From fd0986071c082170fa154ce6d7695805ad513515 Mon Sep 17 00:00:00 2001
From: Andrew Mayorov <encube.ul@gmail.com>
Date: Fri, 4 Aug 2023 19:27:43 +0400
Subject: [PATCH] perf(topicidx): implement fast-forwarding prefixes

This should give less `ets:next/2` calls in general and much less
when index has relatively small number of long non-wildcard topics.
---
 apps/emqx/src/emqx_topic_index.erl        | 20 ++++++++++++++++----
 apps/emqx/test/emqx_topic_index_SUITE.erl | 10 ++++++++++
 2 files changed, 26 insertions(+), 4 deletions(-)

diff --git a/apps/emqx/src/emqx_topic_index.erl b/apps/emqx/src/emqx_topic_index.erl
index 29d8694af..44e88e659 100644
--- a/apps/emqx/src/emqx_topic_index.erl
+++ b/apps/emqx/src/emqx_topic_index.erl
@@ -73,7 +73,13 @@ match(K, Prefix, Words, RPrefix, Tab) ->
             match_rest(Matched, Words, RPrefix, Tab)
     end.
 
-match_rest(false, [W | Rest], RPrefix, Tab) ->
+match_rest([W1 | [W2 | _] = SLast], [W1 | [W2 | _] = Rest], RPrefix, Tab) ->
+    % NOTE
+    % Fast-forward through identical words in the topic and the last key suffixes.
+    % This should save us a few redundant `ets:next` calls at the cost of slightly
+    % more complex match patterns.
+    match_rest(SLast, Rest, [W1 | RPrefix], Tab);
+match_rest(SLast, [W | Rest], RPrefix, Tab) when is_list(SLast) ->
     match(Rest, [W | RPrefix], Tab);
 match_rest(plus, [W | Rest], RPrefix, Tab) ->
     case match(Rest, ['+' | RPrefix], Tab) of
@@ -115,7 +121,13 @@ matches(K, Prefix, Words, RPrefix, Acc, Tab) ->
             matches_rest(Matched, Words, RPrefix, Acc, Tab)
     end.
 
-matches_rest(false, [W | Rest], RPrefix, Acc, Tab) ->
+matches_rest([W1 | [W2 | _] = SLast], [W1 | [W2 | _] = Rest], RPrefix, Acc, Tab) ->
+    % NOTE
+    % Fast-forward through identical words in the topic and the last key suffixes.
+    % This should save us a few redundant `ets:next` calls at the cost of slightly
+    % more complex match patterns.
+    matches_rest(SLast, Rest, [W1 | RPrefix], Acc, Tab);
+matches_rest(SLast, [W | Rest], RPrefix, Acc, Tab) when is_list(SLast) ->
     matches(Rest, [W | RPrefix], Acc, Tab);
 matches_rest(plus, [W | Rest], RPrefix, Acc, Tab) ->
     NAcc = matches(Rest, ['+' | RPrefix], Acc, Tab),
@@ -143,8 +155,8 @@ match_filter([], ['#'], _Suffix) ->
     true;
 match_filter([], ['+' | _], _Suffix) ->
     plus;
-match_filter([], [_H | _], _Suffix) ->
-    false;
+match_filter([], [_H | _] = Rest, _Suffix) ->
+    Rest;
 match_filter([H | T1], [H | T2], Suffix) ->
     match_filter(T1, T2, Suffix);
 match_filter([H1 | _], [H2 | _], _Suffix) when H2 > H1 ->
diff --git a/apps/emqx/test/emqx_topic_index_SUITE.erl b/apps/emqx/test/emqx_topic_index_SUITE.erl
index f2b263a5a..80ca536b4 100644
--- a/apps/emqx/test/emqx_topic_index_SUITE.erl
+++ b/apps/emqx/test/emqx_topic_index_SUITE.erl
@@ -133,6 +133,16 @@ t_match7(_) ->
     emqx_topic_index:insert(W, t_match7, <<>>, Tab),
     ?assertEqual(W, topic(match(T, Tab))).
 
+t_match_fast_forward(_) ->
+    Tab = emqx_topic_index:new(),
+    emqx_topic_index:insert(<<"a/b/1/2/3/4/5/6/7/8/9/#">>, id1, <<>>, Tab),
+    emqx_topic_index:insert(<<"z/y/x/+/+">>, id2, <<>>, Tab),
+    emqx_topic_index:insert(<<"a/b/c/+">>, id3, <<>>, Tab),
+    % dbg:tracer(),
+    % dbg:p(all, c),
+    % dbg:tpl({ets, next, '_'}, x),
+    ?assertEqual([id1], [id(M) || M <- matches(<<"a/b/1/2/3/4/5/6/7/8/9/0">>, Tab)]).
+
 t_match_unique(_) ->
     Tab = emqx_topic_index:new(),
     emqx_topic_index:insert(<<"a/b/c">>, t_match_id1, <<>>, Tab),