From 516a220d9f6e27784f4eed807df8692859f8bc44 Mon Sep 17 00:00:00 2001
From: Joshua Peek <josh@joshpeek.com>
Date: Wed, 20 Jun 2012 15:48:46 -0500
Subject: [PATCH] Verify classifer counts

---
 lib/linguist/classifier.rb | 9 +++++++++
 test/test_classifier.rb    | 4 ++++
 2 files changed, 13 insertions(+)

diff --git a/lib/linguist/classifier.rb b/lib/linguist/classifier.rb
index fe2c5a14..d3731c2d 100644
--- a/lib/linguist/classifier.rb
+++ b/lib/linguist/classifier.rb
@@ -68,6 +68,15 @@ module Linguist
       nil
     end
 
+    # Public: Verify internal counts are consistent.
+    #
+    # Returns Boolean.
+    def verify
+      @languages.inject(0) { |n, (l, c)| n += c } == @languages_total &&
+        @language_tokens.inject(0) { |n, (l, c)| n += c } == @tokens_total &&
+        @tokens.inject(0) { |n, (l, ts)| n += ts.inject(0) { |m, (t, c)| m += c } } == @tokens_total
+    end
+
     # Public: Prune infrequent tokens.
     #
     # Returns receiver Classifier instance.
diff --git a/test/test_classifier.rb b/test/test_classifier.rb
index c329364d..b39fb20c 100644
--- a/test/test_classifier.rb
+++ b/test/test_classifier.rb
@@ -55,6 +55,10 @@ class TestClassifier < Test::Unit::TestCase
     assert results.first[1] < 0.5, results.first.inspect
   end
 
+  def test_verify
+    assert Classifier.instance.verify
+  end
+
   def test_gc
     Classifier.instance.gc
   end