refactor Globalize::Locale::Rfc4646 and add Globalize::Locale::Langua…

…geTag

lib/globalize/locale/fallbacks.rb

@@ -1,4 +1,4 @@
-require 'globalize/locale/rfc4646'
+require 'globalize/locale/language_tag'
 
 module Globalize
   module Locale
@@ -9,7 +9,7 @@ def initialize(rules = nil)
 
       # TODO make this code handle real cases
       def compute(tag)
-        rfc_tag = Rfc4646::tag(tag)
+        rfc_tag = LanguageTag::tag(tag)
         [ rfc_tag, rfc_tag.parent ].compact.map {|rt| rt.to_s } + (@map[tag] || [])
       end
 

lib/globalize/locale/rfc4646.rb → lib/globalize/locale/language_tag.rb

@@ -1,14 +1,18 @@
-# TODO rename this according to http://en.wikipedia.org/wiki/IETF_language_tag
+# for specifications see http://en.wikipedia.org/wiki/IETF_language_tag
+#
+# SimpleParser does not implement advanced usages such as grandfathered tags
 
 module Globalize
   module Locale
     module Rfc4646
       SUBTAGS = [:language, :script, :region, :variant, :extension, :privateuse, :grandfathered]
       FORMATS = {:language => :downcase, :script => :capitalize, :region => :upcase, :variant => :downcase}
-
+    end
+
+    class LanguageTag < Struct.new(*Rfc4646::SUBTAGS)
       class << self
         def parser
-          @@parser ||= Simple
+          @@parser ||= SimpleParser
         end
 
         def parser=(parser)
@@ -17,30 +21,28 @@ def parser=(parser)
 
         def tag(tag)
           matches = parser.match(tag)
-          Tag.new *matches if matches
+          new *matches if matches
         end
       end
 
-      class Tag < Struct.new(*SUBTAGS)
-        FORMATS.each do |name, format|
-          define_method(name) { self[name].send(format) unless self[name].nil? }
-        end
-
-        def to_s
-          @tag ||= to_a.compact.join("-")
-        end
-
-        def to_a
-          members.collect {|attr| self.send(attr) }
-        end
-
-        def parent
-          segs = to_a.compact
-          segs.length < 2 ? nil : Rfc4646.tag(segs[0..(segs.length-2)].join('-'))
-        end
+      Rfc4646::FORMATS.each do |name, format|
+        define_method(name) { self[name].send(format) unless self[name].nil? }
+      end
+
+      def to_s
+        @tag ||= to_a.compact.join("-")
+      end
+
+      def to_a
+        members.collect {|attr| self.send(attr) }
+      end
+
+      def parent
+        segs = to_a.compact
+        segs.length < 2 ? nil : LanguageTag.tag(segs[0..(segs.length-2)].join('-'))
       end
 
-      module Simple
+      module SimpleParser
         PATTERN = %r{\A(?:
           ([a-z]{2,3}(?:(?:-[a-z]{3}){0,3})?|[a-z]{4}|[a-z]{5,8}) # language
           (?:-([a-z]{4}))?                                        # script

spec/locale/fallbacks_spec.rb

@@ -1,6 +1,6 @@
 require File.dirname(__FILE__) + '/../spec/helper.rb'
 require 'globalize/locale/fallbacks'
-require 'globalize/locale/rfc4646'
+require 'globalize/locale/language_tag'
 
 include Globalize::Locale
 

spec/locale/rfc4646_spec.rb → spec/locale/language_tag_spec.rb

@@ -2,42 +2,42 @@
 
 include Globalize::Locale
 
-describe Globalize::Locale::Rfc4646 do
-  it "given a valid tag 'de' returns an Rfc4646::Tag from #tag" do
-    Rfc4646.tag('de').should be_instance_of(Rfc4646::Tag)
+describe Globalize::Locale::LanguageTag do
+  it "given a valid tag 'de' returns an LanguageTag from #tag" do
+    LanguageTag.tag('de').should be_instance_of(LanguageTag)
   end
 end
 
-describe Rfc4646::Simple, '#match' do
+describe LanguageTag::SimpleParser, '#match' do
   it "given a valid tag 'de' returns an array of subtags" do
-    Rfc4646::Simple.match('de').should == ['de', nil, nil, nil, nil, nil, nil]
+    LanguageTag::SimpleParser.match('de').should == ['de', nil, nil, nil, nil, nil, nil]
   end
 
   it "given a valid tag 'de' returns an array of subtags" do
-    Rfc4646::Simple.match('de-DE').should == ['de', nil, 'DE', nil, nil, nil, nil]
+    LanguageTag::SimpleParser.match('de-DE').should == ['de', nil, 'DE', nil, nil, nil, nil]
   end
 
   it "given a valid lowercase tag 'de-latn-de-variant-x-phonebk' returns an array of subtags" do
-    Rfc4646::Simple.match('de-latn-de-variant-x-phonebk').should == ['de', 'latn', 'de', 'variant', nil, 'x-phonebk', nil]
+    LanguageTag::SimpleParser.match('de-latn-de-variant-x-phonebk').should == ['de', 'latn', 'de', 'variant', nil, 'x-phonebk', nil]
   end
 
   it "given a valid uppercase tag 'DE-LATN-DE-VARIANT-X-PHONEBK' returns an array of subtags" do
-    Rfc4646::Simple.match('DE-LATN-DE-VARIANT-X-PHONEBK').should == ['DE', 'LATN', 'DE', 'VARIANT', nil, 'X-PHONEBK', nil]
+    LanguageTag::SimpleParser.match('DE-LATN-DE-VARIANT-X-PHONEBK').should == ['DE', 'LATN', 'DE', 'VARIANT', nil, 'X-PHONEBK', nil]
   end
 
   it "given an invalid tag 'a-DE' it returns false" do
-    Rfc4646::Simple.match('a-DE').should == false
+    LanguageTag::SimpleParser.match('a-DE').should == false
   end
 
   it "given an invalid tag 'de-419-DE' it returns false" do
-    Rfc4646::Simple.match('de-419-DE').should == false
+    LanguageTag::SimpleParser.match('de-419-DE').should == false
   end
 end
 
-describe "Rfc4646::Tag for the locale 'DE-latn-de-Variant-a-ext-x-phonebk-i-klingon'" do
+describe "LanguageTag for the locale 'DE-latn-de-Variant-a-ext-x-phonebk-i-klingon'" do
   before :each do
     subtags = %w(de Latn DE variant a-ext x-phonebk i-klingon)
-    @tag = Rfc4646::Tag.new *subtags
+    @tag = LanguageTag.new *subtags
   end
 
   it "returns 'de' as the language subtag in lowercase" do
@@ -77,41 +77,41 @@
   end
 end
 
-describe "Rfc4646::Tag inheritance" do
+describe "LanguageTag inheritance" do
   it "returns 'de-Latn-DE-variant-a-ext-x-phonebk' as the parent of 'de-Latn-DE-variant-a-ext-x-phonebk-i-klingon'" do
-    tag = Rfc4646::Tag.new *%w(de Latn DE variant a-ext x-phonebk i-klingon)
+    tag = LanguageTag.new *%w(de Latn DE variant a-ext x-phonebk i-klingon)
     tag.parent.to_s.should == 'de-Latn-DE-variant-a-ext-x-phonebk'
   end
 
   it "returns 'de-Latn-DE-variant-a-ext' as the parent of 'de-Latn-DE-variant-a-ext-x-phonebk'" do
-    tag = Rfc4646::Tag.new *%w(de Latn DE variant a-ext x-phonebk)
+    tag = LanguageTag.new *%w(de Latn DE variant a-ext x-phonebk)
     tag.parent.to_s.should == 'de-Latn-DE-variant-a-ext'
   end
 
   it "returns 'de-Latn-DE-variant' as the parent of 'de-Latn-DE-variant-a-ext'" do
-    tag = Rfc4646::Tag.new *%w(de Latn DE variant a-ext)
+    tag = LanguageTag.new *%w(de Latn DE variant a-ext)
     tag.parent.to_s.should == 'de-Latn-DE-variant'
   end
 
   it "returns 'de-Latn-DE' as the parent of 'de-Latn-DE-variant'" do
-    tag = Rfc4646::Tag.new *%w(de Latn DE variant)
+    tag = LanguageTag.new *%w(de Latn DE variant)
     tag.parent.to_s.should == 'de-Latn-DE'
   end
 
   it "returns 'de-Latn' as the parent of 'de-Latn-DE'" do
-    tag = Rfc4646::Tag.new *%w(de Latn DE)
+    tag = LanguageTag.new *%w(de Latn DE)
     tag.parent.to_s.should == 'de-Latn'
   end
 
   it "returns 'de' as the parent of 'de-Latn'" do
-    tag = Rfc4646::Tag.new *%w(de Latn)
+    tag = LanguageTag.new *%w(de Latn)
     tag.parent.to_s.should == 'de'
   end
 
   # TODO RFC4647 says: "If no language tag matches the request, the "default" value is returned."
   # where should we set the default language?
   # it "returns '' as the parent of 'de'" do
-  #   tag = Rfc4646::Tag.new *%w(de)
+  #   tag = LanguageTag.new *%w(de)
   #   tag.parent.to_s.should == ''
   # end
 end