From b9779e805e16b7097a5883334f6ed712abb10314 Mon Sep 17 00:00:00 2001
From: Joshua Peek <josh@joshpeek.com>
Date: Mon, 23 Jul 2012 13:21:30 -0500
Subject: [PATCH] Move outdated check to samples

---
 lib/linguist/samples.rb | 25 ++++++++++++++++++-------
 test/test_classifier.rb |  6 +-----
 2 files changed, 19 insertions(+), 12 deletions(-)

diff --git a/lib/linguist/samples.rb b/lib/linguist/samples.rb
index f85f9ae0..85d3193a 100644
--- a/lib/linguist/samples.rb
+++ b/lib/linguist/samples.rb
@@ -1,33 +1,44 @@
 require 'set'
 require 'yaml'
+require 'linguist/md5'
 
 module Linguist
   # Model for accessing classifier training data.
   module Samples
-    # Samples live in test/ for now, we'll eventually move them out
-    PATH = File.expand_path("../../../samples", __FILE__)
+    # Path to samples root directory
+    ROOT = File.expand_path("../../../samples", __FILE__)
 
-    YML = File.expand_path('../samples.yml', __FILE__)
-    if File.exist?(YML)
-      DATA = YAML.load_file(YML)
+    # Path for serialized samples db
+    PATH = File.expand_path('../samples.yml', __FILE__)
+
+    # Hash of serialized samples object
+    if File.exist?(PATH)
+      DATA = YAML.load_file(PATH)
     else
       DATA = nil
     end
 
+    # Check if serialized db is out of sync from db directory.
+    #
+    # Returns Boolean.
+    def self.outdated?
+      MD5.hexdigest(DATA) != MD5.hexdigest(classifier.to_hash)
+    end
+
     # Public: Iterate over each sample.
     #
     # &block - Yields Sample to block
     #
     # Returns nothing.
     def self.each(&block)
-      Dir.entries(PATH).each do |category|
+      Dir.entries(ROOT).each do |category|
         next if category == '.' || category == '..'
 
         # Skip text and binary for now
         # Possibly reconsider this later
         next if category == 'text' || category == 'binary'
 
-        dirname = File.join(PATH, category)
+        dirname = File.join(ROOT, category)
         Dir.entries(dirname).each do |filename|
           next if filename == '.' || filename == '..'
 
diff --git a/test/test_classifier.rb b/test/test_classifier.rb
index 90a7d239..890696fb 100644
--- a/test/test_classifier.rb
+++ b/test/test_classifier.rb
@@ -2,7 +2,6 @@ require 'linguist/classifier'
 require 'linguist/language'
 require 'linguist/samples'
 require 'linguist/tokenizer'
-require 'linguist/md5'
 
 require 'test/unit'
 
@@ -18,11 +17,8 @@ class TestClassifier < Test::Unit::TestCase
   end
 
   def test_instance_freshness
-    serialized = Linguist::MD5.hexdigest(Samples::DATA)
-    latest     = Linguist::MD5.hexdigest(Linguist::Samples.classifier.to_hash)
-
     # Just warn, it shouldn't scare people off by breaking the build.
-    if serialized != latest
+    if Samples.outdated?
       warn "Classifier database is out of date. Run `bundle exec rake classifier`."
     end
   end