cwninja-has_related 0.1.0

data/README.markdown

@@ -0,0 +1,38 @@
+Has related is used by Reevoo to create similar product recommendations.
+
+Example Index Builder:
+
+    namespace :apache do
+      desc "Parse the last apache logs for similar products"
+      task :find_similar_products => :environment do
+        log_files = ENV["LOG_FILES"] || "/var/log/httpd/access.*.gz"
+
+        prefs = Hash.new{|h,k| h[k] = Hash.new(0) }
+
+        Dir.glob(log_files) do |filename|
+          File.open(filename) do |f|
+            Zlib::GzipReader.new(f).each_line do |line|
+              product_id, ip = parse_log_line(line)
+              if product_id and ip
+                prefs[product_id][ip] += 1
+              end
+            end
+          end
+        end
+
+        HasRelated.dump_dataset(prefs, "Product")
+      end
+    end
+
+(You will need to define your own `parse_log_line` method)
+
+
+Example Usage:
+
+    class Product < ActiveRecord::Base
+      has_related "related_products"
+    end
+
+    most_related_product = Product.related_products.first
+
+Happy hunting.

data/lib/has_related.rb

@@ -0,0 +1,110 @@
+require 'fileutils'
+module HasRelated
+  def self.included(base)
+    base.extend(ClassMethods)
+  end
+
+  module ClassMethods
+    def has_related(method_name = "related_items")
+      class_eval do
+        define_method method_name do |count|
+          HasRelated.similar_items(self, count)
+        end
+      end
+    end
+  end
+
+  class << self
+
+    def file_for_class(klass)
+      File.join(Rails.root, "db", "similar_items_datasets", klass.to_s.underscore + ".bin")
+    end
+
+    def similar_items(item, count)
+      ids = similar_item_ids(item, count)
+      (item.class.find_all_by_id(ids) || []).sort_by{|item| ids.index(item.id) }.first(count)
+    end
+
+
+    def similar_item_ids(item, count = nil)
+      dataset = similar_items_dataset(item.class.name) || {}
+      rankings = dataset[item.id] || []
+      rankings = rankings.first(count) if count
+      rankings.map{|_, id| id}
+    end
+
+    def similarity(item1, prefs_item1, item2, prefs_item2, people, all_people)
+      people.inject(0){|acc, person|
+        acc + prefs_item2[person] * prefs_item1[person]
+      } / all_people.size.to_f
+    end
+
+    def generate_dataset(prefs, &block)
+      all_results = {}
+      items = prefs.keys
+
+      prefs.each do |item1, item1_prefs|
+        agregated_recomendation_map = []
+
+        item1_people = item1_prefs.keys
+
+        prefs.each do |item2, item2_prefs|
+          item2_people = item2_prefs.keys
+          common_people = item1_people & item2_people
+          all_people = item1_people | item2_people
+
+          unless item1 == item2
+            item_similarity = similarity(item1, item1_prefs, item2, item2_prefs, common_people, all_people)
+            agregated_recomendation_map << [item_similarity, item2] if item_similarity > 0
+          end
+        end
+
+        all_results[item1] = agregated_recomendation_map.sort_by{|count, item1| -count}.first(16) if agregated_recomendation_map.any?
+
+        yield [item1, agregated_recomendation_map] if block_given?
+      end
+
+      return all_results
+    end
+
+    def dump_dataset(prefs, klass, &block)
+      ensure_data_dir_exists!(klass)
+      dataset = generate_dataset(prefs, &block)
+      write_dataset_to_disk(dataset, klass)
+    end
+
+    def dump_grouped_datasets(grouped_prefs, klass, &block)
+      ensure_data_dir_exists!(klass)
+      dataset = Hash.new
+      grouped_prefs.each do |id, prefs|
+        dataset.merge! generate_dataset(prefs, &block)
+      end
+      write_dataset_to_disk(dataset, klass)
+    end
+
+    def similar_items_dataset(klass_name)
+      @similar_items_dataset ||= {}
+      return @similar_items_dataset[klass_name] if @similar_items_dataset[klass_name]
+
+      if File.readable? file_for_class(klass_name)
+        @similar_items_dataset[klass_name] = Marshal.load(File.open(file_for_class(klass_name)))
+      else
+        @similar_items_dataset[klass_name] = {}
+      end
+
+      return @similar_items_dataset[klass_name]
+    end
+
+  private
+
+    def ensure_data_dir_exists!(klass)
+      FileUtils.mkdir_p(File.dirname(file_for_class(klass))) unless File.directory? File.dirname(file_for_class(klass))
+    end
+
+    def write_dataset_to_disk(dataset, klass)
+      File.open(file_for_class(klass), "w") do |io|
+        Marshal.dump(dataset, io)
+      end
+    end
+  end
+end

data/rails/init.rb

@@ -0,0 +1,5 @@
+require File.dirname(__FILE__) + '/../lib/has_related'
+if defined? ActiveRecord
+  ActiveRecord::Base.send(:include, HasRelated)
+end
+

data/test/dataset_generation_test.rb

@@ -0,0 +1,32 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+class DatasetGenerationTest < Test::Unit::TestCase
+  def setup
+    prefs = {
+      :p1 => {:u1 => 1, :u2 => 1, :u3 => 1},
+      :p2 => {:u2 => 1},
+      :p3 => {:u3 => 1, :u5 => 1},
+      :p4 => {:u1 => 1, :u4 => 1},
+      :p5 => {:u1 => 1, :u4 => 1, :u5 => 1}
+    }
+    @rv = HasRelated.generate_dataset(prefs)
+  end
+
+  def test_tied_game
+    assert [:p2, :p3, :p4].include?( @rv[:p1].first.last )
+  end
+
+  def test_one_of_many
+    assert_equal :p1, @rv[:p2].first.last
+  end
+
+  def test_clear_winner
+    assert_equal :p4, @rv[:p5].first.last
+    assert_equal :p3, @rv[:p5][1].last
+  end
+
+  def test_no_relation
+    assert !@rv[:p3].map{|v| v.last }.include?( :p4 )
+  end
+
+end

data/test/test_helper.rb

@@ -0,0 +1,22 @@
+$:.unshift(File.dirname(__FILE__) + '/../lib')
+require 'rubygems'
+require 'test/unit'
+require 'fileutils'
+
+class Test::Unit::TestCase #:nodoc:
+end
+
+module Rails
+  def root
+    if @tmpdir.nil?
+      @tmpdir = File.join(ENV["TMPDIR"], "has_related")
+      Dir.mkdir(@tmpdir) unless File.directory? @tmpdir
+      ObjectSpace.define_finalizer(@tmpdir, proc{|id| FileUtils.remove_entry_secure @tmpdir })
+    end
+    @tmpdir
+  end
+
+  extend self
+end
+
+require "#{File.dirname(__FILE__)}/../init"

metadata

@@ -0,0 +1,58 @@
+--- !ruby/object:Gem::Specification 
+name: cwninja-has_related
+version: !ruby/object:Gem::Version 
+  version: 0.1.0
+platform: ruby
+authors: 
+- Tom Lea
+autorequire: 
+bindir: bin
+cert_chain: []
+
+date: 2009-05-18 00:00:00 -07:00
+default_executable: 
+dependencies: []
+
+description: 
+email: contrib@tomlea.co.uk
+executables: []
+
+extensions: []
+
+extra_rdoc_files: 
+- README.markdown
+files: 
+- README.markdown
+- test/dataset_generation_test.rb
+- test/test_helper.rb
+- lib/has_related.rb
+- rails/init.rb
+has_rdoc: true
+homepage: http://tomlea.co.uk
+post_install_message: 
+rdoc_options: 
+- --main
+- README.markdown
+require_paths: 
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+requirements: []
+
+rubyforge_project: has_related
+rubygems_version: 1.2.0
+signing_key: 
+specification_version: 2
+summary: Finds similar items based on user demand.
+test_files: []
+