twitter_cldr 6.0.2 → 6.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e16faa9cf2e7fda6f36008b520e7f18cc1cd64c9e5c0a9c01cc3c787cfddf38d
-  data.tar.gz: 0750c3375803f85d1c126b473d15b11e91429af41ca6940da670732ebe57ce8d
+  metadata.gz: 72981bad7d4ed54141a4e9dd7dc94cb49035766eed724834916ac772c48f7eed
+  data.tar.gz: 96889477a704251878f8f863caef717a7e371831cb909a4724ebc32aedcc447b
 SHA512:
-  metadata.gz: ffa52ad4c2ae417fca4e1d87c23272845dbfd7069f14984b4c123ebdf770410cef4ae38c552d1fe225727897f926fb671a3b7624a48a3429e6ba8518722ba412
-  data.tar.gz: f839fa10c1623407fa95a8f7e044376b3f139ba2bd602d0d57c6982899455c046f0e947d9f047a6f7150bbbc4b75bf1010562c720656afec25fde2ebe2b55853
+  metadata.gz: 64876d8266dbe0e629b21bd8e62a49aea758046e14a74f31afc4e751e7c93d9081ac1f73e4a1ff64899a2c9dda519e14fab83d8ed017e953ebeb42967f3040cf
+  data.tar.gz: e953e933a7a0c6a1a45167613591e5a2e93653ef64aff35066474bc3f371511ae5f1d06f06ab6683500960d01aaebea7714dbb8d1d535a5724fe8122b309a95c

data/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

data/lib/twitter_cldr.rb

@@ -112,9 +112,28 @@ module TwitterCldr
     end
 
     def convert_locale(locale)
-      locale = locale.to_sym if locale.respond_to?(:to_sym)
+      locale = normalize_locale(locale)
+
+      unless supported_locale?(locale)
+        loc = TwitterCldr::Shared::Locale.parse(locale)
+        max_supported = loc.max_supported
+
+        if loc.dasherized == 'und' || !max_supported
+          return locale
+        end
+
+        locale = normalize_locale(max_supported.dasherized)
+      end
+
+      locale
+    end
+
+    def normalize_locale(locale)
+      return locale unless (locale.is_a?(String) || locale.is_a?(Symbol))
+
+      locale = locale.to_sym
       locale = lowercase_locales_map.fetch(locale, locale)
-      TWITTER_LOCALE_MAP.fetch(locale, locale)
+      TWITTER_LOCALE_MAP.fetch(locale.downcase, locale)
     end
 
     def twitter_locale(locale)
@@ -127,7 +146,7 @@ module TwitterCldr
     end
 
     def supported_locale?(locale)
-      !!locale && supported_locales.include?(convert_locale(locale))
+      !!locale && supported_locales.include?(normalize_locale(locale))
     end
 
     protected

data/lib/twitter_cldr/resources.rb

@@ -20,6 +20,7 @@ module TwitterCldr
     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 @@ module TwitterCldr
           LanguageCodesImporter,
           LocalesResourcesImporter,
           NumberFormatsImporter,
+          ParentLocalesImporter,
           PostalCodesImporter,
           RbnfTestImporter,
           SegmentDictionariesImporter,

data/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

data/lib/twitter_cldr/shared/locale.rb

@@ -209,6 +209,10 @@ module TwitterCldr
           @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 @@ module TwitterCldr
 
       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 @@ module TwitterCldr
         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

data/lib/twitter_cldr/version.rb

@@ -4,5 +4,5 @@
 # http://www.apache.org/licenses/LICENSE-2.0
 
 module TwitterCldr
-  VERSION = '6.0.2'
+  VERSION = '6.1.0'
 end

data/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

data/spec/formatters/plurals/plural_formatter_spec.rb

@@ -195,7 +195,7 @@ describe TwitterCldr::Formatters::PluralFormatter do
 
   describe '#pluralization_rule' do
     it 'delegates pluralization rule fetching to Rules.rule_for method' do
-      expect(TwitterCldr::Formatters::Plurals::Rules).to receive(:rule_for).with(42, :jp).and_return('result')
+      expect(TwitterCldr::Formatters::Plurals::Rules).to receive(:rule_for).with(42, :ja).and_return('result')
       expect(described_class.new(:jp).send(:pluralization_rule, 42)).to eq('result')
     end
   end

data/spec/shared/calendar_spec.rb

@@ -28,7 +28,7 @@ describe TwitterCldr::Shared::Calendar do
     end
 
     it 'returns calendar for a specific locale' do
-      expect(described_class.new(:jp).locale).to eq(:jp)
+      expect(described_class.new(:jp).locale).to eq(:ja)
     end
 
     it 'uses TwitterCldr.convert_locale' do

data/spec/shared/locale_spec.rb

@@ -136,7 +136,25 @@ describe TwitterCldr::Shared::Locale do
     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 @@ describe TwitterCldr::Shared::Locale do
         )
       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

data/spec/shared/numbers_spec.rb

@@ -10,8 +10,8 @@ describe TwitterCldr::Shared::Numbers do
     let(:symbols) { { nan: 'NaN', minus_sign: '-' } }
 
     it 'returns numerical symbols for default locale' do
-      allow(TwitterCldr).to receive(:locale).and_return(:jp)
-      allow(TwitterCldr).to receive(:get_locale_resource).with(:jp, :numbers).and_return(jp: { numbers: { symbols: symbols } })
+      allow(TwitterCldr).to receive(:locale).and_return(:ja)
+      allow(TwitterCldr).to receive(:get_locale_resource).with(:ja, :numbers).and_return(ja: { numbers: { symbols: symbols } })
       expect(described_class.symbols).to eq(symbols)
     end
 

data/spec/twitter_cldr_spec.rb

@@ -53,10 +53,31 @@ describe TwitterCldr do
       expect(TwitterCldr.convert_locale(:msa)).to eq(:ms)
       expect(TwitterCldr.convert_locale(:'zh-cn')).to eq(:zh)
       expect(TwitterCldr.convert_locale(:'zh-tw')).to eq(:'zh-Hant')
+      expect(TwitterCldr.convert_locale(:'zh-TW')).to eq(:'zh-Hant')
+    end
+
+    it "should fallback to language if locale is unsupported but language is" do
+      expect(TwitterCldr.convert_locale(:'pt-BR')).to eq(:pt)
+      expect(TwitterCldr.convert_locale(:'zh-Hans-CN')).to eq(:zh)
+    end
+
+    it "should leave known locales alone" do
+      expect(TwitterCldr.convert_locale(:fr)).to eq(:fr)
+      expect(TwitterCldr.convert_locale(:'fr-ca')).to eq(:'fr-CA')
+      expect(TwitterCldr.convert_locale(:'fr-CA')).to eq(:'fr-CA')
+      expect(TwitterCldr.convert_locale(:'es-419')).to eq(:'es-419')
     end
 
     it "should leave unknown locales alone" do
       expect(TwitterCldr.convert_locale(:blarg)).to eq(:blarg)
+      expect(TwitterCldr.convert_locale(:'zz-ZZ')).to eq(:'zz-ZZ')
+    end
+
+    it "shouldn't blowup on bad input" do
+      expect(TwitterCldr.convert_locale(5)).to eq(5)
+      expect(TwitterCldr.convert_locale([])).to eq([])
+      expect(TwitterCldr.convert_locale('')).to eq(:'')
+      expect(TwitterCldr.convert_locale(nil)).to eq(nil)
     end
   end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: twitter_cldr
 version: !ruby/object:Gem::Version
-  version: 6.0.2
+  version: 6.1.0
 platform: ruby
 authors:
 - Cameron Dutro
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-07-28 00:00:00.000000000 Z
+date: 2020-08-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: camertron-eprun
@@ -148,6 +148,7 @@ files:
 - lib/twitter_cldr/resources/loader.rb
 - lib/twitter_cldr/resources/locales_resources_importer.rb
 - lib/twitter_cldr/resources/number_formats_importer.rb
+- lib/twitter_cldr/resources/parent_locales_importer.rb
 - lib/twitter_cldr/resources/postal_codes_importer.rb
 - lib/twitter_cldr/resources/properties.rb
 - lib/twitter_cldr/resources/properties/age_property_importer.rb
@@ -1833,6 +1834,7 @@ files:
 - resources/shared/likely_subtags.yml
 - resources/shared/metazones.yml
 - resources/shared/numbering_systems.yml
+- resources/shared/parent_locales.yml
 - resources/shared/postal_codes.yml
 - resources/shared/rbnf_root.yml
 - resources/shared/segments/dictionaries/burmesedict.dump