Chris@1295: if RUBY_VERSION < '1.9'
Chris@1295:   require 'iconv'
Chris@1295: end
Chris@1295: 
Chris@1295: module Redmine
Chris@1295:   module CodesetUtil
Chris@1295: 
Chris@1295:     def self.replace_invalid_utf8(str)
Chris@1295:       return str if str.nil?
Chris@1295:       if str.respond_to?(:force_encoding)
Chris@1295:         str.force_encoding('UTF-8')
Chris@1295:         if ! str.valid_encoding?
Chris@1295:           str = str.encode("US-ASCII", :invalid => :replace,
Chris@1295:                 :undef => :replace, :replace => '?').encode("UTF-8")
Chris@1295:         end
Chris@1295:       elsif RUBY_PLATFORM == 'java'
Chris@1295:         begin
Chris@1295:           ic = Iconv.new('UTF-8', 'UTF-8')
Chris@1295:           str = ic.iconv(str)
Chris@1295:         rescue
Chris@1295:           str = str.gsub(%r{[^\r\n\t\x20-\x7e]}, '?')
Chris@1295:         end
Chris@1295:       else
Chris@1295:         ic = Iconv.new('UTF-8', 'UTF-8')
Chris@1295:         txtar = ""
Chris@1295:         begin
Chris@1295:           txtar += ic.iconv(str)
Chris@1295:         rescue Iconv::IllegalSequence
Chris@1295:           txtar += $!.success
Chris@1295:           str = '?' + $!.failed[1,$!.failed.length]
Chris@1295:           retry
Chris@1295:         rescue
Chris@1295:           txtar += $!.success
Chris@1295:         end
Chris@1295:         str = txtar
Chris@1295:       end
Chris@1295:       str
Chris@1295:     end
Chris@1295: 
Chris@1295:     def self.to_utf8(str, encoding)
Chris@1295:       return str if str.nil?
Chris@1295:       str.force_encoding("ASCII-8BIT") if str.respond_to?(:force_encoding)
Chris@1295:       if str.empty?
Chris@1295:         str.force_encoding("UTF-8") if str.respond_to?(:force_encoding)
Chris@1295:         return str
Chris@1295:       end
Chris@1295:       enc = encoding.blank? ? "UTF-8" : encoding
Chris@1295:       if str.respond_to?(:force_encoding)
Chris@1295:         if enc.upcase != "UTF-8"
Chris@1295:           str.force_encoding(enc)
Chris@1295:           str = str.encode("UTF-8", :invalid => :replace,
Chris@1295:                 :undef => :replace, :replace => '?')
Chris@1295:         else
Chris@1295:           str.force_encoding("UTF-8")
Chris@1295:           if ! str.valid_encoding?
Chris@1295:             str = str.encode("US-ASCII", :invalid => :replace,
Chris@1295:                   :undef => :replace, :replace => '?').encode("UTF-8")
Chris@1295:           end
Chris@1295:         end
Chris@1295:       elsif RUBY_PLATFORM == 'java'
Chris@1295:         begin
Chris@1295:           ic = Iconv.new('UTF-8', enc)
Chris@1295:           str = ic.iconv(str)
Chris@1295:         rescue
Chris@1295:           str = str.gsub(%r{[^\r\n\t\x20-\x7e]}, '?')
Chris@1295:         end
Chris@1295:       else
Chris@1295:         ic = Iconv.new('UTF-8', enc)
Chris@1295:         txtar = ""
Chris@1295:         begin
Chris@1295:           txtar += ic.iconv(str)
Chris@1295:         rescue Iconv::IllegalSequence
Chris@1295:           txtar += $!.success
Chris@1295:           str = '?' + $!.failed[1,$!.failed.length]
Chris@1295:           retry
Chris@1295:         rescue
Chris@1295:           txtar += $!.success
Chris@1295:         end
Chris@1295:         str = txtar
Chris@1295:       end
Chris@1295:       str
Chris@1295:     end
Chris@1295: 
Chris@1295:     def self.to_utf8_by_setting(str)
Chris@1295:       return str if str.nil?
Chris@1295:       str = self.to_utf8_by_setting_internal(str)
Chris@1295:       if str.respond_to?(:force_encoding)
Chris@1295:         str.force_encoding('UTF-8')
Chris@1295:       end
Chris@1295:       str
Chris@1295:     end
Chris@1295: 
Chris@1295:     def self.to_utf8_by_setting_internal(str)
Chris@1295:       return str if str.nil?
Chris@1295:       if str.respond_to?(:force_encoding)
Chris@1295:         str.force_encoding('ASCII-8BIT')
Chris@1295:       end
Chris@1295:       return str if str.empty?
Chris@1295:       return str if /\A[\r\n\t\x20-\x7e]*\Z/n.match(str) # for us-ascii
Chris@1295:       if str.respond_to?(:force_encoding)
Chris@1295:         str.force_encoding('UTF-8')
Chris@1295:       end
Chris@1295:       encodings = Setting.repositories_encodings.split(',').collect(&:strip)
Chris@1295:       encodings.each do |encoding|
Chris@1295:         if str.respond_to?(:force_encoding)
Chris@1295:           begin
Chris@1295:             str.force_encoding(encoding)
Chris@1295:             utf8 = str.encode('UTF-8')
Chris@1295:             return utf8 if utf8.valid_encoding?
Chris@1295:           rescue
Chris@1295:             # do nothing here and try the next encoding
Chris@1295:           end
Chris@1295:         else
Chris@1295:           begin
Chris@1295:             return Iconv.conv('UTF-8', encoding, str)
Chris@1295:           rescue Iconv::Failure
Chris@1295:             # do nothing here and try the next encoding
Chris@1295:           end
Chris@1295:         end
Chris@1295:       end
Chris@1295:       str = self.replace_invalid_utf8(str)
Chris@1295:       if str.respond_to?(:force_encoding)
Chris@1295:         str.force_encoding('UTF-8')
Chris@1295:       end
Chris@1295:       str
Chris@1295:     end
Chris@1295: 
Chris@1295:     def self.from_utf8(str, encoding)
Chris@1295:       str ||= ''
Chris@1295:       if str.respond_to?(:force_encoding)
Chris@1295:         str.force_encoding('UTF-8')
Chris@1295:         if encoding.upcase != 'UTF-8'
Chris@1295:           str = str.encode(encoding, :invalid => :replace,
Chris@1295:                            :undef => :replace, :replace => '?')
Chris@1295:         else
Chris@1295:           str = self.replace_invalid_utf8(str)
Chris@1295:         end
Chris@1295:       elsif RUBY_PLATFORM == 'java'
Chris@1295:         begin
Chris@1295:           ic = Iconv.new(encoding, 'UTF-8')
Chris@1295:           str = ic.iconv(str)
Chris@1295:         rescue
Chris@1295:           str = str.gsub(%r{[^\r\n\t\x20-\x7e]}, '?')
Chris@1295:         end
Chris@1295:       else
Chris@1295:         ic = Iconv.new(encoding, 'UTF-8')
Chris@1295:         txtar = ""
Chris@1295:         begin
Chris@1295:           txtar += ic.iconv(str)
Chris@1295:         rescue Iconv::IllegalSequence
Chris@1295:           txtar += $!.success
Chris@1295:           str = '?' + $!.failed[1, $!.failed.length]
Chris@1295:           retry
Chris@1295:         rescue
Chris@1295:           txtar += $!.success
Chris@1295:         end
Chris@1295:         str = txtar
Chris@1295:       end
Chris@1295:     end
Chris@1295:   end
Chris@1295: end