fix(topicidx): allow to return matches unique by record id

This commit is contained in:
Andrew Mayorov 2023-07-21 20:06:46 +02:00 committed by JianBo He
parent 3304e32ea1
commit 6432c9c8fc
2 changed files with 109 additions and 77 deletions

View File

@ -16,29 +16,29 @@
%% @doc Topic index for matching topics to topic filters.
%%
%% Works on top of ETS ordered_set table. Keys are parsed topic filters
%% with record ID appended to the end, wrapped in a tuple to disambiguate from
%% topic filter words. Existing table may be used if existing keys will not
%% collide with index keys.
%% Works on top of ETS ordered_set table. Keys are tuples constructed from
%% parsed topic filters and record IDs, wrapped in a tuple to order them
%% strictly greater than unit tuple (`{}`). Existing table may be used if
%% existing keys will not collide with index keys.
%%
%% Designed to effectively answer questions like:
%% 1. Does any topic filter match given topic?
%% 2. Which records are associated with topic filters matching given topic?
%%
%% Questions like these are _only slightly_ less effective:
%% 1. Which topic filters match given topic?
%% 2. Which record IDs are associated with topic filters matching given topic?
%% 3. Which topic filters match given topic?
%% 4. Which record IDs are associated with topic filters matching given topic?
-module(emqx_rule_index).
-export([insert/4]).
-export([delete/3]).
-export([match/2]).
-export([matches/2]).
-export([matches/3]).
-export([get_id/1]).
-export([get_topic/1]).
-export([get_record/2]).
-type key(ID) :: [binary() | '+' | '#' | {ID}].
-type key(ID) :: {[binary() | '+' | '#'], {ID}}.
-type match(ID) :: key(ID).
-ifdef(TEST).
@ -46,11 +46,10 @@
-endif.
insert(Filter, ID, Record, Tab) ->
%% TODO: topic compact. see also in emqx_trie.erl
ets:insert(Tab, {emqx_topic:words(Filter) ++ [{ID}], Record}).
ets:insert(Tab, {{emqx_topic:words(Filter), {ID}}, Record}).
delete(Filter, ID, Tab) ->
ets:delete(Tab, emqx_topic:words(Filter) ++ [{ID}]).
ets:delete(Tab, {emqx_topic:words(Filter), {ID}}).
-spec match(emqx_types:topic(), ets:table()) -> match(_ID) | false.
match(Topic, Tab) ->
@ -59,8 +58,8 @@ match(Topic, Tab) ->
match(Words, RPrefix, Tab) ->
Prefix = lists:reverse(RPrefix),
K = ets:next(Tab, Prefix),
case match_filter(Prefix, K, Words =/= []) of
K = ets:next(Tab, {Prefix, {}}),
case match_filter(Prefix, K, Words == []) of
true ->
K;
stop ->
@ -73,7 +72,7 @@ match_rest(false, [W | Rest], RPrefix, Tab) ->
match(Rest, [W | RPrefix], Tab);
match_rest(plus, [W | Rest], RPrefix, Tab) ->
case match(Rest, ['+' | RPrefix], Tab) of
Match when is_list(Match) ->
Match = {_, _} ->
Match;
false ->
match(Rest, [W | RPrefix], Tab)
@ -81,48 +80,71 @@ match_rest(plus, [W | Rest], RPrefix, Tab) ->
match_rest(_, [], _RPrefix, _Tab) ->
false.
-spec matches(emqx_types:topic(), ets:table()) -> [match(_ID)].
matches(Topic, Tab) ->
-spec matches(emqx_types:topic(), ets:table(), _Opts :: [unique]) -> [match(_ID)].
matches(Topic, Tab, Opts) ->
{Words, RPrefix} = match_init(Topic),
matches(Words, RPrefix, Tab).
matches(Words, RPrefix, Tab) ->
Prefix = lists:reverse(RPrefix),
matches(ets:next(Tab, Prefix), Prefix, Words, RPrefix, Tab).
matches(K, Prefix, Words, RPrefix, Tab) ->
case match_filter(Prefix, K, Words =/= []) of
true ->
[K | matches(ets:next(Tab, K), Prefix, Words, RPrefix, Tab)];
stop ->
[];
Matched ->
matches_rest(Matched, Words, RPrefix, Tab)
AccIn =
case Opts of
[unique | _] -> #{};
[] -> []
end,
Matches = matches(Words, RPrefix, AccIn, Tab),
case Matches of
#{} -> maps:values(Matches);
_ -> Matches
end.
matches_rest(false, [W | Rest], RPrefix, Tab) ->
matches(Rest, [W | RPrefix], Tab);
matches_rest(plus, [W | Rest], RPrefix, Tab) ->
matches(Rest, ['+' | RPrefix], Tab) ++ matches(Rest, [W | RPrefix], Tab);
matches_rest(_, [], _RPrefix, _Tab) ->
[].
matches(Words, RPrefix, Acc, Tab) ->
Prefix = lists:reverse(RPrefix),
matches(ets:next(Tab, {Prefix, {}}), Prefix, Words, RPrefix, Acc, Tab).
match_filter([], [{_ID}], _IsPrefix = false) ->
% NOTE: exact match is `true` only if we match whole topic, not prefix
true;
match_filter([], ['#', {_ID}], _IsPrefix) ->
matches(K, Prefix, Words, RPrefix, Acc, Tab) ->
case match_filter(Prefix, K, Words == []) of
true ->
matches(ets:next(Tab, K), Prefix, Words, RPrefix, match_add(K, Acc), Tab);
stop ->
Acc;
Matched ->
matches_rest(Matched, Words, RPrefix, Acc, Tab)
end.
matches_rest(false, [W | Rest], RPrefix, Acc, Tab) ->
matches(Rest, [W | RPrefix], Acc, Tab);
matches_rest(plus, [W | Rest], RPrefix, Acc, Tab) ->
NAcc = matches(Rest, ['+' | RPrefix], Acc, Tab),
matches(Rest, [W | RPrefix], NAcc, Tab);
matches_rest(_, [], _RPrefix, Acc, _Tab) ->
Acc.
match_add(K = {_Filter, ID}, Acc = #{}) ->
Acc#{ID => K};
match_add(K, Acc) ->
[K | Acc].
match_filter(Prefix, {Filter, _ID}, NotPrefix) ->
case match_filter(Prefix, Filter) of
exact ->
% NOTE: exact match is `true` only if we match whole topic, not prefix
NotPrefix;
Match ->
Match
end;
match_filter(_, '$end_of_table', _) ->
stop.
match_filter([], []) ->
exact;
match_filter([], ['#']) ->
% NOTE: naturally, '#' < '+', so this is already optimal for `match/2`
true;
match_filter([], ['+' | _], _) ->
match_filter([], ['+' | _]) ->
plus;
match_filter([], [_H | _], _) ->
match_filter([], [_H | _]) ->
false;
match_filter([H | T1], [H | T2], IsPrefix) ->
match_filter(T1, T2, IsPrefix);
match_filter([H1 | _], [H2 | _], _) when H2 > H1 ->
match_filter([H | T1], [H | T2]) ->
match_filter(T1, T2);
match_filter([H1 | _], [H2 | _]) when H2 > H1 ->
% NOTE: we're strictly past the prefix, no need to continue
stop;
match_filter(_, '$end_of_table', _) ->
stop.
match_init(Topic) ->
@ -135,6 +157,14 @@ match_init(Topic) ->
{Words, []}
end.
-spec get_id(match(ID)) -> ID.
get_id({_Filter, {ID}}) ->
ID.
-spec get_topic(match(_ID)) -> emqx_types:topic().
get_topic({Filter, _ID}) ->
emqx_topic:join(Filter).
-spec get_record(match(_ID), ets:table()) -> _Record.
get_record(K, Tab) ->
ets:lookup_element(Tab, K, 2).

View File

@ -29,8 +29,8 @@ t_insert(_) ->
true = emqx_rule_index:insert(<<"sensor/1/metric/2">>, t_insert_1, <<>>, Tab),
true = emqx_rule_index:insert(<<"sensor/+/#">>, t_insert_2, <<>>, Tab),
true = emqx_rule_index:insert(<<"sensor/#">>, t_insert_3, <<>>, Tab),
?assertEqual(<<"sensor/#">>, get_topic(match(<<"sensor">>, Tab))),
?assertEqual(t_insert_3, get_id(match(<<"sensor">>, Tab))),
?assertEqual(<<"sensor/#">>, topic(match(<<"sensor">>, Tab))),
?assertEqual(t_insert_3, id(match(<<"sensor">>, Tab))),
true = ets:delete(Tab).
t_match(_) ->
@ -40,7 +40,7 @@ t_match(_) ->
true = emqx_rule_index:insert(<<"sensor/#">>, t_match_3, <<>>, Tab),
?assertMatch(
[<<"sensor/#">>, <<"sensor/+/#">>],
[get_topic(M) || M <- matches(<<"sensor/1">>, Tab)]
[topic(M) || M <- matches(<<"sensor/1">>, Tab)]
),
true = ets:delete(Tab).
@ -51,7 +51,7 @@ t_match2(_) ->
true = emqx_rule_index:insert(<<"+/+/#">>, t_match2_3, <<>>, Tab),
?assertEqual(
[<<"#">>, <<"+/#">>, <<"+/+/#">>],
[get_topic(M) || M <- matches(<<"a/b/c">>, Tab)]
[topic(M) || M <- matches(<<"a/b/c">>, Tab)]
),
?assertEqual(
false,
@ -79,7 +79,7 @@ t_match3(_) ->
end,
?assertEqual(
t_match3_sys,
get_id(match(<<"$SYS/a/b/c">>, Tab))
id(match(<<"$SYS/a/b/c">>, Tab))
),
true = ets:delete(Tab).
@ -92,11 +92,11 @@ t_match4(_) ->
),
?assertEqual(
[<<"/#">>, <<"/+">>],
[get_topic(M) || M <- matches(<<"/">>, Tab)]
[topic(M) || M <- matches(<<"/">>, Tab)]
),
?assertEqual(
[<<"/#">>, <<"/+/a/b/c">>],
[get_topic(M) || M <- matches(<<"/0/a/b/c">>, Tab)]
[topic(M) || M <- matches(<<"/0/a/b/c">>, Tab)]
),
true = ets:delete(Tab).
@ -114,11 +114,11 @@ t_match5(_) ->
),
?assertEqual(
[<<"#">>, <<T/binary, "/#">>],
[get_topic(M) || M <- matches(T, Tab)]
[topic(M) || M <- matches(T, Tab)]
),
?assertEqual(
[<<"#">>, <<T/binary, "/#">>, <<T/binary, "/+">>],
[get_topic(M) || M <- matches(<<T/binary, "/1">>, Tab)]
[topic(M) || M <- matches(<<T/binary, "/1">>, Tab)]
),
true = ets:delete(Tab).
@ -127,7 +127,7 @@ t_match6(_) ->
T = <<"a/b/c/d/e/f/g/h/i/j/k/l/m/n/o/p/q/r/s/t/u/v/w/x/y/z">>,
W = <<"+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/+/#">>,
emqx_rule_index:insert(W, ID = t_match6, <<>>, Tab),
?assertEqual(ID, get_id(match(T, Tab))),
?assertEqual(ID, id(match(T, Tab))),
true = ets:delete(Tab).
t_match7(_) ->
@ -135,9 +135,23 @@ t_match7(_) ->
T = <<"a/b/c/d/e/f/g/h/i/j/k/l/m/n/o/p/q/r/s/t/u/v/w/x/y/z">>,
W = <<"a/+/c/+/e/+/g/+/i/+/k/+/m/+/o/+/q/+/s/+/u/+/w/+/y/+/#">>,
emqx_rule_index:insert(W, t_match7, <<>>, Tab),
?assertEqual(W, get_topic(match(T, Tab))),
?assertEqual(W, topic(match(T, Tab))),
true = ets:delete(Tab).
t_match_unique(_) ->
Tab = new(),
emqx_rule_index:insert(<<"a/b/c">>, t_match_id1, <<>>, Tab),
emqx_rule_index:insert(<<"a/b/+">>, t_match_id1, <<>>, Tab),
emqx_rule_index:insert(<<"a/b/c/+">>, t_match_id2, <<>>, Tab),
?assertEqual(
[t_match_id1, t_match_id1],
[id(M) || M <- emqx_rule_index:matches(<<"a/b/c">>, Tab, [])]
),
?assertEqual(
[t_match_id1],
[id(M) || M <- emqx_rule_index:matches(<<"a/b/c">>, Tab, [unique])]
).
new() ->
ets:new(?MODULE, [public, ordered_set, {write_concurrency, true}]).
@ -145,22 +159,10 @@ match(T, Tab) ->
emqx_rule_index:match(T, Tab).
matches(T, Tab) ->
lists:sort(emqx_rule_index:matches(T, Tab)).
lists:sort(emqx_rule_index:matches(T, Tab, [])).
-spec get_id(emqx_rule_index:match(ID)) -> ID.
get_id([{ID}]) ->
ID;
get_id([_ | Rest]) ->
get_id(Rest).
id(Match) ->
emqx_rule_index:get_id(Match).
-spec get_topic(emqx_rule_index:match(_ID)) -> emqx_types:topic().
get_topic(K) ->
emqx_topic:join(cut_topic(K)).
cut_topic(K) ->
cut_topic(K, []).
cut_topic([{_ID}], Acc) ->
lists:reverse(Acc);
cut_topic([W | Rest], Acc) ->
cut_topic(Rest, [W | Acc]).
topic(Match) ->
emqx_rule_index:get_topic(Match).