Make parsing locales more robust.

Library/Homebrew/locale.rb

@@ -1,20 +1,28 @@
 class Locale
-  class ParserError < ::RuntimeError
+  class ParserError < StandardError
   end
 
-  LANGUAGE_REGEX = /(?:[a-z]{2})/
+  LANGUAGE_REGEX = /(?:[a-z]{2,3})/    # ISO 639-1 or ISO 639-2
-  REGION_REGEX = /(?:[A-Z]{2})/
+  REGION_REGEX   = /(?:[A-Z]{2})/      # ISO 3166-1
-  SCRIPT_REGEX = /(?:[A-Z][a-z]{3})/
+  SCRIPT_REGEX   = /(?:[A-Z][a-z]{3})/ # ISO 15924
 
-  LOCALE_REGEX = /^(#{LANGUAGE_REGEX})?(?:(?:^|-)(#{REGION_REGEX}))?(?:(?:^|-)(#{SCRIPT_REGEX}))?$/
+  LOCALE_REGEX = /\A((?:#{LANGUAGE_REGEX}|#{REGION_REGEX}|#{SCRIPT_REGEX})(?:\-|$)){1,3}\Z/
 
   def self.parse(string)
-    language, region, script = string.to_s.scan(LOCALE_REGEX)[0]
+    string = string.to_s
 
-    if language.nil? && region.nil? && script.nil?
+    if string !~ LOCALE_REGEX
-      raise ParserError, "'#{string}' cannot be parsed to a #{self.class}"
+      raise ParserError, "'#{string}' cannot be parsed to a #{self}"
     end
 
+    scan = proc do |regex|
+      string.scan(/(?:\-|^)(#{regex})(?:\-|$)/).flatten.first
+    end
+
+    language = scan.call(LANGUAGE_REGEX)
+    region   = scan.call(REGION_REGEX)
+    script   = scan.call(SCRIPT_REGEX)
+
     new(language, region, script)
   end
 

Library/Homebrew/test/test_os_mac_language.rb

@@ -1,16 +1,19 @@
 require "testing_env"
+require "locale"
 require "os/mac"
 
 class OSMacLanguageTests < Homebrew::TestCase
-  LANGUAGE_REGEX = /\A[a-z]{2}(-[A-Z]{2})?(-[A-Z][a-z]{3})?\Z/
-
   def test_languages_format
     OS::Mac.languages.each do |language|
-      assert_match LANGUAGE_REGEX, language
+      assert_nothing_raised do
+        Locale.parse(language)
+      end
     end
   end
 
   def test_language_format
-    assert_match LANGUAGE_REGEX, OS::Mac.language
+    assert_nothing_raised do
+      Locale.parse(OS::Mac.language)
+    end
   end
 end