Answer 2

E.g. rated R horror comedies
You can join all the tables together, aggregate by movie and filter with a HAVING clause:
select m.id, m.title
from movies m
inner join movie_genre mg on mg.movid_id = m.id
inner join genre g on g.id = mg.genre_id
inner join movie_rating mr on mr.movie_id = m.id
inner join rating r on r.id = mr.rating_id
group by m.id, m.title
having
max(r.title = 'R') = 1
and max(g.title = 'Horror') = 1
and max(g.title = 'Comedy') = 1
You can also use a couple of exists conditions along with correlated subqueries:
select m.*
from movie m
where
exists (
select 1
from movie_genre mg
inner join genre g on g.id = mg.genre_id
where mg.movie_id = m.id and g.title = 'R')
and exists (
select 1
from movie_rating mr
inner join rating r on r.id = mr.rating_id
where mr.movie_id = m.id and r.title = 'Horror'
)
and exists (
select 1
from movie_rating mr
inner join rating r on r.id = mr.rating_id
where mr.movie_id = m.id and r.title = 'Comedy'
)