fix bugs with charset handling

Fixes #520181, #521169

Author: 375gnu

lib/locale/driver/env.rb

@@ -24,11 +24,11 @@ module Driver
     module Env
       module_function
 
-      # Gets the locale from environment variable. (LC_ALL > LC_MESSAGES > LANG)
+      # Gets the locale from environment variable. (LC_ALL > LC_CTYPE > LANG)
       # Returns: the locale as Locale::Tag::Posix.
       def locale
         # At least one environment valiables should be set on *nix system.
-        [ENV["LC_ALL"], ENV["LC_MESSAGES"], ENV["LANG"]].each do |loc|
+        [ENV["LC_ALL"], ENV["LC_CTYPE"], ENV["LANG"]].each do |loc|
           if loc != nil and loc.size > 0
             return Locale::Tag::Posix.parse(loc)
           end

lib/locale/tag/posix.rb

@@ -30,9 +30,9 @@ def initialize(language, region = nil, charset = nil, modifier = nil)
       end
 
       def self.parse(tag)
-        if tag =~ /^(C|POSIX)$/
-          ret = self.new("en", "US")
-          ret.tag = tag
+        if tag =~ /\A(C|POSIX)(?:\.([^@]+))?\Z/
+          ret = self.new("en", "US", $2)
+          ret.tag = $1
           ret
         elsif tag =~ TAG_RE
           ret = self.new($1, $2, $3, $4)
@@ -47,10 +47,15 @@ def self.parse(tag)
       #   <language>_<COUNTRY>.<CHARSET>@<MODIFIER>
       #   (e.g.) "ja_JP.EUC-JP@Modifier"
       def to_s
-        s = @language.dup
-        s << "_#{@region}" if @region
-        s << ".#{@charset}" if @charset
-        s << "@#{@modifier}" if @modifier
+        if posix?
+          s = tag.dup
+          s << ".#{@charset}" if @charset
+        else
+          s = @language.dup
+          s << "_#{@region}" if @region
+          s << ".#{@charset}" if @charset
+          s << "@#{@modifier}" if @modifier
+        end
         s
       end
 
@@ -92,6 +97,10 @@ def convert_to(klass)
         end
       end
 
+      def posix?
+        ['POSIX', 'C'].include? tag
+      end
+
     end
   end
 end

lib/locale/taglist.rb

@@ -46,11 +46,7 @@ def script
     end
     # Returns the top priority charset. (posix)
     def charset
-      if self[0].respond_to? :charset
-        self[0].charset
-      else
-        ::Locale.driver_module.charset
-      end
+      self[0].respond_to?(:charset) and self[0].charset or ::Locale.driver_module.charset
     end
     memoize :charset
 

test/test_detect_general.rb

@@ -6,14 +6,14 @@ class TestDetectGeneral < Test::Unit::TestCase
   def setup
     Locale.clear_all
     ENV["LC_ALL"] = nil
-    ENV["LC_MESSAGES"] = nil
+    ENV["LC_CTYPE"] = nil
     ENV["LANG"] = nil
     ENV["LANGUAGE"] = nil
   end
 
   def test_lc_all
     ENV["LC_ALL"] = "ja_JP.eucJP"
-    ENV["LC_MESSAGES"] = "zh_CN.UTF-8"  #Ignored.
+    ENV["LC_CTYPE"] = "zh_CN.UTF-8"  #Ignored.
     ENV["LANG"] = "ko_KR.UTF-8"  #Ignored.
     ENV["LANGUAGE"] = nil
 
@@ -29,7 +29,7 @@ def test_lc_all
 
   def test_lc_messages
     ENV["LC_ALL"] = nil
-    ENV["LC_MESSAGES"] = "ja_JP.eucJP"
+    ENV["LC_CTYPE"] = "ja_JP.eucJP"
     ENV["LANG"] = "ko_KR.UTF-8"  #Ignored.
     ENV["LANGUAGE"] = nil
 
@@ -45,7 +45,7 @@ def test_lc_messages
 
   def test_lang
     ENV["LC_ALL"] = nil
-    ENV["LC_MESSAGES"] = nil
+    ENV["LC_CTYPE"] = nil
     ENV["LANG"] = "ja_JP.eucJP"
     ENV["LANGUAGE"] = nil
 
@@ -61,7 +61,7 @@ def test_lang
 
   def test_lang_complex
     ENV["LC_ALL"] = "zh_CN.UTF-8"  # Ignored.
-    ENV["LC_MESSAGES"] = "ko_KR.UTF-8" #Ingored.
+    ENV["LC_CTYPE"] = "ko_KR.UTF-8" #Ingored.
     ENV["LANG"] = "en_US.UTF-8"  # Ignored.
     ENV["LANGUAGE"] ="ja_JP.eucJP:zh_CN.UTF-8"