Add sample gathering class

lib/linguist/sample.rb

@@ -0,0 +1,46 @@
+require 'linguist/classifier'
+require 'linguist/language'
+
+module Linguist
+  class Sample
+    # Samples live in test/ for now, we'll eventually move them out
+    PATH = File.expand_path("../../../test/fixtures", __FILE__)
+
+    def self.each(&block)
+      Dir.entries(PATH).each do |category|
+        next if category == '.' || category == '..'
+
+        # Skip text and binary for now
+        next if category == 'text' || category == 'binary'
+
+        language = Linguist::Language.find_by_alias(category)
+        raise "No language for #{category.inspect}" unless language
+
+        dirname = File.join(PATH, category)
+        Dir.entries(dirname).each do |filename|
+          next if filename == '.' || filename == '..'
+          yield new(File.join(dirname, filename), language)
+        end
+      end
+
+      nil
+    end
+
+    def self.classifier
+      classifier = Classifier.new
+      each { |sample| classifier.train(sample.language, sample.data) }
+      classifier
+    end
+
+    def initialize(path, language)
+      @path     = path
+      @language = language
+    end
+
+    def data
+      File.read(path)
+    end
+
+    attr_reader :path, :language
+  end
+end

test/test_blob.rb

@@ -1,4 +1,5 @@
 require 'linguist/file_blob'
+require 'linguist/sample'
 
 require 'test/unit'
 require 'mime/types'
@@ -24,23 +25,6 @@ class TestBlob < Test::Unit::TestCase
     blob
   end
 
-  def each_language_fixture
-    Dir["#{fixtures_path}/*"].each do |path|
-      name = File.basename(path)
-
-      if name == 'text' || name == 'binary'
-        next
-      else
-        assert language = Language.find_by_alias(name), "No language alias for #{name.inspect}"
-      end
-
-      Dir.entries(path).each do |filename|
-        next if filename == '.' || filename == '..'
-        yield language, blob(File.join(path, filename))
-      end
-    end
-  end
-
   def test_name
     assert_equal "foo.rb", blob("foo.rb").name
   end
@@ -287,9 +271,9 @@ class TestBlob < Test::Unit::TestCase
   end
 
   def test_language
-    # Drop any files under test/fixtures/LANGUAGE
+    Sample.each do |sample|
-    each_language_fixture do |language, blob|
+      blob = blob(sample.path)
-      assert_equal language, blob.language, blob.name
+      assert_equal sample.language, blob.language, blob.name
     end
   end