Choose a supported locale based on ancestry (#237)

Rakefile

@@ -141,6 +141,11 @@ namespace :update do
     TwitterCldr::Resources::CasefolderClassGenerator.new.import
   end
 
+  desc 'Import parent locale data'
+  task :parent_locales do
+    TwitterCldr::Resources::ParentLocalesImporter.new.import
+  end
+
   desc 'Import postal codes resource'
   task :postal_codes do
     TwitterCldr::Resources::PostalCodesImporter.new.import

lib/twitter_cldr/resources.rb

@@ -20,6 +20,7 @@ module Resources
     autoload :Loader,                         'twitter_cldr/resources/loader'
     autoload :LocalesResourcesImporter,       'twitter_cldr/resources/locales_resources_importer'
     autoload :NumberFormatsImporter,          'twitter_cldr/resources/number_formats_importer'
+    autoload :ParentLocalesImporter,          'twitter_cldr/resources/parent_locales_importer'
     autoload :PostalCodesImporter,            'twitter_cldr/resources/postal_codes_importer'
     autoload :Properties,                     'twitter_cldr/resources/properties'
     autoload :SegmentDictionariesImporter,    'twitter_cldr/resources/segment_dictionaries_importer'
@@ -59,6 +60,7 @@ def standard_importer_classes
           LanguageCodesImporter,
           LocalesResourcesImporter,
           NumberFormatsImporter,
+          ParentLocalesImporter,
           PostalCodesImporter,
           RbnfTestImporter,
           SegmentDictionariesImporter,

lib/twitter_cldr/resources/parent_locales_importer.rb

@@ -0,0 +1,54 @@
+# encoding: UTF-8
+
+# Copyright 2012 Twitter, Inc
+# http://www.apache.org/licenses/LICENSE-2.0
+
+require 'nokogiri'
+
+module TwitterCldr
+  module Resources
+
+    class ParentLocalesImporter < Importer
+
+      requirement :cldr, Versions.cldr_version
+      output_path 'shared'
+      ruby_engine :mri
+
+      private
+
+      def execute
+        output_file = File.join(output_path, 'parent_locales.yml')
+        File.write(output_file, YAML.dump(parent_locales))
+      end
+
+      def parent_locales
+        @parent_locales ||= supplemental_data.xpath('//parentLocales/parentLocale').each_with_object({}) do |node, ret|
+          parent = node.attr('parent')
+          locales = node.attr('locales').split(' ')
+
+          locales.each do |locale|
+            ret[locale] = parent
+          end
+        end
+      end
+
+      def output_path
+        params.fetch(:output_path)
+      end
+
+      def supplemental_data
+        @supplemental_data ||= Nokogiri.XML(
+          File.read(
+            File.join(
+              requirements[:cldr].common_path,
+              'supplemental',
+              'supplementalData.xml'
+            )
+          )
+        )
+      end
+
+    end
+
+  end
+end

lib/twitter_cldr/shared/locale.rb

@@ -209,6 +209,10 @@ def validity_resource
           @validity_resource ||=
             TwitterCldr.get_resource('shared', 'validity_data')[:validity_data]
         end
+
+        def parent_locales
+          @parent_locales ||= TwitterCldr.get_resource('shared', 'parent_locales')
+        end
       end
 
       attr_accessor :language, :script, :region, :variants
@@ -239,11 +243,9 @@ def max_supported
 
       def supported
         @supported ||= begin
-          found = permutations('-').find do |perm|
-            TwitterCldr.supported_locale?(perm)
+          ancestor_chain.sort.find do |loc|
+            TwitterCldr.supported_locale?(loc.dasherized)
           end
-
-          self.class.new(found) if found
         end
       end
 
@@ -273,6 +275,52 @@ def permutations(delimiter = '_')
         perms.uniq
       end
 
+      def ==(other)
+        language == other.language &&
+          script == other.script &&
+          region == other.region &&
+          variants == other.variants
+      end
+
+      alias eql? ==
+
+      def hash
+        to_a.hash
+      end
+
+      def sort_key
+        k = 0
+        k += 3 if language
+        k += 2 if script
+        k += 1 if region
+        k
+      end
+
+      def <=>(other)
+        other.sort_key <=> sort_key
+      end
+
+      def ancestor_chain
+        ancestry = [self]
+        remaining = [self]
+
+        until remaining.empty?
+          locale = remaining.pop
+
+          if parent = self.class.send(:parent_locales)[locale.to_s]
+            parent = self.class.parse(parent)
+            ancestry << parent
+            remaining << parent
+          else
+            parents = locale.permutations.map { |p| self.class.parse(p) }
+            remaining += parents - ancestry
+            ancestry += parents - ancestry
+          end
+        end
+
+        ancestry
+      end
+
     end
   end
 end

resources/shared/parent_locales.yml

@@ -0,0 +1,172 @@
+---
+az_Arab: root
+az_Cyrl: root
+blt_Latn: root
+bm_Nkoo: root
+bs_Cyrl: root
+byn_Latn: root
+cu_Glag: root
+dje_Arab: root
+dyo_Arab: root
+en_Dsrt: root
+en_Shaw: root
+ff_Adlm: root
+ff_Arab: root
+ha_Arab: root
+hi_Latn: root
+iu_Latn: root
+kk_Arab: root
+ks_Deva: root
+ku_Arab: root
+ky_Arab: root
+ky_Latn: root
+ml_Arab: root
+mn_Mong: root
+mni_Mtei: root
+ms_Arab: root
+pa_Arab: root
+sat_Deva: root
+sd_Deva: root
+sd_Khoj: root
+sd_Sind: root
+shi_Latn: root
+so_Arab: root
+sr_Latn: root
+sw_Arab: root
+tg_Arab: root
+ug_Cyrl: root
+uz_Arab: root
+uz_Cyrl: root
+vai_Latn: root
+wo_Arab: root
+yo_Arab: root
+yue_Hans: root
+zh_Hant: root
+en_150: en_001
+en_AG: en_001
+en_AI: en_001
+en_AU: en_001
+en_BB: en_001
+en_BM: en_001
+en_BS: en_001
+en_BW: en_001
+en_BZ: en_001
+en_CA: en_001
+en_CC: en_001
+en_CK: en_001
+en_CM: en_001
+en_CX: en_001
+en_CY: en_001
+en_DG: en_001
+en_DM: en_001
+en_ER: en_001
+en_FJ: en_001
+en_FK: en_001
+en_FM: en_001
+en_GB: en_001
+en_GD: en_001
+en_GG: en_001
+en_GH: en_001
+en_GI: en_001
+en_GM: en_001
+en_GY: en_001
+en_HK: en_001
+en_IE: en_001
+en_IL: en_001
+en_IM: en_001
+en_IN: en_001
+en_IO: en_001
+en_JE: en_001
+en_JM: en_001
+en_KE: en_001
+en_KI: en_001
+en_KN: en_001
+en_KY: en_001
+en_LC: en_001
+en_LR: en_001
+en_LS: en_001
+en_MG: en_001
+en_MO: en_001
+en_MS: en_001
+en_MT: en_001
+en_MU: en_001
+en_MW: en_001
+en_MY: en_001
+en_NA: en_001
+en_NF: en_001
+en_NG: en_001
+en_NR: en_001
+en_NU: en_001
+en_NZ: en_001
+en_PG: en_001
+en_PH: en_001
+en_PK: en_001
+en_PN: en_001
+en_PW: en_001
+en_RW: en_001
+en_SB: en_001
+en_SC: en_001
+en_SD: en_001
+en_SG: en_001
+en_SH: en_001
+en_SL: en_001
+en_SS: en_001
+en_SX: en_001
+en_SZ: en_001
+en_TC: en_001
+en_TK: en_001
+en_TO: en_001
+en_TT: en_001
+en_TV: en_001
+en_TZ: en_001
+en_UG: en_001
+en_VC: en_001
+en_VG: en_001
+en_VU: en_001
+en_WS: en_001
+en_ZA: en_001
+en_ZM: en_001
+en_ZW: en_001
+en_AT: en_150
+en_BE: en_150
+en_CH: en_150
+en_DE: en_150
+en_DK: en_150
+en_FI: en_150
+en_NL: en_150
+en_SE: en_150
+en_SI: en_150
+es_AR: es_419
+es_BO: es_419
+es_BR: es_419
+es_BZ: es_419
+es_CL: es_419
+es_CO: es_419
+es_CR: es_419
+es_CU: es_419
+es_DO: es_419
+es_EC: es_419
+es_GT: es_419
+es_HN: es_419
+es_MX: es_419
+es_NI: es_419
+es_PA: es_419
+es_PE: es_419
+es_PR: es_419
+es_PY: es_419
+es_SV: es_419
+es_US: es_419
+es_UY: es_419
+es_VE: es_419
+pt_AO: pt_PT
+pt_CH: pt_PT
+pt_CV: pt_PT
+pt_FR: pt_PT
+pt_GQ: pt_PT
+pt_GW: pt_PT
+pt_LU: pt_PT
+pt_MO: pt_PT
+pt_MZ: pt_PT
+pt_ST: pt_PT
+pt_TL: pt_PT
+zh_Hant_MO: zh_Hant_HK

spec/shared/locale_spec.rb

@@ -136,7 +136,25 @@
     end
   end
 
-  context 'with a locale instance' do
+  context 'with a locale with interesting ancestry' do
+    let(:locale) { described_class.new('es', nil, 'CR') }
+
+    describe '#ancestor_chain' do
+      it 'identifies the correct ancestors' do
+        expect(locale.ancestor_chain.map(&:dasherized)).to eq(
+          ['es-CR', 'es-419', 'es']
+        )
+      end
+    end
+
+    describe '#max_supported' do
+      it 'identifies the correct supported parent locale' do
+        expect(locale.max_supported.dasherized).to eq('es-419')
+      end
+    end
+  end
+
+  context 'with a locale instance (Korean)' do
     let(:locale) { described_class.new('ko', nil, 'KR') }
 
     describe '#full_script' do
@@ -212,5 +230,19 @@
         )
       end
     end
+
+    describe '#ancestor_chain' do
+      it 'identifies the correct ancestors' do
+        expect(locale.ancestor_chain.map(&:dasherized)).to eq(
+          ['ko-KR', 'ko']
+        )
+      end
+    end
+
+    describe '#max_supported' do
+      it 'identifies the correct supported locale' do
+        expect(locale.max_supported.dasherized).to eq('ko')
+      end
+    end
   end
 end