Python - use of 'self' - noob here going crazy trying to understand it.

Question

I have a simple socket class: MySocketLib.py ..

import socket

class socklib():
    def create_tcp_serversocket(self,port):
        serversocket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
        serversocket.bind((socket.gethostname(), port))
        serversocket.listen(5) 
        return serversocket

    def send_msg(self, sock, msg, msglen):
         ... blah blah ...  

    def recv_msg(self, sock, msglen):
         ... blah blah .... 

My main file is Server.py as follows:

import MySocketLib

class serverclass(): 

    def __init__(self,port):
        self.servsock = 0
        self.port = port

    def run(self):
        self.servsock = self.create_tcp_serversocket(self.port)
        (clientsock, address) = self.servsock.accept() # ERROR: 'SELF' not defined
        ... blah blah ... 

#############################
###   Main startup code

if __name__ == '__main__':
    server = serverclass(2000) # server is on port 2000
    server.run()

I keep getting the below error:

File "D:\CodeLearn\Server.py", line 14, in serverclass
(clientsock, address) = self.servsock.accept()
NameError: name 'self' is not defined

I can't seem to get a handle on the concept of "self". Some times, it seems to be needed and sometimes not. Why would i need to define 'self' here when it is a python keyword?

Answer 1

It maybe indentation error. File

"D:\CodeLearn\Server.py", line 14, in serverclass

This meen that

(clientsock, address) = self.servsock.accept()

is not inside run function, but in serverclass itself

Answer 2

Don't use tabs in Python source code. Configure your editor to always use spaces.

self is not a Python keyword, it's a convention. It's the usual name for the "instance" of a class you're using. Example:

class X:
    def __init__(self, v): self.v = v

a = X(1)
b = X(2)
print a.v, b.v

When this code runs, the Python runtime will eventually allocate memory for two instances of X which it will assign to a and b respectively. If there wasn't something like self, you would have to write:

a = X()
a.v = 1
b = X()
b.b = 2
print a.v, b.v

And you would get an error because you wrote b.b instead of b.v. Moreover, calling methods would be outright ugly:

class X:
    def set(v): ???.v = v

How would you say "access the reference v which was allocated in __init__ of X"?

One solution would be to pass the instance reference (the pointer to the memory which was allocated for X) as a parameter:

class X:
    def set(a, v): a.v = v

a.set(a, 1) # Holy ugly!

Everyone would use different names and we would all violate DRY and it would be a mess. So what Python did is:

class X:
    def set(self, v): self.v = v

a.set(1) # while "set" runs, "self" == "a"
b.set(2) # while "set" runs, "self" == "b"

That said, I have no idea why the code above fails. self is obviously defined or the error would already happen in the first line of your run() method. So my guess is that you mixed spaces and tabs for indentation and that confuses Python and it sees:

def run(self):
    self.servsock = self.create_tcp_serversocket(self.port)
(clientsock, address) = self.servsock.accept() # ERROR: 'SELF' not defined

Then it would try to evaluate self.servsock.accept() while it parses the class definition. At that time, there is no instance yet (the class doesn't exist!) so self is not available either.

Conclusion: Never mix tabs and spaces.

Answer 3

Your indentation is wrong, so __init__() and run() are being interpreted as top-level functions -- with no automatic self parameter. What you need is:

class serverclass:

    def __init__(self,port):
        self.servsock = 0
        self.port = port

    def run(self):
        self.servsock = self.create_tcp_serversocket(self.port)
        (clientsock, address) = self.servsock.accept() # ERROR: 'SELF' not defined
        ... blah blah ... 

Answer 4

Thanks everyone ... it was indeed the tabs and spacing issues. I use eclipse and in preferences I have set "Editor" preferences to use spaces when tabbing.