Validate that emails are valid on SMTP server.

validate_emails.py

#!/usr/bin/env python3
# Dependencies python3 and dnspython
import os
import re
import sys
import csv
import time
import string
import random
import smtplib
import argparse

import dns.resolver

from dns.resolver import NXDOMAIN
from socket import gaierror


def valid_email_syntax(addressToVerify):
    # Simple Regex for syntax checking
    regex = "^[A-Za-z0-9\.\+_-]+@[A-Za-z0-9\._-]+\.[a-zA-Z]*$"
    return not (re.match(regex, addressToVerify) is None)


def verify_mailbox(addressToVerify):
    name = ''.join(random.choice(string.ascii_lowercase) for x in range(6))
    fromAddress = '{}@gmail.com'.format(name)
    catchall_servers = ['google.com', 'microsoft.com', 'adobe.com', 'humanlongevity.com']

    # Get domain for DNS lookup
    splitAddress = addressToVerify.split('@')
    domain = str(splitAddress[1])

    if domain in catchall_servers:
        return -2

    # MX record lookup
    try:
        answers = dns.resolver.query(domain, 'MX')
    except NXDOMAIN as e:
        return -1
    except Exception as e:
        print(e)
        return -666

    code = 0
    for rdata in sorted(answers):
        server = smtplib.SMTP()
        try:
            code, message = server.connect(rdata.exchange.to_text())
            if code != 220:
                # Could not connect
                continue

            server.ehlo_or_helo_if_needed()
            code, message = server.mail(fromAddress)
            if code != 250:
                # Could not create fake mail
                continue

            code, message = server.rcpt(addressToVerify)
            break

        except gaierror as e:
            # Cant connect to MX, try the next one
            continue
        except ConnectionRefusedError as e:
            # Cant connect to MX, try the next one
            continue
        except TimeoutError as e:
            # Cant connect to MX, try the next one
            continue
        except OSError as e:
            if e.errno == 101:
                # Can't find network, wait ant try the next MX
                time.sleep(1)
                continue
            else:
                raise
        except smtplib.SMTPServerDisconnected as e:
            # Disconnected from MX, try the next one
            time.sleep(1)
            continue
        server.quit()

    return code


def get_args():
    parser = argparse.ArgumentParser(description="Validate either an email list (coma separated no space), a file of emails (one per line) or a CMT3 invite file (.tsv).")
    parser.add_argument('emails')
    return parser.parse_args()

if __name__ == '__main__':
    args = get_args()

    # Load emails
    if '@' in args.emails:
        emails = args.emails.split(',')
    elif os.path.isfile(args.emails):

        with open(args.emails, 'r') as f:
            if args.emails.endswith('.tsv'):
                email_reader = csv.reader(f, dialect='excel-tab')
                emails = []
                for name, middle_name, last_name, email, instution in email_reader:
                    emails.append(email)
            else:
                emails = f.readlines()
    else:
        print("File ({}) does not exists.".format(args.emails))
        exit(1)

    # Validate emails
    for email in filter(str.strip, emails):
        email = email.lower().strip()

        if not valid_email_syntax(email):
            print("INVALID (syntax) - {}".format(email))
            continue

        code = verify_mailbox(email)
        if code == 250:
            print("OK - {}".format(email))
        elif code == 550:
            print("INVALID - {}".format(email))
        elif code == -1:
            print("INVALID (domain) - {}".format(email))
        elif code == 450 or code == 451:
            print("Unknown (greylisted) - {}".format(email))
        elif code == -2:
            print("Unknown (catchall) - {}".format(email))
        else:
            print("Unknown ({}) - {}".format(code, email))

        time.sleep(random.uniform(0, 5))