Transfer of python files, client-server

Question

Guys I have a problem in my code, I need to develop an application that is able to send a file to a server and also be able to recover files from the server I’m using a tuple to send the data to the server

nome = input('Digite o nome do arquivo a ser enviado: ')
arquivo = open(nome, 'r')
message = arquivo.read()
tupla = (opcao, message)

# Enviando ao servidor
socket_object.sendall(tupla)

But I’m having this mistake

TypeError: a bytes-like object is required, not 'tuple'

Is there any way to transform the tuple into bytes to send it through the socket?

PS: I am using the standard Ubuntu terminal python3

Answer 1

I suggest to transfer data structures use json or pickle, although with json the serialization can be faster.

That being said, this question unintentionally becomes a bit of a blatant/challenging.

The description of the error itself talks about the problem, that is, you need to send bytes and not a tuple, and how you cannot convert a tuple in bytes directly but can do it with strings, you can send a json (although in this case it becomes more laborious and so I will not do it, I explain below the reason).

However we still have a problem, you cannot send the bytes of the file to json because there are problems with serialization (list of the types of data that json understands), you’d have to send two separate messages, one with the adjacent information (name, file size etc. in the form of a Dict for example) and another with the bytes that compose it, and VERY IMPORTANT, you would have to send a flag (null byte for ex, b'\x00') at the end of the first to indicate that it has come to an end and that the server can then start receiving binary data from the file.

This flag must exist because TCP is a Protocol streaming there is no guarantee whatsoever that the method recv (server side in this case) received exactly the json with the information before, may contain only a portion of json or may include some of the binary information from the file to the mix.

Here I put a basic example of file upoload for the server with pickle, where we can convert a data structure (objects, dictionaries, tuples, etc...) into a data structure byte stream, so with pickle, we can immediately send the necessary data "at once", so in this case it makes it easier:

Client:

import socket, pickle

f = input('escreva o nome do ficheiro')
f_bin = open(f, 'rb').read()
info = {'name': f, 'file': f_bin, 'opcao': 1} # ajustar opcao para adequar ao teu codigo
with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
    s.connect(('', 9005))
    s.sendall(pickle.dumps(info))

Server:

import socket, threading, pickle, time

def run(conn):
    info_bin = b''
    st = time.time()
    while True:
        c = conn.recv(2048)
        if not c:
            break
        info_bin += c
        if time.time() - st >= 2: # opcional, informacao sobre o total ja carregado
            print('bytes downloaded:', len(info_bin))
            st = time.time()
    info = pickle.loads(info_bin)
    if info['file']:
        dest = 'files/{}'.format(info['name'])
        with open(dest, 'wb') as f:
            f.write(info['file'])
        print('success on receiving and saving {} for {}'.format(info['name'], conn.getpeername()))
        conn.close()

host, port = ('', 9005)
with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as sock:
    sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
    sock.bind((host, port))
    sock.listen(5)
    while True:
        conn, addr = sock.accept()
        print('conn', addr)
        threading.Thread(target=run, args=(conn,)).start()

I have not tested with multiple clients but in principle will also work.

You must adjust your connection and where the server will write the file