octocat
diff --git a/bin/linguist b/bin/linguist
@@ -23,7 +23,6 @@ elsif File.file?(path)
 
   puts "#{blob.name}: #{blob.loc} lines (#{blob.sloc} sloc)"
   puts "  type:      #{type}"
-  puts "  extension: #{blob.pathname.extname}"
   puts "  mime type: #{blob.mime_type}"
   puts "  language:  #{blob.language}"
 

diff --git a/github-linguist.gemspec b/github-linguist.gemspec
@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name    = 'github-linguist'
-  s.version = '2.2.0'
+  s.version = '2.3.3'
   s.summary = "GitHub Language detection"
 
   s.authors = "GitHub"
@@ -10,8 +10,9 @@ Gem::Specification.new do |s|
 
   s.add_dependency 'charlock_holmes', '~> 0.6.6'
   s.add_dependency 'escape_utils',    '~> 0.2.3'
-  s.add_dependency 'mime-types',      '~> 1.18'
+  s.add_dependency 'mime-types',      '~> 1.19'
   s.add_dependency 'pygments.rb',     '>= 0.2.13'
+  s.add_development_dependency 'mocha'
   s.add_development_dependency 'json'
   s.add_development_dependency 'rake'
   s.add_development_dependency 'yajl-ruby'

diff --git a/lib/linguist.rb b/lib/linguist.rb
@@ -1,6 +1,5 @@
 require 'linguist/blob_helper'
 require 'linguist/generated'
 require 'linguist/language'
-require 'linguist/mime'
 require 'linguist/repository'
 require 'linguist/samples'
diff --git a/lib/linguist/blob_helper.rb b/lib/linguist/blob_helper.rb
@@ -1,9 +1,9 @@
 require 'linguist/generated'
 require 'linguist/language'
-require 'linguist/mime'
 
 require 'charlock_holmes'
 require 'escape_utils'
+require 'mime/types'
 require 'pygments'
 require 'yaml'
 
@@ -23,6 +23,22 @@ def extname
       File.extname(name.to_s)
     end
 
+    # Internal: Lookup mime type for extension.
+    #
+    # Returns a MIME::Type
+    def _mime_type
+      if defined? @_mime_type
+        @_mime_type
+      else
+        guesses = ::MIME::Types.type_for(extname.to_s)
+
+        # Prefer text mime types over binary
+        @_mime_type = guesses.detect { |type| type.ascii? } ||
+          # Otherwise use the first guess
+          guesses.first
+      end
+    end
+
     # Public: Get the actual blob mime type
     #
     # Examples
@@ -32,7 +48,14 @@ def extname
     #
     # Returns a mime type String.
     def mime_type
-      @mime_type ||= Mime.mime_for(extname.to_s)
+      _mime_type ? _mime_type.to_s : 'text/plain'
+    end
+
+    # Internal: Is the blob binary according to its mime type
+    #
+    # Return true or false
+    def binary_mime_type?
+      _mime_type ? _mime_type.binary? : false
     end
 
     # Public: Get the Content-Type header value
@@ -83,15 +106,6 @@ def detect_encoding
       @detect_encoding ||= CharlockHolmes::EncodingDetector.new.detect(data) if data
     end
 
-    # Public: Is the blob binary according to its mime type
-    #
-    # Return true or false
-    def binary_mime_type?
-      if mime_type = Mime.lookup_mime_type_for(extname)
-        mime_type.binary?
-      end
-    end
-
     # Public: Is the blob binary?
     #
     # Return true or false
@@ -146,7 +160,7 @@ def large?
     #
     # Return true or false
     def safe_to_colorize?
-      text? && !large? && !high_ratio_of_long_lines?
+      !large? && text? && !high_ratio_of_long_lines?
     end
 
     # Internal: Does the blob have a ratio of long lines?
@@ -236,7 +250,9 @@ def generated?
     #
     # Return true or false
     def indexable?
-      if binary?
+      if size > 100 * 1024
+        false
+      elsif binary?
         false
       elsif extname == '.txt'
         true
@@ -246,8 +262,6 @@ def indexable?
         false
       elsif generated?
         false
-      elsif size > 100 * 1024
-        false
       else
         true
       end
@@ -259,11 +273,15 @@ def indexable?
     #
     # Returns a Language or nil if none is detected
     def language
-      if defined? @language
-        @language
-      elsif !binary_mime_type?
-        @language = Language.detect(name.to_s, lambda { data }, mode)
+      return @language if defined? @language
+
+      if defined?(@data) && @data.is_a?(String)
+        data = @data
+      else
+        data = lambda { binary_mime_type? ? "" : self.data }
       end
+
+      @language = Language.detect(name.to_s, data, mode)
     end
 
     # Internal: Get the lexer of the blob.

diff --git a/lib/linguist/generated.rb b/lib/linguist/generated.rb
@@ -37,7 +37,8 @@ def data
     #
     # Returns an Array of lines
     def lines
-      @lines ||= data.split("\n", -1)
+      # TODO: data should be required to be a String, no nils
+      @lines ||= data ? data.split("\n", -1) : []
     end
 
     # Internal: Is the blob a generated file?

diff --git a/lib/linguist/languages.yml b/lib/linguist/languages.yml
@@ -68,9 +68,10 @@ Apex:
   primary_extension: .cls
 
 AppleScript:
+  type: programming
   aliases:
   - osascript
-  primary_extension: .scpt
+  primary_extension: .applescript
 
 Arc:
   type: programming

diff --git a/lib/linguist/mime.rb b/lib/linguist/mime.rb
diff --git a/lib/linguist/mimes.yml b/lib/linguist/mimes.yml