Check a list of domains from a text file. To see if they resolve.

dns_checker.py

from collections import OrderedDict
import socket
import sys
import re

def extract_first_hostname(text):
    """
    Extract the first hostname from a given string.
    
    Args:
        text (str): The input string to search for a hostname.
    
    Returns:
        str or None: The first hostname found, or None if no hostname is detected.
    """
    # Regular expression pattern to match hostnames
    # Matches typical hostname formats like:
    # - example.com
    # - www.example.com
    # - subdomain.example.co.uk
    hostname_pattern = r'\b((?:[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?\.)+[a-zA-Z]{2,6})\b'
    
    # Find the first match
    match = re.search(hostname_pattern, text)
    
    # Return the matched hostname or None
    return match.group(1) if match else None

def dns_lookup(url):
    try:
        result = socket.gethostbyname(url)
        return result
    except socket.gaierror:
        return None

def process_urls(input_file):
    domain_ip_map = OrderedDict()
    dead_domains = []

    with open(input_file, 'r') as f:
        for line in f:
            domain = extract_first_hostname(line.strip())
            if domain is None:
                continue

            ip_address = dns_lookup(domain)
            
            if ip_address:
                domain_ip_map[domain] = ip_address
                print(f"  {domain}: {ip_address}")
            else:
                dead_domains.append(domain)
                print(f"  - {domain}", file=sys.stderr)

    # print("Ordered Dictionary of Hosts with DNS Records:")
    # for url, ip in url_ip_map.items():
    #     print(f"  {url}: {ip}")
    
    # print("\nList of Dead Hosts:")
    # for domain in dead_domains:
    #     print(f"  - {domain}", file=sys.stderr)

    return domain_ip_map, dead_domains

# Usage
input_file = sys.argv[1] if len(sys.argv) > 1 else ""

url_ip_map, dead_domains = process_urls(input_file)

print(f"Number of valid domains: {len(url_ip_map)}")
print(f"Number of dead domains: {len(dead_domains)}", file=sys.stderr)