Merge pull request #1555 from github/vendored-ignored-gitattributes

Vendored ignored gitattributes

Rakefile

@@ -2,7 +2,6 @@ require 'json'
 require 'rake/clean'
 require 'rake/testtask'
 require 'yaml'
-require 'pry'
 
 task :default => :test
 

lib/linguist/lazy_blob.rb

@@ -4,7 +4,7 @@ require 'rugged'
 
 module Linguist
   class LazyBlob
-    GIT_ATTR = ['linguist-ignore', 'linguist-lang']
+    GIT_ATTR = ['linguist-language', 'linguist-vendored', 'linguist-generated']
     GIT_ATTR_OPTS = { :priority => [:index], :skip_system => true }
     GIT_ATTR_FLAGS = Rugged::Repository::Attributes.parse_opts(GIT_ATTR_OPTS)
 
@@ -29,18 +29,30 @@ module Linguist
         name, GIT_ATTR, GIT_ATTR_FLAGS)
     end
 
-    def ignored?
-      !!git_attributes['linguist-ignore']
+    def vendored?
+      if attr = git_attributes['linguist-vendored']
+        return boolean_attribute(attr)
+      else
+        return super
+      end
     end
 
-    def overriden_language
-      if lang = git_attributes['linguist-lang']
-        Language.find_by_name(lang)
+    def generated?
+      if attr = git_attributes['linguist-generated']
+        return boolean_attribute(attr)
+      else
+        return super
       end
     end
 
     def language
-      @language ||= (overriden_language || Language.detect(self))
+      return @language if defined?(@language)
+
+      @language = if lang = git_attributes['linguist-language']
+        Language.find_by_name(lang)
+      else
+        super
+      end
     end
 
     def data
@@ -54,6 +66,12 @@ module Linguist
     end
 
     protected
+
+    # Returns true if the attribute is present and not the string "false".
+    def boolean_attribute(attr)
+      attr != "false"
+    end
+
     def load_blob!
       @data, @size = Rugged::Blob.to_buffer(repository, oid, MAX_SIZE) if @data.nil?
     end

lib/linguist/repository.rb

@@ -110,22 +110,30 @@ module Linguist
         if @old_commit_oid == @commit_oid
           @old_stats
         else
-          compute_stats(@old_commit_oid, @commit_oid, @old_stats)
+          compute_stats(@old_commit_oid, @old_stats)
         end
       end
     end
 
+    def read_index
+      attr_index = Rugged::Index.new
+      attr_index.read_tree(current_tree)
+      repository.index = attr_index
+    end
+
+    def current_tree
+      @tree ||= Rugged::Commit.lookup(repository, @commit_oid).tree
+    end
+
     protected
-    def compute_stats(old_commit_oid, commit_oid, cache = nil)
+
+    def compute_stats(old_commit_oid, cache = nil)
       file_map = cache ? cache.dup : {}
       old_tree = old_commit_oid && Rugged::Commit.lookup(repository, old_commit_oid).tree
-      new_tree = Rugged::Commit.lookup(repository, commit_oid).tree
 
-      diff = Rugged::Tree.diff(repository, old_tree, new_tree)
+      read_index
 
-      attr_index = Rugged::Index.new
-      attr_index.read_tree(new_tree)
-      repository.index = attr_index
+      diff = Rugged::Tree.diff(repository, old_tree, current_tree)
 
       diff.each_delta do |delta|
         old = delta.old_file[:path]
@@ -142,7 +150,7 @@ module Linguist
           blob = Linguist::LazyBlob.new(repository, delta.new_file[:oid], new, mode.to_s(8))
 
           # Skip vendored or generated blobs
-          next if blob.ignored? || blob.vendored? || blob.generated? || blob.language.nil?
+          next if blob.vendored? || blob.generated? || blob.language.nil?
 
           # Only include programming languages and acceptable markup languages
           if blob.language.type == :programming || Language.detectable_markup.include?(blob.language.name)

test/fixtures/foo.rb

@@ -0,0 +1,3 @@
+def foo
+  return "BAR"
+end

test/test_repository.rb

@@ -1,4 +1,5 @@
 require 'linguist/repository'
+require 'linguist/lazy_blob'
 require 'test/unit'
 
 class TestRepository < Test::Unit::TestCase
@@ -47,13 +48,17 @@ class TestRepository < Test::Unit::TestCase
     assert_equal linguist_repo.cache, new_repo.cache
   end
 
-  def test_git_attributes
-    # See https://github.com/github/linguist/blob/525304738ebdb7ab3b7d2bf9a7514cc428faa273/.gitattributes
+  def test_repo_git_attributes
+    # See https://github.com/github/linguist/blob/7ee006cbcb2d7261f9e648510a684ee9ac64126b/.gitattributes
     #
     # It looks like this:
-    # test/*.rb linguist-ignore
-    # lib/linguist.rb linguist-lang=Java
-    attr_commit = '525304738ebdb7ab3b7d2bf9a7514cc428faa273'
+    # 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 = '7ee006cbcb2d7261f9e648510a684ee9ac64126b'
     repo = linguist_repo(attr_commit)
 
     assert repo.breakdown_by_file.has_key?("Java")
@@ -61,8 +66,36 @@ class TestRepository < Test::Unit::TestCase
 
     assert repo.breakdown_by_file.has_key?("Ruby")
     assert !repo.breakdown_by_file["Ruby"].empty?
-    repo.breakdown_by_file["Ruby"].each do |file|
-      assert !file.start_with?("test/")
-    end
+  end
+
+  def test_linguist_override_generated?
+    attr_commit = '7ee006cbcb2d7261f9e648510a684ee9ac64126b'
+    linguist_repo(attr_commit).read_index
+
+    file = Linguist::LazyBlob.new(rugged_repository, attr_commit, 'Rakefile')
+
+    # overridden in .gitattributes
+    assert file.generated?
+  end
+
+  def test_linguist_override_vendored?
+    attr_commit = '7ee006cbcb2d7261f9e648510a684ee9ac64126b'
+    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 = '7ee006cbcb2d7261f9e648510a684ee9ac64126b'
+    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