Be explicit about ambiguous extensions

lib/linguist/language.rb

@@ -9,12 +9,27 @@ module Linguist
   # Languages are defined in `lib/linguist/languages.yml`.
   class Language
     @languages       = []
+    @overrides       = {}
     @index           = {}
     @name_index      = {}
     @alias_index     = {}
     @extension_index = {}
     @filename_index  = {}
 
+    # Internal: Test if extension maps to multiple Languages.
+    #
+    # Returns true or false.
+    def self.ambiguous?(extension)
+      @overrides.include?(extension)
+    end
+
+    # Include?: Return overridden extensions.
+    #
+    # Returns extensions Array.
+    def self.overridden_extensions
+      @overrides.keys
+    end
+
     # Internal: Create a new Language object
     #
     # attributes - A hash of attributes
@@ -47,17 +62,21 @@ module Linguist
           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)
-          warn "Duplicate extension: #{extension}"
+        unless ambiguous?(extension)
+          # Index the extension with a leading ".": ".rb"
+          @extension_index[extension] = language
+
+          # Index the extension without a leading ".": "rb"
+          @extension_index[extension.sub(/^\./, '')] = language
+        end
+      end
+
+      language.overrides.each do |extension|
+        if extension !~ /^\./
+          warn "Extension is missing a '.': #{extension.inspect}"
         end
 
-        # Index the extension with a leading ".": ".rb"
-        @extension_index[extension] = language
-
-        # Index the extension without a leading ".": "rb"
-        @extension_index[extension.sub(/^\./, '')] = language
+        @overrides[extension] = language
       end
 
       language.filenames.each do |filename|
@@ -191,6 +210,7 @@ module Linguist
 
       # Set extensions or default to [].
       @extensions = attributes[:extensions] || []
+      @overrides  = attributes[:overrides]  || []
       @filenames  = attributes[:filenames]  || []
 
       # Set popular, major, and searchable flags
@@ -260,6 +280,11 @@ module Linguist
     # Returns the extensions Array
     attr_reader :extensions
 
+    # Internal: Get overridden extensions.
+    #
+    # Returns the extensions Array.
+    attr_reader :overrides
+
     # Public: Get filenames
     #
     # Examples
@@ -381,6 +406,7 @@ module Linguist
       :searchable  => options.key?('searchable') ? options['searchable'] : true,
       :search_term => options['search_term'],
       :extensions  => options['extensions'],
+      :overrides   => options['overrides'],
       :filenames   => options['filenames'],
       :major       => options['major'],
       :popular     => popular.include?(name)

lib/linguist/languages.yml

@@ -9,6 +9,7 @@
 # aliases     - An Array of additional aliases (implicitly
 #               includes name.downcase)
 # extension   - An Array of associated extensions
+# overrides   - An Array of extensions that takes precedence over conflicts
 # major       - Boolean flag major programming languages. Please leave
 #               this option to GitHub staff to decide.
 # searchable  - Boolean flag to enable searching (defaults to true)
@@ -477,6 +478,7 @@ Markdown:
 
 Matlab:
   extensions:
+  - .m
   - .matlab
 
 Max/MSP:
@@ -542,6 +544,8 @@ ObjDump:
 
 Objective-C:
   major: true
+  overrides:
+  - .m
   extensions:
   - .m
   - .mm
@@ -580,6 +584,8 @@ Parrot Internal Representation:
 
 Perl:
   major: true
+  overrides:
+  - .pl
   extensions:
   - .pl
   - .ph
@@ -593,6 +599,7 @@ Perl:
 Prolog:
   major: true
   extensions:
+  - .pl
   - .pro
   - .prolog
 
@@ -620,6 +627,8 @@ Python traceback:
 R:
   major: true
   lexer: S
+  overrides:
+  - .r
   extensions:
   - .r
   - .R
@@ -648,9 +657,10 @@ Raw token data:
 Rebol:
   lexer: REBOL
   extensions:
-  - .rebol
+  - .r
   - .r2
   - .r3
+  - .rebol
 
 Redcode:
   extensions:

test/test_language.rb

@@ -5,6 +5,17 @@ require 'test/unit'
 class TestLanguage < Test::Unit::TestCase
   include Linguist
 
+  def test_ambiguous_extensions
+    assert Language.ambiguous?('.m')
+    assert_equal Language['Objective-C'], Language.find_by_extension('m')
+
+    assert Language.ambiguous?('.pl')
+    assert_equal Language['Perl'], Language.find_by_extension('pl')
+
+    assert Language.ambiguous?('.r')
+    assert_equal Language['R'], Language.find_by_extension('r')
+  end
+
   def test_lexer
     # Add an assertion to this list if you add/change any lexers
     # in languages.yml. Please keep this list alphabetized.
@@ -383,7 +394,9 @@ class TestLanguage < Test::Unit::TestCase
   def test_find_all_by_extension
     Language.all.each do |language|
       language.extensions.each do |extension|
-        assert_equal language, Language.find_by_extension(extension)
+        unless Language.ambiguous?(extension)
+          assert_equal language, Language.find_by_extension(extension)
+        end
       end
     end
   end