Add special mime type overrides

lib/linguist/mime.rb

@@ -0,0 +1,22 @@
+require 'mime/types'
+
+module Linguist
+  module Mime
+    Special = YAML.load_file(File.expand_path("../special_mime_types.yml", __FILE__))
+
+    def self.lookup(ext)
+      ext ||= ''
+
+      guesses = ::MIME::Types.type_for(ext)
+      orginal_type = guesses.first ? guesses.first.simplified : 'text/plain'
+
+      type = Special[orginal_type] ||
+        Special[ext.sub(/^\./, '')] ||
+        orginal_type
+
+      type += '; charset=utf-8' if type =~ /^text\//
+
+      type
+    end
+  end
+end

lib/linguist/pathname.rb

@@ -1,5 +1,5 @@
 require 'linguist/language'
-require 'mime/types'
+require 'linguist/mime'
 
 module Linguist
   class Pathname
@@ -30,10 +30,7 @@ module Linguist
     end
 
     def mime_type
-      @mime_type ||= begin
+      @mime_type ||= Mime.lookup(extname)
-        guesses = MIME::Types.type_for(extname)
-        guesses.first ? guesses.first.simplified : 'text/plain'
-      end
     end
 
     def to_s

lib/linguist/special_mime_types.yml

@@ -0,0 +1,63 @@
+# mime types
+text/html: text/plain
+application/sh: text/plain
+application/rdf+xml: text/plain
+application/xml: text/plain
+application/xhtml+xml: text/plain
+application/tex: text/plain
+
+# binary files
+a: application/octet-stream
+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
+icns: application/octet-stream
+ipa: application/octet-stream
+lib: application/octet-stream
+mcz: application/octet-stream
+mov: application/octet-stream
+mp3: application/octet-stream
+nib: application/octet-stream
+o: application/octet-stream
+odp: application/octet-stream
+ods: application/octet-stream
+odt: application/octet-stream
+ogg: application/octet-stream
+ogv: application/octet-stream
+otf: application/octet-stream
+pfx: application/octet-stream
+pigx: application/octet-stream
+plgx: application/octet-stream
+pptx: application/octet-stream
+psd: application/octet-stream
+sib: application/octet-stream
+so: application/octet-stream
+spl: application/octet-stream
+sqlite3: application/octet-stream
+swc: application/octet-stream
+swf: application/octet-stream
+tar: application/octet-stream
+ucode: application/octet-stream
+xpi: application/octet-stream
+zip: application/octet-stream
+
+# plaintext
+latex: text/plain
+ms: text/plain
+nc: text/plain
+nim: text/plain
+ps: text/plain
+sc: text/plain
+sh: text/plain
+src: text/plain
+tcl: text/plain
+texi: text/plain
+texinfo: text/plain
+xul: text/plain

test/test_mime.rb

@@ -0,0 +1,20 @@
+require 'linguist/mime'
+
+require 'test/unit'
+
+class TestMime < Test::Unit::TestCase
+  include Linguist
+
+  def test_lookup
+    assert_equal 'application/ruby', Mime.lookup(".rb")
+    assert_equal 'application/javascript', Mime.lookup(".js")
+    assert_equal 'application/python', Mime.lookup(".py")
+
+    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".kt")
+    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".html")
+    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".sh")
+    assert_equal 'text/plain; charset=utf-8', Mime.lookup(".latex")
+
+    assert_equal 'application/octet-stream', Mime.lookup(".dmg")
+  end
+end

test/test_pathname.rb

@@ -55,7 +55,6 @@ class TestPathname < Test::Unit::TestCase
     assert_equal 'application/ruby', Pathname.new("file.rb").mime_type
     assert_equal 'application/javascript', Pathname.new("file.js").mime_type
     assert_equal 'application/python', Pathname.new("itty.py").mime_type
-    assert_equal 'text/plain', Pathname.new("itty.nu").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
 end