answer-factory 0.1.2 → 0.1.3.4

data/Rakefile

@@ -12,7 +12,7 @@ begin
     
     gemspec.required_ruby_version = '>= 1.9.1'
     
-    gemspec.add_dependency('nudge', '>= 0.2')
+    gemspec.add_dependency('nudge', '>= 0.2.9')
     gemspec.add_dependency('thor', '>= 0.13')
     gemspec.add_dependency('couchrest', '>= 0.33')
     gemspec.add_dependency('configatron', '>= 2.6.2')

data/VERSION

@@ -1 +1 @@
-0.1.2
+0.1.3.4

data/answer-factory.gemspec

@@ -1,15 +1,15 @@
 # Generated by jeweler
 # DO NOT EDIT THIS FILE DIRECTLY
-# Instead, edit Jeweler::Tasks in rakefile, and run the gemspec command
+# Instead, edit Jeweler::Tasks in Rakefile, and run the gemspec command
 # -*- encoding: utf-8 -*-
 
 Gem::Specification.new do |s|
   s.name = %q{answer-factory}
-  s.version = "0.1.2"
+  s.version = "0.1.3.4"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Bill Tozier", "Trek Glowacki", "Jesse Sielaff"]
-  s.date = %q{2010-05-12}
+  s.date = %q{2010-05-27}
   s.default_executable = %q{answer-factory}
   s.description = %q{The pragmaticgp gem provides a simple framework for building, running and managing genetic programming experiments which automatically discover algorithms and equations to solve user-defined problems.}
   s.email = %q{bill@vagueinnovation.com}
@@ -34,20 +34,25 @@ Gem::Specification.new do |s|
      "lib/machines/any_one.rb",
      "lib/machines/build_random.rb",
      "lib/machines/evaluate_simple_score.rb",
+     "lib/machines/evaluate_with_test_cases.rb",
      "lib/machines/infrastructure.rb",
      "lib/machines/mutate_footnotes.rb",
+     "lib/machines/select_by_summed_rank.rb",
      "lib/machines/select_nondominated.rb",
      "readme.md",
      "spec/answers/answer_spec.rb",
      "spec/answers/batch_spec.rb",
      "spec/factories/factory_spec.rb",
      "spec/factories/workstation_spec.rb",
+     "spec/fixtures/my_data_source.csv",
      "spec/integration_specs/couch_db_integration.rspec",
      "spec/machines/any_one_spec.rb",
      "spec/machines/build_random_spec.rb",
      "spec/machines/evaluate_simple_score_spec.rb",
+     "spec/machines/evaluate_with_test_cases_spec.rb",
      "spec/machines/infrastructure_spec.rb",
      "spec/machines/mutate_footnotes_spec.rb",
+     "spec/machines/select_by_summed_rank_spec.rb",
      "spec/machines/select_nondominated_spec.rb",
      "spec/spec_helper.rb",
      "tasks/setup_factory.thor",
@@ -58,7 +63,7 @@ Gem::Specification.new do |s|
   s.rdoc_options = ["--charset=UTF-8"]
   s.require_paths = ["lib"]
   s.required_ruby_version = Gem::Requirement.new(">= 1.9.1")
-  s.rubygems_version = %q{1.3.6}
+  s.rubygems_version = %q{1.3.7}
   s.summary = %q{Genetic Programming in the Nudge language}
   s.test_files = [
     "spec/answers/answer_spec.rb",
@@ -68,8 +73,10 @@ Gem::Specification.new do |s|
      "spec/machines/any_one_spec.rb",
      "spec/machines/build_random_spec.rb",
      "spec/machines/evaluate_simple_score_spec.rb",
+     "spec/machines/evaluate_with_test_cases_spec.rb",
      "spec/machines/infrastructure_spec.rb",
      "spec/machines/mutate_footnotes_spec.rb",
+     "spec/machines/select_by_summed_rank_spec.rb",
      "spec/machines/select_nondominated_spec.rb",
      "spec/spec_helper.rb"
   ]
@@ -78,8 +85,8 @@ Gem::Specification.new do |s|
     current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
     s.specification_version = 3
 
-    if Gem::Version.new(Gem::RubyGemsVersion) >= Gem::Version.new('1.2.0') then
-      s.add_runtime_dependency(%q<nudge>, [">= 0.2"])
+    if Gem::Version.new(Gem::VERSION) >= Gem::Version.new('1.2.0') then
+      s.add_runtime_dependency(%q<nudge>, [">= 0.2.9"])
       s.add_runtime_dependency(%q<thor>, [">= 0.13"])
       s.add_runtime_dependency(%q<couchrest>, [">= 0.33"])
       s.add_runtime_dependency(%q<configatron>, [">= 2.6.2"])
@@ -87,7 +94,7 @@ Gem::Specification.new do |s|
       s.add_runtime_dependency(%q<sinatra>, [">= 0.9.4"])
       s.add_runtime_dependency(%q<activesupport>, [">= 2.3.5"])
     else
-      s.add_dependency(%q<nudge>, [">= 0.2"])
+      s.add_dependency(%q<nudge>, [">= 0.2.9"])
       s.add_dependency(%q<thor>, [">= 0.13"])
       s.add_dependency(%q<couchrest>, [">= 0.33"])
       s.add_dependency(%q<configatron>, [">= 2.6.2"])
@@ -96,7 +103,7 @@ Gem::Specification.new do |s|
       s.add_dependency(%q<activesupport>, [">= 2.3.5"])
     end
   else
-    s.add_dependency(%q<nudge>, [">= 0.2"])
+    s.add_dependency(%q<nudge>, [">= 0.2.9"])
     s.add_dependency(%q<thor>, [">= 0.13"])
     s.add_dependency(%q<couchrest>, [">= 0.33"])
     s.add_dependency(%q<configatron>, [">= 2.6.2"])

data/lib/answer-factory.rb

@@ -13,7 +13,9 @@ require 'machines/infrastructure'
 require 'machines/any_one'
 require 'machines/build_random'
 require 'machines/evaluate_simple_score'
+require 'machines/evaluate_with_test_cases'
 require 'machines/select_nondominated'
+require 'machines/select_by_summed_rank'
 require 'machines/mutate_footnotes'
 
 

data/lib/factories/factory.rb

@@ -26,6 +26,8 @@ module AnswerFactory
       @couchdb_server = options[:couchdb_server] ||
         configatron.factory.couchdb.retrieve(:server, nil) ||
         "http://127.0.0.1:5984"
+        
+        
       
       @couchdb_name = options[:couchdb_name] ||
         configatron.factory.couchdb.retrieve(:name, nil) ||
@@ -54,6 +56,7 @@ module AnswerFactory
       configatron.factory.workstation_names = @workstation_names
       configatron.factory.couchdb.server = @couchdb_server
       configatron.factory.couchdb.name = @couchdb_name
+      configatron.factory.training_datasource = self.training_datasource
     end
     
     
@@ -63,5 +66,9 @@ module AnswerFactory
     rescue StandardError
       false 
     end
+    
+    def training_datasource
+      "#{@couchdb_server}/#{@name}_training"
+    end
   end
 end

data/lib/factories/workstation.rb

@@ -52,6 +52,22 @@ module AnswerFactory
     end
     
     
+    def couchdb_viewdoc
+      {'_id' => "_design/#{@name.to_s}",
+        views: { 
+          current: {
+            map: "function(doc) { if(doc.location == '#{@name.to_s}') { emit(doc._id, doc); } }"
+          }
+        }
+      }
+    end
+    
+    
+    def couchdb_create_view(db_uri = couchdb_uri)
+      CouchRest.database!(db_uri).save_doc(self.couchdb_viewdoc, use_uuids=false)
+    end
+    
+    
     def ship!
       # Workstation is a superclass; the default behavior (doing nothing)
       # should be overridden in a subclass definition

data/lib/machines/evaluate_with_test_cases.rb

@@ -0,0 +1,140 @@
+#encoding: utf-8
+module AnswerFactory
+  module Machines
+    
+    
+    class EvaluateWithTestCases < Machine
+      attr_accessor :sensors
+      attr_accessor :test_cases
+      attr_reader :name
+      attr_reader :csv_filename
+      
+      def initialize(options = {})
+        super
+        @name = options[:name] || "evaluator"
+        @sensors = options[:sensors] || {}
+        @csv_filename = options[:training_data_csv]
+        
+        self.load_training_data!
+      end
+      
+      
+      def score(batch, overridden_options = {})
+        all_options = @options.merge(overridden_options)
+        name = all_options[:name]
+        
+        
+        raise ArgumentError, "EvaluateWithTestCases#score cannot process a #{batch.class}" unless
+          batch.kind_of?(Batch)
+        raise ArgumentError, "EvaluateWithTestCases: Undefined #name attribute" if
+          name.nil?
+        
+        load_training_data!
+        
+        batch.each do |answer|
+          
+          raw_results = Hash.new {|hash, key| hash[key] = []}
+          
+          @test_cases.each do |t|
+            
+            interpreter = Interpreter.new(answer.blueprint,all_options)
+            
+            t.inputs.each do |variable_header, variable_value|
+              variable_name, variable_type = variable_header.split(":")
+              interpreter.bind_variable(variable_name, ValuePoint.new(variable_type, variable_value))
+            end
+            
+            @sensors.each do |sensor_name, sensor_block|
+              interpreter.register_sensor(sensor_name, &sensor_block)
+            end
+            
+            interpreter.run.each do |sensor_name, sensor_result|
+              raw_results[sensor_name] << (t.outputs[sensor_name].to_i - sensor_result)
+            end
+          end
+          
+          @sensors.each do |sensor_name, sensor_block|
+            answer.scores[sensor_name] = raw_results[sensor_name].inject(0) do |sum, measurement|
+              sum + measurement.abs
+            end
+          end
+        end
+        
+        
+        return batch
+      end
+      
+      
+      def training_datasource
+        configatron.factory.training_datasource
+      end
+      
+      
+      def training_data_view
+       "#{configatron.factory.training_datasource}/_design/#{@name}/_view/test_cases"
+      end
+      
+      
+      def header_prep(header_string)
+        raise ArgumentError, "Header must match /reference_name:nudge_type/" unless
+          header_string.match /[\p{Alpha}][\p{Alnum}_]*:[\p{Alpha}][\p{Alnum}_]/
+        header_string.strip
+      end
+      
+      
+      def build_sensor(name, &block)
+        @sensors[name] = block
+      end
+      
+      
+      def install_training_data_from_csv(csv_filename = @csv_filename)
+        reader = CSV.new(File.open(csv_filename), headers: true)
+        reader.readline
+        split_point = reader.headers.find_index(nil)
+        
+        input_headers = reader.headers[0...split_point].collect {|head| header_prep(head)}
+        output_headers = reader.headers[split_point+1..-1].collect {|head| head.strip}
+        
+        reader.rewind
+        
+        offset = input_headers.length+1
+        db = CouchRest.database!(training_datasource)
+        
+        reader.each do |row|
+          inputs = {}
+          input_headers.each_with_index {|header,i| inputs[header] = row[i].strip}
+          outputs = {}
+          output_headers.each_with_index {|header,i| outputs[header] = row[i+offset].strip}
+          db.bulk_save_doc( {:inputs => inputs, :outputs => outputs})
+        end
+        
+        db.bulk_save
+        
+      end
+      
+      
+      def load_training_data!
+        db = CouchRest.database!(training_datasource)
+        result = db.view("#{@name}/test_cases")
+        @test_cases = 
+          result["rows"].collect {|r| TestCase.new(
+              inputs: r["value"]["inputs"], outputs: r["value"]["outputs"])}
+      end
+      
+      
+      alias :generate :score
+    end
+    
+    
+    
+    
+    class TestCase
+      attr_accessor :inputs, :outputs
+      
+      def initialize(options = {})
+        @inputs = options[:inputs] || {}
+        @outputs = options[:outputs] || {}
+      end
+    end
+  end
+end

data/lib/machines/infrastructure.rb

@@ -1,6 +1,8 @@
 module AnswerFactory
   module Machines
     
+    require 'csv'
+    
     class Machine
       
       attr_reader :options

data/lib/machines/select_by_summed_rank.rb

@@ -0,0 +1,54 @@
+module AnswerFactory
+  module Machines
+    
+    
+    
+    
+    class SelectBySummedRank < Machine
+      
+      def screen(batch, overriden_options={})
+        raise ArgumentError, "SelectBySummedRank#screen cannot process class #{batch.class}" unless
+          batch.kind_of?(Batch)
+        all_options = @options.merge(overriden_options)
+        
+        criteria = all_options[:comparison_criteria] || shared_goals(batch)
+        return batch if criteria.empty?
+        
+        scored = Hash.new(0)
+        incomparable = Set.new
+        
+        criteria.each do |criterion|
+          scorable, unscorable = batch.partition {|a| a.scores.include?(criterion)}
+          incomparable += unscorable
+          ranked = scorable.sort_by {|a| a.scores[criterion]}
+          ranked.each_with_index do |a, index|
+            scored[a] += index
+          end
+        end
+        
+        incomparable.each {|a| scored.delete(a)}
+        
+        lowest_sum = scored.values.min
+        winners = batch.find_all {|a| scored[a] == lowest_sum }
+        
+        result = Batch.new
+        incomparable.each {|a| result << a}
+        
+        winners.each {|a| result << a unless result.include?(a)}
+        return result
+      end
+      
+      
+      def all_goals(batch)
+        (batch.collect {|a| a.scores.keys}).flatten.to_set.to_a
+      end
+      
+      
+      def shared_goals(batch)
+        batch.inject(all_goals(batch)) {|intersection, answer| intersection &= answer.scores.keys}
+      end
+      
+      alias :generate :screen
+    end
+  end
+end

data/spec/factories/factory_spec.rb

@@ -159,7 +159,7 @@ describe "Factory" do
             :body => "Go away!", :status => [404, "Not Found"])
           Factory.couch_available?.should == false
           
-          f1 = Factory.new(name:"boo", couchdb_server:"http://127.0.0.1:9991/place")
+          f1 = Factory.new(name:"boo", couchdb_server:"http://127.0.0.1:5984/place")
           Factory.couch_available?.should == false
         end
       end
@@ -174,6 +174,18 @@ describe "Factory" do
         it "should require authorization"
       end
       
+      describe "training data" do
+        before(:each) do
+          FakeWeb.allow_net_connect = false
+        end
+        
+        it "should get the name of the :training_datasource from configatron" do
+          Factory.new(name:"baz", couchdb_server:"http://127.0.0.1:5984").
+            training_datasource.should == "http://127.0.0.1:5984/baz_training"
+        end
+        
+      end
+      
       describe "authorization" do
         it "should be necessary to set up an admin account in couch"
         

data/spec/factories/workstation_spec.rb

@@ -24,6 +24,25 @@ describe "Workstation" do
     end
   end
   
+  describe "couchdb views documents" do
+    it "should respond to :couchdb_viewdoc" do
+      Workstation.new(:foo).should respond_to(:couchdb_viewdoc)
+    end
+    
+    it "should return a Hash" do
+      Workstation.new(:foo).couchdb_viewdoc.should be_a_kind_of(Hash)
+    end
+    
+    it "should have the right _id value" do
+      Workstation.new(:foo).couchdb_viewdoc["_id"].should == "_design/foo"
+    end
+    
+    it "should include the correct map code for this Workstation" do
+      Workstation.new(:foo).couchdb_viewdoc[:views][:current][:map].should ==
+        "function(doc) { if(doc.location == 'foo') { emit(doc._id, doc); } }"
+    end
+  end
+  
   
   describe "capacity" do
     it "should accept a #capacity attribute in initialization" do

data/spec/fixtures/my_data_source.csv

@@ -0,0 +1,4 @@
+x1:int, x2:int, x3:int,,y1, y2
+1,2,3,, 9991, 8812
+2,3,4,, 88, -12
+3,4,5,, boo, far

data/spec/integration_specs/couch_db_integration.rspec

@@ -15,7 +15,6 @@ end
 
 describe "CouchDB stuff" do
   before(:each) do
-    
     configatron.factory.couchdb.server = "http://127.0.0.1:5984"
     configatron.factory.couchdb.name = "integration_test_db"
     break("CouchDB is offline") unless Factory.couch_available?
@@ -24,6 +23,7 @@ describe "CouchDB stuff" do
   
   after(:all) do
     CouchRest.database!(@db_uri).delete!
+    CouchRest.database!("#{@db_uri}_training").delete!
   end
   
   
@@ -37,7 +37,52 @@ describe "CouchDB stuff" do
       Factory.couch_available?.should be false
       FakeWeb.allow_net_connect = true
     end
+  end
+  
+  
+  describe "Machine CouchDB interaction" do
+    before(:each) do
+      the_database_name = "#{@db_uri}_training"
+      @db = CouchRest.database!(the_database_name)
+    end
     
+    it "should accept :install_training_data_from_csv to populate a db from csv" do
+      configatron.temp do
+        configatron.factory.training_datasource = "http://127.0.0.1:5984/integration_test_db_training"
+        @db.save_doc(
+          {'_id' => "_design/tester",
+          views: { test_cases: { map: 
+            "function(doc) { emit(doc._id, doc); }"}}})
+        
+        @my_machine = Machines::EvaluateWithTestCases.new(name: :tester)
+        @my_machine.install_training_data_from_csv('./spec/fixtures/my_data_source.csv')
+        saved = @db.documents
+        saved["rows"].length.should == 4
+      end
+    end
+    
+    it "should strip whitespace from all the training value strings" do
+      configatron.temp do
+        configatron.factory.training_datasource = "http://127.0.0.1:5984/integration_test_db_training"
+        @my_machine = Machines::EvaluateWithTestCases.new(name: :tester)
+        @my_machine.install_training_data_from_csv('./spec/fixtures/my_data_source.csv')
+        first_doc = @db.get(@db.documents["rows"][0]["id"])
+        first_doc["outputs"].each {|k,v| k.strip.should == k}
+        first_doc["outputs"].each {|k,v| v.strip.should == v}
+        
+        first_doc["inputs"].each {|k,v| k.strip.should == k}
+        first_doc["inputs"].each {|k,v| v.strip.should == v}
+      end
+    end
+    
+    it "should use load_training_data! to access the right view" do
+      configatron.temp do
+        configatron.factory.training_datasource = "http://127.0.0.1:5984/integration_test_db_training"
+        @my_machine = Machines::EvaluateWithTestCases.new(name: :tester)
+        @my_machine.load_training_data!
+        @my_machine.test_cases[0].inputs["x1:int"].should == "3"
+      end
+    end
   end
   
   describe "Batch reading and writing" do
@@ -87,6 +132,19 @@ describe "CouchDB stuff" do
   
   describe "Workstations passing Batches around" do
     
+    describe "couchdb_create_view" do
+      before(:each) do
+        @ws_viewer = Workstation.new(:view_maker)
+      end
+      
+      it "should save a view doc" do
+        response = @ws_viewer.couchdb_create_view
+        # "ok"=>true, "id"=>"_design/view_maker", "rev"=>"1-a813357133c1538c86168966b36f97bd"}
+        response["ok"].should == true
+        response["id"].should == "_design/view_maker"
+      end
+    end
+    
     describe "gathering" do
       before(:each) do
         @ws1 = ShipperStation.new(:this_place)