From b7f58d96cbc5ff4c2954976f65adbf18930ce810 Mon Sep 17 00:00:00 2001
From: Joshua Peek <josh@joshpeek.com>
Date: Mon, 23 Jul 2012 12:17:32 -0500
Subject: [PATCH] Compare md5s of dbs

---
 lib/linguist/classifier.rb | 26 +++++++++++++-------------
 test/test_classifier.rb    |  6 +++++-
 2 files changed, 18 insertions(+), 14 deletions(-)

diff --git a/lib/linguist/classifier.rb b/lib/linguist/classifier.rb
index 7a348aa4..843ffa3c 100644
--- a/lib/linguist/classifier.rb
+++ b/lib/linguist/classifier.rb
@@ -29,19 +29,6 @@ module Linguist
       @languages       = attrs['languages'] || {}
     end
 
-    # Public: Compare Classifier objects.
-    #
-    # other - Classifier object to compare to.
-    #
-    # Returns Boolean.
-    def eql?(other)
-      # Lazy fast check counts only
-      other.is_a?(self.class) &&
-        @tokens_total == other.instance_variable_get(:@tokens_total) &&
-        @languages_total == other.instance_variable_get(:@languages_total)
-    end
-    alias_method :==, :eql?
-
     # Public: Train classifier that data is a certain language.
     #
     # language - String language of data
@@ -146,6 +133,19 @@ module Linguist
       Math.log(@languages[language].to_f / @languages_total.to_f)
     end
 
+    # Public: Returns serializable hash representation.
+    #
+    # Returns Hash.
+    def to_hash
+      {
+        'tokens_total'    => @tokens_total,
+        'languages_total' => @languages_total,
+        'tokens'          => @tokens,
+        'language_tokens' => @language_tokens,
+        'languages'       => @languages
+      }
+    end
+
     # Public: Serialize classifier to YAML.
     #
     # opts - Hash of YAML options.
diff --git a/test/test_classifier.rb b/test/test_classifier.rb
index 80df7e77..0571f85e 100644
--- a/test/test_classifier.rb
+++ b/test/test_classifier.rb
@@ -2,6 +2,7 @@ require 'linguist/classifier'
 require 'linguist/language'
 require 'linguist/sample'
 require 'linguist/tokenizer'
+require 'linguist/md5'
 
 require 'test/unit'
 
@@ -17,8 +18,11 @@ class TestClassifier < Test::Unit::TestCase
   end
 
   def test_instance_freshness
+    serialized = Linguist::MD5.hexdigest(Classifier.instance.to_hash)
+    latest     = Linguist::MD5.hexdigest(Linguist::Sample.classifier.to_hash)
+
     # Just warn, it shouldn't scare people off by breaking the build.
-    unless Classifier.instance.eql?(Linguist::Sample.classifier)
+    if serialized != latest
       warn "Classifier database is out of date. Run `bundle exec rake classifier`."
     end
   end