Use Charlock all the time for binary detection

lib/linguist/blob_helper.rb

@@ -53,8 +53,7 @@ module Linguist
     #
     # Returns a content type String.
     def content_type
-      @content_type ||= binary? ?
+      @content_type ||= binary? ? mime_type :
-        mime_type :
         (encoding ? "text/plain; charset=#{encoding.downcase}" : "text/plain")
     end
 
@@ -93,11 +92,8 @@ module Linguist
     #
     # Return true or false
     def binary?
-      if mime_type = Mime.lookup_mime_type_for(pathname.extname)
+      return false if data.empty?
-        mime_type.binary?
+      encoding.nil? || detect_encoding[:type] == :binary
-      else
-        encoding.nil? || detect_encoding[:type] == :binary
-      end
     end
 
     # Public: Is the blob text?

lib/linguist/mime.rb

@@ -58,26 +58,6 @@ module Linguist
       mime_type ? mime_type.simplified : 'text/plain'
     end
 
-    # Internal: Determine if extension or mime type is binary.
-    #
-    # ext_or_mime_type - A file extension ".exe" or
-    #                    mime type "application/octet-stream".
-    #
-    # Returns true or false
-    def self.binary?(ext_or_mime_type)
-      mime_type = lookup_mime_type_for(ext_or_mime_type)
-      mime_type ? mime_type.binary? : false
-    end
-
-    # Internal: Determine if extension or mime type is plain text.
-    #
-    # ext_or_mime_type - A file extension ".txt" or mime type "text/plain".
-    #
-    # Returns true or false
-    def self.text?(ext_or_mime_type)
-      !binary?(ext_or_mime_type)
-    end
-
     # Internal: Lookup mime type for extension or mime type
     #
     # ext_or_mime_type - A file extension ".txt" or mime type "text/plain".

lib/linguist/pathname.rb

@@ -77,27 +77,6 @@ module Linguist
       @mime_type ||= Mime.mime_for(extname)
     end
 
-    # Public: Get the Content-Type header
-    #
-    # This value is used when serving raw blobs.
-    #
-    # Examples
-    #
-    #   Pathname.new('file.txt').content_type
-    #   # => 'text/plain; charset=utf-8'
-    #
-    # Returns a content type String.
-    def content_type
-      @content_type ||= binary? ? mime_type : 'text/plain; charset=utf8'
-    end
-
-    # Public: Is the path binary?
-    #
-    # Return true or false
-    def binary?
-      @binary ||= Mime.binary?(extname)
-    end
-
     # Public: Return self as String
     #
     # Returns a String

test/test_mime.rb

@@ -66,120 +66,4 @@ class TestMime < Test::Unit::TestCase
     assert_equal 'text/rust', Mime.mime_for('.rc')
     assert_equal 'video/quicktime', Mime.mime_for('.mov')
   end
-
-  def test_binary
-    assert Mime.binary?('application/octet-stream')
-
-    # Add an assertion for any binary mime types added to mimes.yml.
-    # Please keep this list alphabetized.
-    assert Mime.binary?('application/java-archive')
-    assert Mime.binary?('application/ogg')
-    assert Mime.binary?('application/pdf')
-    assert Mime.binary?('application/postscript')
-    assert Mime.binary?('application/x-gzip')
-    assert Mime.binary?('application/x-shockwave-flash')
-    assert Mime.binary?('application/x-silverlight-app')
-    assert Mime.binary?('application/zip')
-    assert Mime.binary?('audio/mp4')
-    assert Mime.binary?('image/gif')
-    assert Mime.binary?('image/jpeg')
-    assert Mime.binary?('image/png')
-
-    # Legacy. Prefer testing mime types instead of extensions.
-    assert Mime.binary?('.a')
-    assert Mime.binary?('.air')
-    assert Mime.binary?('.blend')
-    assert Mime.binary?('.crx')
-    assert Mime.binary?('.deb')
-    assert Mime.binary?('.dmg')
-    assert Mime.binary?('.exe')
-    assert Mime.binary?('.gem')
-    assert Mime.binary?('.graffle')
-    assert Mime.binary?('.gz')
-    assert Mime.binary?('.icns')
-    assert Mime.binary?('.ipa')
-    assert Mime.binary?('.key')
-    assert Mime.binary?('.lib')
-    assert Mime.binary?('.mcz')
-    assert Mime.binary?('.mov')
-    assert Mime.binary?('.mp3')
-    assert Mime.binary?('.nib')
-    assert Mime.binary?('.numbers')
-    assert Mime.binary?('.o')
-    assert Mime.binary?('.odp')
-    assert Mime.binary?('.ods')
-    assert Mime.binary?('.odt')
-    assert Mime.binary?('.ogg')
-    assert Mime.binary?('.ogv')
-    assert Mime.binary?('.otf')
-    assert Mime.binary?('.pages')
-    assert Mime.binary?('.pfx')
-    assert Mime.binary?('.pigx')
-    assert Mime.binary?('.plgx')
-    assert Mime.binary?('.pptx')
-    assert Mime.binary?('.psd')
-    assert Mime.binary?('.sib')
-    assert Mime.binary?('.so')
-    assert Mime.binary?('.spl')
-    assert Mime.binary?('.sqlite3')
-    assert Mime.binary?('.swc')
-    assert Mime.binary?('.swf')
-    assert Mime.binary?('.tar')
-    assert Mime.binary?('.ucode')
-    assert Mime.binary?('.xpi')
-    assert Mime.binary?('.zip')
-  end
-
-  def test_text
-    # By default, assume the plain text.
-    assert Mime.text?(nil)
-    assert Mime.text?('')
-
-    assert Mime.text?('text/plain')
-
-    # Add an assertion for any text mime types added to mimes.yml.
-    # Please keep this list alphabetized.
-    assert Mime.text?('application/atom+xml')
-    assert Mime.text?('application/javascript')
-    assert Mime.text?('application/json')
-    assert Mime.text?('application/perl')
-    assert Mime.text?('application/rdf+xml')
-    assert Mime.text?('application/sh')
-    assert Mime.text?('application/x-ms-xbap')
-    assert Mime.text?('application/x-perl')
-    assert Mime.text?('application/x-python')
-    assert Mime.text?('application/x-ruby')
-    assert Mime.text?('application/xaml+xml')
-    assert Mime.text?('application/xhtml+xml')
-    assert Mime.text?('application/xml')
-    assert Mime.text?('text/cache-manifest')
-    assert Mime.text?('text/css')
-    assert Mime.text?('text/csv')
-    assert Mime.text?('text/html')
-    assert Mime.text?('text/javascript')
-    assert Mime.text?('text/plain')
-    assert Mime.text?('text/x-nimrod')
-    assert Mime.text?('text/x-nemerle')
-
-    # Legacy. Prefer testing mime types instead of extensions.
-    assert Mime.text?('.cu')
-    assert Mime.text?('.js')
-    assert Mime.text?('.latex')
-    assert Mime.text?('.ms')
-    assert Mime.text?('.nc')
-    assert Mime.text?('.pl')
-    assert Mime.text?('.pm')
-    assert Mime.text?('.py')
-    assert Mime.text?('.rb')
-    assert Mime.text?('.sc')
-    assert Mime.text?('.sh')
-    assert Mime.text?('.sig')
-    assert Mime.text?('.sml')
-    assert Mime.text?('.sps')
-    assert Mime.text?('.src')
-    assert Mime.text?('.tcl')
-    assert Mime.text?('.texi')
-    assert Mime.text?('.texinfo')
-    assert Mime.text?('.xul')
-  end
 end

test/test_pathname.rb

@@ -59,10 +59,4 @@ class TestPathname < Test::Unit::TestCase
     assert_equal 'application/python', Pathname.new("itty.py").mime_type
     assert_equal 'text/plain', Pathname.new("defun.kt").mime_type
   end
-
-  def test_content_type
-    assert_equal 'text/plain; charset=utf8', Pathname.new("file.txt").content_type
-    assert_equal 'text/plain; charset=utf8', Pathname.new("file.rb").content_type
-    assert_equal 'image/png', Pathname.new("octocat.png").content_type
-  end
 end