first pass on import/export

.gitignore

@@ -1,6 +1,8 @@
 Gemfile.lock
 pkg
+
 *.rdb
 docs/_site/
 docs/.sass-cache/
 docs/.jekyll-metadata
+test/fixtures/export.yml

.rubocop.yml

@@ -1,6 +1,3 @@
-Style/IfUnlessModifier:
-  MaxLineLength: 150
-
 Metrics/LineLength:
   Max: 146
 
@@ -28,7 +25,7 @@ Metrics/ClassLength:
 SingleLineBlockParams:
   Enabled: false
 
-Lint/Eval:
+Security/Eval:
   Enabled: false
 
 Lint/AssignmentInCondition:
@@ -37,10 +34,10 @@ Lint/AssignmentInCondition:
 SignalException:
   Enabled: false
 
-Style/FileName:
+Naming/FileName:
   Enabled: false
 
-Style/MethodName:
+Naming/MethodName:
   Enabled: false
 
 Lint/UnusedBlockArgument:

lib/classifier-reborn/backends/bayes_memory_backend.rb

@@ -1,6 +1,9 @@
+require_relative 'data_handler'
+
 module ClassifierReborn
   class BayesMemoryBackend
     attr_reader :total_words, :total_trainings
+    include DataHandler
 
     # This class provides Memory as the storage backend for the classifier data structures
     def initialize

lib/classifier-reborn/backends/bayes_redis_backend.rb

@@ -1,3 +1,4 @@
+require_relative 'data_handler'
 require_relative 'no_redis_error'
 # require redis when we run #intialize. This way only people using this backend
 # will need to install and load the backend without having to
@@ -6,6 +7,7 @@
 module ClassifierReborn
   # This class provides Redis as the storage backend for the classifier data structures
   class BayesRedisBackend
+    include DataHandler
     # The class can be created with the same arguments that the redis gem accepts
     # E.g.,
     #      b = ClassifierReborn::BayesRedisBackend.new

lib/classifier-reborn/backends/data_handler.rb

@@ -0,0 +1,22 @@
+module DataHandler
+  # Read the data and populate the backend in use
+  def import!(data)
+    data[:categories].keys.each { |category| add_category(category) }
+    categories = data[:categories]
+    categories.each_key do |category|
+      categories[category].each do |word, diff|
+        update_category_word_frequency(category, word, diff)
+      end
+    end
+    update_total_words(data[:total_words])
+  end
+
+  def export
+    {
+      categories: @categories,
+      category_counts: @category_counts,
+      category_word_count: @category_word_count,
+      total_words: @total_words
+    }
+  end
+end

lib/classifier-reborn/bayes.rb

@@ -1,7 +1,6 @@
 # Author::    Lucas Carlson  (mailto:[email protected])
 # Copyright:: Copyright (c) 2005 Lucas Carlson
 # License::   LGPL
-
 require 'set'
 
 require_relative 'extensions/tokenizer/whitespace'
@@ -261,6 +260,27 @@ def reset
       populate_initial_categories
     end
 
+    def import!(data)
+      @auto_categorize = data[:auto_categorize]
+      @enable_stemmer = data[:enable_stemmer]
+      @enable_threshold = data[:enable_threshold]
+      @initial_categories = data[:categories].keys.map(&:to_s)
+      @language = data[:language]
+      @threshold = data[:threshold]
+      @backend.import!(data)
+    end
+
+    def export
+      backend_data = @backend.export
+      {
+        auto_categorize: @auto_categorize,
+        enable_stemmer: @enable_stemmer,
+        enable_threshold: @enable_threshold,
+        language: @language,
+        threshold: @threshold
+      }.merge(backend_data)
+    end
+
     private
 
     def populate_initial_categories

test/bayes/bayesian_common_tests.rb

@@ -1,4 +1,5 @@
 # encoding: utf-8
+require 'yaml'
 
 module BayesianCommonTests
   def test_good_training
@@ -191,6 +192,25 @@ def test_reset
     assert classifier.categories.empty?
   end
 
+  def test_export
+    classifier = another_classifier
+    classifier.train_interesting %"Dutch painting of the Golden Age is included in the general European 
+                                   period of Baroque painting, and often shows many of its characteristics 
+                                   most lacks the idealization"
+    classifier.train_uninteresting %"Grasslands such as savannah and prairie where grasses are dominant are 
+                                    estimated to constitute forty percent of the land area of the Earth"
+    exported_data = classifier.export
+    reference_data = YAML.load(File.read('test/fixtures/reference.yml'))
+    assert_equal(exported_data, reference_data)
+  end
+
+  def test_import
+    classifier = ClassifierReborn::Bayes.new backend: @alternate_backend
+    reference_data = YAML.load(File.read('test/fixtures/reference.yml'))
+    classifier.import!(reference_data)
+    assert_equal('Interesting', classifier.classify('Dutch painting of the Golden Age'))
+  end
+
   private
 
   def another_classifier

test/fixtures/reference.yml

@@ -0,0 +1,49 @@
+---
+:auto_categorize: false
+:enable_stemmer: true
+:enable_threshold: false
+:language: en
+:threshold: 0.0
+:categories:
+  :Interesting:
+    :dutch: 1
+    :paint: 2
+    :golden: 1
+    :ag: 1
+    :includ: 1
+    :gener: 1
+    :european: 1
+    :period: 1
+    :baroqu: 1
+    :often: 1
+    :show: 1
+    :mani: 1
+    :it: 1
+    :characterist: 1
+    :lack: 1
+    :ideal: 1
+    :,: 1
+  :Uninteresting:
+    :grassland: 1
+    :such: 1
+    :savannah: 1
+    :prairi: 1
+    :where: 1
+    :grass: 1
+    :domin: 1
+    :estim: 1
+    :constitut: 1
+    :forti: 1
+    :percent: 1
+    :land: 1
+    :area: 1
+    :earth: 1
+:category_counts:
+  :Interesting:
+    :training: 1
+    :word: 18
+  :Uninteresting:
+    :training: 1
+    :word: 14
+:category_word_count: 
+:total_words: 32