Python脚本，仅使用主要实体收集IP地址的所有主机名

Question

I have a Python script to collect hostnames of ip address with primes as byte entities. Eg, 211.13.17.2 is a valid ip according to my problem set where every byte entity(decimal representation) is a prime.

Code:

from itertools import product
import socket


# prime or not
def prime(n):
    if n > 1:
        p = 0
        for i in range(2, n-1):
            if divmod(n, i)[1] == 0:
                p = 1
                break
        if p == 0:
            return True


def get_host_name(b1, b2, b3, b4):
    addr = str(b1) + '.' + str(b2) + '.' + str(b3) + '.' + str(b4)
    try:
        return socket.gethostbyaddr(addr)
    except socket.herror:
        pass


# find host names whose ip addresses are all primes
byte = [b for b in range(0, 256) if prime(b)]
ips = list(product(byte, byte, byte, byte))
print 'Total ips = ', len(ips)

for ip in ips:
    if get_host_name(*ip):
        print get_host_name(*ip)

The problem is my script is too slow. I need expert help to optimize this code. Please pinpoint all mistakes and ways to make it behave faster.

Answer 1

for the prime numbers, you can use something like this,

import numpy as np
isprime = lambda x: np.all(np.mod(x, range(2, 1 + int(np.sqrt(x)))))
primes = np.array([ x for x in range(2, 255) if isprime(x) ])

and you can have a generator for ip addresses by

('{}.{}.{}.{}'.format(*x) for x in itertools.product(primes, repeat=4))

but most likely the code is slow in the socket part, and because of the number of combinations that it needs to check; for that you may try parallelism, by using a pool of worker processes; something like this:

from multiprocessing import Pool
from socket import gethostbyaddr

def gethost(addr):
    try:
        return gethostbyaddr(addr)
    except:
        pass

if __name__ == '__main__':

    p = Pool(3)
    print (p.map(gethost,['74.125.228.137',
                          '11.222.333.444',
                          '17.149.160.49',
                          '98.139.183.24']))

edit : for only prime numbers less than 50, (50K+ combinations) and 20 worker processes it takes almost 6 minutes on my machines and it finds 16K+ results. so, with this huge number of combinations parallelism cannot help much.