File

local cache = require "util.cache";
local ip = require "util.ip";
local jid_bare = require "util.jid".bare;
local jid_host = require "util.jid".host;
local jid_split = require "util.jid".split;
local set = require "util.set";
local sha256 = require "util.hashes".sha256;
local st = require"util.stanza";
local is_contact_subscribed = require "core.rostermanager".is_contact_subscribed;
local full_sessions = prosody.full_sessions;

local user_exists = require "core.usermanager".user_exists;

local new_rtbl_subscription = module:require("rtbl").new_rtbl_subscription;
local trie = module:require("trie");
local pset = module:require("pset");

-- { [service_jid] = set, ... }
local spam_source_domains_by_service = {};
local spam_source_ips_by_service = {};
local spam_source_jids_by_service = {};

local service_probabilities = {
	-- if_present = probability the address is a spammer if they are on the list
	-- if_absent (optional): probability the address is a spammer if they are not on the list
	-- [service_jid] = { if_present = 0.9, if_absent = 0.5 };
};


-- These "probabilistic sets" combine the multiple lists according to their weights
local p_spam_source_domains = pset.new(spam_source_domains_by_service, service_probabilities);
local p_spam_source_ips = pset.new(spam_source_ips_by_service, service_probabilities);
local p_spam_source_jids = pset.new(spam_source_jids_by_service, service_probabilities);

local domain_local_report_threshold = module:get_option_number("anti_spam_local_report_threshold", 2);

local count_spam_blocked = module:metric("counter", "anti_spam_blocked", "stanzas", "Stanzas blocked as spam", {"reason"});

local hosts = prosody.hosts;

local reason_messages = {
	default = "Rejected as spam";
	["known-spam-source"] = "Rejected as spam. Your server is listed as a known source of spam. Please contact your server operator.";
};

function block_spam(event, reason, action)
	event.spam_reason = reason;
	event.spam_action = action;
	if module:fire_event("spam-blocked", event) == false then
		module:log("debug", "Spam allowed by another module");
		return;
	end

	count_spam_blocked:with_labels(reason):add(1);

	if action == "bounce" then
		module:log("debug", "Bouncing likely spam %s from %s (%s)", event.stanza.name, event.stanza.attr.from, reason);
		event.origin.send(st.error_reply("cancel", "policy-violation", reason_messages[reason] or reason_messages.default));
	else
		module:log("debug", "Discarding likely spam %s from %s (%s)", event.stanza.name, event.stanza.attr.from, reason);
	end

	return true;
end

function is_from_stranger(from_jid, event)
	local stanza = event.stanza;
	local to_user, to_host, to_resource = jid_split(stanza.attr.to);

	if not to_user then return false; end

	local to_session = full_sessions[stanza.attr.to];
	if to_session then return false; end

	if not is_contact_subscribed(to_user, to_host, from_jid) then
		-- Allow all messages from your own jid
		if from_jid == to_user.."@"..to_host then
			return false; -- Pass through
		end
		if to_resource and stanza.attr.type == "groupchat" then
			return false; -- Pass through
		end
		return true; -- Stranger danger
	end
end

function is_spammy_server(session)
	if p_spam_source_domains:contains(session.from_host) then
		return true;
	end
	local raw_ip = session.ip;
	local parsed_ip = raw_ip and ip.new_ip(session.ip);
	-- Not every session has an ip - for example, stanzas sent from a
	-- local host session
	if parsed_ip and p_spam_source_ips:contains_ip(parsed_ip) then
		return true;
	end
end

function is_spammy_sender(sender_jid)
	return p_spam_source_jids:contains(sha256(sender_jid, true));
end

local spammy_strings = module:get_option_array("anti_spam_block_strings");
local spammy_patterns = module:get_option_array("anti_spam_block_patterns");

function is_spammy_content(stanza)
	-- Only support message content
	if stanza.name ~= "message" then return; end
	if not (spammy_strings or spammy_patterns) then return; end

	local body = stanza:get_child_text("body");
	if not body then return; end

	if spammy_strings then
		for _, s in ipairs(spammy_strings) do
			if body:find(s, 1, true) then
				return true;
			end
		end
	end
	if spammy_patterns then
		for _, s in ipairs(spammy_patterns) do
			if body:find(s) then
				return true;
			end
		end
	end
end

-- Set up RTBLs

local anti_spam_services = module:get_option_array("anti_spam_services", {});

for _, rtbl_service_jid in ipairs(anti_spam_services) do
	service_probabilities[rtbl_service_jid] = { if_present = 0.95 };

	local spam_source_domains = set.new();
	local spam_source_ips = trie.new();
	local spam_source_jids = set.new();

	spam_source_domains_by_service[rtbl_service_jid] = spam_source_domains;
	spam_source_ips_by_service[rtbl_service_jid] = spam_source_ips;
	spam_source_jids_by_service[rtbl_service_jid] = spam_source_jids;

	new_rtbl_subscription(rtbl_service_jid, "spam_source_domains", {
		added = function (item)
			spam_source_domains:add(item);
		end;
		removed = function (item)
			spam_source_domains:remove(item);
		end;
	});
	new_rtbl_subscription(rtbl_service_jid, "spam_source_ips", {
		added = function (item)
			local subnet_ip, subnet_bits = ip.parse_cidr(item);
			if not subnet_ip then
				return;
			end
			spam_source_ips:add_subnet(subnet_ip, subnet_bits);
		end;
		removed = function (item)
			local subnet_ip, subnet_bits = ip.parse_cidr(item);
			if not subnet_ip then
				return;
			end
			spam_source_ips:remove_subnet(subnet_ip, subnet_bits);
		end;
	});
	new_rtbl_subscription(rtbl_service_jid, "spam_source_jids_sha256", {
		added = function (item)
			spam_source_jids:add(item);
		end;
		removed = function (item)
			spam_source_jids:remove(item);
		end;
	});
end

-- And local reports...

do
	local spam_source_domains = set.new();
	local spam_source_ips = set.new();

	local domain_counts = cache.new(100);

	service_probabilities[module.host] = { if_present = 0.6, if_absent = 0.4 };

	module:hook("mod_spam_reporting/spam-report", function (event)
		-- TODO: check for >= prosody:member
		local reported_jid = event.jid;
		local reported_domain = jid_host(reported_jid);
		local report_count = (domain_counts:get(reported_domain) or 0) + 1;
		domain_counts:set(reported_domain, report_count);

		if report_count >= domain_local_report_threshold then
			spam_source_domains:add(reported_domain);
		end
	end);

	module:add_item("shell-command", {
		section = "antispam";
		section_desc = "Anti-spam management commands";
		name = "filter_domain";
		desc = "Restrict interactions from a remote domain to a virtual host";
		args = {
			{ name = "host", type = "string" };
			{ name = "remote_domain", type = "string" };
		};
		host_selector = "host";
		handler = function(self, host, remote_domain) --luacheck: ignore 212/self 212/host
			spam_source_domains:add(remote_domain);
			return true, "Remote domain now restricted: "..remote_domain;
		end;
	});

	module:add_item("shell-command", {
		section = "antispam";
		section_desc = "Anti-spam management commands";
		name = "filter_ip";
		desc = "Restrict interactions from a remote IP/CIDR to a virtual host";
		args = {
			{ name = "host", type = "string" };
			{ name = "remote_ip", type = "string" };
		};
		host_selector = "host";
		handler = function(self, host, remote_ip) --luacheck: ignore 212/self 212/host
			local subnet_ip, subnet_bits = ip.parse_cidr(remote_ip);
			if not subnet_ip then
				return false, subnet_bits; -- false, err
			end

			spam_source_ips:add_subnet(subnet_ip, subnet_bits);

			return true, "Remote IP now restricted: "..remote_ip;
		end;
	});

end

module:hook("message/bare", function (event)
	local to_user, to_host = jid_split(event.stanza.attr.to);

	if not hosts[to_host] then
		module:log("warn", "Skipping filtering of message to unknown host <%s>", to_host);
		return;
	end

	if not user_exists(to_user, to_host) then return; end

	local from_bare = jid_bare(event.stanza.attr.from);
	if not is_from_stranger(from_bare, event) then return; end

	module:log("debug", "Processing message from stranger...");

	if is_spammy_server(event.origin) then
		return block_spam(event, "known-spam-source", "drop");
	end

	if is_spammy_sender(from_bare) then
		return block_spam(event, "known-spam-jid", "drop");
	end

	if is_spammy_content(event.stanza) then
		return block_spam(event, "spam-content", "drop");
	end

	module:log("debug", "Allowing message through");
end, 500);

module:hook("presence/bare", function (event)
	if event.stanza.attr.type ~= "subscribe" then
		return;
	end

	module:log("debug", "Processing subscription request...");

	if is_spammy_server(event.origin) then
		return block_spam(event, "known-spam-source", "drop");
	end

	module:log("debug", "Not from known spam source server");

	if is_spammy_sender(jid_bare(event.stanza.attr.from)) then
		return block_spam(event, "known-spam-jid", "drop");
	end

	module:log("debug", "Not from known spam source JID");

	module:log("debug", "Allowing subscription request through");
end, 500);