Merge branch 'master' into 1233-local

Conflicts:
	lib/linguist/language.rb
	lib/linguist/languages.yml
	lib/linguist/samples.json

lib/linguist/blob_helper.rb

@@ -321,6 +321,11 @@ module Linguist
       language ? language.lexer : Pygments::Lexer.find_by_name('Text only')
     end
 
+    # Internal: Get the TextMate compatible scope for the blob
+    def tm_scope
+      language && language.tm_scope
+    end
+
     # Public: Highlight syntax of blob
     #
     # options - A Hash of options (defaults to {})

lib/linguist/generated.rb

@@ -63,6 +63,7 @@ module Linguist
         generated_jni_header? ||
         composer_lock? ||
         node_modules? ||
+        godeps? ||
         vcr_cassette? ||
         generated_by_zephir?
     end
@@ -231,6 +232,14 @@ module Linguist
       !!name.match(/node_modules\//)
     end
 
+    # Internal: Is the blob part of Godeps/,
+    # which are not meant for humans in pull requests.
+    #
+    # Returns true or false.
+    def godeps?
+      !!name.match(/Godeps\//)
+    end
+
     # Internal: Is the blob a generated php composer lock file?
     #
     # Returns true or false.

lib/linguist/heuristics.rb

@@ -19,6 +19,15 @@ module Linguist
         if languages.all? { |l| ["ECL", "Prolog"].include?(l) }
           result = disambiguate_ecl(data, languages)
         end
+        if languages.all? { |l| ["IDL", "Prolog"].include?(l) }
+          result = disambiguate_pro(data, languages)
+        end
+        if languages.all? { |l| ["Common Lisp", "OpenCL"].include?(l) }
+          result = disambiguate_cl(data, languages)
+        end
+        if languages.all? { |l| ["AsciiDoc", "AGS Script"].include?(l) }
+          result = disambiguate_asc(data, languages)
+        end
         return result
       end
     end
@@ -48,6 +57,16 @@ module Linguist
       matches
     end
 
+    def self.disambiguate_pro(data, languages)
+      matches = []
+      if (data.include?(":-"))
+        matches << Language["Prolog"]
+      else
+        matches << Language["IDL"]
+      end
+      matches
+    end
+
     def self.disambiguate_ts(data, languages)
       matches = []
       if (data.include?("</translation>"))
@@ -72,6 +91,12 @@ module Linguist
       matches
     end
 
+    def self.disambiguate_asc(data, languages)
+      matches = []
+      matches << Language["AsciiDoc"] if /^=+(\s|\n)/.match(data)
+      matches
+    end
+
     def self.active?
       !!ACTIVE
     end

lib/linguist/language.rb

@@ -2,7 +2,7 @@ require 'escape_utils'
 require 'pygments'
 require 'yaml'
 begin
-  require 'json'
+  require 'yajl'
 rescue LoadError
 end
 
@@ -150,7 +150,7 @@ module Linguist
           if (determined = Heuristics.find_by_heuristics(data, possible_language_names)) && !determined.empty?
             return determined.first
           # Lastly, fall back to the probabilistic classifier.
-          elsif classified = Classifier.classify(Samples::DATA, data, possible_language_names).first
+          elsif classified = Classifier.classify(Samples.cache, data, possible_language_names).first
             # Return the actual Language object based of the string language name (i.e., first element of `#classify`)
             return Language[classified[0]]
           end
@@ -302,6 +302,16 @@ module Linguist
       @lexer = Pygments::Lexer.find_by_name(attributes[:lexer] || name) ||
         raise(ArgumentError, "#{@name} is missing lexer")
 
+      @tm_scope = attributes[:tm_scope] || begin
+        context = case @type
+                  when :data, :markup, :prose
+                    'text'
+                  when :programming, nil
+                    'source'
+                  end
+        "#{context}.#{@name.downcase}"
+      end
+
       @ace_mode = attributes[:ace_mode]
       @wrap = attributes[:wrap] || false
 
@@ -375,6 +385,11 @@ module Linguist
     # Returns the Lexer
     attr_reader :lexer
 
+    # Public: Get the name of a TextMate-compatible scope
+    #
+    # Returns the scope
+    attr_reader :tm_scope
+
     # Public: Get Ace mode
     #
     # Examples
@@ -522,16 +537,16 @@ module Linguist
     end
   end
 
-  extensions = Samples::DATA['extnames']
-  interpreters = Samples::DATA['interpreters']
-  filenames = Samples::DATA['filenames']
+  extensions = Samples.cache['extnames']
+  interpreters = Samples.cache['interpreters']
+  filenames = Samples.cache['filenames']
   popular = YAML.load_file(File.expand_path("../popular.yml", __FILE__))
 
   languages_yml = File.expand_path("../languages.yml", __FILE__)
   languages_json = File.expand_path("../languages.json", __FILE__)
 
-  if File.exist?(languages_json) && defined?(JSON)
-    languages = JSON.load(File.read(languages_json))
+  if File.exist?(languages_json) && defined?(Yajl)
+    languages = Yajl.load(File.read(languages_json))
   else
     languages = YAML.load_file(languages_yml)
   end
@@ -576,6 +591,7 @@ module Linguist
       :type              => options['type'],
       :aliases           => options['aliases'],
       :lexer             => options['lexer'],
+      :tm_scope          => options['tm_scope'],
       :ace_mode          => options['ace_mode'],
       :wrap              => options['wrap'],
       :group_name        => options['group'],

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 +110,37 @@ 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
 
-    protected
-    def compute_stats(old_commit_oid, 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
+    def read_index
+      attr_index = Rugged::Index.new
+      attr_index.read_tree(current_tree)
+      repository.index = attr_index
+    end
 
-      diff = Rugged::Tree.diff(repository, old_tree, new_tree)
+    def current_tree
+      @tree ||= Rugged::Commit.lookup(repository, @commit_oid).tree
+    end
+
+    protected
+
+    def compute_stats(old_commit_oid, cache = nil)
+      old_tree = old_commit_oid && Rugged::Commit.lookup(repository, old_commit_oid).tree
+
+      read_index
+
+      diff = Rugged::Tree.diff(repository, old_tree, current_tree)
+
+      # Clear file map and fetch full diff if any .gitattributes files are changed
+      if cache && diff.each_delta.any? { |delta| File.basename(delta.new_file[:path]) == ".gitattributes" }
+        diff = Rugged::Tree.diff(repository, old_tree = nil, current_tree)
+        file_map = {}
+      else
+        file_map = cache ? cache.dup : {}
+      end
 
       diff.each_delta do |delta|
         old = delta.old_file[:path]

lib/linguist/samples.rb

@@ -1,5 +1,5 @@
 begin
-  require 'json'
+  require 'yajl'
 rescue LoadError
   require 'yaml'
 end
@@ -17,9 +17,11 @@ module Linguist
     PATH = File.expand_path('../samples.json', __FILE__)
 
     # Hash of serialized samples object
-    if File.exist?(PATH)
-      serializer = defined?(JSON) ? JSON : YAML
-      DATA = serializer.load(File.read(PATH))
+    def self.cache
+      @cache ||= begin
+        serializer = defined?(Yajl) ? Yajl : YAML
+        serializer.load(File.read(PATH))
+      end
     end
 
     # Public: Iterate over each sample.

lib/linguist/vendor.yml

@@ -33,6 +33,9 @@
 # Erlang bundles
 - ^rebar$
 
+# Go dependencies
+- Godeps/_workspace/
+
 # Bootstrap minified css and js
 - (^|/)bootstrap([^.]*)(\.min)?\.(js|css)$
 
@@ -235,3 +238,11 @@
 - octicons.css
 - octicons.min.css
 - sprockets-octicons.scss
+
+# Typesafe Activator
+- (^|/)activator$
+- (^|/)activator\.bat$
+
+# ProGuard
+- proguard.pro
+- proguard-rules.pro

lib/linguist/version.rb

@@ -1,3 +1,3 @@
 module Linguist
-  VERSION = "3.1.5"
+  VERSION = "3.4.1"
 end