TidyBytesInRequest Rack Middleware

tidy_bytes_in_request.rb

require "active_support/core_ext/string/multibyte"

class TidyBytesInRequest

  TIDY_BYTES_ENV_KEYS = %w(
    HTTP_USER_AGENT
    HTTP_REFERER
    PATH_INFO
    QUERY_STRING
    REQUEST_PATH
    REQUEST_URI
    HTTP_COOKIE
  ).freeze

  URI_ENCODABLE_ENV_KEYS = %w(
    HTTP_REFERER
    PATH_INFO
    QUERY_STRING
    REQUEST_PATH
    REQUEST_URI
    HTTP_COOKIE
  ).freeze

  def initialize(app)
    @app = app
  end

  def call(given_env)
    @app.call(tidy_bytes_in_env given_env)
  end

  private

    def tidy_bytes_in_env(given_env)
      tidy_bytes_in_env_keys given_env
      tidy_bytes_in_rack_input given_env
      given_env
    end

    def tidy_bytes_in_env_keys(given_env)
      TIDY_BYTES_ENV_KEYS.each do |key|
        next unless value = given_env[key]

        # Tidy Bytes unless value contains ascii caracters only:
        given_env[key] = unless value.ascii_only?

          tidied_value = value.mb_chars.tidy_bytes.to_s

          # URI encode the tidied_value, to avoid Rack::Utils complaining about
          # a regexp match against a UTF-8 string:

          if URI_ENCODABLE_ENV_KEYS.include?(key)
            URI.encode tidied_value
          else
            tidied_value
          end

        else
          value
        end
      end
    end

    def tidy_bytes_in_rack_input(given_env)
      case given_env['CONTENT_TYPE']
      when 'application/x-www-form-urlencoded'

        value = given_env['rack.input'].read

        # Tidy Bytes unless the rack input contains ascii caracters only:
        unless value.ascii_only?

          tidied_value = value.mb_chars.tidy_bytes.to_s

          # URI encode the tidied_value, to avoid Rack::Utils complaining about
          # a regexp match against a UTF-8 string:
          given_env['rack.input'] = StringIO.new(URI.encode tidied_value)
        end

        # Rewind the string IO:
        given_env['rack.input'].rewind
      when 'multipart/form-data'
        # Don't process the data since it may contain binary content
      else
        # Unknown content type. Leave it alone
      end
    end

end