Test against real gist data

Rakefile

@@ -17,3 +17,52 @@ end
 CLOBBER.include 'lib/linguist/classifier.yml'
 
 task :classifier => ['lib/linguist/classifier.yml']
+
+namespace :classifier do
+  LIMIT = 1_000
+
+  desc "Run classifier against #{LIMIT} public gists"
+  task :test do
+    require 'linguist/classifier'
+
+    total, correct, incorrect = 0, 0, 0
+    $stdout.sync = true
+
+    each_public_gist do |gist_url, file_url, file_language|
+      next if file_language.nil? || file_language == 'Text'
+      begin
+        data = open(file_url).read
+        guessed_language, score = Linguist::Classifier.instance.classify(data).first
+
+        total += 1
+        guessed_language.name == file_language ? correct += 1 : incorrect += 1
+
+        print "\r\e[0K%d:%d  %g%%" % [correct, incorrect, (correct.to_f/total.to_f)*100]
+        $stdout.flush
+      rescue URI::InvalidURIError
+      else
+        break if total >= LIMIT
+      end
+    end
+    puts ""
+  end
+
+  def each_public_gist
+    require 'open-uri'
+    require 'json'
+
+    url = "https://api.github.com/gists/public"
+
+    loop do
+      resp = open(url)
+      url = resp.meta['link'][/<([^>]+)>; rel="next"/, 1]
+      gists = JSON.parse(resp.read)
+
+      for gist in gists
+        for filename, attrs in gist['files']
+          yield gist['url'], attrs['raw_url'], attrs['language']
+        end
+      end
+    end
+  end
+end

linguist.gemspec

@@ -12,5 +12,6 @@ Gem::Specification.new do |s|
   s.add_dependency 'escape_utils',    '~> 0.2.3'
   s.add_dependency 'mime-types',      '~> 1.18'
   s.add_dependency 'pygments.rb',     '~> 0.2.11'
+  s.add_development_dependency 'json'
   s.add_development_dependency 'rake'
 end