annotate mod_muc_rtbl/mod_muc_rtbl.lua @ 4813:0a257d1402c3

mod_muc_rtbl: Optimize case with zero hashes On the assumption that during quiet times between torrents of spam, the hash set would be empty. There would be no point in doing the operations and hashes to check for a match in that case.
author Kim Alvefur <zash@zash.se>
date Mon, 06 Dec 2021 18:19:19 +0100
parents 9cdbb1b5e6f2
children 460f78654864
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4808
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
1 local array = require "util.array";
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
2 local it = require "util.iterators";
4807
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
3 local jid = require "util.jid";
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
4 local sha256 = require "util.hashes".sha256;
4808
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
5 local set = require "util.set";
4807
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
6 local st = require "util.stanza";
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
7
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
8 local rtbl_service_jid = assert(module:get_option_string("muc_rtbl_jid"), "No RTBL JID supplied");
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
9 local rtbl_node = module:get_option_string("muc_rtbl_node", "muc_bans_sha256");
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
10
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
11 local banned_hashes = module:shared("banned_hashes");
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
12
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
13 module:depends("pubsub_subscription");
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
14
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
15 module:add_item("pubsub-subscription", {
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
16 service = rtbl_service_jid;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
17 node = rtbl_node;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
18
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
19 -- Callbacks:
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
20 on_subscribed = function()
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
21 module:log("info", "RTBL active");
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
22 end;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
23
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
24 on_error = function(err)
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
25 module:log("error", "Failed to subscribe to RTBL: %s::%s: %s", err.type, err.condition, err.text);
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
26 end;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
27
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
28 on_item = function(event)
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
29 local hash = event.item.attr.id;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
30 if not hash then return; end
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
31 module:log("debug", "Received new hash: %s", hash);
4808
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
32 banned_hashes[hash] = true;
4807
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
33 end;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
34
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
35 on_retract = function (event)
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
36 local hash = event.item.attr.id;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
37 if not hash then return; end
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
38 module:log("debug", "Retracted hash: %s", hash);
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
39 banned_hashes[hash] = nil;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
40 end;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
41 });
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
42
4808
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
43 function request_list()
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
44 local items_request = st.iq({ to = rtbl_service_jid, from = module.host, type = "get", id = "rtbl-request" })
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
45 :tag("pubsub", { xmlns = "http://jabber.org/protocol/pubsub" })
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
46 :tag("items", { node = rtbl_node }):up()
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
47 :up();
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
48
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
49 module:send(items_request);
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
50 end
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
51
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
52 function update_list(event)
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
53 local from_jid = event.stanza.attr.from;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
54 if from_jid ~= rtbl_service_jid then
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
55 module:log("debug", "Ignoring RTBL response from unknown sender");
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
56 return;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
57 end
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
58 local items_el = event.stanza:find("{http://jabber.org/protocol/pubsub}pubsub/items");
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
59 if not items_el then
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
60 module:log("warn", "Invalid items response from RTBL service");
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
61 return;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
62 end
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
63
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
64 local old_entries = set.new(array.collect(it.keys(banned_hashes)));
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
65
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
66 local n_added, n_removed, n_total = 0, 0, 0;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
67 for item in items_el:childtags("item") do
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
68 local hash = item.attr.id;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
69 if hash then
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
70 n_total = n_total + 1;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
71 if not old_entries:contains(hash) then
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
72 -- New entry
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
73 n_added = n_added + 1;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
74 banned_hashes[hash] = true;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
75 else
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
76 -- Entry already existed
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
77 old_entries:remove(hash);
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
78 end
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
79 end
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
80 end
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
81
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
82 -- Remove old entries that weren't in the received list
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
83 for hash in old_entries do
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
84 n_removed = n_removed + 1;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
85 banned_hashes[hash] = nil;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
86 end
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
87
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
88 module:log("info", "%d RTBL entries received from %s (%d added, %d removed)", n_total, from_jid, n_added, n_removed);
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
89 return true;
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
90 end
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
91
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
92 module:hook("iq-result/host/rtbl-request", update_list);
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
93
4807
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
94 module:hook("muc-occupant-pre-join", function (event)
4813
0a257d1402c3 mod_muc_rtbl: Optimize case with zero hashes
Kim Alvefur <zash@zash.se>
parents: 4812
diff changeset
95 if next(banned_hashes) == nil then return end
0a257d1402c3 mod_muc_rtbl: Optimize case with zero hashes
Kim Alvefur <zash@zash.se>
parents: 4812
diff changeset
96
4807
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
97 local from_bare = jid.bare(event.stanza.attr.from);
4810
181738ae4117 mod_muc_rtbl: Skip check if user has any explicit affiliation with the MUC
Matthew Wild <mwild1@gmail.com>
parents: 4809
diff changeset
98
181738ae4117 mod_muc_rtbl: Skip check if user has any explicit affiliation with the MUC
Matthew Wild <mwild1@gmail.com>
parents: 4809
diff changeset
99 local affiliation = event.room:get_affiliation(from_bare);
4811
a1fe59c06c48 mod_muc_rtbl: Fix typo in variable name in previous commit (thanks luacheck)
Matthew Wild <mwild1@gmail.com>
parents: 4810
diff changeset
100 if affiliation and affiliation ~= "none" then
4810
181738ae4117 mod_muc_rtbl: Skip check if user has any explicit affiliation with the MUC
Matthew Wild <mwild1@gmail.com>
parents: 4809
diff changeset
101 -- Skip check for affiliated users
181738ae4117 mod_muc_rtbl: Skip check if user has any explicit affiliation with the MUC
Matthew Wild <mwild1@gmail.com>
parents: 4809
diff changeset
102 return;
181738ae4117 mod_muc_rtbl: Skip check if user has any explicit affiliation with the MUC
Matthew Wild <mwild1@gmail.com>
parents: 4809
diff changeset
103 end
181738ae4117 mod_muc_rtbl: Skip check if user has any explicit affiliation with the MUC
Matthew Wild <mwild1@gmail.com>
parents: 4809
diff changeset
104
4812
9cdbb1b5e6f2 mod_muc_rtbl: Check joining user's host against RTBL
Matthew Wild <mwild1@gmail.com>
parents: 4811
diff changeset
105 local bare_hash = sha256(jid.bare(event.stanza.attr.from), true);
9cdbb1b5e6f2 mod_muc_rtbl: Check joining user's host against RTBL
Matthew Wild <mwild1@gmail.com>
parents: 4811
diff changeset
106 local host_hash = sha256(jid.host(event.stanza.attr.from), true);
9cdbb1b5e6f2 mod_muc_rtbl: Check joining user's host against RTBL
Matthew Wild <mwild1@gmail.com>
parents: 4811
diff changeset
107 if banned_hashes[bare_hash] or banned_hashes[host_hash] then
4807
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
108 module:log("info", "Blocked user <%s> from room <%s> due to RTBL match", from_bare, event.stanza.attr.to);
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
109 local error_reply = st.error_reply(event.stanza, "cancel", "forbidden", "You are banned from this service", event.room.jid);
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
110 event.origin.send(error_reply);
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
111 return true;
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
112 end
62a65c52c3f5 mod_muc_rtbl: Real-time blocklist checks for MUC services
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
113 end);
4808
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
114
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
115 if prosody.start_time then
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
116 request_list();
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
117 else
4809
9e9ec0f0b128 mod_muc_rtbl: Fix to hook server-started globally, to fetch entries at startup (thanks mirux)
Matthew Wild <mwild1@gmail.com>
parents: 4808
diff changeset
118 module:hook_global("server-started", function ()
9e9ec0f0b128 mod_muc_rtbl: Fix to hook server-started globally, to fetch entries at startup (thanks mirux)
Matthew Wild <mwild1@gmail.com>
parents: 4808
diff changeset
119 request_list();
9e9ec0f0b128 mod_muc_rtbl: Fix to hook server-started globally, to fetch entries at startup (thanks mirux)
Matthew Wild <mwild1@gmail.com>
parents: 4808
diff changeset
120 end);
4808
8a63a0daf129 mod_muc_rtbl: Sync existing list entries when first loaded
Matthew Wild <mwild1@gmail.com>
parents: 4807
diff changeset
121 end