Ruby on Rails: Search one table where multiple rows must be present in another table

Question

I'm trying to create a search where a single record must have multiple records in another table (linked by id's and has_many statements) in order to be included as a result.

I have tables users, skill_lists, skill_maps.

users are mapped to individual skills through single entries in the skill_maps table. Many user can share a single skill and single user can have many skills trough multiple entries in the skill_maps table.

e.g.

User_id | Skill_list_id
2       | 9
2       | 15
3       | 9

user 2 has skills 9 and 15
user 3 has only skill 9

I'm trying to create a search that returns a hash of all users which have a set of skills. The set of required skill_ids appear as an array in the params.

Here's the code that I'm using:

skill_selection_user_ids = SkillMap.find_all_by_skill_list_id(params[:skill_ids]).map(&:user_id) 

@results = User.find(:all, :conditions => {:id => skill_selection_user_ids})

The problem is that this returns all users that have ANY of these skills not users that have ALL of them.

Also, my users table is linked to the skill_lists table :through => :skill_maps and visa versa so that i can call @user.skill_list etc...

I'm sure this is a real newbie question, I'm totally new to rails (and programming). I searched and searched for a solution but couldn't find anything. I don't really know how to explain the problem in a single search term.

Answer 1

I personally don't know how to do this using ActiveRecord's query interface. The easiest thing to do would be to retrieve lists of users who have each individual skill, and then take the intersection of those lists, perhaps using Set:

require 'set'

skills = [5, 10, 19] # for example
user_ids = skills.map { |s| Set.new(SkillMap.find_all_by_skill_list_id(s).map(&:user_id)) }.reduce(:&)
users = User.where(:id => user_ids.to_a)

For (likely) higher performance, you could "roll your own" SQL and let the DB engine do the work. I may be able to come up with some SQL for you, if you need high performance here. (Or if anyone else can, please edit this answer!)

By the way, you should probably put an index on skill_maps.skill_list_id to ensure good performance even if the skill_maps table gets very large. See the ActiveMigration documentation: http://api.rubyonrails.org/classes/ActiveRecord/Migration.html

Answer 2

You'll probably have to use some custom SQL to get the user IDs. I tested this query on a similar HABTM relationship and it seems to work:

SELECT DISTINCT(user_id) FROM skill_maps AS t1 WHERE (SELECT COUNT(skill_list_id) FROM skill_maps AS t2 WHERE t2.user_id = t1.user_id AND t2.skill_list_id IN (1,2,3)) = 3

The trick is in the subquery. For each row in the outer query, it finds a count of records for that row that match any of the skills that you're interested in. Then it checks whether that count matches the total number of skills you're interested in. If there's a match, then the user must possess all of the skills you searched for.

You could execute this in Rails using find_by_sql:

sql = 'SELECT DISTINCT(user_id) FROM skill_maps AS t1 WHERE (SELECT COUNT(skill_list_id) FROM skill_maps AS t2 WHERE t2.user_id = t1.user_id AND t2.skill_list_id IN (?)) = ?'
skill_ids = params[:skill_ids]
user_ids = SkillMap.find_by_sql([sql, skill_ids, skill_ids.size])

Sorry if the table and column names aren't exactly right, but hopefully this is in the ballpark.