Separate filename and extension configuration

lib/linguist/blob_helper.rb

@@ -260,7 +260,7 @@ module Linguist
           language
 
         # See if there is a Language for the extension
-        elsif language = Language.find_by_extension(extname)
+        elsif language = Language.find_by_filename(pathname.to_s)
           language
 
         # Try to detect Language from shebang line

lib/linguist/language.rb

@@ -8,6 +8,7 @@ module Linguist
     @name_index      = {}
     @alias_index     = {}
     @extension_index = {}
+    @filename_index  = {}
 
     # Internal: Create a new Language object
     #
@@ -37,6 +38,10 @@ module Linguist
       end
 
       language.extensions.each do |extension|
+        if extension !~ /^\./
+          warn "Extension is missing a '.': #{extension.inspect}"
+        end
+
         # All Language extensions should be unique. Warn if there is a
         # duplicate.
         if @extension_index.key?(extension)
@@ -47,7 +52,11 @@ module Linguist
         @extension_index[extension] = language
 
         # Index the extension without a leading ".": "rb"
-        @extension_index[extension.sub(/^./, '')] = language
+        @extension_index[extension.sub(/^\./, '')] = language
+      end
+
+      language.filenames.each do |filename|
+        @filename_index[filename] = language
       end
 
       language
@@ -102,6 +111,21 @@ module Linguist
       @extension_index[extension]
     end
 
+    # Public: Look up Language by filename.
+    #
+    # filename - The path String.
+    #
+    # Examples
+    #
+    #   Language.find_by_filename('foo.rb')
+    #   # => #<Language name="Ruby">
+    #
+    # Returns the Language or nil if none was found.
+    def self.find_by_filename(filename)
+      basename, extname = File.basename(filename), File.extname(filename)
+      @filename_index[basename] || @extension_index[extname]
+    end
+
     # Public: Look up Language by its name or lexer.
     #
     # name - The case-insensitive String name of the Language
@@ -163,6 +187,7 @@ module Linguist
       # Set extensions or default to [].
       # Consider using `@lexer.extensions`
       @extensions = attributes[:extensions] || []
+      @filenames  = attributes[:filenames]  || []
 
       # Set popular, common, and searchable flags
       @popular    = attributes.key?(:popular)    ? attributes[:popular]    : false
@@ -211,11 +236,20 @@ module Linguist
     #
     # Examples
     #
-    #   # => ['.rb', '.rake', 'Rakefile', ...]
+    #   # => ['.rb', '.rake', ...]
     #
     # Returns the extensions Array
     attr_reader :extensions
 
+    # Public: Get filenames
+    #
+    # Examples
+    #
+    #   # => ['Rakefile', ...]
+    #
+    # Returns the extensions Array
+    attr_reader :filenames
+
     # Internal: Get default alias name
     #
     # Returns the alias name String
@@ -302,6 +336,7 @@ module Linguist
       :searchable  => options.key?(:searchable) ? options[:searchable] : true,
       :search_term => options[:search_term],
       :extensions  => options[:ext],
+      :filenames   => options[:filenames],
       :popular     => popular.include?(name),
       :common      => common.include?(name)
     )

lib/linguist/languages.yml

@@ -105,6 +105,7 @@ Clojure:
 CoffeeScript:
   :ext:
   - .coffee
+  :filenames:
   - Cakefile
 ColdFusion:
   :lexer: ColdFusion HTML
@@ -249,9 +250,10 @@ Haskell:
 INI:
   :ext:
   - .cfg
-  - .gitconfig
   - .ini
   - .properties
+  :filenames:
+  - .gitconfig
 IRC log:
   :lexer: IRC logs
   :search_term: irc
@@ -277,11 +279,12 @@ JavaScript:
   :ext:
   - .js
   - .sjs
-  - Jakefile
   - .jss
   - .ssjs
   - .jsx
   - .jake
+  :filenames:
+  - Jakefile
 JSON:
   :lexer: javascript
   :search_term: javascript
@@ -303,6 +306,7 @@ Lua:
 Makefile:
   :ext:
   - .mak
+  :filenames:
   - Makefile
 Mako:
   :ext:
@@ -341,6 +345,7 @@ Nu:
   - nush
   :ext:
   - .nu
+  :filenames:
   - Nukefile
 NumPy:
   :ext:
@@ -451,16 +456,17 @@ Ruby:
   - .rb
   - .ru
   - .builder
-  - Capfile
   - .rbw
   - .rbx
   - .god
   - .rake
   - .gemspec
-  - Rakefile
   - .irbrc
-  - Thorfile
   - .thor
+  :filenames:
+  - Capfile
+  - Rakefile
+  - Thorfile
   - Gemfile
 SQL:
   :searchable: false
@@ -491,10 +497,11 @@ Shell:
   - zsh
   :ext:
   - .bash
+  - .sh
+  :filenames:
   - .bash_profile
   - .bashrc
   - .profile
-  - .sh
   - .zlogin
   - .zsh
   - .zshrc
@@ -548,6 +555,7 @@ VimL:
   - vim
   :ext:
   - .vim
+  :filenames:
   - .vimrc
   - .gvimrc
 Visual Basic:
@@ -583,6 +591,7 @@ YAML:
   :ext:
   - .yml
   - .yaml
+  :filenames:
   - .gemrc
 Java Server Pages:
   :lexer: Java Server Page

lib/linguist/pathname.rb

@@ -39,18 +39,9 @@ module Linguist
     #   Pathname.new('file.rb').extname
     #   # => '.rb'
     #
-    #   Pathname.new('Rakefile').extname
-    #   # => 'Rakefile'
-    #
     # Returns a String.
     def extname
-      if basename[0] == ?.
-        basename
-      elsif basename.include?('.')
-        File.extname(basename)
-      else
-        basename
-      end
+      File.extname(@path)
     end
 
     # Public: Get the language of the path
@@ -65,7 +56,7 @@ module Linguist
     #
     # Returns a Langauge.
     def language
-      Language.find_by_extension(extname) || Language['Text']
+      Language.find_by_filename(@path) || Language['Text']
     end
 
     # Internal: Has a language.
@@ -74,7 +65,7 @@ module Linguist
     #
     # Returns true or false.
     def language?
-      Language.find_by_extension(extname) ? true : false
+      Language.find_by_filename(@path) ? true : false
     end
 
     # Deprecated: Get the lexer of the path

test/test_blob.rb

@@ -92,7 +92,11 @@ class TestBlob < Test::Unit::TestCase
     assert blob("README").text?
     assert blob("file.json").text?
     assert blob("file.txt").text?
+    assert blob("md").text?
     assert blob("script.sh").text?
+    assert blob("tender.md").text?
+    assert blob("txt").text?
+    assert blob("zip").text?
   end
 
   def test_image

test/test_language.rb

@@ -108,6 +108,15 @@ class TestLanguage < Test::Unit::TestCase
     end
   end
 
+  def test_find_by_filename
+    assert_equal Language['Ruby'], Language.find_by_filename('foo.rb')
+    assert_equal Language['Ruby'], Language.find_by_filename('foo/bar.rb')
+    assert_equal Language['Ruby'], Language.find_by_filename('Rakefile')
+    assert_nil Language.find_by_filename('rb')
+    assert_nil Language.find_by_filename('.rb')
+    assert_nil Language.find_by_filename('.kt')
+  end
+
   def test_find
     assert_equal "Ruby", Language['Ruby'].name
     assert_equal "Ruby", Language['ruby'].name

test/test_pathname.rb

@@ -13,21 +13,19 @@ class TestPathname < Test::Unit::TestCase
     assert_equal 'file.rb', Pathname.new("file.rb").basename
     assert_equal 'file.rb', Pathname.new("./file.rb").basename
     assert_equal 'file.rb', Pathname.new("sub/dir/file.rb").basename
+    assert_equal '.profile', Pathname.new(".profile").basename
   end
 
   def test_extname
-    assert_equal '.rb', Pathname.new(".rb").extname
     assert_equal '.rb', Pathname.new("file.rb").extname
     assert_equal '.rb', Pathname.new("./file.rb").extname
     assert_equal '.rb', Pathname.new("sub/dir/file.rb").extname
-
-    assert_equal 'Rakefile', Pathname.new("Rakefile").extname
-    assert_equal 'Rakefile', Pathname.new("./Rakefile").extname
-    assert_equal 'Rakefile', Pathname.new("vendor/Rakefile").extname
+    assert_equal '',    Pathname.new(".profile").extname
   end
 
   def test_language
-    assert_equal Language['Ruby'], Pathname.new(".rb").language
+    assert_equal Language['Text'], Pathname.new(".rb").language
+
     assert_equal Language['Ruby'], Pathname.new("file.rb").language
     assert_equal Language['Ruby'], Pathname.new("./file.rb").language
     assert_equal Language['Ruby'], Pathname.new("sub/dir/file.rb").language