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