Merge branch 'master' into 1036-local

Conflicts:
	lib/linguist/heuristics.rb
	lib/linguist/samples.json

test/test_blob.rb

@@ -262,6 +262,10 @@ class TestBlob < Test::Unit::TestCase
 
 
     assert Linguist::Generated.generated?("node_modules/grunt/lib/grunt.js", nil)
+
+    # Godep saved dependencies
+    assert blob("Godeps/Godeps.json").generated?
+    assert blob("Godeps/_workspace/src/github.com/kr/s3/sign.go").generated?
   end
 
   def test_vendored
@@ -279,6 +283,10 @@ class TestBlob < Test::Unit::TestCase
     assert blob("app/bower_components/custom/custom.js").vendored?
     assert blob("vendor/assets/bower_components/custom/custom.js").vendored?
 
+    # Go dependencies
+    assert !blob("Godeps/Godeps.json").vendored?
+    assert blob("Godeps/_workspace/src/github.com/kr/s3/sign.go").vendored?
+
     # Rails vendor/
     assert blob("vendor/plugins/will_paginate/lib/will_paginate.rb").vendored?
 
@@ -437,6 +445,12 @@ class TestBlob < Test::Unit::TestCase
     assert blob("octicons.css").vendored?
     assert blob("public/octicons.min.css").vendored?
     assert blob("public/octicons/sprockets-octicons.scss").vendored?
+
+    # Typesafe Activator
+    assert blob("activator").vendored?
+    assert blob("activator.bat").vendored?
+    assert blob("subproject/activator").vendored?
+    assert blob("subproject/activator.bat").vendored?
   end
 
   def test_language

test/test_classifier.rb

@@ -44,12 +44,12 @@ class TestClassifier < Test::Unit::TestCase
   end
 
   def test_instance_classify_empty
-    results = Classifier.classify(Samples::DATA, "")
+    results = Classifier.classify(Samples.cache, "")
     assert results.first[1] < 0.5, results.first.inspect
   end
 
   def test_instance_classify_nil
-    assert_equal [], Classifier.classify(Samples::DATA, nil)
+    assert_equal [], Classifier.classify(Samples.cache, nil)
   end
 
   def test_classify_ambiguous_languages
@@ -58,7 +58,7 @@ class TestClassifier < Test::Unit::TestCase
       languages = Language.find_by_filename(sample[:path]).map(&:name)
       next unless languages.length > 1
 
-      results = Classifier.classify(Samples::DATA, File.read(sample[:path]), languages)
+      results = Classifier.classify(Samples.cache, File.read(sample[:path]), languages)
       assert_equal language.name, results.first[0], "#{sample[:path]}\n#{results.inspect}"
     end
   end

test/test_heuristics.rb

@@ -74,6 +74,18 @@ class TestHeuristcs < Test::Unit::TestCase
     assert_equal Language["ECL"], results.first
   end
 
+  def test_pro_prolog_by_heuristics
+    languages = ["IDL", "Prolog"]
+    results = Heuristics.disambiguate_pro(fixture("Prolog/logic-problem.pro"), languages)
+    assert_equal Language["Prolog"], results.first
+  end
+
+  def test_pro_idl_by_heuristics
+    languages = ["IDL", "Prolog"]
+    results = Heuristics.disambiguate_pro(fixture("IDL/mg_acosh.pro"), languages)
+    assert_equal Language["IDL"], results.first
+  end
+
   def test_ts_typescript_by_heuristics
     languages = ["TypeScript", "XML"]
     results = Heuristics.disambiguate_ts(fixture("TypeScript/classes.ts"), languages)

test/test_language.rb

@@ -33,6 +33,7 @@ class TestLanguage < Test::Unit::TestCase
     assert_equal Lexer['Java'], Language['ChucK'].lexer
     assert_equal Lexer['Java'], Language['Java'].lexer
     assert_equal Lexer['JavaScript'], Language['JavaScript'].lexer
+    assert_equal Lexer['LSL'], Language['LSL'].lexer
     assert_equal Lexer['MOOCode'], Language['Moocode'].lexer
     assert_equal Lexer['MuPAD'], Language['mupad'].lexer
     assert_equal Lexer['NASM'], Language['Assembly'].lexer
@@ -186,6 +187,7 @@ class TestLanguage < Test::Unit::TestCase
 
   def test_programming
     assert_equal :programming, Language['JavaScript'].type
+    assert_equal :programming, Language['LSL'].type
     assert_equal :programming, Language['Perl'].type
     assert_equal :programming, Language['PowerShell'].type
     assert_equal :programming, Language['Python'].type
@@ -326,6 +328,7 @@ class TestLanguage < Test::Unit::TestCase
     assert_equal '#3581ba', Language['Python'].color
     assert_equal '#f1e05a', Language['JavaScript'].color
     assert_equal '#31859c', Language['TypeScript'].color
+    assert_equal '#3d9970', Language['LSL'].color
   end
 
   def test_colors
@@ -338,6 +341,7 @@ class TestLanguage < Test::Unit::TestCase
     assert_equal 'coffee', Language['CoffeeScript'].ace_mode
     assert_equal 'csharp', Language['C#'].ace_mode
     assert_equal 'css', Language['CSS'].ace_mode
+    assert_equal 'lsl', Language['LSL'].ace_mode
     assert_equal 'javascript', Language['JavaScript'].ace_mode
   end
 
@@ -352,6 +356,7 @@ class TestLanguage < Test::Unit::TestCase
   end
 
   def test_extensions
+    assert Language['LSL'].extensions.include?('.lsl')
     assert Language['Perl'].extensions.include?('.pl')
     assert Language['Python'].extensions.include?('.py')
     assert Language['Ruby'].extensions.include?('.rb')

test/test_repository.rb

@@ -1,5 +1,5 @@
 require 'linguist/repository'
-
+require 'linguist/lazy_blob'
 require 'test/unit'
 
 class TestRepository < Test::Unit::TestCase
@@ -47,4 +47,58 @@ class TestRepository < Test::Unit::TestCase
 
     assert_equal linguist_repo.cache, new_repo.cache
   end
+
+  def test_repo_git_attributes
+    # See https://github.com/github/linguist/blob/351c1cc8fd57340839bdb400d7812332af80e9bd/.gitattributes
+    #
+    # It looks like this:
+    # Gemfile linguist-vendored=true
+    # lib/linguist.rb linguist-language=Java
+    # test/*.rb linguist-language=Java
+    # Rakefile linguist-generated
+    # test/fixtures/* linguist-vendored=false
+
+    attr_commit = '351c1cc8fd57340839bdb400d7812332af80e9bd'
+    repo = linguist_repo(attr_commit)
+
+    assert repo.breakdown_by_file.has_key?("Java")
+    assert repo.breakdown_by_file["Java"].include?("lib/linguist.rb")
+
+    assert repo.breakdown_by_file.has_key?("Ruby")
+    assert !repo.breakdown_by_file["Ruby"].empty?
+  end
+
+  def test_commit_with_git_attributes_data
+    # Before we had any .gitattributes data
+    old_commit = '4a017d9033f91b2776eb85275463f9613cc371ef'
+    old_repo = linguist_repo(old_commit)
+
+    # With some .gitattributes data
+    attr_commit = '7ee006cbcb2d7261f9e648510a684ee9ac64126b'
+    # It's incremental but should bust the cache
+    new_repo = Linguist::Repository.incremental(rugged_repository, attr_commit, old_commit, old_repo.cache)
+
+    assert new_repo.breakdown_by_file["Java"].include?("lib/linguist.rb")
+  end
+
+  def test_linguist_override_vendored?
+    attr_commit = '351c1cc8fd57340839bdb400d7812332af80e9bd'
+    repo = linguist_repo(attr_commit).read_index
+
+    override_vendored = Linguist::LazyBlob.new(rugged_repository, attr_commit, 'Gemfile')
+
+    # overridden .gitattributes
+    assert override_vendored.vendored?
+  end
+
+  def test_linguist_override_unvendored?
+    attr_commit = '351c1cc8fd57340839bdb400d7812332af80e9bd'
+    repo = linguist_repo(attr_commit).read_index
+
+    # lib/linguist/vendor.yml defines this as vendored.
+    override_unvendored = Linguist::LazyBlob.new(rugged_repository, attr_commit, 'test/fixtures/foo.rb')
+
+    # overridden .gitattributes
+    assert !override_unvendored.vendored?
+  end
 end

test/test_samples.rb

@@ -8,7 +8,7 @@ class TestSamples < Test::Unit::TestCase
   include Linguist
 
   def test_up_to_date
-    assert serialized = Samples::DATA
+    assert serialized = Samples.cache
     assert latest = Samples.data
 
     # Just warn, it shouldn't scare people off by breaking the build.
@@ -29,7 +29,7 @@ class TestSamples < Test::Unit::TestCase
   end
 
   def test_verify
-    assert data = Samples::DATA
+    assert data = Samples.cache
 
     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 }
@@ -38,7 +38,7 @@ class TestSamples < Test::Unit::TestCase
 
   # Check that there aren't samples with extensions that aren't explicitly defined in languages.yml
   def test_parity
-    extensions = Samples::DATA['extnames']
+    extensions = Samples.cache['extnames']
     languages_yml = File.expand_path("../../lib/linguist/languages.yml", __FILE__)
     languages = YAML.load_file(languages_yml)