import subprocess
import re
import json
import ipaddress
from datetime import datetime
from sys import hash_info
import pprint

def run_traceroute(host):
    timestamp = datetime.now().timestamp()
    result = subprocess.run(['traceroute', host], stdout=subprocess.PIPE)
    return result.stdout.decode(), timestamp

def parse_traceroute_output(output, timestamp):

    lines = output.strip().split('\n')
    hops = []

    ip_regex = r"\((.*?)\)" # ipaddress are in ()

    target = output.strip().split('\n')[0].split()[2]

    for line in lines[1:]:
        hop = {}
        hop_info = line.split()
        hop_number = int(hop_info[0])
        hop_name = None
        hop_ip = None
        hop_latency = None

        latencies = []

        #print("##### "+str(hop_info))
        count = 0
        for part in hop_info[1:]:
            count += 1

            # source node drops or blocks icmp packages
            # We will give funny to name to hop for not answering and move on.
            if part == '*':
                hop_name = 'hop.'+str(count)+'.'+str(target)
                break


            # If first colum is either name or ip-address
            if count == 1:
                match = re.search(ip_regex, part)
                if  match:
                    hop_ip = part.strip('()')
                else:
                    hop_name = part

            # Second colum is ip-address first latency reading
            if count == 2:
                if re.search(ip_regex, part):
                    try:
                        _ip = ipaddress.ip_address(part.strip('()'))
                        hop_ip = part.strip('()')
                    except ValueError:
                        pass  # Ignore if it's not a valid IP address

            # Rest of the input colums are either latency floats, 'ms' or 
            # reruns of the hop_name and hop_ip...
            # We only need the latency floats anymore.
            else:
                try:
                    latency = float(part)
                    latencies.append(latency)
                except ValueError:
                    pass

        hop_latency = sum(latencies) / len(latencies) if latencies else None

        hop['timestamp'] = timestamp
        hop['hop_number'] = hop_number
        if not hop_name == None:
            hop['hop_name'] = hop_name
        hop['hop_ip'] = hop_ip
        hop['hop_latency'] = hop_latency
        

        hops.append(hop)

    return target, hops

if __name__ == '__main__':

    target='8.8.8.8'
    traceroute_output, timestamp = run_traceroute(target)
    
    target, hops = parse_traceroute_output(traceroute_output, timestamp)
    print('>> '+target)
    pprint.pprint(hops)