Split mime type and content type

lib/linguist/blob_helper.rb

@@ -11,18 +11,15 @@ module Linguist
     end
 
     def mime_type
-      @mime_type ||= begin
+      @mime_type ||= pathname.mime_type
-        guesses = MIME::Types.type_for(pathname.extname)
-        guesses.first ? guesses.first.simplified : 'text/plain'
-      end
     end
 
-    def special_mime_type
+    def content_type
-      Mime.lookup(pathname.extname)
+      pathname.content_type
     end
 
     def disposition
-      case special_mime_type
+      case content_type
       when 'application/octet-stream', 'application/java-archive'
         "attachment; filename=#{EscapeUtils.escape_url(pathname.basename)}"
       else
@@ -43,7 +40,7 @@ module Linguist
     end
 
     def binary?
-      special_mime_type == 'octet-stream' || !(text? || image?)
+      content_type.include?('octet') || !(text? || image?)
     end
 
     def file?
@@ -51,8 +48,7 @@ module Linguist
     end
 
     def text?
-      pathname.media_type == 'text' ||
+      content_type[/(text|json)/]
-        pathname.mime_type == 'application/json'
     end
 
     def image?

lib/linguist/content_types.yml

@@ -12,9 +12,6 @@ air: application/octet-stream
 blend: application/octet-stream
 crx: application/octet-stream
 deb: application/octet-stream
-dll: application/octet-stream
-dmg: application/octet-stream
-exe: application/octet-stream
 gem: application/octet-stream
 graffle: application/octet-stream
 gz: application/octet-stream

lib/linguist/mime.rb

@@ -1,5 +1,11 @@
 require 'mime/types'
 
+# Register additional binary extensions
+binary = MIME::Types['application/octet-stream'].first
+binary.extensions << 'dmg'
+binary.extensions << 'dll'
+MIME::Types.index_extensions(binary)
+
 # Register 'ear' and 'war' as java
 java = MIME::Types['application/java-archive'].first
 java.extensions << 'ear'
@@ -8,20 +14,19 @@ MIME::Types.index_extensions(java)
 
 module Linguist
   module Mime
-    Special = YAML.load_file(File.expand_path("../special_mime_types.yml", __FILE__))
+    Special = YAML.load_file(File.expand_path("../content_types.yml", __FILE__))
 
-    def self.lookup(ext)
+    def self.mime_for(ext)
       ext ||= ''
-
       guesses = ::MIME::Types.type_for(ext)
-      orginal_type = guesses.first ? guesses.first.simplified : 'text/plain'
+      guesses.first ? guesses.first.simplified : 'text/plain'
-
+    end
-      type = Special[orginal_type] ||
-        Special[ext.sub(/^\./, '')] ||
-        orginal_type
 
+    def self.content_type_for(ext)
+      ext ||= ''
+      type = mime_for(ext)
+      type = Special[type] || Special[ext.sub(/^\./, '')] || type
       type += '; charset=utf-8' if type =~ /^text\//
-
       type
     end
   end

lib/linguist/pathname.rb

@@ -34,11 +34,19 @@ module Linguist
     end
 
     def mime_type
-      @mime_type ||= Mime.lookup(extname)
+      @mime_type ||= Mime.mime_for(extname)
     end
 
     def media_type
-      mime_type.split('/').first
+      mime_type.split('/')[0]
+    end
+
+    def sub_type
+      mime_type.split('/')[1]
+    end
+
+    def content_type
+      @content_type ||= Mime.content_type_for(extname)
     end
 
     def to_s

test/test_blob.rb

@@ -41,12 +41,14 @@ class TestBlob < Test::Unit::TestCase
     assert_equal "text/plain", blob("grit.rb").mime_type
     assert_equal "application/xml", blob("bar.xml").mime_type
     assert_equal "text/plain", blob("dog.o").mime_type
+    assert_equal "application/sh", blob("script.sh").mime_type
   end
 
-  def test_special_mime_type
+  def test_content_type
-    assert_equal "text/plain; charset=utf-8", blob("grit.rb").special_mime_type
+    assert_equal "text/plain; charset=utf-8", blob("grit.rb").content_type
-    assert_equal "text/plain; charset=utf-8", blob("bar.xml").special_mime_type
+    assert_equal "text/plain; charset=utf-8", blob("bar.xml").content_type
-    assert_equal "application/octet-stream", blob("dog.o").special_mime_type
+    assert_equal "application/octet-stream", blob("dog.o").content_type
+    assert_equal "text/plain; charset=utf-8", blob("script.sh").content_type
   end
 
   def test_disposition
@@ -93,6 +95,7 @@ class TestBlob < Test::Unit::TestCase
   def test_text
     assert blob("file.txt").text?
     assert blob("file.json").text?
+    assert blob("script.sh").text?
   end
 
   def test_image

test/test_mime.rb

@@ -5,22 +5,45 @@ require 'test/unit'
 class TestMime < Test::Unit::TestCase
   include Linguist
 
-  def test_lookup
+  def test_mime
-    assert_equal 'text/plain; charset=utf-8', Mime.lookup(nil)
+    assert_equal 'text/plain', Mime.mime_for(nil)
 
-    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".rb")
+    assert_equal 'text/plain', Mime.mime_for(".rb")
-    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".js")
+    assert_equal 'text/plain', Mime.mime_for(".js")
-    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".py")
+    assert_equal 'text/plain', Mime.mime_for(".py")
 
-    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".kt")
+    assert_equal 'text/plain', Mime.mime_for(".kt")
-    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".html")
+    assert_equal 'text/html', Mime.mime_for(".html")
-    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".sh")
+    assert_equal 'application/sh', Mime.mime_for(".sh")
-    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".latex")
+    assert_equal 'application/latex', Mime.mime_for(".latex")
 
-    assert_equal 'application/octet-stream', Mime.lookup(".dmg")
+    assert_equal 'application/octet-stream', Mime.mime_for(".dmg")
+    assert_equal 'application/octet-stream', Mime.mime_for(".exe")
+    assert_equal 'application/octet-stream', Mime.mime_for(".dll")
 
-    assert_equal 'application/java-archive', Mime.lookup(".jar")
+    assert_equal 'application/java-archive', Mime.mime_for(".jar")
-    assert_equal 'application/java-archive', Mime.lookup(".ear")
+    assert_equal 'application/java-archive', Mime.mime_for(".ear")
-    assert_equal 'application/java-archive', Mime.lookup(".war")
+    assert_equal 'application/java-archive', Mime.mime_for(".war")
+  end
+
+  def test_content_type
+    assert_equal 'text/plain; charset=utf-8', Mime.content_type_for(nil)
+
+    assert_equal 'text/plain; charset=utf-8', Mime.content_type_for(".rb")
+    assert_equal 'text/plain; charset=utf-8', Mime.content_type_for(".js")
+    assert_equal 'text/plain; charset=utf-8', Mime.content_type_for(".py")
+
+    assert_equal 'text/plain; charset=utf-8', Mime.content_type_for(".kt")
+    assert_equal 'text/plain; charset=utf-8', Mime.content_type_for(".html")
+    assert_equal 'text/plain; charset=utf-8', Mime.content_type_for(".sh")
+    assert_equal 'text/plain; charset=utf-8', Mime.content_type_for(".latex")
+
+    assert_equal 'application/octet-stream', Mime.content_type_for(".dmg")
+    assert_equal 'application/octet-stream', Mime.content_type_for(".exe")
+    assert_equal 'application/octet-stream', Mime.content_type_for(".dll")
+
+    assert_equal 'application/java-archive', Mime.content_type_for(".jar")
+    assert_equal 'application/java-archive', Mime.content_type_for(".ear")
+    assert_equal 'application/java-archive', Mime.content_type_for(".war")
   end
 end

test/test_pathname.rb

@@ -61,10 +61,16 @@ class TestPathname < Test::Unit::TestCase
   end
 
   def test_mime_type
-    assert_equal 'text/plain; charset=utf-8', Pathname.new("file.rb").mime_type
+    assert_equal 'text/plain', Pathname.new("file.rb").mime_type
-    assert_equal 'text/plain; charset=utf-8', Pathname.new("file.js").mime_type
+    assert_equal 'text/plain', Pathname.new("file.js").mime_type
-    assert_equal 'text/plain; charset=utf-8', Pathname.new("itty.py").mime_type
+    assert_equal 'text/plain', Pathname.new("itty.py").mime_type
-    assert_equal 'text/plain; charset=utf-8', Pathname.new("defun.kt").mime_type
+    assert_equal 'text/plain', Pathname.new("defun.kt").mime_type
+  end
+
+  def test_content_type
+    assert_equal 'text/plain; charset=utf-8', Pathname.new("file.txt").content_type
+    assert_equal 'text/plain; charset=utf-8', Pathname.new("file.rb").content_type
+    assert_equal 'image/png', Pathname.new("octocat.png").content_type
   end
 
   def test_media_type