whois.erl

%% Simple(?) WHOIS Client in Erlang
%% Copyright 2020 İlteriş Yağıztegin Eroğlu (linuxgemini) <new-iwhois@linuxgemini.space>
%% SPDX-License-Identifier: AGPL-3.0-or-later

-module(whois).
-export([query_iana/1,
	query_ripe/1,
	query_arin/1,
	query_afrinic/1,
	query_apnic/1,
	query_lacnic/1,
	query_radb/1,
	query_host/2,
	query_recursive/1]).

receive_data(Socket, SoFar) ->
	receive
		{tcp, Socket, Bin} ->
			receive_data(Socket, [Bin|SoFar]);
		{tcp_closed, Socket} ->
			list_to_binary(lists:reverse(SoFar))
	end.

string_clean(Str) ->
	Final = try re:replace(Str, "\r\n", "\n", [global, caseless]) of
		_ ->	re:replace(Str, "\r\n", "\n", [global, caseless])
	catch
		_:_ ->	""
	end,
	FinalTwo = unicode:characters_to_list(Final),
	FinalThree = string:trim(FinalTwo),
	FinalThree.

remove_dupes(List) ->
	Set = sets:from_list(List),
	sets:to_list(Set).

merge_refs(L, S) -> lists:append(L, S).

remove(X, L) ->
	[Y || Y <- L, Y =/= X].

regex_first_pass(S) ->
	R = re:replace(S, "^whois:|^refer:|^ReferralServer:", "", [caseless]),
	RTwo = re:replace(R, "Registrar WHOIS Server: ", "", [caseless]),
	string_clean(unicode:characters_to_list(RTwo)).

regex_second_pass(S) ->
	R = re:replace(S, "^whois:\/\/", "", [caseless]),
	RTwo = re:replace(R, ":[0-9]+$", "", [caseless]),
	string_clean(unicode:characters_to_list(RTwo)).

whois_string_clean(S) ->
	R = regex_first_pass(S),
	RTwo = regex_second_pass(R),
	RThree = case re:run(RTwo, "^rwhois:\/\/", [caseless, {capture, none}]) of
		match -> "";
		nomatch -> RTwo
	end,
	RThree.

get_refer(S) ->
	A = string_clean(S),
	R = case re:run(A, "^whois:|^refer:|^ReferralServer:", [caseless, {capture, none}]) of
		    match -> whois_string_clean(A);
		    nomatch ->
			    case re:run(A, "^Registrar WHOIS Server: ", [caseless, {capture, none}]) of
				    match -> whois_string_clean(A);
				    nomatch -> ""
			    end
	end,
	RTwo = case re:run(R, "^#|^%|:", [caseless, global, {capture, none}]) of
		       match -> "";
		       nomatch -> R
	end,
	RTwo.


query_host(QueryString, Host) ->
	{ok, Socket} = gen_tcp:connect(Host, 43,
				       [binary, {packet, 0}]),
	%% I hate quirks
	QS = case Host of
		"whois.ripe.net" -> string:concat(QueryString, " -B");
		"whois.arin.net" ->
			case re:run(QueryString, "^AS[0-9]+", [caseless, {capture, none}]) of
				match -> unicode:characters_to_list(re:replace(QueryString, "^AS", "a ", [caseless]));
				nomatch -> QueryString
			end;
		"whois.afrinic.net" -> string:concat(QueryString, " -B");
		_ -> QueryString
	end,
	ok = gen_tcp:send(Socket, string:concat(QS, "\r\n")),
	Result = receive_data(Socket, []),
	ResultTwo = binary_to_list(Result),
	ResultThree = unicode:characters_to_list(ResultTwo),
	string_clean(ResultThree).

query_host_print(QS, H) -> io:fwrite(string:concat(query_host(QS, H), "\n\n")).

just_print(R) -> io:fwrite(string:concat(R, "\n\n")).

get_proper_refs(R, H) ->
	RefsRaw = lists:map(fun(X) -> get_refer(X) end, R),
	RefProc = case item_exists("", RefsRaw) of
			  true -> ProcOne = remove("", RefsRaw),
				   remove([], ProcOne);
			  false -> remove([], RefsRaw)
		  end,
	RefFinal = case RefProc == [] of
			   true -> merge_refs([], [H]);
			   false -> RefProc
		   end,
	remove_dupes(RefFinal).

item_exists(I, L) -> lists:member(I, L).

query_recursive(QS, PrevRefs) ->
	H = lists:nth(1, PrevRefs),
	A = query_host(QS, H),

	just_print(A),

	Refs = get_proper_refs(string:split(A, "\n", all), H),
	IsServerListed = item_exists(H, Refs),

	case IsServerListed of
		true ->	ok;
		false ->
			AllRefs = merge_refs(Refs, remove_dupes(PrevRefs)),
			query_recursive(QS, AllRefs)
	end.

query_recursive(QS) ->
	H = "whois.iana.org",
	A = query_host(QS, H),
	just_print(A),
	Refs = get_proper_refs(string:split(A, "\n", all), H),
	query_recursive(QS, merge_refs(remove_dupes(Refs), [H])).

query_iana(QS) -> query_host_print(QS, "whois.iana.org").

query_ripe(QS) -> query_host_print(QS, "whois.ripe.net").

query_arin(QS) -> query_host_print(QS, "whois.arin.net").

query_afrinic(QS) -> query_host_print(QS, "whois.afrinic.net").

query_apnic(QS) -> query_host_print(QS, "whois.apnic.net").

query_lacnic(QS) -> query_host_print(QS, "whois.lacnic.net").

query_radb(QS) -> query_host_print(QS, "whois.radb.net").