import re
from urllib.parse import parse_qs, unquote, urlparse

import demistomock as demisto
from CommonServerPython import *
from tld import get_fld
from validate_email import validate_email

# ============================================================================================================== #
# This script is highly similar to 'ExtractFQDNFromUrlAndEmail', they are not unified due to run time performance.
# Please change both scripts respectively.
# ============================================================================================================== #

PROOFPOINT_PREFIXES = ["https://urldefense.proofpoint.com/v1/url?u=", "https://urldefense.proofpoint.com/v2/url?u="]
ATP_LINK_REG = r"(https:\/\/\w*|\w*)\.safelinks\.protection\.outlook\.com\/.*\?url="


def atp_get_original_url(safe_url):
    split_url = urlparse(safe_url)
    query = split_url.query
    query_dict = parse_qs(query)
    encoded_url_list = query_dict.get("url", [])
    encoded_url = encoded_url_list[0] if len(encoded_url_list) >= 1 else None
    if not encoded_url:
        error_msg = "Could not decode ATP Safe Link. Returning original URL."
        demisto.info(error_msg)
        return safe_url
    decoded_url = unquote(encoded_url)
    return decoded_url


def proofpoint_get_original_url(safe_url):
    regex = r"&.*$"
    split_url = urlparse(safe_url)
    query = split_url.query
    query_dict = parse_qs(query)
    encoded_url_list = query_dict.get("u", [])
    encoded_url = encoded_url_list[0] if len(encoded_url_list) >= 1 else None
    clean = encoded_url.replace("-", "%").replace("_", "/").replace(regex, "") if encoded_url else None
    clean = unquote(clean) if clean else None
    return clean


def unescape_url(escaped_url):
    # Normalize: 1) [.] --> . 2) hxxp --> http 3) &amp --> & 4) http:\\ --> http://
    url = escaped_url.lower().replace("[.]", ".").replace("hxxp", "http").replace("&", "&").replace("http:\\\\", "http://")
    # Normalize the URL with http prefix
    if url.find("http:") == 0 and url.find("http://") == -1:
        url = url.replace("http:", "http://")
    if url.find("http") != 0 and url.find("ftp") != 0:
        return "http://" + url
    return url


def extract_domain(the_input):
    is_url = None
    domain_from_mail = None
    is_email = validate_email(the_input)
    if is_email:
        # Take the entire part after the @ of the email
        domain_from_mail = the_input.split("@")[1]
    else:
        # Test if URL, else proceed as domain

        # Check if it is a Microsoft ATP Safe Link
        if re.match(ATP_LINK_REG, the_input):
            the_input = atp_get_original_url(the_input)
        # Check if it is a Proofpoint URL
        elif the_input.find(PROOFPOINT_PREFIXES[0]) == 0 or the_input.find(PROOFPOINT_PREFIXES[1]) == 0:
            the_input = proofpoint_get_original_url(the_input)
        # Not ATP Link or Proofpoint URL so just unescape
        else:
            the_input = unescape_url(the_input)
        is_url = domain = get_fld(the_input, fail_silently=True)

    # Extract domain itself from a potential subdomain
    if domain_from_mail or not is_url:
        full_domain = "https://"
        full_domain += domain_from_mail if domain_from_mail else the_input
        # get_tld fails to parse subdomain since it is not URL, over-ride error by injecting protocol.
        domain = get_fld(full_domain, fail_silently=True)

    # convert None to empty string if needed
    domain = domain if domain else ""
    return domain


def main():
    domains = []
    the_input = demisto.args().get("input")
    # argToList returns the argument as is if it's already a list so no need to check here
    the_input = argToList(the_input)

    # Otherwise assumes it's already an array
    for item in the_input:
        domains.append(extract_domain(item))
    demisto.results(domains)


# python2 uses __builtin__ python3 uses builtins
if __name__ == "__builtin__" or __name__ == "builtins":
    main()