Amass output cleaner

amass2nuclei.py

subdomains = open('subdomains.txt', 'r').read()
removable_chars = [
    '', '[92m', '[0m', '[91m', '[93m', '[94m', '[95m', '[96m', '[97m', '[90m', '[1m', '[4m', '[5m', '[7m', '[8m',
    '[0;0m', '[0;1m', '[0;2m', '[0;3m', '[0;4m', '[0;5m', '[0;6m', '[0;7m', '[0;8m', '[0;9m', '[0;10m', '[0;11m',
    '[0;12m', '[0;13m', '[0;14m', '[0;15m', '[0;16m', '[0;17m', '[0;18m', '[0;19m', '[0;20m', '[0;21m', '[0;22m',
    '[0;23m', '[0;24m', '[0;25m', '[0;26m', '[0;27m', '[0;28m', '[0;29m', '[0;30m', '[0;31m', '[0;32m', '[0;33m',
    '[0;34m', '[0;35m', '[0;36m', '[0;37m', '[0;38m', '[0;39m', '[0;40m', '[0;41m', '[0;42m', '[0;43m', '[0;44m',
    '[0;45m', '[0;46m', '[0;47m', '[0;48m', '[0;49m', '[0;50m', '[0;51m', '[0;52m', '[0;53m', '[0;54m', '[0;55m',
    '[0;56m', '[0;57m', '[0;58m', '[0;59m', '[0;60m', '[0;61m', '[0;62m', '[0;63m', '[0;64m', '[0;65m', '[0;66m',
    '[0;67m', '[0;68m', '[0;69m', '[0;70m', '[0;71m', '[0;72m', '[0;73m',
]

# CHANGE THIS TO YOUR DOMAIN EXTENTION
domain_extention = '.id'

for char in removable_chars:
    subdomains = subdomains.replace(char, '')

clean_subdomains = []

for subdomain in subdomains.split('\n'):
    whitelists = [
        'a_record', 'aaaa_record',
        'cname_record',
    ]

    blacklist_subdomains = [
        'webmail', 'cpanel', 'whm', 'webdisk', 'autodiscover',
        'autoconfig', 'cpcalendars', 'cpcontacts', 'mail'
    ]

    if any(whitelist in subdomain for whitelist in whitelists):
        subdomain = subdomain.split(' --> ')[0].replace(' (FQDN)', '').replace('www.', '')
        if subdomain.endswith(domain_extention) and not any(blacklist_subdomain in subdomain for blacklist_subdomain in blacklist_subdomains):
            clean_subdomains.append(subdomain)

clean_subdomains = list(set(clean_subdomains))

with open('clean_subdomains.txt', 'w') as f:
    for subdomain in clean_subdomains:
        f.write(f'{subdomain}\n')

bash version grep -oP '(?<=--> )[a-zA-Z0-9.-]+(?= \(FQDN\))' amass_output.txt | sort -u