Mongoid query referencing an ID but not using an embedded document

Question

I have a Mongo collection that simply references an ID to another collection. Hypothetically the collection I'm specifically referring to might be called:

Walks. Walks has a reference to owner_id. The owner takes many walks with his many pets every day. What I want to do is query Walks for a list of N owner_ids and get only the last walk they took for each owner and group that by owner_id. To get a list of all walks by said list we'd do something like.

Walk.any_in(:owner_id => list_of_ids)

My question is, is there a way to query that list_of_ids, get only one walk per owner_id (the last one they took which can be sorted by the field created_at and returned in a hash where each walk is pointed to by an owner_id such as:

{ 5 => {..walk data..}, 10 => {.. walk data ..}}

Answer 1

Here's an answer that uses MongoDB's group command. For the purposes of testing, I've used walk_time instead of created_at . Hope that this helps and that you like it.

class Owner
  include Mongoid::Document
  field :name, type: String
  has_many :walks
end

class Walk
  include Mongoid::Document
  field :pet_name, type: String
  field :walk_time, type: Time
  belongs_to :owner
end

test/unit/walk_test.rb

require 'test_helper'

class WalkTest < ActiveSupport::TestCase
  def setup
    Owner.delete_all
    Walk.delete_all
  end

  test "group in Ruby" do
    walks_input = {
        'George' => [ ['Fido',  2.days.ago], ['Fifi',  1.day.ago],  ['Fozzy',    3.days.ago] ],
        'Helen'  => [ ['Gerty', 4.days.ago], ['Gilly', 2.days.ago], ['Garfield', 3.days.ago] ],
        'Ivan'   => [ ['Happy', 2.days.ago], ['Harry', 6.days.ago], ['Hipster',  4.days.ago] ]
    }
    owners = walks_input.map do |owner_name, pet_walks|
      owner = Owner.create(name: owner_name)
      pet_walks.each do |pet_name, time|
        owner.walks << Walk.create(pet_name: pet_name, walk_time: time)
      end
      owner
    end
    assert_equal(3, Owner.count)
    assert_equal(9, Walk.count)
    condition = { owner_id: { '$in' => owners[0..1].map(&:id) } } # don't use all owners for testing
    reduce = <<-EOS
      function(doc, out) {
        if (out.last_walk == undefined || out.last_walk.walk_time < doc.walk_time)
          out.last_walk = doc;
      }
    EOS
    last_walk_via_group = Walk.collection.group(key: :owner_id, cond: condition, initial: {}, reduce: reduce)
    p last_walk_via_group.collect{|r|[Owner.find(r['owner_id']).name, r['last_walk']['pet_name']]}
    last_walk = last_walk_via_group.collect{|r|Walk.new(r['last_walk'])}
    p last_walk
  end
end

test output

Run options: --name=test_group_in_Ruby

# Running tests:

[["George", "Fifi"], ["Helen", "Gilly"]]
[#<Walk _id: 4fbfa7a97f11ba53b3000003, _type: nil, pet_name: "Fifi", walk_time: 2012-05-24 15:39:21 UTC, owner_id: BSON::ObjectId('4fbfa7a97f11ba53b3000001')>, #<Walk _id: 4fbfa7a97f11ba53b3000007, _type: nil, pet_name: "Gilly", walk_time: 2012-05-23 15:39:21 UTC, owner_id: BSON::ObjectId('4fbfa7a97f11ba53b3000005')>]
.

Finished tests in 0.051868s, 19.2797 tests/s, 38.5594 assertions/s.

1 tests, 2 assertions, 0 failures, 0 errors, 0 skips