i18n_data 0.2.4

data/i18n_data.gemspec

@@ -0,0 +1,212 @@
+# Generated by jeweler
+# DO NOT EDIT THIS FILE DIRECTLY
+# Instead, edit Jeweler::Tasks in Rakefile, and run the gemspec command
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = %q{i18n_data}
+  s.version = "0.2.4"
+
+  s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
+  s.authors = ["Michael Grosser"]
+  s.date = %q{2009-11-16}
+  s.email = %q{grosser.michael@gmail.com}
+  s.extra_rdoc_files = [
+    "README.markdown"
+  ]
+  s.files = [
+    ".gitignore",
+     "CHANGELOG",
+     "README.markdown",
+     "Rakefile",
+     "VERSION",
+     "cache/file_data_provider/countries-AF.txt",
+     "cache/file_data_provider/countries-AM.txt",
+     "cache/file_data_provider/countries-AR.txt",
+     "cache/file_data_provider/countries-AZ.txt",
+     "cache/file_data_provider/countries-BE.txt",
+     "cache/file_data_provider/countries-BG.txt",
+     "cache/file_data_provider/countries-BN.txt",
+     "cache/file_data_provider/countries-BR.txt",
+     "cache/file_data_provider/countries-BS.txt",
+     "cache/file_data_provider/countries-CA.txt",
+     "cache/file_data_provider/countries-CS.txt",
+     "cache/file_data_provider/countries-CY.txt",
+     "cache/file_data_provider/countries-DA.txt",
+     "cache/file_data_provider/countries-DE.txt",
+     "cache/file_data_provider/countries-DZ.txt",
+     "cache/file_data_provider/countries-EL.txt",
+     "cache/file_data_provider/countries-EN.txt",
+     "cache/file_data_provider/countries-EO.txt",
+     "cache/file_data_provider/countries-ES.txt",
+     "cache/file_data_provider/countries-ET.txt",
+     "cache/file_data_provider/countries-EU.txt",
+     "cache/file_data_provider/countries-FA.txt",
+     "cache/file_data_provider/countries-FI.txt",
+     "cache/file_data_provider/countries-FO.txt",
+     "cache/file_data_provider/countries-FR.txt",
+     "cache/file_data_provider/countries-GA.txt",
+     "cache/file_data_provider/countries-GL.txt",
+     "cache/file_data_provider/countries-GU.txt",
+     "cache/file_data_provider/countries-HE.txt",
+     "cache/file_data_provider/countries-HI.txt",
+     "cache/file_data_provider/countries-HR.txt",
+     "cache/file_data_provider/countries-HU.txt",
+     "cache/file_data_provider/countries-HY.txt",
+     "cache/file_data_provider/countries-ID.txt",
+     "cache/file_data_provider/countries-IS.txt",
+     "cache/file_data_provider/countries-IT.txt",
+     "cache/file_data_provider/countries-JA.txt",
+     "cache/file_data_provider/countries-KA.txt",
+     "cache/file_data_provider/countries-KK.txt",
+     "cache/file_data_provider/countries-KM.txt",
+     "cache/file_data_provider/countries-KN.txt",
+     "cache/file_data_provider/countries-KO.txt",
+     "cache/file_data_provider/countries-KU.txt",
+     "cache/file_data_provider/countries-LT.txt",
+     "cache/file_data_provider/countries-LV.txt",
+     "cache/file_data_provider/countries-MI.txt",
+     "cache/file_data_provider/countries-MK.txt",
+     "cache/file_data_provider/countries-ML.txt",
+     "cache/file_data_provider/countries-MN.txt",
+     "cache/file_data_provider/countries-MR.txt",
+     "cache/file_data_provider/countries-MS.txt",
+     "cache/file_data_provider/countries-MT.txt",
+     "cache/file_data_provider/countries-NB.txt",
+     "cache/file_data_provider/countries-NE.txt",
+     "cache/file_data_provider/countries-NL.txt",
+     "cache/file_data_provider/countries-NN.txt",
+     "cache/file_data_provider/countries-OC.txt",
+     "cache/file_data_provider/countries-PA.txt",
+     "cache/file_data_provider/countries-PL.txt",
+     "cache/file_data_provider/countries-PS.txt",
+     "cache/file_data_provider/countries-PT.txt",
+     "cache/file_data_provider/countries-RO.txt",
+     "cache/file_data_provider/countries-RU.txt",
+     "cache/file_data_provider/countries-RW.txt",
+     "cache/file_data_provider/countries-SK.txt",
+     "cache/file_data_provider/countries-SL.txt",
+     "cache/file_data_provider/countries-SO.txt",
+     "cache/file_data_provider/countries-SQ.txt",
+     "cache/file_data_provider/countries-SR.txt",
+     "cache/file_data_provider/countries-SV.txt",
+     "cache/file_data_provider/countries-SW.txt",
+     "cache/file_data_provider/countries-TA.txt",
+     "cache/file_data_provider/countries-TH.txt",
+     "cache/file_data_provider/countries-TI.txt",
+     "cache/file_data_provider/countries-TK.txt",
+     "cache/file_data_provider/countries-TL.txt",
+     "cache/file_data_provider/countries-TR.txt",
+     "cache/file_data_provider/countries-TT.txt",
+     "cache/file_data_provider/countries-UK.txt",
+     "cache/file_data_provider/countries-VE.txt",
+     "cache/file_data_provider/countries-VI.txt",
+     "cache/file_data_provider/countries-WA.txt",
+     "cache/file_data_provider/countries-WO.txt",
+     "cache/file_data_provider/countries-XH.txt",
+     "cache/file_data_provider/countries-ZU.txt",
+     "cache/file_data_provider/languages-AF.txt",
+     "cache/file_data_provider/languages-AM.txt",
+     "cache/file_data_provider/languages-AR.txt",
+     "cache/file_data_provider/languages-AZ.txt",
+     "cache/file_data_provider/languages-BE.txt",
+     "cache/file_data_provider/languages-BG.txt",
+     "cache/file_data_provider/languages-BR.txt",
+     "cache/file_data_provider/languages-BS.txt",
+     "cache/file_data_provider/languages-CA.txt",
+     "cache/file_data_provider/languages-CS.txt",
+     "cache/file_data_provider/languages-CY.txt",
+     "cache/file_data_provider/languages-DA.txt",
+     "cache/file_data_provider/languages-DE.txt",
+     "cache/file_data_provider/languages-EL.txt",
+     "cache/file_data_provider/languages-EN.txt",
+     "cache/file_data_provider/languages-EO.txt",
+     "cache/file_data_provider/languages-ES.txt",
+     "cache/file_data_provider/languages-ET.txt",
+     "cache/file_data_provider/languages-EU.txt",
+     "cache/file_data_provider/languages-FA.txt",
+     "cache/file_data_provider/languages-FI.txt",
+     "cache/file_data_provider/languages-FR.txt",
+     "cache/file_data_provider/languages-GA.txt",
+     "cache/file_data_provider/languages-GL.txt",
+     "cache/file_data_provider/languages-HE.txt",
+     "cache/file_data_provider/languages-HI.txt",
+     "cache/file_data_provider/languages-HR.txt",
+     "cache/file_data_provider/languages-HU.txt",
+     "cache/file_data_provider/languages-ID.txt",
+     "cache/file_data_provider/languages-IS.txt",
+     "cache/file_data_provider/languages-IT.txt",
+     "cache/file_data_provider/languages-JA.txt",
+     "cache/file_data_provider/languages-KO.txt",
+     "cache/file_data_provider/languages-LT.txt",
+     "cache/file_data_provider/languages-LV.txt",
+     "cache/file_data_provider/languages-MI.txt",
+     "cache/file_data_provider/languages-MK.txt",
+     "cache/file_data_provider/languages-MN.txt",
+     "cache/file_data_provider/languages-MR.txt",
+     "cache/file_data_provider/languages-MS.txt",
+     "cache/file_data_provider/languages-MT.txt",
+     "cache/file_data_provider/languages-NB.txt",
+     "cache/file_data_provider/languages-NL.txt",
+     "cache/file_data_provider/languages-NN.txt",
+     "cache/file_data_provider/languages-OC.txt",
+     "cache/file_data_provider/languages-PL.txt",
+     "cache/file_data_provider/languages-PS.txt",
+     "cache/file_data_provider/languages-PT.txt",
+     "cache/file_data_provider/languages-RO.txt",
+     "cache/file_data_provider/languages-RU.txt",
+     "cache/file_data_provider/languages-RW.txt",
+     "cache/file_data_provider/languages-SK.txt",
+     "cache/file_data_provider/languages-SL.txt",
+     "cache/file_data_provider/languages-SR.txt",
+     "cache/file_data_provider/languages-SV.txt",
+     "cache/file_data_provider/languages-TA.txt",
+     "cache/file_data_provider/languages-TH.txt",
+     "cache/file_data_provider/languages-TI.txt",
+     "cache/file_data_provider/languages-TR.txt",
+     "cache/file_data_provider/languages-TT.txt",
+     "cache/file_data_provider/languages-UK.txt",
+     "cache/file_data_provider/languages-VE.txt",
+     "cache/file_data_provider/languages-VI.txt",
+     "cache/file_data_provider/languages-WA.txt",
+     "cache/file_data_provider/languages-XH.txt",
+     "cache/file_data_provider/languages-ZU.txt",
+     "example_output/all_names_for_DE.txt",
+     "example_output/all_names_for_FR.txt",
+     "example_output/all_names_for_GB.txt",
+     "example_output/all_names_for_US.txt",
+     "i18n_data.gemspec",
+     "lib/i18n_data.rb",
+     "lib/i18n_data/file_data_provider.rb",
+     "lib/i18n_data/live_data_provider.rb",
+     "spec/i18n_data/file_data_provider_spec.rb",
+     "spec/i18n_data/live_data_provider_spec.rb",
+     "spec/i18n_data_spec.rb",
+     "spec/spec_helper.rb"
+  ]
+  s.homepage = %q{http://github.com/grosser/i18n_data}
+  s.rdoc_options = ["--charset=UTF-8"]
+  s.require_paths = ["lib"]
+  s.rubygems_version = %q{1.3.5}
+  s.summary = %q{country/language names and 2-letter-code pairs, in 85 languages}
+  s.test_files = [
+    "spec/spec_helper.rb",
+     "spec/i18n_data/file_data_provider_spec.rb",
+     "spec/i18n_data/live_data_provider_spec.rb",
+     "spec/i18n_data_spec.rb"
+  ]
+
+  if s.respond_to? :specification_version then
+    current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
+    s.specification_version = 3
+
+    if Gem::Version.new(Gem::RubyGemsVersion) >= Gem::Version.new('1.2.0') then
+      s.add_runtime_dependency(%q<activesupport>, [">= 0"])
+    else
+      s.add_dependency(%q<activesupport>, [">= 0"])
+    end
+  else
+    s.add_dependency(%q<activesupport>, [">= 0"])
+  end
+end
+

data/lib/i18n_data/file_data_provider.rb

@@ -0,0 +1,56 @@
+module I18nData
+  module FileDataProvider
+    DATA_SEPERATOR = ";;"
+    extend self
+
+    def codes(type,language_code)
+      unless data = read_from_file(type,language_code)
+        raise NoTranslationAvailable.new("#{type}-#{language_code}")
+      end
+      data
+    end
+
+    def write_cache(provider)
+      languages = provider.codes(:languages,'EN').keys
+      languages.each{|language_code|
+        [:languages,:countries].each {|type|
+          begin
+            data = provider.send(:codes,type,language_code)
+            write_to_file(data,type,language_code)
+          rescue NoTranslationAvailable
+          end
+        }
+      }
+    end
+
+  private
+
+    def read_from_file(type,language_code)
+      file = cache_for(type,language_code)
+      return nil unless File.exist?(file)
+      data = {}
+      IO.read(file).split("\n").each{|line|
+        code, translation = line.split(DATA_SEPERATOR)
+        data[code] = translation
+      }
+      data
+    end
+
+    def write_to_file(data,type,language_code)
+      return if data.empty?
+      file = cache_for(type,language_code)
+      FileUtils.mkdir_p(File.dirname(file))
+      File.open(file,'w') do |f|
+        f.puts data.map{|code,translation|"#{code}#{DATA_SEPERATOR}#{translation}"} * "\n"
+      end
+    end
+
+    def cache_for(type,language_code)
+      cache("#{type}-#{language_code}")
+    end
+
+    def cache(file)
+      File.join(File.dirname(__FILE__),'..','..','cache','file_data_provider',"#{file}.txt")
+    end
+  end
+end

data/lib/i18n_data/live_data_provider.rb

@@ -0,0 +1,97 @@
+require 'open-uri'
+require 'rexml/document'
+gem 'activesupport', '> 2.2'
+require 'activesupport'
+
+module I18nData
+  # fetches data online from debian svn
+  module LiveDataProvider
+    extend ActiveSupport::Memoizable
+    extend self
+    
+    XML_CODES = {
+      :countries => 'http://svn.debian.org/viewsvn/*checkout*/pkg-isocodes/trunk/iso-codes/iso_3166/iso_3166.xml',
+      :languages => 'http://svn.debian.org/viewsvn/*checkout*/pkg-isocodes/trunk/iso-codes/iso_639/iso_639.xml'
+    }
+    TRANSLATIONS = {
+      :languages => 'http://svn.debian.org/viewsvn/*checkout*/pkg-isocodes/trunk/iso-codes/iso_639/',
+      :countries => 'http://svn.debian.org/viewsvn/*checkout*/pkg-isocodes/trunk/iso-codes/iso_3166/'
+    }
+
+    def codes(type,language_code)
+      language_code = language_code.upcase
+      if language_code == 'EN'
+        send("english_#{type}")
+      else
+        translated(type,language_code)
+      end
+    end
+
+  private
+
+    def translate(type,language,to_language_code)
+      translated = translations(type,to_language_code)[language]
+      translated.to_s.empty? ? nil : translated
+    end
+
+    def translated(type,language_code)
+      translations = {}
+      send("english_#{type}").each do |code,name|
+        translation = translate(type,name,language_code)
+        translations[code] = translation ? translation : name
+      end
+      translations
+    end
+    memoize :translated
+
+    def translations(type,language_code)
+      begin
+        data = open(TRANSLATIONS[type]+"#{language_code.downcase}.po").readlines
+      rescue
+        raise NoTranslationAvailable.new("for #{type} and language code = #{language_code}")
+      end
+
+      po_to_hash(data)
+    end
+    memoize :translations
+
+    def english_languages
+      codes = {}
+      xml(:languages).elements.each('*/iso_639_entry') do |entry|
+        name = entry.attributes['name'].to_s.gsub("'","\\'")
+        code = entry.attributes['iso_639_1_code'].to_s.upcase
+        next if code.empty? or name.empty?
+        codes[code]=name
+      end
+      codes
+    end
+    memoize :english_languages
+
+    def english_countries
+      codes = {}
+      xml(:countries).elements.each('*/iso_3166_entry') do |entry|
+        name = entry.attributes['name'].to_s.gsub("'","\\'")
+        code = entry.attributes['alpha_2_code'].to_s.upcase
+        codes[code]=name
+      end
+      codes
+    end
+    memoize :english_countries
+
+    def po_to_hash(data)
+      names = data.select{|l| l =~ /^msgid/}.map{|line| line.match(/^msgid "(.*?)"/)[1]}
+      translations = data.select{|l| l =~ /^msgstr/}.map{|line| line.match(/^msgstr "(.*?)"/)[1]}
+
+      translated = {}
+      names.each_with_index do |name,index|
+        translated[name]=translations[index]
+      end
+      translated
+    end
+
+    def xml(type)
+      xml = open(XML_CODES[type]).read
+      REXML::Document.new(xml)
+    end
+  end
+end

data/lib/i18n_data.rb

@@ -0,0 +1,34 @@
+require 'activesupport'
+
+module I18nData
+  VERSION = File.read( File.join(File.dirname(__FILE__),'..','VERSION') ).strip
+  
+  extend self
+  
+  def languages(language_code='EN')
+    data_provider.codes(:languages,language_code.to_s.upcase)
+  end
+
+  def countries(language_code='EN')
+    data_provider.codes(:countries,language_code.to_s.upcase)
+  end
+
+  def data_provider
+    if @data_provider
+      @data_provider
+    else
+      require 'i18n_data/file_data_provider'
+      FileDataProvider
+    end
+  end
+
+  def data_provider=(provider)
+    @data_provider = provider
+  end
+  
+  class NoTranslationAvailable < Exception
+    def to_s
+      "NoTranslationAvailable -- #{super}"
+    end
+  end
+end

data/spec/i18n_data/file_data_provider_spec.rb

@@ -0,0 +1,23 @@
+require File.join(File.dirname(__FILE__),'..',"spec_helper")
+require 'i18n_data/file_data_provider'
+
+describe I18nData::FileDataProvider do
+  before do
+    `rm -f #{I18nData::FileDataProvider.send(:cache_for,"XX","YY")}`
+  end
+
+  def read(x,y)
+    I18nData::FileDataProvider.codes(x,y)
+  end
+
+  it "preserves data when writing and then reading" do
+    data = {"x"=>"y","z"=>"w"}
+    I18nData::FileDataProvider.send(:write_to_file,data,"XX","YY")
+    read("XX","YY").should == data
+  end
+
+  it "does not write empty data sets" do
+    I18nData::FileDataProvider.send(:write_to_file,{},"XX","YY")
+    lambda{read("XX","YY")}.should raise_error I18nData::NoTranslationAvailable
+  end
+end

data/spec/i18n_data/live_data_provider_spec.rb

@@ -0,0 +1,44 @@
+require File.join(File.dirname(__FILE__),'..',"spec_helper")
+require 'i18n_data/live_data_provider'
+
+describe I18nData::LiveDataProvider do
+  describe :po_to_hash do
+    def po_to_hash(text)
+      I18nData::LiveDataProvider.send(:po_to_hash,text)
+    end
+
+    it "parses po file into translations" do
+      text = <<EOF
+# come comment msgstr
+msgid "one"
+msgstr "1"
+EOF
+      po_to_hash(text).should == {"one"=>"1"}
+    end
+
+    it "keeps order of translations" do
+      text = <<EOF
+msgid "one"
+msgstr "1"
+msgid "two"
+msgstr ""
+msgid "three"
+msgstr "3"
+EOF
+      po_to_hash(text).should == {"one"=>"1","two"=>"","three"=>"3"}
+    end
+
+    it "finds x-line long translations" do
+      pending
+      text = <<EOF
+#. name for chu, cu
+msgid ""
+"Church Slavic; Old Slavonic; Church Slavonic; Old Bulgarian; Old Church "
+"Slavonic"
+msgstr "Kirchenslavisch"
+EOF
+      long_name = "Church Slavic; Old Slavonic; Church Slavonic; Old Bulgarian; Old Church Slavonic"
+      po_to_hash(text)[long_name].should == "Kirchenslavisch"
+    end
+  end
+end

data/spec/i18n_data_spec.rb

@@ -0,0 +1,101 @@
+require File.expand_path("spec_helper", File.dirname(__FILE__))
+
+NUM_2_LETTER_LANGUAGES = 185
+NUM_COUNTRIES = 246
+
+describe I18nData do
+  require "i18n_data/live_data_provider"
+  require "i18n_data/file_data_provider"
+  
+  [I18nData::LiveDataProvider,I18nData::FileDataProvider].each do |provider|
+    describe "using #{provider}" do
+      before :all do
+        I18nData.data_provider = provider
+      end
+
+      describe :languages do
+        it "raises NoTranslationAvailable for unavailable languages" do
+          lambda{I18nData.languages('XX')}.should raise_error(I18nData::NoTranslationAvailable)
+        end
+
+        describe :english do
+          it "does not contain blanks" do
+            I18nData.languages.detect {|k,v| k.blank? or v.blank?}.should == nil
+          end
+
+          it "has english as default" do
+            I18nData.languages['DE'].should == 'German'
+          end
+
+          it "contains all languages" do
+            I18nData.languages.size.should == NUM_2_LETTER_LANGUAGES
+          end
+        end
+
+        describe :translated do
+          it "is translated" do
+            I18nData.languages('DE')['DE'].should == 'Deutsch'
+          end
+
+          it "contains all languages" do
+            I18nData.languages('DE').size.should == NUM_2_LETTER_LANGUAGES
+          end
+
+          it "has english names for not-translateable languages" do
+            I18nData.languages('IS')['HA'].should == I18nData.languages['HA']
+          end
+
+          it "does not contain blanks" do
+            I18nData.languages('GL').detect {|k,v| k.blank? or v.blank?}.should == nil
+          end
+
+          it "is written in unicode" do
+            I18nData.languages('DE')['DA'].should == 'Dänisch'
+          end
+        end
+      end
+
+      describe :countries do
+        describe :english do
+          it "has english as default" do
+            I18nData.countries['DE'].should == 'Germany'
+          end
+
+          it "does not contain blanks" do
+            I18nData.countries.detect {|k,v| k.blank? or v.blank?}.should == nil
+          end
+
+          it "contains all countries" do
+            I18nData.countries.size.should == NUM_COUNTRIES
+          end
+        end
+
+        describe :translated do
+          it "is translated" do
+            I18nData.countries('DE')['DE'].should == 'Deutschland'
+          end
+
+          it "contains all countries" do
+            I18nData.countries('DE').size.should == NUM_COUNTRIES
+          end
+
+          it "has english names for not-translateable countries" do
+            I18nData.countries('IS')['PK'].should == I18nData.countries['PK']
+          end
+
+          it "does not contain blanks" do
+            I18nData.countries('GL').detect {|k,v| k.blank? or v.blank?}.should == nil
+          end
+
+          it "is written in unicode" do
+            I18nData.countries('DE')['DK'].should == 'Dänemark'
+          end
+        end
+      end
+    end
+  end
+
+  it "has a VERSION" do
+    I18nData::VERSION.should =~ /^\d+\.\d+\.\d+$/
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,2 @@
+$LOAD_PATH << 'lib'
+require 'i18n_data'