Working but basic implementation of facets

cucumber.yml

@@ -1 +1 @@
-default: "--require features/support/env.rb --require features/support/db/mysql.rb --require features/support/db/active_record.rb --require features/support/post_database.rb --require features/step_definitions/alpha_steps.rb --require features/step_definitions/beta_steps.rb --require features/step_definitions/cat_steps.rb --require features/step_definitions/common_steps.rb --require features/step_definitions/datetime_delta_steps.rb --require features/step_definitions/delayed_delta_indexing_steps.rb --require features/step_definitions/find_arguments_steps.rb --require features/step_definitions/gamma_steps.rb --require features/step_definitions/search_steps.rb --require features/step_definitions/sphinx_steps.rb"
+default: "--require features/support/env.rb --require features/support/db/mysql.rb --require features/support/db/active_record.rb --require features/support/post_database.rb --require features/step_definitions/alpha_steps.rb --require features/step_definitions/beta_steps.rb --require features/step_definitions/cat_steps.rb --require features/step_definitions/common_steps.rb --require features/step_definitions/datetime_delta_steps.rb --require features/step_definitions/delayed_delta_indexing_steps.rb --require features/step_definitions/facet_steps.rb --require features/step_definitions/find_arguments_steps.rb --require features/step_definitions/gamma_steps.rb --require features/step_definitions/search_steps.rb --require features/step_definitions/sphinx_steps.rb"

features/datetime_deltas.feature

@@ -51,5 +51,5 @@ Feature: Datetime Delta Indexing
     And I search for thirteen
     Then I should get 1 result
 
-    When I search for the specific id of 107 in the theta_core index
+    When I search for the document id of theta thirteen in the theta_core index
     Then it should exist

features/facets.feature

@@ -0,0 +1,7 @@
+Feature: Search and browse models by their defined facets
+
+  Scenario: Test
+    Given Sphinx is running
+    And I am searching on developers
+    When I am requesting facet results
+    Then I should see facet results

features/searching_across_models.feature

@@ -11,14 +11,9 @@ Feature: Searching across multiple model
 
   Scenario: Confirming existance of a document id in a given index
     Given Sphinx is running
-    When I search for the specific id of 51 in the person_core index
+    When I search for the document id of alpha one in the alpha_core index
     Then it should exist
 
-  Scenario: Unsuccessfully confirming existance of a document id in a given index
-    Given Sphinx is running
-    When I search for the specific id of 52 in the person_core index
-    Then it should not exist
-
   Scenario: Retrieving results from multiple models
     Given Sphinx is running
     When I search for ten

features/step_definitions/facet_steps.rb

@@ -0,0 +1,8 @@
+When "I am requesting facet results" do
+  @method = :facets
+end
+
+Then "I should see facet results" do
+  results.should be_kind_of(Hash)
+  results.values.each { |value| value.should be_kind_of(Hash) }
+end

features/step_definitions/search_steps.rb

@@ -32,10 +32,10 @@
 end
 
 Then /^I can iterate by result and group and count$/ do
-  results.each_with_group_and_count do |result, group, count|
+  results.each_with_groupby_and_count do |result, group, count|
     result.should be_kind_of(@model)
     count.should  be_kind_of(Integer)
-    group.should  be_kind_of(Integer) unless group.nil?
+    group.should  be_kind_of(Integer)
   end
 end
 

features/support/db/migrations/create_developers.rb

@@ -0,0 +1,39 @@
+require 'faker'
+
+ActiveRecord::Base.connection.create_table :developers, :force => true do |t|
+  t.column :name,     :string,  :null => false
+  t.column :city,     :string
+  t.column :state,    :string
+  t.column :country,  :string
+  t.column :age,      :integer
+end
+
+Developer.create :name => "Pat Allan", :city => "Melbourne", :state => "Victoria", :country => "Australia", :age => 26
+
+2.times do
+  Developer.create :name => Faker::Name.name, :city => "Melbourne", :state => "Victoria", :country => "Australia", :age => 30
+end
+
+2.times do
+  Developer.create :name => Faker::Name.name, :city => "Sydney", :state => "New South Wales", :country => "Australia", :age => 28
+end
+
+2.times do
+  Developer.create :name => Faker::Name.name, :city => "Adelaide", :state => "South Australia", :country => "Australia", :age => 32
+end
+
+2.times do
+  Developer.create :name => Faker::Name.name, :city => "Bendigo", :state => "Victoria", :country => "Australia", :age => 30
+end
+
+2.times do
+  Developer.create :name => Faker::Name.name, :city => "Goulburn", :state => "New South Wales", :country => "Australia", :age => 28
+end
+
+2.times do
+  Developer.create :name => Faker::Name.name, :city => "Auckland", :state => "North Island", :country => "New Zealand", :age => 32
+end
+
+2.times do
+  Developer.create :name => Faker::Name.name, :city => "Christchurch", :state => "South Island", :country => "New Zealand", :age => 30
+end

features/support/models/developer.rb

@@ -0,0 +1,7 @@
+class Developer < ActiveRecord::Base
+  define_index do
+    indexes country,  :facet => true
+    indexes state,    :facet => true
+    has age,          :facet => true
+  end
+end

lib/thinking_sphinx.rb

@@ -12,6 +12,7 @@
 require 'thinking_sphinx/attribute'
 require 'thinking_sphinx/collection'
 require 'thinking_sphinx/configuration'
+require 'thinking_sphinx/facet'
 require 'thinking_sphinx/field'
 require 'thinking_sphinx/index'
 require 'thinking_sphinx/rails_additions'

lib/thinking_sphinx/active_record.rb

@@ -10,7 +10,7 @@ module ThinkingSphinx
   module ActiveRecord
     def self.included(base)
       base.class_eval do
-        class_inheritable_array :sphinx_indexes
+        class_inheritable_array :sphinx_indexes, :sphinx_facets
         class << self
           # Allows creation of indexes for Sphinx. If you don't do this, there
           # isn't much point trying to search (or using this plugin at all,

lib/thinking_sphinx/active_record/search.rb

@@ -42,6 +42,13 @@ def search_for_id(*args)
               args << options
               ThinkingSphinx::Search.search_for_id(*args)
             end
+
+            def facets(*args)
+              options = args.extract_options!
+              options[:class] = self
+              args << options
+              ThinkingSphinx::Search.facets(*args)
+            end
           end
         end
       end

lib/thinking_sphinx/attribute.rb

@@ -9,7 +9,7 @@ module ThinkingSphinx
   # associations. Which can get messy. Use Index.link!, it really helps.
   # 
   class Attribute
-    attr_accessor :alias, :columns, :associations, :model
+    attr_accessor :alias, :columns, :associations, :model, :faceted
 
     # To create a new attribute, you'll need to pass in either a single Column
     # or an array of them, and some (optional) options.
@@ -59,8 +59,9 @@ def initialize(columns, options = {})
 
       raise "Cannot define a field with no columns. Maybe you are trying to index a field with a reserved name (id, name). You can fix this error by using a symbol rather than a bare name (:id instead of id)." if @columns.empty? || @columns.any? { |column| !column.respond_to?(:__stack) }
 
-      @alias        = options[:as]
-      @type         = options[:type]
+      @alias    = options[:as]
+      @type     = options[:type]
+      @faceted  = options[:facet]
     end
 
     # Get the part of the SELECT clause related to this attribute. Don't forget
@@ -133,6 +134,27 @@ def unique_name
       end
     end
 
+    # Returns the type of the column. If that's not already set, it returns
+    # :multi if there's the possibility of more than one value, :string if
+    # there's more than one association, otherwise it figures out what the
+    # actual column's datatype is and returns that.
+    def type
+      @type ||= case
+      when is_many?
+        :multi
+      when @associations.values.flatten.length > 1
+        :string
+      else
+        translated_type_from_database
+      end
+    end
+
+    def to_facet
+      return nil unless @faceted
+
+      ThinkingSphinx::Facet.new(unique_name, @columns, self)
+    end
+
     private
 
     def adapter
@@ -190,21 +212,6 @@ def is_string?
       columns.all? { |col| col.is_string? }
     end
 
-    # Returns the type of the column. If that's not already set, it returns
-    # :multi if there's the possibility of more than one value, :string if
-    # there's more than one association, otherwise it figures out what the
-    # actual column's datatype is and returns that.
-    def type
-      @type ||= case
-      when is_many?
-        :multi
-      when @associations.values.flatten.length > 1
-        :string
-      else
-        translated_type_from_database
-      end
-    end
-
     def all_ints?
       @columns.all? { |col|
         klasses = @associations[col].empty? ? [@model] :

lib/thinking_sphinx/collection.rb

@@ -113,9 +113,9 @@ def method_missing(method, *args, &block)
       each_with_attribute method.to_s.gsub(/^each_with_/, ''), &block
     end
 
-    def each_with_group_and_count(&block)
+    def each_with_groupby_and_count(&block)
       results[:matches].each_with_index do |match, index|
-        yield self[index], match[:attributes]["@group"], match[:attributes]["@count"]
+        yield self[index], match[:attributes]["@groupby"], match[:attributes]["@count"]
       end
     end
 

lib/thinking_sphinx/facet.rb

@@ -0,0 +1,49 @@
+module ThinkingSphinx
+  class Facet
+    attr_reader :name, :column, :reference
+
+    def initialize(name, columns, reference)
+      @name = name
+      @columns = columns
+      @reference = reference
+    end
+
+    def attribute_name
+      @attribute_name ||= case @reference
+      when Attribute
+        @reference.unique_name.to_s
+      when Field
+        @reference.unique_name.to_s + "_sort"
+      end
+    end
+
+    def value(object, attribute_value)
+      return translate(object, attribute_value) if @reference.is_a?(Field)
+
+      case @reference.type
+      when :string, :multi
+        translate(object, attribute_value)
+      when :datetime
+        Time.at(attribute_value)
+      when :boolean
+        attribute_value > 0
+      else
+        attribute_value
+      end
+    end
+
+    private
+
+    def translate(object, attribute_value)
+      if @columns.length > 1
+        raise "Can't translate Facets on multiple-column field or attribute"
+      end
+
+      column  = @columns.first
+      column.__stack.each { |method|
+        object = object.send(method)
+      }
+      object.send(column.__name)
+    end
+  end
+end

lib/thinking_sphinx/field.rb

@@ -8,7 +8,8 @@ module ThinkingSphinx
   # associations. Which can get messy. Use Index.link!, it really helps.
   # 
   class Field
-    attr_accessor :alias, :columns, :sortable, :associations, :model, :infixes, :prefixes
+    attr_accessor :alias, :columns, :sortable, :associations, :model, :infixes,
+      :prefixes, :faceted
 
     # To create a new field, you'll need to pass in either a single Column
     # or an array of them, and some (optional) options. The columns are
@@ -58,10 +59,11 @@ def initialize(columns, options = {})
 
       raise "Cannot define a field with no columns. Maybe you are trying to index a field with a reserved name (id, name). You can fix this error by using a symbol rather than a bare name (:id instead of id)." if @columns.empty? || @columns.any? { |column| !column.respond_to?(:__stack) }
 
-      @alias        = options[:as]
-      @sortable     = options[:sortable] || false
-      @infixes      = options[:infixes]  || false
-      @prefixes     = options[:prefixes] || false
+      @alias    = options[:as]
+      @sortable = options[:sortable] || false
+      @infixes  = options[:infixes]  || false
+      @prefixes = options[:prefixes] || false
+      @faceted  = options[:facet]    || false
     end
 
     # Get the part of the SELECT clause related to this field. Don't forget
@@ -110,6 +112,12 @@ def unique_name
       end
     end
 
+    def to_facet
+      return nil unless @faceted
+
+      ThinkingSphinx::Facet.new(unique_name, @columns, self)
+    end
+
     private
 
     def adapter

lib/thinking_sphinx/index.rb

@@ -167,6 +167,14 @@ def initialize_from_builder(&block)
       @delta_object = ThinkingSphinx::Deltas.parse self, builder.properties
       @options      = builder.properties
 
+      @model.sphinx_facets ||= []
+      @fields.select { |field| field.faceted }.each { |field|
+        @model.sphinx_facets << field.to_facet
+      }
+      @attributes.select { |attrib| attrib.faceted }.each { |attrib|
+        @model.sphinx_facets << attrib.to_facet
+      }
+
       # We want to make sure that if the database doesn't exist, then Thinking
       # Sphinx doesn't mind when running non-TS tasks (like db:create, db:drop
       # and db:migrate). It's a bit hacky, but I can't think of a better way.

lib/thinking_sphinx/index/builder.rb

@@ -89,13 +89,13 @@ def indexes(*args)
           args.each do |columns|
             fields << Field.new(FauxColumn.coerce(columns), options)
 
-            if fields.last.sortable
+            if fields.last.sortable || fields.last.faceted
               attributes << Attribute.new(
                 fields.last.columns.collect { |col| col.clone },
                 options.merge(
                   :type => :string,
                   :as => fields.last.unique_name.to_s.concat("_sort").to_sym
-                )
+                ).except(:facet)
               )
             end
           end

lib/thinking_sphinx/search.rb

@@ -352,6 +352,23 @@ def search_for_id(*args)
         end
       end
 
+      def facets(*args)
+        options = args.extract_options!.merge! :group_function => :attr
+
+        options[:class].sphinx_facets.inject({}) do |hash, facet|
+          facet_result = {}
+          options[:group_by] = facet.attribute_name
+
+          results = search *(args + [options])
+          results.each_with_groupby_and_count do |result, group, count|
+            facet_result[facet.value(result, group)] = count
+          end
+          hash[facet.name] = facet_result
+
+          hash
+        end
+      end
+
       private
 
       # This method handles the common search functionality, and returns both