twitter_cldr 3.0.10 → 3.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: bac470ce4e512aac2ad6466f8dd297e7f82a3f03
-  data.tar.gz: b12587637965f791f2c8bedb6fee338d4631aeaf
+  metadata.gz: accacbf47c61c3008a7cdae91acde6d4bb848bc9
+  data.tar.gz: 112cdf6e94bc614b5677289ee345b4ea38bd55e0
 SHA512:
-  metadata.gz: 2d166f1a1383b5c950edef4bcf2fd1f750fc568c3b29ec896f923b22258150d227b19210e3b6e3e53700703f3b34ef3499b0b5d799ace421dff15b924dd04f6a
-  data.tar.gz: d8e91fbcd4cb123bd8eacc3e802c8e9d69563462cb9d0f14d3678ed389c4ca7b5457c3d50566460e474121be96e0e8b1b3ed59cab11862d140c2afcb51f57e93
+  metadata.gz: f923179204c1627ebe184f5fe38d2249e0016b1c20a524c0c0b7e06c3382f19240b054866daf0019edf9215e6034c1bc65d942e4914c8ed4332894883fdd6402
+  data.tar.gz: 48dd5f10034e32e5900abec0f02a1ff11e0e98bb14dd6921f355b1e171f2fa03bc8c1f68bfb885244abdd9c0b8b1f3fa16cefc62f582c77f661f9db8fbd91050

data/Gemfile

@@ -20,13 +20,14 @@ group :development, :test do
     gem 'oniguruma'
   end
 
-  gem 'regexp_parser', '~> 0.1.5'
+  gem 'regexp_parser', '~> 0.1.6'
 end
 
 group :development do
   gem 'nokogiri', "~> 1.5.9"
 
-  gem 'ruby-cldr', :github => 'svenfuchs/ruby-cldr', :ref => 'cddf44b885'
+  gem 'ruby-cldr', :github => 'kl-7/ruby-cldr', :branch => 'kl_nubmber_patttern_pluralization'
+  gem 'cldr-plurals', '~> 1.0.0'
 end
 
 group :test do

data/History.txt

@@ -1,3 +1,12 @@
+== 3.1.0
+
+* Updated resources from CLDR v26 (except the collation data).
+* Added support for ordinal plurals.
+* Added PostalCodes#find_all method.
+* Added negative numbers abbreviation.
+* Fixed pluralization for abbreviated numbers.
+* Fixed pluralization rules by not merging 'en' rules into every locale.
+
 == 3.0.10
 
 * Adding Date back in as a localizable object.

data/README.md

@@ -456,6 +456,14 @@ postal_code = TwitterCldr::Shared::PostalCodes.for_territory(:ca)
 postal_code.valid?("V3H 1Z7")   # true
 ```
 
+Match all valid postal codes in a string with the `#find_all` method:
+
+```ruby
+# United States
+postal_code = TwitterCldr::Shared::PostalCodes.for_territory(:us) 
+postal_code.find_all("12345 23456")    # ["12345", "23456"]
+```
+
 Get a list of supported territories by using the `#territories` method:
 
 ```ruby

data/Rakefile

@@ -96,7 +96,7 @@ end
 # TODO: 'add_locale' task that creates a new directory and runs all necessary 'update' tasks (+ suggests to run those that depend on JRuby)
 
 namespace :update do
-  ICU_JAR = './vendor/icu4j-52_1.jar'
+  ICU_JAR = './vendor/icu4j.jar'
 
   desc 'Import locales resources'
   task :locales_resources, :cldr_path do |_, args|

data/lib/twitter_cldr.rb

@@ -33,7 +33,7 @@ module TwitterCldr
   extend SingleForwardable
 
   # version of CLDR that was used for generating YAML files in the resources/ directory
-  CLDR_VERSION = '21.0' # release date: 2012-02-10
+  CLDR_VERSION = '26.0' # release date: 2014-09-18
 
   DEFAULT_LOCALE = :en
   DEFAULT_CALENDAR_TYPE = :gregorian

data/lib/twitter_cldr/data_readers/number_data_reader.rb

@@ -32,6 +32,8 @@ module TwitterCldr
       DEFAULT_FORMAT = :default
       DEFAULT_SIGN = :positive
 
+      REDIRECT_REGEX = /\Anumbers\.formats\.\w+\.patterns\.\w+\z/
+
       attr_reader :type, :format
 
       def self.types
@@ -49,7 +51,20 @@ module TwitterCldr
         @format = options[:format] || DEFAULT_FORMAT
       end
 
-      def pattern(number = nil)
+      def format_number(number, options = [])
+        precision = options[:precision] || 0
+        pattern_for_number = pattern(number, precision == 0)
+
+        if pattern_for_number == 0
+          # there's no specific formatting for the number in the current locale
+          number.to_s
+        else
+          tokens = tokenizer.tokenize(pattern_for_number)
+          formatter.format(tokens, number, options)
+        end
+      end
+
+      def pattern(number = nil, decimal = true)
         sign = number < 0 ? :negative : :positive 
         path = BASE_PATH + if valid_type_for?(number, type)
           TYPE_PATHS[type]
@@ -57,14 +72,14 @@ module TwitterCldr
           TYPE_PATHS[:default]
         end
 
-        pattern = traverse(path)
+        pattern = get_pattern_data(path)
 
         if pattern[format]
           pattern = pattern[format]
         end
 
         if number
-          pattern = pattern_for_number(pattern, number)
+          pattern = pattern_for_number(pattern, number, decimal)
         end
 
         if pattern.is_a?(String)
@@ -96,6 +111,16 @@ module TwitterCldr
 
       private
 
+      def get_pattern_data(path)
+        data = traverse(path)
+
+        if data.is_a?(Symbol) && data.to_s =~ REDIRECT_REGEX
+          get_pattern_data(data.to_s.split('.').map(&:to_sym))
+        else
+          data
+        end
+      end
+
       def abbreviated?(type)
         ABBREVIATED_TYPES.include?(type)
       end
@@ -109,17 +134,43 @@ module TwitterCldr
       end
 
       def get_key_for(number)
-        "1#{"0" * (number.to_i.to_s.size - 1)}".to_i
+        "1#{"0" * (number.to_i.abs.to_s.size - 1)}".to_i
       end
 
-      def pattern_for_number(pattern, number)
+      def pattern_for_number(pattern, number, decimal)
         if pattern.is_a?(Hash)
-          pattern[get_key_for(number)] || pattern
+          range_pattern = pattern[get_key_for(number)]
+
+          if range_pattern
+            pattern_sample = range_pattern.values.first
+
+            if pattern_sample != 0
+              rule = pluralization_rule(number, pattern_sample, decimal)
+              # fall back to :other and raise an exception if it's missing as well
+              range_pattern.fetch(rule, range_pattern.fetch(:other))
+            else
+              0
+            end
+          else
+            pattern
+          end
         else
           pattern
         end
       end
 
+      def truncate_number(number, pattern)
+        formatter.truncate_number(number, tokenizer.tokenize(pattern)[1].value.size)
+      end
+
+      def pluralization_rule(number, pattern, decimal)
+        truncated_number = truncate_number(number, pattern)
+        # decimal and fractional numbers might have different pluralization,
+        # so it's important to convert to integer if we want a decimal result
+        truncated_number = truncated_number.to_i if decimal
+        TwitterCldr::Formatters::Plurals::Rules.rule_for(truncated_number, locale)
+      end
+
       def pattern_for_sign(pattern, sign)
         if pattern.include?(";")
           positive, negative = pattern.split(";")
@@ -127,7 +178,7 @@ module TwitterCldr
         else
           case sign
             when :negative
-              "#{symbols[:minus] || '-'}#{pattern}"
+              "#{symbols[:minus_sign] || '-'}#{pattern}"
             else
               pattern
           end
@@ -142,7 +193,8 @@ module TwitterCldr
       end
 
       def self.within_abbreviation_range?(number)
-        NUMBER_MIN <= number && number < NUMBER_MAX
+        abs_value = number.abs
+        NUMBER_MIN <= abs_value && abs_value < NUMBER_MAX
       end
     end
   end

data/lib/twitter_cldr/formatters/numbers/abbreviated/abbreviated_number_formatter.rb

@@ -7,18 +7,10 @@ module TwitterCldr
   module Formatters
     class AbbreviatedNumberFormatter < NumberFormatter
 
-      FORMAT_REGEX = /^0+$/
-
-      protected
-
-      def truncate_number(number, integer_format)
+      def truncate_number(number, decimal_digits)
         if TwitterCldr::DataReaders::NumberDataReader.within_abbreviation_range?(number)
-          if integer_format.format =~ FORMAT_REGEX
-            factor = [0, number.to_i.to_s.length - integer_format.format.length].max
-            number / (10.0 ** factor)
-          else
-            raise ArgumentError.new("unexpected format string #{integer_format.inspect}")
-          end
+          factor = [0, number.to_i.abs.to_s.length - decimal_digits].max
+          number / (10.0 ** factor)
         else
           number
         end

data/lib/twitter_cldr/formatters/numbers/number_formatter.rb

@@ -7,13 +7,6 @@ module TwitterCldr
   module Formatters
     class NumberFormatter < Formatter
 
-      DEFAULT_SYMBOLS = {
-        :group => ',',
-        :decimal => '.',
-        :plus_sign => '+',
-        :minus_sign => '-'
-      }
-
       attr_reader :data_reader
 
       def initialize(data_reader)
@@ -25,7 +18,7 @@ module TwitterCldr
         options[:type] ||= :decimal
 
         prefix, suffix, integer_format, fraction_format = *partition_tokens(tokens)
-        number = truncate_number(number, integer_format)
+        number = truncate_number(number, integer_format.format.length)
 
         int, fraction = parse_number(number, options)
         result =  integer_format.apply(int, options)
@@ -36,6 +29,10 @@ module TwitterCldr
         )
       end
 
+      def truncate_number(number, decimal_digits)
+        number # noop for base class
+      end
+
       protected
 
       # data readers should encapsulate formatting options, and when they do, this "type"
@@ -46,10 +43,6 @@ module TwitterCldr
         )
       end
 
-      def truncate_number(number, integer_format)
-        number  # noop for base class
-      end
-
       def partition_tokens(tokens)
         [
           token_val_from(tokens[0]),

data/lib/twitter_cldr/formatters/numbers/rbnf.rb

@@ -18,6 +18,7 @@ module TwitterCldr
       autoload :RuleGroup,    "twitter_cldr/formatters/numbers/rbnf/rule_group"
       autoload :RuleParser,   "twitter_cldr/formatters/numbers/rbnf/rule_parser"
       autoload :Substitution, "twitter_cldr/formatters/numbers/rbnf/substitution"
+      autoload :Plural,       "twitter_cldr/formatters/numbers/rbnf/plural"
 
       class PrivateRuleSetError < StandardError; end
 
@@ -100,7 +101,7 @@ module TwitterCldr
         def rule_set_from_resource(rule_set_data)
           RuleSet.new(
             rule_set_data[:rules].map do |rule|
-              Rule.new(rule[:value], rule[:rule], rule[:radix])
+              Rule.new(rule[:value], rule[:rule], rule[:radix], locale)
             end,
             rule_set_data[:type],
             rule_set_data[:access] || "public"

data/lib/twitter_cldr/formatters/numbers/rbnf/formatters.rb

@@ -16,7 +16,6 @@ module TwitterCldr
 
           def format(number, rule_set, rule_group, locale)
             rule = rule_set.rule_for(number)
-            # puts "#{number}, #{rule.base_value}: '" + rule.rule_text + "'"
             formatter = formatter_for(rule, rule_set, rule_group, locale)
             result = formatter.format(number, rule)
             keep_soft_hyphens ? result : remove_soft_hyphens(result)
@@ -120,6 +119,10 @@ module TwitterCldr
           ""
         end
 
+        def plural(number, rule, token)
+          token.render(number / rule.divisor)
+        end
+
         protected
 
         def invalid_token_error(token)

data/lib/twitter_cldr/formatters/numbers/rbnf/plural.rb

@@ -0,0 +1,49 @@
+# encoding: UTF-8
+
+# Copyright 2012 Twitter, Inc
+# http://www.apache.org/licenses/LICENSE-2.0
+
+module TwitterCldr
+  module Formatters
+    module Rbnf
+
+      class Plural
+        def self.from_string(locale, str)
+          # $(cardinal,one{ tysiąc}few{ tysiące}other{ tysięcy})$
+          plural_type, cases = str.gsub(/\$\((.*)\)\$/, '\1').split(',')
+
+          # one{ tysiąc}few{ tysiące}other{ tysięcy}
+          cases = cases.scan(/([\w]+)\{([^}]+)\}/).inject({}) do |ret, case_arr|
+            ret[case_arr.first.to_sym] = case_arr.last
+            ret
+          end
+
+          new(locale, plural_type.to_sym, cases)
+        end
+
+        attr_reader :locale, :plural_type, :cases
+
+        # plural_type = cardinal, etc
+        # cases = hash of form one: "foo", two: "bar"
+        def initialize(locale, plural_type, cases)
+          @locale = locale
+          @plural_type = plural_type
+          @cases = cases
+        end
+
+        def render(number)
+          rule_name = TwitterCldr::Formatters::Plurals::Rules.rule_for(
+            number, locale, plural_type
+          )
+
+          cases[rule_name] || cases[:other]
+        end
+
+        def type
+          :plural
+        end
+      end
+
+    end
+  end
+end

data/lib/twitter_cldr/formatters/numbers/rbnf/rule.rb

@@ -14,12 +14,13 @@ module TwitterCldr
         PROPER_FRACTION = "0.x"
         NEGATIVE = "-x"
 
-        attr_reader :base_value, :rule_text, :radix
+        attr_reader :base_value, :rule_text, :radix, :locale
 
-        def initialize(base_value, rule_text, radix)
+        def initialize(base_value, rule_text, radix, locale)
           @base_value = base_value
           @rule_text = rule_text
           @radix = radix
+          @locale = locale
         end
 
         def divisor
@@ -76,7 +77,7 @@ module TwitterCldr
         private
 
         def inline_substitutions(tokens)
-          parser.parse(tokens)
+          parser.parse(tokens, :locale => locale)
         end
 
         def parser

data/lib/twitter_cldr/formatters/numbers/rbnf/rule_parser.rb

@@ -12,6 +12,7 @@ module TwitterCldr
         private
 
         def do_parse(options)
+          @locale = options.fetch(:locale, TwitterCldr.locale)
           switch([])
         end
 
@@ -48,6 +49,16 @@ module TwitterCldr
           switch(list)
         end
 
+        def plural(list)
+          sub = Plural.from_string(
+            @locale, current_token.value
+          )
+
+          list << sub
+          next_token(:plural)
+          switch(list)
+        end
+
         def plaintext(list)
           add_and_advance(list)
         end
@@ -83,4 +94,4 @@ module TwitterCldr
       end
     end
   end
-end
+end

data/lib/twitter_cldr/formatters/numbers/rbnf/substitution.rb

@@ -27,4 +27,4 @@ module TwitterCldr
 
     end
   end
-end
+end

data/lib/twitter_cldr/formatters/plurals/rules.rb

@@ -3,6 +3,8 @@
 # Copyright 2012 Twitter, Inc
 # http://www.apache.org/licenses/LICENSE-2.0
 
+require 'cldr-plurals/ruby_runtime'
+
 module TwitterCldr
   module Formatters
     module Plurals
@@ -10,18 +12,18 @@ module TwitterCldr
 
         class << self
 
-          def all
-            all_for(TwitterCldr.locale)
+          DEFAULT_TYPE = :cardinal
+
+          def all(type = DEFAULT_TYPE)
+            all_for(TwitterCldr.locale, type)
           end
 
-          def all_for(locale)
-            get_resource(locale)[:keys]
-          rescue
-            nil
+          def all_for(locale, type = DEFAULT_TYPE)
+            names(locale, type)
           end
 
-          def rule_for(number, locale = TwitterCldr.locale)
-            get_resource(locale)[:rule].call(number)
+          def rule_for(number, locale = TwitterCldr.locale, type = DEFAULT_TYPE)
+            rule(locale, type).call(number.to_s, CldrPlurals::RubyRuntime)
           rescue
             :other
           end
@@ -31,7 +33,21 @@ module TwitterCldr
           def get_resource(locale)
             locale = TwitterCldr.convert_locale(locale)
             cache_key = TwitterCldr::Utils.compute_cache_key(locale)
-            locale_cache[cache_key] ||= eval(TwitterCldr.get_locale_resource(locale, :plurals)[locale])[locale][:i18n][:plural]
+            locale_cache[cache_key] ||= begin
+              rsrc = TwitterCldr.get_locale_resource(locale, :plurals)[locale]
+              rsrc.inject({}) do |ret, (rule_type, rule_data)|
+                ret[rule_type] = rule_data.merge(:rule => eval(rule_data[:rule]))
+                ret
+              end
+            end
+          end
+
+          def rule(locale, type)
+            get_resource(locale)[type][:rule]
+          end
+
+          def names(locale, type)
+            get_resource(locale)[type][:names]
           end
 
           def locale_cache
@@ -43,4 +59,4 @@ module TwitterCldr
       end
     end
   end
-end
+end