Separate registry content type and content disposition

lib/linguist/blob_helper.rb

@@ -2,7 +2,6 @@ require 'linguist/language'
 require 'linguist/mime'
 require 'linguist/pathname'
 
-require 'escape_utils'
 require 'yaml'
 
 module Linguist
@@ -63,12 +62,7 @@ module Linguist
     #
     # Returns a content disposition String.
     def disposition
-      case content_type
-      when 'application/octet-stream', 'application/java-archive'
-        "attachment; filename=#{EscapeUtils.escape_url(pathname.basename)}"
-      else
-        'inline'
-      end
+      pathname.disposition
     end
 
     # Public: Is the blob text?

lib/linguist/content_types.yml

@@ -1,7 +1,11 @@
 # MIME types and extensions to override when the raw blob is served
 
 # mime types
+application/javascript: text/plain
+application/perl: text/plain
+application/python: text/plain
 application/rdf+xml: text/plain
+application/ruby: text/plain
 application/sh: text/plain
 application/tex: text/plain
 application/xhtml+xml: text/plain

lib/linguist/mime.rb

@@ -1,11 +1,42 @@
 require 'mime/types'
 require 'yaml'
 
+module MIME
+  class Type
+    attr_accessor :binary
+
+    undef_method :binary?
+    def binary?
+      if defined? @binary
+        @binary
+      else
+        @encoding == 'base64'
+      end
+    end
+
+    attr_accessor :attachment
+
+    def attachment?
+      if defined? @attachment
+        @attachment
+      else
+        binary?
+      end
+    end
+  end
+end
+
 # Register additional mime type extensions
 mime_extensions = YAML.load_file(File.expand_path("../mimes.yml", __FILE__))
-mime_extensions.each do |mime_type, exts|
+mime_extensions.each do |mime_type, options|
   mime = MIME::Types[mime_type].first || MIME::Type.new(mime_type)
-  exts.each { |ext| mime.extensions << ext }
+
+  (options['extensions'] || []).each { |ext| mime.extensions << ext }
+
+  mime.binary     = options['binary']     if options.key?('binary')
+  mime.attachment = options['attachment'] if options.key?('attachment')
+
+  MIME::Types.add_type_variant(mime)
   MIME::Types.index_extensions(mime)
 end
 
@@ -59,5 +90,49 @@ module Linguist
 
       type
     end
+
+    # Internal: Determine if extension or mime type is binary.
+    #
+    # ext_or_mime_type - A file extension ".txt" or mime type "text/plain".
+    #
+    # Returns true or false
+    def self.binary?(ext_or_mime_type)
+      mime_type = lookup_mime_type_for(ext_or_mime_type)
+      mime_type.nil? || mime_type.binary?
+    end
+
+    # Internal: Determine if extension or mime type is an attachment.
+    #
+    # ext_or_mime_type - A file extension ".txt" or mime type "text/plain".
+    #
+    # Attachments are files that should be downloaded rather than be
+    # displayed in the browser.
+    #
+    # This is used to set our Content-Disposition headers.
+    #
+    # Attachment files should generally binary files but non-
+    # attachments do not imply plain text. For an example Images are
+    # not treated as attachments.
+    #
+    # Returns true or false
+    def self.attachment?(ext_or_mime_type)
+      mime_type = lookup_mime_type_for(ext_or_mime_type)
+      mime_type.nil? || mime_type.attachment?
+    end
+
+    # Internal: Lookup mime type for extension or mime type
+    #
+    # Returns a MIME::Type
+    def self.lookup_mime_type_for(ext_or_mime_type)
+      ext_or_mime_type ||= ''
+
+      if ext_or_mime_type =~ /\w+\/\w+/
+        guesses = ::MIME::Types[ext_or_mime_type]
+      else
+        guesses = ::MIME::Types.type_for(ext_or_mime_type)
+      end
+
+      guesses.first
+    end
   end
 end

lib/linguist/mimes.yml

@@ -2,13 +2,75 @@
 #
 # Review this list if we ever upgrade from mime-types 1.15 to 1.16
 
+application/atom+xml:
+  binary: false
+
+application/javascript:
+  binary: false
+  extensions:
+  - js
+
+application/json:
+  binary: false
+  extensions:
+  - json
+
+application/rdf+xml:
+  binary: false
+
+application/sh:
+  binary: false
+
+application/x-troff-ms:
+  binary: false
+
+application/netcdf:
+  binary: false
+
+application/x-perl:
+  binary: false
+  extensions:
+  - pl
+  - pm
+
+application/x-python:
+  binary: false
+  extensions:
+  - py
+
+application/x-ruby:
+  binary: false
+  extensions:
+  - rb
+
+application/x-wais-source:
+  binary: false
+
+application/vnd.mozilla.xul+xml:
+  binary: false
+
 application/octet-stream:
-- dmg
-- dll
+  binary: true
+  extensions:
+  - dmg
+  - dll
 
 application/java-archive:
-- ear
-- war
+  binary: true
+  extensions:
+  - ear
+  - war
 
 application/x-shockwave-flash:
-- swf
+  binary: true
+  extensions:
+  - swf
+
+image/gif:
+  attachment: false
+
+image/jpeg:
+  attachment: false
+
+image/png:
+  attachment: false

lib/linguist/pathname.rb

@@ -1,6 +1,8 @@
 require 'linguist/language'
 require 'linguist/mime'
 
+require 'escape_utils'
+
 module Linguist
   # Similar to ::Pathname, Linguist::Pathname wraps a path string and
   # provides helpful query methods. Its useful when you only have a
@@ -99,6 +101,30 @@ module Linguist
       @content_type ||= Mime.content_type_for(extname)
     end
 
+    # Public: Determine if the Pathname should be served as an
+    # attachment.
+    #
+    # Returns true or false.
+    def attachment?
+      @attachment ||= Mime.attachment?(extname)
+    end
+
+    # Public: Get the Content-Disposition header value
+    #
+    # This value is used when serving raw blobs.
+    #
+    #   # => "attachment; filename=file.tar"
+    #   # => "inline"
+    #
+    # Returns a content disposition String.
+    def disposition
+      if attachment?
+        "attachment; filename=#{EscapeUtils.escape_url(basename)}"
+      else
+        'inline'
+      end
+    end
+
     def to_s
       @path.dup
     end

test/test_blob.rb

@@ -23,7 +23,7 @@ class TestBlob < Test::Unit::TestCase
   end
 
   def test_mime_type
-    assert_equal "text/plain", blob("grit.rb").mime_type
+    assert_equal "application/ruby", 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
@@ -31,6 +31,7 @@ class TestBlob < Test::Unit::TestCase
 
   def test_content_type
     assert_equal "text/plain; charset=utf-8", blob("grit.rb").content_type
+    assert_equal "text/plain; charset=utf-8", blob("foo.pl").content_type
     assert_equal "text/plain; charset=utf-8", blob("bar.xml").content_type
     assert_equal "application/octet-stream", blob("dog.o").content_type
     assert_equal "text/plain; charset=utf-8", blob("script.sh").content_type
@@ -69,7 +70,9 @@ class TestBlob < Test::Unit::TestCase
     assert blob("git.deb").binary?
     assert blob("git.exe").binary?
     assert !blob("file.txt").binary?
+    assert !blob("foo.rb").binary?
     assert !blob("octocat.png").binary?
+    assert !blob("script.pl").binary?
   end
 
   def test_text
@@ -88,6 +91,7 @@ class TestBlob < Test::Unit::TestCase
 
   def test_viewable
     assert blob("foo.rb").viewable?
+    assert blob("script.pl").viewable?
     assert !blob("octocat.png").viewable?
     assert !blob("linguist.gem").viewable?
   end

test/test_mime.rb

@@ -8,9 +8,9 @@ class TestMime < Test::Unit::TestCase
   def test_mime
     assert_equal 'text/plain', Mime.mime_for(nil)
 
-    assert_equal 'text/plain', Mime.mime_for(".rb")
-    assert_equal 'text/plain', Mime.mime_for(".js")
-    assert_equal 'text/plain', Mime.mime_for(".py")
+    assert_equal 'application/ruby', Mime.mime_for(".rb")
+    assert_equal 'application/javascript', Mime.mime_for(".js")
+    assert_equal 'application/python', Mime.mime_for(".py")
 
     assert_equal 'text/plain', Mime.mime_for(".kt")
     assert_equal 'text/html', Mime.mime_for(".html")
@@ -49,4 +49,116 @@ class TestMime < Test::Unit::TestCase
     assert_equal 'application/java-archive', Mime.content_type_for(".ear")
     assert_equal 'application/java-archive', Mime.content_type_for(".war")
   end
+
+  def test_binary
+    assert Mime.binary?("application/octet-stream")
+    assert !Mime.binary?("text/plain")
+
+    assert Mime.binary?("image/gif")
+    assert Mime.binary?("image/jpeg")
+    assert Mime.binary?("image/png")
+    assert Mime.binary?("java-archive")
+    assert Mime.binary?("x-shockwave-flash")
+
+    assert !Mime.binary?("application/atom+xml")
+    assert !Mime.binary?("application/javascript")
+    assert !Mime.binary?("application/json")
+    assert !Mime.binary?("application/rdf+xml")
+    assert !Mime.binary?("application/sh")
+    assert !Mime.binary?("application/x-perl")
+    assert !Mime.binary?("application/x-python")
+    assert !Mime.binary?("application/x-ruby")
+
+    assert !Mime.binary?(".ms")
+    assert !Mime.binary?(".nc")
+    assert !Mime.binary?(".src")
+    assert !Mime.binary?(".xul")
+  end
+
+  def test_attachment
+    assert Mime.attachment?("application/octet-stream")
+    assert !Mime.attachment?("text/plain")
+
+    assert Mime.attachment?("application/java-archive")
+    assert Mime.attachment?("application/ogg")
+    assert Mime.attachment?("application/pdf")
+    assert Mime.attachment?("application/x-gzip")
+    assert Mime.attachment?("application/zip")
+    assert Mime.attachment?("audio/mp4")
+
+    assert Mime.attachment?(".a")
+    assert Mime.attachment?(".air")
+    assert Mime.attachment?(".blend")
+    assert Mime.attachment?(".crx")
+    assert Mime.attachment?(".deb")
+    assert Mime.attachment?(".dmg")
+    assert Mime.attachment?(".exe")
+    assert Mime.attachment?(".gem")
+    assert Mime.attachment?(".graffle")
+    assert Mime.attachment?(".gz")
+    assert Mime.attachment?(".icns")
+    assert Mime.attachment?(".ipa")
+    assert Mime.attachment?(".lib")
+    assert Mime.attachment?(".mcz")
+    assert Mime.attachment?(".mov")
+    assert Mime.attachment?(".mp3")
+    assert Mime.attachment?(".nib")
+    assert Mime.attachment?(".o")
+    assert Mime.attachment?(".odp")
+    assert Mime.attachment?(".ods")
+    assert Mime.attachment?(".odt")
+    assert Mime.attachment?(".ogg")
+    assert Mime.attachment?(".ogv")
+    assert Mime.attachment?(".otf")
+    assert Mime.attachment?(".pfx")
+    assert Mime.attachment?(".pigx")
+    assert Mime.attachment?(".plgx")
+    assert Mime.attachment?(".pptx")
+    assert Mime.attachment?(".psd")
+    assert Mime.attachment?(".sib")
+    assert Mime.attachment?(".so")
+    assert Mime.attachment?(".spl")
+    assert Mime.attachment?(".sqlite3")
+    assert Mime.attachment?(".swc")
+    assert Mime.attachment?(".swf")
+    assert Mime.attachment?(".tar")
+    assert Mime.attachment?(".ucode")
+    assert Mime.attachment?(".xpi")
+    assert Mime.attachment?(".zip")
+
+    assert !Mime.attachment?("application/atom+xml")
+    assert !Mime.attachment?("application/javascript")
+    assert !Mime.attachment?("application/json")
+    assert !Mime.attachment?("application/rdf+xml")
+    assert !Mime.attachment?("application/sh")
+    assert !Mime.attachment?("application/xhtml+xml")
+    assert !Mime.attachment?("application/xml")
+    assert !Mime.attachment?("image/gif")
+    assert !Mime.attachment?("image/jpeg")
+    assert !Mime.attachment?("image/png")
+    assert !Mime.attachment?("text/css")
+    assert !Mime.attachment?("text/csv")
+    assert !Mime.attachment?("text/html")
+    assert !Mime.attachment?("text/javascript")
+    assert !Mime.attachment?("text/plain")
+
+    assert !Mime.attachment?(".gif")
+    assert !Mime.attachment?(".jpeg")
+    assert !Mime.attachment?(".jpg")
+    assert !Mime.attachment?(".js")
+    assert !Mime.attachment?(".latex")
+    assert !Mime.attachment?(".ms")
+    assert !Mime.attachment?(".nc")
+    assert !Mime.attachment?(".pl")
+    assert !Mime.attachment?(".png")
+    assert !Mime.attachment?(".ps")
+    assert !Mime.attachment?(".py")
+    assert !Mime.attachment?(".rb")
+    assert !Mime.attachment?(".sh")
+    assert !Mime.attachment?(".src")
+    assert !Mime.attachment?(".tcl")
+    assert !Mime.attachment?(".texi")
+    assert !Mime.attachment?(".texinfo")
+    assert !Mime.attachment?(".xul")
+  end
 end

test/test_pathname.rb

@@ -52,9 +52,9 @@ class TestPathname < Test::Unit::TestCase
   end
 
   def test_mime_type
-    assert_equal 'text/plain', Pathname.new("file.rb").mime_type
-    assert_equal 'text/plain', Pathname.new("file.js").mime_type
-    assert_equal 'text/plain', Pathname.new("itty.py").mime_type
+    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("defun.kt").mime_type
   end