gistfile1.rb

require 'socket'
require 'webrick/httprequest'
require 'stringio'

# Super-primitive transparent, recording, single-threaded HTTP Proxy
class TransparentHttpProxy
  attr_reader :port

  def initialize(port, recorder)
    @port = port
    @recorder = recorder
  end

  def start
    proxy_socket = TCPServer.open('0.0.0.0', @port)
    $stderr.puts "#{self.class.name} started on #{@port}"
    loop {
      cycle = ProxyCycle.new(proxy_socket.accept).start
      @recorder.call(cycle.request_text, cycle.response_text)
    }
  end

  class ProxyCycle
    attr_accessor :request_text, :response_text

    def initialize(client)
      @client = client
    end

    def start
      read_request
      propagate_request
      propagate_response
      self
    end

    def request
      @request ||= begin
        r = WEBrick::HTTPRequest.new({})
        r.parse(StringIO.new(request_text))
        r
      end
    end

    private

    def debug(msg=nil)
      if ENV['VERBOSE']
        if block_given?
          $stderr.puts yield
        else
          $stderr.puts msg
        end
      end
    end

    def log_lines(heading, http_message)
      heading + http_message.gsub(/\r\n/, "\r\n#{heading}")
    end

    def read_request
      debug "read_request"
      @request_text = ""
      until @request_text =~ /\r\n\r\n/ || (line = @client.gets).nil?
        @request_text += line
      end
      read_entity_if_necessary
      debug { log_lines("read_request: ", request_text) }
    end

    def read_entity_if_necessary
      if @request_text =~ /Transfer-Encoding:/
        msg = "Lucky you: you get to implement Transfer-Encoding in #{__FILE__}!"
        $stderr.puts '*' * msg.length
        $stderr.puts msg
        $stderr.puts "This request will fail."
        $stderr.puts '*' * msg.length
        # If you are the lucky winner, see section 4.4 of RFC 2616.
      elsif @request_text =~ /Content-Length:\s+(\d+)/
        content_length = $1.to_i
        debug { "Attempting to read #{content_length} bytes" }
        @request_text += @client.read(content_length)
      end
    end

    def propagate_request
      debug { "propagate_request to #{request.host}:#{request.port}" }
      http = TCPSocket.open(request.host, request.port)
      debug { log_lines("corrected_request_text: ", corrected_request_text) }
      http << corrected_request_text
      @response_text = ""
      while line = http.gets
        debug { "recvd: #{line}" }
        @response_text += line
      end
      http.close
    end

    def propagate_response
      debug "propagate_response"
      debug { log_lines("response_text: ", response_text) }
      @client << response_text
      @client.close
    end

    # Remove & replace Connection headers, if any
    def corrected_request_text
      request_text.
        gsub(/^Proxy-Connection:.*\r\n/, ''). # don't care about current value -- always close
        gsub(/^Connection:.*\r\n/, ''). # don't care about current value -- always close
        sub(/\r\n\r\n/, "\r\nConnection: close\r\n\r\n") # add Connection: close as last header
    end
  end
end