Using github's linguist to better detect filetypes and to do syntax highlighting.

Gemfile

@@ -18,6 +18,7 @@ gem "six"
 gem "therubyracer"
 gem "faker"
 gem "seed-fu"
+gem "linguist", "~> 1.0.0", :git => "https://github.com/github/linguist.git"
 gem "pygments.rb", "0.2.7"
 gem "thin"
 gem "unicorn"

Gemfile.lock

@@ -4,6 +4,16 @@ GIT
   specs:
     annotate (2.4.1.beta1)
 
+GIT
+  remote: https://github.com/github/linguist.git
+  revision: c444c25b27131cd2aca6017f2a9ce5eae60dfcd3
+  specs:
+    linguist (1.0.0)
+      charlock_holmes (~> 0.6.6)
+      escape_utils (~> 0.2.3)
+      mime-types (~> 1.18)
+      pygments.rb (~> 0.2.11)
+
 GIT
   remote: https://github.com/gitlabhq/gitolite-client.git
   revision: 36dabd226caa40ff052677719adaacbfe667b36c
@@ -104,6 +114,7 @@ GEM
     diff-lcs (1.1.3)
     drapper (0.8.4)
     erubis (2.7.0)
+    escape_utils (0.2.4)
     eventmachine (0.12.10)
     execjs (1.3.0)
       multi_json (~> 1.0)
@@ -326,6 +337,7 @@ DEPENDENCIES
   kaminari
   launchy
   letter_opener
+  linguist (~> 1.0.0)!
   modularity
   mysql2
   omniauth-ldap

app/views/refs/_tree_file.html.haml

@@ -11,10 +11,10 @@
   - if file.text?
     .view_file_content
       - unless file.empty?
-        %div{:class => current_user.dark_scheme ? "black" : "white"} 
-          = preserve do 
-            = raw file.colorize
-      - else 
+        %div{:class => current_user.dark_scheme ? "black" : "white"}
+          = preserve do
+            = raw file.colorize(options: { linenos: 'True'})
+      - else
         %h3
           %center Empty file
   - elsif file.image?

app/views/snippets/show.html.haml

@@ -13,8 +13,7 @@
       .view_file_header
         %strong= @snippet.file_name
       .view_file_content
-        %div{:class => current_user.dark_scheme ? "black" : ""} 
-          :erb
-            <%= raw @snippet.colorize %>
+        %div{:class => current_user.dark_scheme ? "black" : ""}
+          = raw @snippet.colorize(options: { linenos: 'True'})
 
 = render "notes/notes", :tid => @snippet.id, :tt => "snippet"

config/initializers/gitlabhq/20_grit_ext.rb

@@ -1,10 +1,7 @@
 require 'grit'
-require 'pygments'
-require "utils"
 
 Grit::Blob.class_eval do
-  include Utils::FileHelper
-  include Utils::Colorize
+  include Linguist::BlobHelper
 end
 
 #monkey patch raw_object from string
@@ -15,7 +12,7 @@ Grit::GitRuby::Internal::RawObject.class_eval do
 end
 
 
-Grit::Diff.class_eval do 
+Grit::Diff.class_eval do
   def old_path
     Gitlabhq::Encode.utf8 a_path
   end

lib/utils.rb

@@ -1,44 +0,0 @@
-module Utils
-  module FileHelper
-    def binary?(string)
-      string.each_byte do |x|
-        x.nonzero? or return true
-      end
-      false
-    end
-
-    def image?
-      mime_type =~ /image/
-    end
-
-    def text?
-      mime_type =~ /application|text/ && !binary?(data)
-    end
-  end
-
-  module Colorize
-    def colorize
-      system_colorize(data, name)
-    end
-
-    def system_colorize(data, file_name)
-      options = { :encoding => 'utf-8', :linenos => 'True' }
-
-      # Try detect language with pygments
-      Pygments.highlight data, :filename => file_name, :options => options
-    rescue 
-      # if it fails use manual detection
-      ft = handle_file_type(file_name)
-      Pygments.highlight(data, :lexer => ft, :options => options)
-    end
-
-    def handle_file_type(file_name)
-      case file_name
-      when /(\.ru|Gemfile)$/
-        :ruby
-      else
-        :text
-      end
-    end
-  end
-end