Load classifer db into sample data hash

5 files changed, 52 insertions, 63 deletions

diff --git a/Rakefile b/Rakefile
index d093868..cade0dc 100644
--- a/Rakefile
+++ b/Rakefile
@@ -10,8 +10,7 @@ end
 
 file 'lib/linguist/samples.yml' => Dir['samples/**/*'] do |f|
   require 'linguist/sample'
-  classifier = Linguist::Sample.classifier
-  File.open(f.name, 'w') { |io| classifier.to_yaml(io) }
+  File.open(f.name, 'w') { |io| Linguist::Sample.serialize_to_yaml(Linguist::Sample::DATA, io) }
 end
 
 CLOBBER.include 'lib/linguist/samples.yml'
@@ -32,7 +31,7 @@ namespace :classifier do
       next if file_language.nil? || file_language == 'Text'
       begin
         data = open(file_url).read
-        guessed_language, score = Linguist::Classifier.instance.classify(data).first
+        guessed_language, score = Linguist::Classifier.new(Sample::DATA).classify(data).first
 
         total += 1
         guessed_language == file_language ? correct += 1 : incorrect += 1
diff --git a/lib/linguist/blob_helper.rb b/lib/linguist/blob_helper.rb
index 1ea5b44..b932612 100644
--- a/lib/linguist/blob_helper.rb
+++ b/lib/linguist/blob_helper.rb
@@ -2,6 +2,7 @@ require 'linguist/classifier'
 require 'linguist/language'
 require 'linguist/mime'
 require 'linguist/pathname'
+require 'linguist/sample'
 
 require 'charlock_holmes'
 require 'escape_utils'
@@ -441,7 +442,7 @@ module Linguist
       if Language.ambiguous?(extname)
         possible_languages = Language.all.select { |l| l.extensions.include?(extname) }.map(&:name)
         if possible_languages.any?
-          if result = Classifier.instance.classify(data, possible_languages).first
+          if result = Classifier.new(Sample::DATA).classify(data, possible_languages).first
             Language[result[0]]
           end
         end
diff --git a/lib/linguist/classifier.rb b/lib/linguist/classifier.rb
index 2cbd1bd..d50f75d 100644
--- a/lib/linguist/classifier.rb
+++ b/lib/linguist/classifier.rb
@@ -3,23 +3,6 @@ require 'linguist/tokenizer'
 module Linguist
   # Language bayesian classifier.
   class Classifier
-    # Internal: Path to persisted classifier db.
-    PATH = File.expand_path('../samples.yml', __FILE__)
-
-    # Public: Check if persisted db exists on disk.
-    #
-    # Returns Boolean.
-    def self.exist?
-      File.exist?(PATH)
-    end
-
-    # Public: Get persisted Classifier instance.
-    #
-    # Returns Classifier.
-    def self.instance
-      @instance ||= new(YAML.load_file(PATH))
-    end
-
     # Public: Initialize a Classifier.
     def initialize(attrs = {})
       @tokens_total    = attrs['tokens_total'] || 0
@@ -129,42 +112,5 @@ module Linguist
         'languages'       => @languages
       }
     end
-
-    # Public: Serialize classifier to YAML.
-    #
-    # opts - Hash of YAML options.
-    #
-    # Returns nothing.
-    def to_yaml(io)
-      data = ""
-      escape = lambda { |s| s.inspect.gsub(/\\#/, "\#") }
-
-      data << "languages_total: #{@languages_total}\n"
-      data << "tokens_total: #{@tokens_total}\n"
-
-      data << "languages:\n"
-      @languages.sort.each do |language, count|
-        data << "  #{escape.call(language)}: #{count}\n"
-      end
-
-      data << "language_tokens:\n"
-      @language_tokens.sort.each do |language, count|
-        data << "  #{escape.call(language)}: #{count}\n"
-      end
-
-      data << "tokens:\n"
-      @tokens.sort.each do |language, tokens|
-        data << "  #{escape.call(language)}:\n"
-        tokens.sort.each do |token, count|
-          data << "    #{escape.call(token)}: #{count}\n"
-        end
-      end
-
-      io.write data
-      nil
-    end
   end
-
-  # Eager load instance
-  Classifier.instance if Classifier.exist?
 end
diff --git a/lib/linguist/sample.rb b/lib/linguist/sample.rb
index 77b1102..4c3f5e9 100644
--- a/lib/linguist/sample.rb
+++ b/lib/linguist/sample.rb
@@ -1,4 +1,5 @@
 require 'set'
+require 'yaml'
 
 module Linguist
   # Model for accessing classifier training data.
@@ -6,6 +7,13 @@ module Linguist
     # Samples live in test/ for now, we'll eventually move them out
     PATH = File.expand_path("../../../samples", __FILE__)
 
+    YML = File.expand_path('../samples.yml', __FILE__)
+    if File.exist?(YML)
+      DATA = YAML.load_file(YML)
+    else
+      DATA = nil
+    end
+
     # Public: Iterate over each sample.
     #
     # &block - Yields Sample to block
@@ -91,5 +99,40 @@ module Linguist
       }
       classifier
     end
+    
+    # Public: Serialize samples data to YAML.
+    #
+    # data - Hash
+    # io   - IO object to write to
+    #
+    # Returns nothing.
+    def self.serialize_to_yaml(data, io)
+      data = ""
+      escape = lambda { |s| s.inspect.gsub(/\\#/, "\#") }
+
+      data << "languages_total: #{data['languages_total']}\n"
+      data << "tokens_total: #{data['tokens_total']}\n"
+
+      data << "languages:\n"
+      data['languages'].sort.each do |language, count|
+        data << "  #{escape.call(language)}: #{count}\n"
+      end
+
+      data << "language_tokens:\n"
+      data['language_tokens'].sort.each do |language, count|
+        data << "  #{escape.call(language)}: #{count}\n"
+      end
+
+      data << "tokens:\n"
+      data['tokens'].sort.each do |language, tokens|
+        data << "  #{escape.call(language)}:\n"
+        tokens.sort.each do |token, count|
+          data << "    #{escape.call(token)}: #{count}\n"
+        end
+      end
+
+      io.write data
+      nil
+    end
   end
 end
diff --git a/test/test_classifier.rb b/test/test_classifier.rb
index e648dce..2d55aa0 100644
--- a/test/test_classifier.rb
+++ b/test/test_classifier.rb
@@ -18,7 +18,7 @@ class TestClassifier < Test::Unit::TestCase
   end
 
   def test_instance_freshness
-    serialized = Linguist::MD5.hexdigest(Classifier.instance.to_hash)
+    serialized = Linguist::MD5.hexdigest(Sample::DATA)
     latest     = Linguist::MD5.hexdigest(Linguist::Sample.classifier.to_hash)
 
     # Just warn, it shouldn't scare people off by breaking the build.
@@ -55,16 +55,16 @@ class TestClassifier < Test::Unit::TestCase
   end
 
   def test_instance_classify_empty
-    results = Classifier.instance.classify("")
+    results = Classifier.new(Sample::DATA).classify("")
     assert results.first[1] < 0.5, results.first.inspect
   end
 
   def test_instance_classify_nil
-    assert_equal [], Classifier.instance.classify(nil)
+    assert_equal [], Classifier.new(Sample::DATA).classify(nil)
   end
 
   def test_verify
-    data = Classifier.instance.to_hash
+    data = Sample::DATA
 
     assert_equal data['languages_total'], data['languages'].inject(0) { |n, (_, c)| n += c }
     assert_equal data['tokens_total'], data['language_tokens'].inject(0) { |n, (_, c)| n += c }
@@ -80,7 +80,7 @@ class TestClassifier < Test::Unit::TestCase
       languages = Language.all.select { |l| l.extensions.include?(extname) }.map(&:name)
       next unless languages.length > 1
 
-      results = Classifier.instance.classify(File.read(sample[:path]), languages)
+      results = Classifier.new(Sample::DATA).classify(File.read(sample[:path]), languages)
       assert_equal language.name, results.first[0], "#{sample[:path]}\n#{results.inspect}"
     end
   end