Merge master

test/fixtures/Python/run_tests.module

@@ -0,0 +1,22 @@
+#!/usr/bin/env python
+import sys, os
+
+# Set the current working directory to the directory where this script is located
+os.chdir(os.path.abspath(os.path.dirname(sys.argv[0])))
+
+#### Set the name of the application here and moose directory relative to the application
+app_name = 'stork'
+
+MODULE_DIR = os.path.abspath('..')
+MOOSE_DIR = os.path.abspath(os.path.join(MODULE_DIR, '..'))
+#### See if MOOSE_DIR is already in the environment instead
+if os.environ.has_key("MOOSE_DIR"):
+  MOOSE_DIR = os.environ['MOOSE_DIR']
+
+sys.path.append(os.path.join(MOOSE_DIR, 'python'))
+import path_tool
+path_tool.activate_module('TestHarness')
+
+from TestHarness import TestHarness
+# Run the tests!
+TestHarness.buildAndRun(sys.argv, app_name, MOOSE_DIR)

test/helper.rb

@@ -0,0 +1,4 @@
+require "bundler/setup"
+require "test/unit"
+require "mocha/setup"
+require "linguist"

test/test_blob.rb

@@ -1,9 +1,4 @@
-require 'linguist/file_blob'
-require 'linguist/samples'
-
-require 'test/unit'
-require 'mocha/setup'
-require 'mime/types'
+require_relative "./helper"
 
 class TestBlob < Test::Unit::TestCase
   include Linguist
@@ -470,6 +465,25 @@ class TestBlob < Test::Unit::TestCase
       assert blob.language, "No language for #{sample[:path]}"
       assert_equal sample[:language], blob.language.name, blob.name
     end
+
+    # Test language detection for files which shouldn't be used as samples
+    root = File.expand_path('../fixtures', __FILE__)
+    Dir.entries(root).each do |language|
+      next unless File.file?(language)
+
+      # Each directory contains test files of a language
+      dirname = File.join(root, language)
+      Dir.entries(dirname).each do |filename|
+        next unless File.file?(filename)
+        
+        # By default blob search the file in the samples;
+        # thus, we need to give it the absolute path
+        filepath = File.join(dirname, filename)
+        blob = blob(filepath)
+        assert blob.language, "No language for #{filepath}"
+        assert_equal language, blob.language.name, blob.name
+      end
+    end
   end
 
   def test_minified_files_not_safe_to_highlight

test/test_classifier.rb

@@ -1,9 +1,4 @@
-require 'linguist/classifier'
-require 'linguist/language'
-require 'linguist/samples'
-require 'linguist/tokenizer'
-
-require 'test/unit'
+require_relative "./helper"
 
 class TestClassifier < Test::Unit::TestCase
   include Linguist

test/test_file_blob.rb

@@ -0,0 +1,10 @@
+require 'linguist/file_blob'
+require 'test/unit'
+
+class TestFileBlob < Test::Unit::TestCase
+  def test_extensions
+    assert_equal [".gitignore"], Linguist::FileBlob.new(".gitignore").extensions
+    assert_equal [".xml"],  Linguist::FileBlob.new("build.xml").extensions
+    assert_equal [".html.erb", ".erb"],  Linguist::FileBlob.new("dotted.dir/index.html.erb").extensions
+  end
+end

test/test_heuristics.rb

@@ -1,9 +1,4 @@
-require 'linguist/heuristics'
-require 'linguist/language'
-require 'linguist/samples'
-require 'linguist/file_blob'
-
-require 'test/unit'
+require_relative "./helper"
 
 class TestHeuristcs < Test::Unit::TestCase
   include Linguist
@@ -132,4 +127,14 @@ class TestHeuristcs < Test::Unit::TestCase
     results = Heuristics.disambiguate_sc(fixture("Scala/node11.sc"))
     assert_equal Language["Scala"], results.first
   end
+
+  def test_fs_by_heuristics
+    languages = ["F#", "Forth", "GLSL"]
+    languages.each do |language|
+      all_fixtures(language).each do |fixture|
+        results = Heuristics.disambiguate_fs(fixture("#{language}/#{File.basename(fixture)}"))
+        assert_equal Language[language], results.first
+      end
+    end
+  end
 end

test/test_language.rb

@@ -1,6 +1,4 @@
-require 'linguist/language'
-require 'test/unit'
-require 'yaml'
+require_relative "./helper"
 
 class TestLanguage < Test::Unit::TestCase
   include Linguist

test/test_md5.rb

@@ -1,6 +1,4 @@
-require 'linguist/md5'
-
-require 'test/unit'
+require_relative "./helper"
 
 class TestMD5 < Test::Unit::TestCase
   include Linguist

test/test_pedantic.rb

@@ -1,5 +1,4 @@
-require 'test/unit'
-require 'yaml'
+require_relative "./helper"
 
 class TestPedantic < Test::Unit::TestCase
   filename = File.expand_path("../../lib/linguist/languages.yml", __FILE__)

test/test_repository.rb

@@ -1,6 +1,4 @@
-require 'linguist/repository'
-require 'linguist/lazy_blob'
-require 'test/unit'
+require_relative "./helper"
 
 class TestRepository < Test::Unit::TestCase
   def rugged_repository

test/test_samples.rb

@@ -1,8 +1,5 @@
-require 'linguist/samples'
-require 'linguist/language'
-require 'tempfile'
-require 'yajl'
-require 'test/unit'
+require_relative "./helper"
+require "tempfile"
 
 class TestSamples < Test::Unit::TestCase
   include Linguist
@@ -34,23 +31,29 @@ class TestSamples < Test::Unit::TestCase
     assert_equal data['languages_total'], data['languages'].inject(0) { |n, (_, c)| n += c }
     assert_equal data['tokens_total'], data['language_tokens'].inject(0) { |n, (_, c)| n += c }
     assert_equal data['tokens_total'], data['tokens'].inject(0) { |n, (_, ts)| n += ts.inject(0) { |m, (_, c)| m += c } }
+    assert !data["interpreters"].empty?
   end
 
-  # Check that there aren't samples with extensions that aren't explicitly defined in languages.yml
-  def test_parity
-    extensions = Samples.cache['extnames']
-    languages_yml = File.expand_path("../../lib/linguist/languages.yml", __FILE__)
-    languages = YAML.load_file(languages_yml)
-
-    languages.each do |name, options|
+  # Check that there aren't samples with extensions or interpreters that
+  # aren't explicitly defined in languages.yml
+  languages_yml = File.expand_path("../../lib/linguist/languages.yml", __FILE__)
+  YAML.load_file(languages_yml).each do |name, options|
+    define_method "test_samples_have_parity_with_languages_yml_for_#{name}" do
       options['extensions'] ||= []
-
-      if extnames = extensions[name]
+      if extnames = Samples.cache['extnames'][name]
         extnames.each do |extname|
           next if extname == '.script!'
           assert options['extensions'].include?(extname), "#{name} has a sample with extension (#{extname}) that isn't explicitly defined in languages.yml"
         end
       end
+
+      options['interpreters'] ||= []
+      if interpreters = Samples.cache['interpreters'][name]
+        interpreters.each do |interpreter|
+          # next if extname == '.script!'
+          assert options['interpreters'].include?(interpreter), "#{name} has a sample with an interpreter (#{interpreter}) that isn't explicitly defined in languages.yml"
+        end
+      end
     end
   end
 
@@ -79,4 +82,9 @@ class TestSamples < Test::Unit::TestCase
       end
     end
   end
+
+  def test_shebang
+    assert_equal "crystal", Linguist.interpreter_from_shebang("#!/usr/bin/env bin/crystal")
+    assert_equal "python2", Linguist.interpreter_from_shebang("#!/usr/bin/python2.4")
+  end
 end

test/test_tokenizer.rb

@@ -1,6 +1,4 @@
-require 'linguist/tokenizer'
-
-require 'test/unit'
+require_relative "./helper"
 
 class TestTokenizer < Test::Unit::TestCase
   include Linguist