Merge pull request #1517 from github/1515-local

3.2.0b3 gem / cc @vmg

.travis.yml

@@ -1,6 +1,7 @@
 before_install:
   - git fetch origin master:master
   - git fetch origin v2.0.0:v2.0.0
+  - git fetch origin test/attributes:test/attributes
   - sudo apt-get install libicu-dev -y
 rvm:
   - 1.9.3

Gemfile

@@ -1,2 +1,3 @@
 source 'https://rubygems.org'
 gemspec
+gem 'rugged', '0.21.1b2'

Rakefile

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

github-linguist.gemspec

@@ -17,7 +17,6 @@ Gem::Specification.new do |s|
   s.add_dependency 'escape_utils',    '~> 1.0.1'
   s.add_dependency 'mime-types',      '~> 1.19'
   s.add_dependency 'pygments.rb',     '~> 0.6.0'
-  s.add_dependency 'rugged',          '~> 0.21.0'
 
   s.add_development_dependency 'json'
   s.add_development_dependency 'mocha'

lib/linguist/lazy_blob.rb

@@ -1,8 +1,13 @@
 require 'linguist/blob_helper'
+require 'linguist/language'
 require 'rugged'
 
 module Linguist
   class LazyBlob
+    GIT_ATTR = ['linguist-language', 'linguist-vendored']
+    GIT_ATTR_OPTS = { :priority => [:index], :skip_system => true }
+    GIT_ATTR_FLAGS = Rugged::Repository::Attributes.parse_opts(GIT_ATTR_OPTS)
+
     include BlobHelper
 
     MAX_SIZE = 128 * 1024
@@ -19,6 +24,29 @@ module Linguist
       @mode = mode
     end
 
+    def git_attributes
+      @git_attributes ||= repository.fetch_attributes(
+        name, GIT_ATTR, GIT_ATTR_FLAGS)
+    end
+
+    def vendored?
+      if attr = git_attributes['linguist-vendored']
+        return boolean_attribute(attr)
+      else
+        return super
+      end
+    end
+
+    def language
+      return @language if defined?(@language)
+
+      @language = if lang = git_attributes['linguist-language']
+        Language.find_by_name(lang)
+      else
+        super
+      end
+    end
+
     def data
       load_blob!
       @data
@@ -30,6 +58,12 @@ 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

@@ -110,18 +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
+
+      diff = Rugged::Tree.diff(repository, old_tree, current_tree)
 
       diff.each_delta do |delta|
         old = delta.old_file[:path]

lib/linguist/version.rb

@@ -1,3 +1,3 @@
 module Linguist
-  VERSION = "3.1.5"
+  VERSION = "3.2.0"
 end

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 +47,45 @@ 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_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