Use same format as mime type lib

lib/linguist/mime.rb

@@ -2,27 +2,42 @@ require 'mime/types'
 require 'yaml'
 
 # Register additional mime type extensions
-mime_extensions = YAML.load_file(File.expand_path("../mimes.yml", __FILE__))
-mime_extensions.each do |mime_type, options|
-  mime = MIME::Types[mime_type].first || MIME::Type.new(mime_type)
+File.read(File.expand_path("../mimes.yml", __FILE__)).lines.each do |line|
+  next unless line =~ %r{^
+    #{MIME::Type::MEDIA_TYPE_RE}
+    (?:\s@([^\s]+))?
+    (?:\s:(#{MIME::Type::ENCODING_RE}))?
+  }x
 
-  (options['extensions'] || []).each { |ext| mime.extensions << ext }
+  mediatype  = $1
+  subtype    = $2
+  extensions = $3
+  encoding   = $4
 
-  (options['exclude_extensions'] || []).each do |ext|
-    mime.extensions.delete(ext)
+  mime_type = MIME::Types["#{mediatype}/#{subtype}"].first ||
+    MIME::Type.new("#{mediatype}/#{subtype}")
+
+  if extensions
+    extensions.split(/,/).each do |extension|
+      if extension =~ /^-(\w+)/
+        extension = $1
+        mime_type.extensions.delete(extension)
 
         MIME::Types.instance_eval do
           @__types__.instance_eval do
-        @extension_index[ext].delete(mime)
+            @extension_index[extension].delete(mime_type)
+          end
+        end
+      else
+        mime_type.extensions << extension
       end
     end
   end
 
-  mime.binary  = options['binary']    if options.key?('binary')
-  mime.encoding = options['encoding'] if options.key?('encoding')
+  mime_type.encoding = encoding
 
-  MIME::Types.add_type_variant(mime)
-  MIME::Types.index_extensions(mime)
+  MIME::Types.add_type_variant(mime_type)
+  MIME::Types.index_extensions(mime_type)
 end
 
 module Linguist

lib/linguist/mimes.yml

@@ -1,92 +1,26 @@
 # Additional types to add to MIME::Types
+#
+#   <type> @<extensions> :<encoding>
+#
+# Follows same format as mime-types data file
+#   https://github.com/halostatue/mime-types/blob/master/lib/mime/types.rb.data
 
-application/x-troff-ms:
-  encoding: 8bit
+# TODO: Lookup actual types
+application/octet-stream @a,blend,gem,graffle,ipa,lib,mcz,nib,o,odp,ods,odt,otf,ogv,pfx,pigx,plgx,pptx,psd,sib,spl,sqlite3,swc,ucode,xpi
 
-application/netcdf:
-  encoding: 8bit
-
-application/x-pagemaker:
-  exclude_extensions:
-  - pm
-
-application/x-wais-source:
-  encoding: 8bit
-
-application/vnd.mozilla.xul+xml:
-  encoding: 8bit
-
-application/vnd.adobe.air-application-installer-package+zip:
-  extensions:
-  - air
-
-application/octet-stream:
-  extensions:
-  # TODO: Lookup actual content type for these extensions
-  - a
-  - blend
-  - gem
-  - graffle
-  - ipa
-  - lib
-  - mcz
-  - nib
-  - o
-  - odp
-  - ods
-  - odt
-  - otf
-  - ogv
-  - pfx
-  - pigx
-  - plgx
-  - pptx
-  - psd
-  - sib
-  - spl
-  - sqlite3
-  - swc
-  - ucode
-  - xpi
-
-application/postscript:
-  encoding: base64
-
-application/java-archive:
-  extensions:
-  - ear
-  - war
-
-application/x-chrome-extension:
-  extensions:
-  - crx
-
-application/x-shockwave-flash:
-  extensions:
-  - swf
-
-image/x-icns:
-  extensions:
-  - icns
-
-text/cache-manifest:
-  extensions:
-  - manifest
-
-application/ogg:
-  extensions:
-  - ogg
-
-application/x-silverlight-app:
-  extensions:
-  - xap
-
-application/xaml+xml:
-  encoding: 8bit
-  extensions:
-  - xaml
-
-application/x-ms-xbap:
-  encoding: 8bit
-  extensions:
-  - xbap
+application/java-archive @ear,war
+application/netcdf :8bit
+application/ogg @ogg
+application/postscript :base64
+application/vnd.adobe.air-application-installer-package+zip @air
+application/vnd.mozilla.xul+xml :8bit
+application/x-chrome-extension @crx
+application/x-ms-xbap @xbap :8bit
+application/x-pagemaker @-pm
+application/x-shockwave-flash @swf
+application/x-silverlight-app @xap
+application/x-troff-ms :8bit
+application/x-wais-source :8bit
+application/xaml+xml @xaml :8bit
+image/x-icns @icns
+text/cache-manifest @manifest

test/test_mime.rb

@@ -120,6 +120,7 @@ class TestMime < Test::Unit::TestCase
     assert !Mime.binary?("application/xaml+xml")
     assert !Mime.binary?("application/xhtml+xml")
     assert !Mime.binary?("application/xml")
+    assert !Mime.binary?("text/cache-manifest")
     assert !Mime.binary?("text/css")
     assert !Mime.binary?("text/csv")
     assert !Mime.binary?("text/html")