# SPDX-FileCopyrightText: 2026 Dennis Fink # # SPDX-License-Identifier: BSD-3-Clause """Validate URL schemes and publicly routable network destinations.""" import ipaddress from urllib.parse import urlsplit import dns.exception import dns.resolver class AddressResolutionError(Exception): pass def is_public_url(url: str) -> bool: """ Check whether a URL resolves exclusively to public IP addresses. :param url: URL whose hostname should be resolved. :return: Whether all resolved addresses are globally routable. :raises ValueError: If the URL does not contain a hostname. :raises AddressResolutionError: If the hostname cannot be resolved. """ hostname = urlsplit(url).hostname if hostname is None: raise ValueError("No hostname was specified") try: answers = dns.resolver.resolve_name(hostname) except dns.exception.DNSException as exc: raise AddressResolutionError( f"Could not resolve hostname {hostname!r}" ) from exc else: for address in answers.addresses(): if not ipaddress.ip_address(address).is_global: return False return True def is_http_url(url: str) -> bool: """ Check whether a URL is an absolute HTTP or HTTPS URL. :param url: URL to validate. :return: Whether the URL uses HTTP or HTTPS and contains a hostname. """ try: parsed = urlsplit(url) except ValueError: return False return parsed.scheme.lower() in {"http", "https"} and parsed.hostname is not None def is_safe_next_url(next_url: str) -> bool: """Allow redirects only to paths on this site.""" # Check the original string too: urlsplit("///evil.example").path is # "/evil.example", even though browsers can treat the original as external. if "\\" in next_url or not next_url.startswith("/") or next_url.startswith("//"): return False try: target = urlsplit(next_url) except ValueError: return False return ( not target.scheme and not target.netloc and target.path.startswith("/") and not target.path.startswith("//") )