Question

I’m trying to do a simple Post/Tags relation in rails 3.
Everything working fine except when I want to query the Posts which are related to several tags.
Basically I’d like to do this kind of thing :

Post.joins(:tags).where('tags.name ILIKE ?', var)

But instead of having just one var, I’d like to use an array.
I tried :

Post.joins(:tags).where('tags.name IN (?)', names_array)

But unfortunately it does a simple LIKE (not ILIKE) and works like a OR condition which sounds perfectly logical.

I also found another solution by using find_by_sql in this post
http://snippets.dzone.com/posts/show/32

But it seems a bit ugly to me.

To better understand the problem.
I’ve got 3 posts :
PostA
PostB
PostC

PostA is related to TagA TagB and TagC tags.
PostB is related to TagA and TagB tags.
PostC is only related to TagA.

If I look for TagA and TagC Posts I’d like to finds PostA because it is related to both Tags.
Using a hash condition returns PostA PostB and PostC.
What I want is the Posts which are related to “at least” all the specified Tags.

So anyone has a better way to handle this ?

Thanks.

Answer 1

I don’t know if it will solve you problem but for complex queries like that I almost always just use Squeel.

Then do something like this:

@posts = Post.joins(:tags)
  .where{tags.name.like_any names_array}
  .group("post_id")
  .having("count(post_id) = #{names_array.size}")

The SQL hopefully looks something like this

SELECT "posts".* FROM "posts"
  INNER JOIN "tags" ON "tags"."post_id" = "posts"."id"
  WHERE (("tags"."name" LIKE "TagA" OR "tags"."name" LIKE "TagB"))
  GROUP BY post_id
  HAVING count(post_id) = 2

If I remember squeel is pretty good at using ILIKE instead of LIKE depending on the database used. (atleast better than AR)

Also you could do this using AR without squeel but I REALLY like some of the ideas and helpers that come with squeel like _all

---

As for an explination…

Assume I searched for TagsA and B.

What that does is finds all the Posts with those tags.

So you’ll have something like:

• PostA TagA
• PostA TagB
• PostB TagA
• PostB TagB
• PostC TagA

Then it will group all those different Post results by the joined tags using post_id.

• PostA TagA TagB
• PostB TagA TagB
• PostC TagA

Then it will check the number of Tags the SQL line has by checking how many forgien_ids are present.
Since A and B have 2 tags you know it matched all you input.