/*
* Validates the syntax of hostnames, ipv4-, and ipv6-addresses.
* Returns an empty string on success, non-empty on failure.
*
* TODO: Implement support for internationalized domain names.
*/
const std::string util::valid_addr(const std::experimental::string_view &addr)
{
if (!valid_ipv46_addr(addr)) {
if (string_contains(addr, ipv6_sep))
return "invalid ipv6 address.";
if (addr.length() > addr_max_length)
return "the address is too long; it's illegal to exceed 255 characters.";
/*
* Split the hostname into its multiple sub-domains
* (seperated by periods) and check them.
*/
std::vector<std::string> tokens = split_string(addr, period);
for (const std::experimental::string_view &s: tokens) {
/*
* In case of "irc..hostname.tld", where the token between '..',
* would be empty.
*/
if (s.empty())
return "a token is empty; does the address contain \"..\" somewhere?";
/* Also as per RFC 1035. */
if (s.front() == hyphen || s.back() == hyphen) {
std::string element = s.data();
return "first or last character in the element \"" +
element + "\" is a hyphen; that's not allowed.";
}
/*
* O̶n̶l̶y̶ ̶[̶A̶-̶Z̶a̶-̶z̶0̶-̶9̶]̶ ̶a̶r̶e̶ ̶a̶l̶l̶o̶w̶e̶d̶.̶
* Okey, no. International domain names is a thing:
* <https://en.wikipedia.org/wiki/Domain_name>
*
* s[first] and s[last] are checked again, but they
* must be checked for non-[A-Za-z0-9], anyway. A check for
* periods could be implemented here, but then we'd lose a specified
* error message and we'd need more checks than what we remove.
*/
if (std::find_if(s.begin(), s.end(),
[](char c) {
return !std::isdigit(c) &&
!std::isalpha(c) &&
c != hyphen; }) != s.end()) {
std::string element = s.data();
return "the element \"" + element +
"\" contains an illegal character (not a [A-Za-z0-9\\-]).";
}
}
}
return "";
}
