Store md5 in samples db

lib/linguist/samples.rb

@@ -18,13 +18,6 @@ module Linguist
       DATA = nil
     end
 
-    # Check if serialized db is out of sync from db directory.
-    #
-    # Returns Boolean.
-    def self.outdated?
-      MD5.hexdigest(DATA) != MD5.hexdigest(data)
-    end
-
     # Public: Iterate over each sample.
     #
     # &block - Yields Sample to block
@@ -108,6 +101,7 @@ module Linguist
         data     = File.read(sample[:path])
         Classifier.train!(db, language.name, data)
       end
+      db['md5'] = MD5.hexdigest(db)
       db
     end
     
@@ -120,6 +114,8 @@ module Linguist
       out = ""
       escape = lambda { |s| s.inspect.gsub(/\\#/, "\#") }
 
+      out << "md5: #{db['md5']}\n"
+
       out << "languages_total: #{db['languages_total']}\n"
       out << "tokens_total: #{db['tokens_total']}\n"
 

lib/linguist/samples.yml

@@ -1,3 +1,4 @@
+md5: b445a8a3e3414d6b628939c347e7a4f3
 languages_total: 243
 tokens_total: 164127
 languages:

test/test_classifier.rb

@@ -18,7 +18,7 @@ class TestClassifier < Test::Unit::TestCase
 
   def test_instance_freshness
     # Just warn, it shouldn't scare people off by breaking the build.
-    if Samples.outdated?
+    if Samples::DATA['md5'] != Samples.data['md5']
       warn "Classifier database is out of date. Run `bundle exec rake classifier`."
     end
   end