Python implementation of traceroute

traceroute.py

# references:
#   Learning by doing: Writing your own traceroute in 8 easy steps (Ksplice Blog)
#   https://blogs.oracle.com/ksplice/entry/learning_by_doing_writing_your

# Edited by: Alejandro Almira <laboral at alejandroalmira.com>

import datetime
import socket
import sys


def traceroute(hostname_or_address, max_hops=30, timeout=2):
    dest_addr = socket.gethostbyname(hostname_or_address)
    proto_icmp = socket.getprotobyname("icmp")
    proto_udp = socket.getprotobyname("udp")
    port = 33434

    for ttl in range(1, max_hops + 1):
        rx = socket.socket(socket.AF_INET, socket.SOCK_RAW, proto_icmp)
        rx.settimeout(timeout)
        rx.bind(("", port))
        tx = socket.socket(socket.AF_INET, socket.SOCK_DGRAM, proto_udp)
        tx.setsockopt(socket.SOL_IP, socket.IP_TTL, ttl)
        start = datetime.datetime.now()
        tx.sendto("".encode(), (dest_addr, port))

        try:
            _, curr_addr = rx.recvfrom(512)
            curr_addr = curr_addr[0]
        except socket.error:
            curr_addr = "*"
        finally:
            end = datetime.datetime.now()
            rx.close()
            tx.close()

        yield curr_addr, (end - start).microseconds

        if curr_addr == dest_addr:
            break


if __name__ == "__main__":
    dest_name = sys.argv[1]
    print(f"traceroute to {dest_name}")
    for i, v in enumerate(traceroute(dest_name)):
        print(f"{i+1}\t{v[0]}\t{v[1]}")

Neat. Here's another implementation that's a bit more robust and doesn't create a new socket connection in each loop:

https://rednafi.com/python/implement_traceroute_in_python/#writing-a-crappier-version-of-traceroute-in-python