answer-factory 0.1.3.4 → 0.1.3.5

data/VERSION

@@ -1 +1 @@
-0.1.3.4
+0.1.3.5

data/answer-factory.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{answer-factory}
-  s.version = "0.1.3.4"
+  s.version = "0.1.3.5"
 
   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{2010-05-27}
+  s.date = %q{2010-06-02}
   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}
@@ -36,6 +36,7 @@ Gem::Specification.new do |s|
      "lib/machines/evaluate_simple_score.rb",
      "lib/machines/evaluate_with_test_cases.rb",
      "lib/machines/infrastructure.rb",
+     "lib/machines/mutate_codeblock.rb",
      "lib/machines/mutate_footnotes.rb",
      "lib/machines/select_by_summed_rank.rb",
      "lib/machines/select_nondominated.rb",
@@ -51,6 +52,7 @@ Gem::Specification.new do |s|
      "spec/machines/evaluate_simple_score_spec.rb",
      "spec/machines/evaluate_with_test_cases_spec.rb",
      "spec/machines/infrastructure_spec.rb",
+     "spec/machines/mutate_codeblock_spec.rb",
      "spec/machines/mutate_footnotes_spec.rb",
      "spec/machines/select_by_summed_rank_spec.rb",
      "spec/machines/select_nondominated_spec.rb",
@@ -75,6 +77,7 @@ Gem::Specification.new do |s|
      "spec/machines/evaluate_simple_score_spec.rb",
      "spec/machines/evaluate_with_test_cases_spec.rb",
      "spec/machines/infrastructure_spec.rb",
+     "spec/machines/mutate_codeblock_spec.rb",
      "spec/machines/mutate_footnotes_spec.rb",
      "spec/machines/select_by_summed_rank_spec.rb",
      "spec/machines/select_nondominated_spec.rb",

data/lib/answer-factory.rb

@@ -16,10 +16,12 @@ require 'machines/evaluate_simple_score'
 require 'machines/evaluate_with_test_cases'
 require 'machines/select_nondominated'
 require 'machines/select_by_summed_rank'
+require 'machines/mutate_codeblock'
 require 'machines/mutate_footnotes'
 
 
 
+
 require 'factories/factory'
 require 'factories/workstation'
 

data/lib/answers/answer.rb

@@ -113,6 +113,7 @@ module AnswerFactory
     end
   end
   
+  
   def move_to(where)
     raise ArgumentError, "#{where} is not a Symbol" unless where.kind_of?(Symbol)
     @location = where.to_sym

data/lib/machines/evaluate_with_test_cases.rb

@@ -14,15 +14,13 @@ module AnswerFactory
         @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]
         
+        score_name = all_options[:name]
         
         raise ArgumentError, "EvaluateWithTestCases#score cannot process a #{batch.class}" unless
           batch.kind_of?(Batch)
@@ -33,6 +31,7 @@ module AnswerFactory
         
         batch.each do |answer|
           
+          
           raw_results = Hash.new {|hash, key| hash[key] = []}
           
           @test_cases.each do |t|
@@ -54,13 +53,12 @@ module AnswerFactory
           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
+            answer.scores["#{score_name}_#{sensor_name}".to_sym] =
+              raw_results[sensor_name].inject(0) do |sum, measurement|
+                sum + measurement.abs
+              end
           end
         end
-        
-        
         return batch
       end
       
@@ -87,24 +85,28 @@ module AnswerFactory
       end
       
       
+      def save_view_doc!
+        db = CouchRest.database!(training_datasource)
+        db.save_doc({'_id' => "_design/#{@name}",
+          views: { test_cases: { map: "function(doc) { emit(doc._id, doc); }"}}})
+      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 = CSV.new(File.open(csv_filename), headers: true)
+        headers = csv_headers(reader)
         
-        reader.rewind
+        save_view_doc!
         
-        offset = input_headers.length+1
+        offset = headers[: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}
+          headers[: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}
+          headers[:output_headers].each_with_index {|header,i| outputs[header] = row[i+offset].strip}
           db.bulk_save_doc( {:inputs => inputs, :outputs => outputs})
         end
         
@@ -113,6 +115,16 @@ module AnswerFactory
       end
       
       
+      def csv_headers(csv_reader)
+        csv_reader.readline
+        split_point = csv_reader.headers.find_index(nil)
+        input_headers = csv_reader.headers[0...split_point].collect {|head| header_prep(head)}
+        output_headers = csv_reader.headers[split_point+1..-1].collect {|head| head.strip}
+        csv_reader.rewind
+        return {input_headers:input_headers, output_headers:output_headers}
+      end
+      
+      
       def load_training_data!
         db = CouchRest.database!(training_datasource)
         result = db.view("#{@name}/test_cases")

data/lib/machines/mutate_codeblock.rb

@@ -0,0 +1,42 @@
+#encoding: utf-8
+module AnswerFactory
+  module Machines
+    
+    
+    
+    
+    class MutateCodeblock < Machine
+      
+      def build(batch, overridden_options={})
+        raise ArgumentError, "MutateCodeblock#build cannot process a #{batch.class}" unless
+          batch.kind_of?(Batch)
+        
+        all_options = @options.merge(overridden_options)
+          
+        replicates = all_options[:replicates] || 1
+        
+        result = Batch.new
+        
+        batch.each do |answer|
+          replicates.times do
+            which_point = rand(answer.points) + 1
+            size = all_options[:size_preserving?] ?
+              answer.program[which_point].points :
+              Kernel.rand(2 * answer.program[which_point].points) + 1
+            new_code = NudgeType::CodeType.any_value(all_options.merge({target_size_in_points:size}))
+            mutated_code = answer.replace_point_or_clone(which_point, new_code)
+            variant = Answer.new(mutated_code, progress:answer.progress + 1)
+            result << variant
+          end
+        end
+        
+        return result
+      end
+      
+      
+      alias :generate :build
+      
+      
+    end
+  end
+end

data/spec/integration_specs/couch_db_integration.rspec

@@ -42,18 +42,19 @@ describe "CouchDB stuff" do
   
   describe "Machine CouchDB interaction" do
     before(:each) do
-      the_database_name = "#{@db_uri}_training"
-      @db = CouchRest.database!(the_database_name)
+      @db_uri = "#{configatron.factory.couchdb.server}/#{configatron.factory.couchdb.name}"
+      training_database_name = "#{@db_uri}_training"
+      @db = CouchRest.database!(training_database_name)
     end
     
-    it "should accept :install_training_data_from_csv to populate a db from csv" do
+    after(:each) do
+      @db.delete!
+    end
+    
+    
+    it "should populate the training db from a csv upon :install_training_data_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
@@ -79,6 +80,7 @@ describe "CouchDB stuff" 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')
         @my_machine.load_training_data!
         @my_machine.test_cases[0].inputs["x1:int"].should == "3"
       end

data/spec/machines/evaluate_with_test_cases_spec.rb

@@ -56,7 +56,7 @@ describe "EvaluateWithTestCases" do
   
   describe "initialization" do
     
-    describe "name" do
+    describe "machine name" do
       it "should have a name" do
         @tester.should respond_to(:name)
       end
@@ -127,7 +127,7 @@ describe "EvaluateWithTestCases" do
       it "should register all the sensors" do
         @tester.sensors = {"y1" => Proc.new{|interpreter| interpreter.peek_value(:int)} }
         i = Interpreter.new
-        Interpreter.should_receive(:new).and_return(i)
+        Interpreter.stub(:new).and_return(i)
         i.should_receive(:register_sensor).exactly(1).times
         @tester.score(@batch)
       end
@@ -138,11 +138,9 @@ describe "EvaluateWithTestCases" do
           "do a", {:name=>:tester, :target_size_in_points=>99}).and_return(i)
         @tester.score(@batch,target_size_in_points:99)
       end
-      
-      
-      
     end
     
+    
     describe "install_training_data_from_csv!" do
       before(:each) do
         FakeWeb.register_uri(:any,
@@ -152,10 +150,15 @@ describe "EvaluateWithTestCases" do
           "http://127.0.0.1:5984/dammit_training/_design/tester/_view/test_cases", 
           :body => @canned, :status => [200, "OK"])
         
+        
+        
         @f1 = Factory.new(name: "dammit")
         @my_csv = "./spec/fixtures/my_data_source.csv"
         @m1 = EvaluateWithTestCases.new(name: :tester, training_data_csv: @my_csv)
         @training_db = "http://127.0.0.1:5984/dammit_training"
+        @my_view = {'_id' => "_design/tester",
+            views: { test_cases: { map: 
+              "function(doc) { emit(doc._id, doc); }"}}}
         
       end
       
@@ -166,22 +169,25 @@ describe "EvaluateWithTestCases" do
       end
       
       it "should open a csv file" do
+        @m1.stub(:save_view_doc!)   # we're just checking the file is touched
         f = File.open(@my_csv)
-        File.should_receive(:open).and_return(f)
+        File.stub(:open).and_return(f)
         c = CSV.new(f, headers: true)
-        CSV.should_receive(:new).with(f, headers: true).and_return(c)
+        CSV.stub(:new).with(f, headers: true).and_return(c)
         @m1.install_training_data_from_csv(@my_csv)
       end
       
       it "should be the training_data default db" do
+        @m1.stub(:save_view_doc!)   # we're just checking the filename
         db = CouchRest.database!(@training_db)
-        CouchRest.should_receive(:database!).with(@training_db).and_return(db)
+        CouchRest.stub(:database!).with(@training_db).and_return(db)
         @m1.install_training_data_from_csv(@my_csv)
       end
       
       it "makes one doc for every row" do
+        @m1.stub(:save_view_doc!)   # we're just checking data row saving
         db = CouchRest.database!(@training_db)
-        CouchRest.should_receive(:database!).with(@training_db).and_return(db)
+        CouchRest.stub(:database!).with(@training_db).and_return(db)
         db.should_receive(:bulk_save_doc).exactly(3).times
         @m1.install_training_data_from_csv(@my_csv)
       end
@@ -194,8 +200,18 @@ describe "EvaluateWithTestCases" do
         lambda{@m1.header_prep(":int")}.should raise_error ArgumentError
       end
       
+      it "should create the appropriate view document" do
+        db = CouchRest.database!(@training_db)
+        CouchRest.stub(:database!).with(@training_db).and_return(db)
+        db.stub(:bulk_save_doc)
+        db.should_receive(:save_doc).exactly(1).times.with(@my_view)
+        @m1.install_training_data_from_csv(@my_csv)
+      end
+      
     end
     
+    
+    
     describe "load_training_data!" do
       
       before(:each) do
@@ -220,7 +236,7 @@ describe "EvaluateWithTestCases" do
           "http://127.0.0.1:5984/dammit_training/_design/tester/_view/test_cases", 
           :body => @canned, :status => [200, "OK"])
         db = CouchRest.database!(@m1.training_datasource)
-        CouchRest.should_receive(:database!).and_return(db)
+        CouchRest.stub(:database!).and_return(db)
         db.should_receive(:view).with(@design_doc).and_return(@expected)
         @m1.load_training_data!
       end
@@ -229,14 +245,14 @@ describe "EvaluateWithTestCases" do
       
       it "should ask for the view document" do
         db = CouchRest.database!(@m1.training_datasource)
-        CouchRest.should_receive(:database!).and_return(db)
+        CouchRest.stub(:database!).and_return(db)
         db.should_receive(:view).with(@design_doc).and_return(@expected)
         @m1.load_training_data!
       end
       
       it "should store Array of TestCases in @test_cases" do
         db = CouchRest.database!(@m1.training_datasource)
-        CouchRest.should_receive(:database!).and_return(db)
+        CouchRest.stub(:database!).and_return(db)
         db.should_receive(:view).with(@design_doc).and_return(@expected)
         
         @m1.load_training_data!
@@ -245,6 +261,7 @@ describe "EvaluateWithTestCases" do
       end
     end
     
+    
     describe "scoring" do
       before(:each) do
         @m1 = EvaluateWithTestCases.new(name: :tester)
@@ -266,28 +283,35 @@ describe "EvaluateWithTestCases" do
       
       it "should run all the Interpreters" do
         @m1.stub!(:load_training_data!)
-        Interpreter.should_receive(:new).exactly(10).times.and_return(@i1)
+        Interpreter.stub(:new).exactly(10).times.and_return(@i1)
         @i1.should_receive(:run).at_least(1).times.and_return({})
         @m1.score(@batch)
       end
       
       it "should register its sensors before each Interpreter run" do
-        Interpreter.stub!(:new).and_return(@i1)
+        Interpreter.stub(:new).and_return(@i1)
         @i1.should_receive(:register_sensor).at_least(1).times
         @m1.score(@batch)
       end
       
       it "should have a score for each sensor" do
         @m1.score(@batch)
-        @batch.first.scores["y1"].should_not == nil
-        @batch.first.scores["y2"].should_not == nil
+        @batch.first.scores[:tester_y1].should_not == nil
+        @batch.first.scores[:tester_y2].should_not == nil
+      end
+      
+      it "should not re-evaluate an Answer's score if :static is true and the score is set" do
+        pending
+        @m1.score(@batch)
+        Interpreter.should_not_receive(:new)
+        @m1.score(@batch, static:true)
       end
       
       it "should return sum of absolute errors" do
         @m1.stub!(:load_training_data!)
         @m1.score(@batch)
-        @batch[0].scores["y1"].should == 777+775+773+771+769+767+765+763+761+759
-        @batch[0].scores["y2"].should == 666+663+660+657+654+651+648+645+642+639
+        @batch[0].scores[:tester_y1].should == 777+775+773+771+769+767+765+763+761+759
+        @batch[0].scores[:tester_y2].should == 666+663+660+657+654+651+648+645+642+639
       end
     end
     

data/spec/machines/mutate_codeblock_spec.rb

@@ -0,0 +1,93 @@
+#encoding: utf-8
+require File.join(File.dirname(__FILE__), "./../spec_helper")
+
+describe "Machines::MutateCodeblock" do
+  before(:each) do
+    @twiddler = Machines::MutateCodeblock.new
+    @two = Batch.[](
+      Answer.new("value «int» \n«int» 1"),
+      Answer.new("block { block {value «code»} ref x1 block { block {ref x2}}} \n«code» value «bool» \n«bool» false"))
+  end
+  
+  
+  describe "#build method" do
+    it "should respond to :build" do
+      @twiddler.should respond_to(:build)
+    end
+    
+    it "should respond to :generate as an alias of :build" do
+      @twiddler.should respond_to(:generate)
+    end
+    
+    it "should raise an error if the argument isn't a Batch" do
+      lambda{@twiddler.build(@two)}.should_not raise_error
+      lambda{@twiddler.build(99)}.should raise_error
+    end
+    
+    it "should produce a Batch" do
+      @twiddler.build(@two).should be_a_kind_of(Batch)
+    end
+    
+    it "should produce a Batch with a new object_id" do
+      @twiddler.build(@two).object_id.should_not == @two.object_id
+    end
+    
+    it "should not modify or include any Answer object from its argument Batch" do
+      original_ids = @two.collect {|a| a.object_id}
+      result_ids = @twiddler.build(@two).collect {|a| a.object_id}
+      (original_ids & result_ids).length.should == 0
+    end
+    
+    describe ":replicates option" do
+      it "should produce a :replicates for each arg Answer, determined by a call option" do
+        @twiddler.build(@two, replicates:2).length.should == 2 * @two.length
+      end
+
+      it "should use the initialization :replicates option if there isn't a call option" do
+        threefer = Machines::MutateCodeblock.new(replicates:3)
+        threefer.build(@two).length.should == 3*@two.length
+      end
+    
+      it "should default to replicates:1 if none was set as an option" do
+        @twiddler.build(@two).length.should == @two.length
+      end
+    end
+    
+    it "should increment the :progress attribute of the derived Answers, regardless of other settings" do
+      @twiddler.build(@two).each {|a| a.progress.should == 1}
+    end
+    
+    describe "mutating code" do
+      before(:each) do
+        @some_code = ReferencePoint.new(:abcde)
+      end
+      
+      it "should replace one program point in each mutant"
+      
+      it "should pick the replaced point with uniform probability" do
+        CodeType.should_receive(:any_value).exactly(2).times.and_return(@some_code)
+        @twiddler.build(@two)
+      end
+      
+      it "should replace a point with one that's the same size if :size_preserving? is true" do
+        Kernel.should_not_receive(:rand)
+        new_dudes = @twiddler.build(@two, size_preserving?:true)
+        new_dudes.each_with_index {|a,idx| a.points.should == @two[idx].points}
+      end
+      
+      it "should replace a point with one between 1 and 2xsize if :sizepreserving? is false" do
+        Kernel.stub!(:rand).and_return(17)
+        CodeType.should_receive(:any_value).at_least(2).times.
+          with(hash_including(size_preserving?:false,target_size_in_points:18)).and_return(@some_code)
+        @twiddler.build(@two, size_preserving?:false)
+      end
+            
+      it "should pass all options through to the any_value call" do
+        Kernel.stub!(:rand).and_return(3)
+        CodeType.should_receive(:any_value).at_least(2).times.
+          with(hash_including(target_size_in_points:4, foo:12)).and_return(@some_code)
+        @twiddler.build(@two, foo:12)
+      end
+    end
+  end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification 
 name: answer-factory
 version: !ruby/object:Gem::Version 
-  hash: 67
+  hash: 65
   prerelease: false
   segments: 
   - 0
   - 1
   - 3
-  - 4
-  version: 0.1.3.4
+  - 5
+  version: 0.1.3.5
 platform: ruby
 authors: 
 - Bill Tozier
@@ -18,7 +18,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-05-27 00:00:00 -04:00
+date: 2010-06-02 00:00:00 -04:00
 default_executable: answer-factory
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -157,6 +157,7 @@ files:
 - lib/machines/evaluate_simple_score.rb
 - lib/machines/evaluate_with_test_cases.rb
 - lib/machines/infrastructure.rb
+- lib/machines/mutate_codeblock.rb
 - lib/machines/mutate_footnotes.rb
 - lib/machines/select_by_summed_rank.rb
 - lib/machines/select_nondominated.rb
@@ -172,6 +173,7 @@ files:
 - spec/machines/evaluate_simple_score_spec.rb
 - spec/machines/evaluate_with_test_cases_spec.rb
 - spec/machines/infrastructure_spec.rb
+- spec/machines/mutate_codeblock_spec.rb
 - spec/machines/mutate_footnotes_spec.rb
 - spec/machines/select_by_summed_rank_spec.rb
 - spec/machines/select_nondominated_spec.rb
@@ -225,6 +227,7 @@ test_files:
 - spec/machines/evaluate_simple_score_spec.rb
 - spec/machines/evaluate_with_test_cases_spec.rb
 - spec/machines/infrastructure_spec.rb
+- spec/machines/mutate_codeblock_spec.rb
 - spec/machines/mutate_footnotes_spec.rb
 - spec/machines/select_by_summed_rank_spec.rb
 - spec/machines/select_nondominated_spec.rb