traduco 0.9.0

data/library/traduco/formatters.rb

@@ -0,0 +1,12 @@
+require 'traduco/formatters/abstract'
+require 'traduco/formatters/android'
+require 'traduco/formatters/apple'
+require 'traduco/formatters/flash'
+require 'traduco/formatters/gettext'
+require 'traduco/formatters/jquery'
+
+module Traduco
+  module Formatters
+    FORMATTERS = [Formatters::Apple, Formatters::Android, Formatters::Gettext, Formatters::JQuery, Formatters::Flash]
+  end
+end

data/library/traduco/formatters/abstract.rb

@@ -0,0 +1,164 @@
+module Traduco
+  module Formatters
+    class Abstract
+      attr_accessor :strings
+      attr_accessor :options
+
+      def self.can_handle_directory?(path)
+        return false
+      end
+
+      def initialize(strings, options)
+        @strings = strings
+        @options = options
+      end
+
+      def iosify_substitutions(str)
+        # 1) use "@" instead of "s" for substituting strings
+        str.gsub!(/%([0-9\$]*)s/, '%\1@')
+
+        # 2) if substitutions are numbered, see if we can remove the numbering safely
+        expectedSub = 1
+        startFound = false
+        foundSub = 0
+        str.each_char do |c|
+          if startFound
+            if c == "%"
+              # this is a literal %, keep moving
+              startFound = false
+            elsif c.match(/\d/)
+              foundSub *= 10
+              foundSub += Integer(c)
+            elsif c == "$"
+              if expectedSub == foundSub
+                # okay to keep going
+                startFound = false
+                expectedSub += 1
+              else
+                # the numbering appears to be important (or non-existent), leave it alone
+                return str
+              end
+            end
+          elsif c == "%"
+            startFound = true
+            foundSub = 0
+          end
+        end
+
+        # if we got this far, then the numbering (if any) is in order left-to-right and safe to remove
+        if expectedSub > 1
+          str.gsub!(/%\d+\$(.)/, '%\1')
+        end
+
+        return str
+      end
+
+      def androidify_substitutions(str)
+        # 1) use "s" instead of "@" for substituting strings
+        str.gsub!(/%([0-9\$]*)@/, '%\1s')
+
+        # 1a) escape strings that begin with a lone "@"
+        str.sub!(/^@ /, '\\@ ')
+
+        # 2) if there is more than one substitution in a string, make sure they are numbered
+        substituteCount = 0
+        startFound = false
+        str.each_char do |c|
+          if startFound
+            if c == "%"
+              # ignore as this is a literal %
+            elsif c.match(/\d/)
+              # leave the string alone if it already has numbered substitutions
+              return str
+            else
+              substituteCount += 1
+            end
+            startFound = false
+          elsif c == "%"
+            startFound = true
+          end
+        end
+
+        if substituteCount > 1
+          currentSub = 1
+          startFound = false
+          newstr = ""
+          str.each_char do |c|
+            if startFound
+              if !(c == "%")
+                newstr = newstr + "#{currentSub}$"
+                currentSub += 1
+              end
+              startFound = false
+            elsif c == "%"
+              startFound = true
+            end
+            newstr = newstr + c
+          end
+          return newstr
+        else
+          return str
+        end
+      end
+
+      def set_translation_for_key(key, lang, value)
+        if @strings.strings_map.include?(key)
+          @strings.strings_map[key].translations[lang] = value
+        elsif @options[:consume_all]
+          STDERR.puts "Adding new string '#{key}' to strings data file."
+          arr = @strings.sections.select { |s| s.name == 'Uncategorized' }
+          current_section = arr ? arr[0] : nil
+          if !current_section
+            current_section = StringsSection.new('Uncategorized')
+            @strings.sections.insert(0, current_section)
+          end
+          current_row = StringsRow.new(key)
+          current_section.rows << current_row
+          @strings.strings_map[key] = current_row
+          @strings.strings_map[key].translations[lang] = value
+        else
+          STDERR.puts "Warning: '#{key}' not found in strings data file."
+        end
+        if !@strings.language_codes.include?(lang)
+          @strings.add_language_code(lang)
+        end
+      end
+
+      def set_comment_for_key(key, comment)
+        if @strings.strings_map.include?(key)
+          @strings.strings_map[key].comment = comment
+        end
+      end
+
+      def default_file_name
+        raise NotImplementedError.new("You must implement default_file_name in your formatter class.")
+      end
+
+      def determine_language_given_path(path)
+        raise NotImplementedError.new("You must implement determine_language_given_path in your formatter class.")
+      end
+
+      def read_file(path, lang)
+        raise NotImplementedError.new("You must implement read_file in your formatter class.")
+      end
+
+      def write_file(path, lang)
+        raise NotImplementedError.new("You must implement write_file in your formatter class.")
+      end
+
+      def write_all_files(path)
+        if !File.directory?(path)
+          raise Traduco::Error.new("Directory does not exist: #{path}")
+        end
+
+        file_name = @options[:file_name] || default_file_name
+        Dir.foreach(path) do |item|
+          lang = determine_language_given_path(item)
+          if lang
+            write_file(File.join(path, item, file_name), lang)
+          end
+        end
+      end
+    end
+  end
+end

data/library/traduco/formatters/android.rb

@@ -0,0 +1,152 @@
+# encoding: utf-8
+require 'cgi'
+require 'rexml/document'
+
+module Traduco
+  module Formatters
+    class Android < Abstract
+      FORMAT_NAME = 'android'
+      EXTENSION = '.xml'
+      DEFAULT_FILE_NAME = 'strings.xml'
+      LANG_CODES = Hash[
+        'zh' => 'zh-Hans',
+        'zh-rCN' => 'zh-Hans',
+        'zh-rHK' => 'zh-Hant',
+        'en-rGB' => 'en-UK',
+        'in' => 'id',
+        'nb' => 'no'
+        # TODO: spanish
+      ]
+      DEFAULT_LANG_CODES = Hash[
+        'zh-TW' => 'zh-Hant' # if we don't have a zh-TW translation, try zh-Hant before en
+      ]
+
+      def self.can_handle_directory?(path)
+        Dir.entries(path).any? { |item| /^values.*$/.match(item) }
+      end
+
+      def default_file_name
+        return DEFAULT_FILE_NAME
+      end
+
+      def determine_language_given_path(path)
+        path_arr = path.split(File::SEPARATOR)
+        path_arr.each do |segment|
+          if segment == 'values'
+            return @strings.language_codes[0]
+          else
+            match = /^values-(.*)$/.match(segment)
+            if match
+              lang = match[1]
+              lang = LANG_CODES.fetch(lang, lang)
+              lang.sub!('-r', '-')
+              return lang
+            end
+          end
+        end
+
+        return
+      end
+
+      def read_file(path, lang)
+        resources_regex = /<resources>(.*)<\/resources>/m
+        key_regex = /<string name="(\w+)">/
+        comment_regex = /<!-- (.*) -->/
+        value_regex = /<string name="\w+">(.*)<\/string>/
+        key = nil
+        value = nil
+        comment = nil
+
+        File.open(path, 'r:UTF-8') do |f|
+          content_match = resources_regex.match(f.read)
+          if content_match
+            for line in content_match[1].split(/\r?\n/)
+              key_match = key_regex.match(line)
+              if key_match
+                key = key_match[1]
+                value_match = value_regex.match(line)
+                if value_match
+                  value = value_match[1]
+                  value = CGI.unescapeHTML(value)
+                  value.gsub!('\\\'', '\'')
+                  value.gsub!('\\"', '"')
+                  value = iosify_substitutions(value)
+                else
+                  value = ""
+                end
+                set_translation_for_key(key, lang, value)
+                if comment and comment.length > 0 and !comment.start_with?("SECTION:")
+                  set_comment_for_key(key, comment)
+                end
+                comment = nil
+              end
+
+              comment_match = comment_regex.match(line)
+              if comment_match
+                comment = comment_match[1]
+              end
+            end
+          end
+        end
+      end
+
+      def write_file(path, lang)
+        default_lang = nil
+        if DEFAULT_LANG_CODES.has_key?(lang)
+          default_lang = DEFAULT_LANG_CODES[lang]
+        end
+        File.open(path, 'w:UTF-8') do |f|
+          f.puts "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n<!-- Android Strings File -->\n<!-- Generated by Traduco #{Traduco::VERSION} -->\n<!-- Language: #{lang} -->"
+          f.write '<resources>'
+          @strings.sections.each do |section|
+            printed_section = false
+            section.rows.each do |row|
+              if row.matches_tags?(@options[:tags], @options[:untagged])
+                if !printed_section
+                  f.puts ''
+                  if section.name && section.name.length > 0
+                    section_name = section.name.gsub('--', '—')
+                    f.puts "\t<!-- SECTION: #{section_name} -->"
+                  end
+                  printed_section = true
+                end
+
+                key = row.key
+
+                value = row.translated_string_for_lang(lang, default_lang)
+                if !value && @options[:include_untranslated]
+                  value = row.translated_string_for_lang(@strings.language_codes[0])
+                end
+
+                if value # if values is nil, there was no appropriate translation, so let Android handle the defaulting
+                  value = String.new(value) # use a copy to prevent modifying the original
+
+                  # Android enforces the following rules on the values
+                  #  1) apostrophes and quotes must be escaped with a backslash
+                  value.gsub!('\'', '\\\\\'')
+                  value.gsub!('"', '\\\\"')
+                  #  2) HTML escape the string
+                  value = CGI.escapeHTML(value)
+                  #  3) fix substitutions (e.g. %s/%@)
+                  value = androidify_substitutions(value)
+
+                  comment = row.comment
+                  if comment
+                    comment = comment.gsub('--', '—')
+                  end
+
+                  if comment && comment.length > 0
+                    f.puts "\t<!-- #{comment} -->\n"
+                  end
+                  f.puts "\t<string name=\"#{key}\">#{value}</string>"
+                end
+              end
+            end
+          end
+
+          f.puts '</resources>'
+        end
+      end
+    end
+  end
+end

data/library/traduco/formatters/apple.rb

@@ -0,0 +1,125 @@
+module Traduco
+  module Formatters
+    class Apple < Abstract
+      FORMAT_NAME = 'apple'
+      EXTENSION = '.strings'
+      DEFAULT_FILE_NAME = 'Localizable.strings'
+
+      def self.can_handle_directory?(path)
+        Dir.entries(path).any? { |item| /^.+\.lproj$/.match(item) }
+      end
+
+      def default_file_name
+        return DEFAULT_FILE_NAME
+      end
+
+      def determine_language_given_path(path)
+        path_arr = path.split(File::SEPARATOR)
+        path_arr.each do |segment|
+          match = /^(.+)\.lproj$/.match(segment)
+          if match
+            return match[1]
+          end
+        end
+
+        return
+      end
+
+      def read_file(path, lang)
+        encoding = Traduco::Encoding.encoding_for_path(path)
+        sep = nil
+        if !encoding.respond_to?(:encode)
+          # This code is not necessary in 1.9.3 and does not work as it did in 1.8.7.
+          if encoding.end_with? 'LE'
+            sep = "\x0a\x00"
+          elsif encoding.end_with? 'BE'
+            sep = "\x00\x0a"
+          else
+            sep = "\n"
+          end
+        end
+
+        if encoding.index('UTF-16')
+          mode = "rb:#{encoding}"
+        else
+          mode = "r:#{encoding}"
+        end
+
+        File.open(path, mode) do |f|
+          last_comment = nil
+          while line = (sep) ? f.gets(sep) : f.gets
+            if encoding.index('UTF-16')
+              if line.respond_to? :encode!
+                line.encode!('UTF-8')
+              else
+                require 'iconv'
+                line = Iconv.iconv('UTF-8', encoding, line).join
+              end
+            end
+            match = /"((?:[^"\\]|\\.)+)"\s*=\s*"((?:[^"\\]|\\.)*)"/.match(line)
+            if match
+              key = match[1]
+              key.gsub!('\\"', '"')
+              value = match[2]
+              value.gsub!('\\"', '"')
+              value = iosify_substitutions(value)
+              set_translation_for_key(key, lang, value)
+              if last_comment
+                set_comment_for_key(key, last_comment)
+              end
+            end
+            if @options[:consume_comments]
+              match = /\/\* (.*) \*\//.match(line)
+              if match
+                last_comment = match[1]
+              else
+                last_comment = nil
+              end
+            end
+          end
+        end
+      end
+
+      def write_file(path, lang)
+        default_lang = @strings.language_codes[0]
+        encoding = @options[:output_encoding] || 'UTF-8'
+        File.open(path, "w:#{encoding}") do |f|
+          f.puts "/**\n * Apple Strings File\n * Generated by Traduco #{Traduco::VERSION}\n * Language: #{lang}\n */"
+          @strings.sections.each do |section|
+            printed_section = false
+            section.rows.each do |row|
+              if row.matches_tags?(@options[:tags], @options[:untagged])
+                f.puts ''
+                if !printed_section
+                  if section.name && section.name.length > 0
+                    f.print "/********** #{section.name} **********/\n\n"
+                  end
+                  printed_section = true
+                end
+
+                key = row.key
+                key = key.gsub('"', '\\\\"')
+
+                value = row.translated_string_for_lang(lang, default_lang)
+                if value
+                  value = value.gsub('"', '\\\\"')
+
+                  comment = row.comment
+                  if comment
+                    comment = comment.gsub('*/', '* /')
+                  end
+
+                  if comment && comment.length > 0
+                    f.print "/* #{comment} */\n"
+                  end
+
+                  f.print "\"#{key}\" = \"#{value}\";\n"
+                end
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end