1. Telnet & SSH

Since the basic functionality offered by these two protocols is the same, I decided to write a common backend for both. Designing this meant that I needed a common high-level API for both the Bees, which would allow me to handle the data flow in an easy way, using a common base class.

Another interesting change that I made was emulating user keystrokes on the Telnet Bee. Using the timing data from this paper, along with Python's great random library, I was able to generate appropriate intervals and emulate the general typing characteristics of a person. This was done in by overriding the write() method of the telnetlib.Telnet class.

def write_human(self, buffer_):
    """ Emulates human typing speed """

    if self.IAC in buffer_:
        buffer_ = buffer_.replace(self.IAC, self.IAC+self.IAC)
    self.msg("send %r", buffer_)
    for char in buffer_:
        delta = random.gauss(80, 20)
        self.sock.sendall(char)
        time.sleep(delta/1000.0) # Convert milliseconds to seconds

In order to make the SDA loop realistic, I created a simple data structure, a sort of mapping, which would help the Bee choose the next command to execute. Something like this:

COMMAND_MAP = {
    'pwd': ['ls', 'uname', 'uptime'],
    'cd': ['ls'],
    'uname': ['uptime', 'ls'],
    'ls': ['cd', 'cat', 'pwd'],
    'cat': ['ls', 'echo', 'sudo', 'pwd'],
    'uptime': ['ls', 'echo', 'sudo', 'uname', 'pwd'],
    'echo': ['ls', 'sudo', 'uname', 'pwd'],
    'sudo': ['logout']
}

Here, the RHS for every command contains commands that can be run after it. So, in general, a cd should be followed by an ls, and so on.

Along with these changes, I also improved and refactored the code a lot. This time, we also have test cases for each command, and I'm beginning to see the benefits of Test-Driven-Development. Next week will be dedicated to the rest of the Bees. Now that I have a clear idea of what needs to be done, I hope I'll be able to finish the work on all of them.