athena

athena.py

import psycopg2
import json
import re


# head -n -1 output.json > output.fixed.json
# tail -n +2 output.fixed.json > fixed.json

DATA_LOCATION = "results.cringe.large"
conn = psycopg2.connect(dbname="athena", user="alex", password="yoloswag", host="localhost", port="5432")

conn.set_client_encoding('UTF8')

cur = conn.cursor()

counter = 0


def get_banner(port_obj):
    if "service" in port_obj:
        if "banner" in port_obj["service"]:
            banner = port_obj["service"]["banner"]
            banner = banner[:2046]
            return re.sub(u'\x00', '', banner)
    return None


with open(DATA_LOCATION) as f:
    for line in f:
        counter += 1
        print('Iteration: %d'
              +
              format(counter, ",d"))

        host = json.loads(line)
        ip = host["ip"]
        banner = None
        port = 0  # Should always be overridden. If I find a zero in the DB there was some kind of error.

        for portObj in host["ports"]:
            banner = get_banner(portObj)
            port = portObj["port"]

        cur.execute(
            "INSERT INTO internet (ip, port, banner) "
            "VALUES (%s, %s, %s)",
            (ip, port, banner))

conn.commit()
cur.close()
conn.close()