(couchdb) 02/03: Add QUERY_CARDINALITY_LIMIT to group_by

[ii...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

iilyak pushed a commit to branch couch-stats-resource-tracker-http-api
in repository https://gitbox.apache.org/repos/asf/couchdb.git

commit b2b4132e70b1fd4d2864739a5f9cb6a5da8abd19
Author: ILYA Khlopotov <ii...@apache.org>
AuthorDate: Tue Jun 10 12:13:07 2025 -0700

    Add QUERY_CARDINALITY_LIMIT to group_by
---
 .../src/couch_stats_resource_tracker.hrl           |  2 +
 src/couch_stats/src/csrt_query.erl                 | 46 ++++++++++++++--------
 2 files changed, 31 insertions(+), 17 deletions(-)

diff --git a/src/couch_stats/src/couch_stats_resource_tracker.hrl b/src/couch_stats/src/couch_stats_resource_tracker.hrl
index 97e38781c..b3b50d7bd 100644
--- a/src/couch_stats/src/couch_stats_resource_tracker.hrl
+++ b/src/couch_stats/src/couch_stats_resource_tracker.hrl
@@ -35,6 +35,8 @@
 -define(ROWS_READ, rows_read).
 -define(FRPC_CHANGES_RETURNED, changes_returned).
 
+-define(QUERY_CARDINALITY_LIMIT, 10_000).
+
 %% Mapping of couch_stat metric names to #rctx{} field names.
 %% These are used for fields that we inc a counter on.
 -define(STATS_TO_KEYS, #{
diff --git a/src/couch_stats/src/csrt_query.erl b/src/couch_stats/src/csrt_query.erl
index 62b03d695..c46345cd3 100644
--- a/src/couch_stats/src/csrt_query.erl
+++ b/src/couch_stats/src/csrt_query.erl
@@ -118,32 +118,44 @@ count_by(KeyFun) ->
 group_by(KeyFun, ValFun) ->
     group_by(KeyFun, ValFun, fun erlang:'+'/2).
 
+group_by(KeyFun, ValFun, AggFun) ->
+    group_by(KeyFun, ValFun, AggFun, ?QUERY_CARDINALITY_LIMIT).
+
 %% eg: group_by(mfa, docs_read).
 %% eg: group_by(fun(#rctx{mfa=MFA,docs_read=DR}) -> {MFA, DR} end, ioq_calls).
 %% eg: ^^ or: group_by([mfa, docs_read], ioq_calls).
 %% eg: group_by([username, dbname, mfa], docs_read).
 %% eg: group_by([username, dbname, mfa], ioq_calls).
 %% eg: group_by([username, dbname, mfa], js_filters).
-group_by(KeyL, ValFun, AggFun) when is_list(KeyL) ->
+group_by(KeyL, ValFun, AggFun, Limit) when is_list(KeyL) ->
     KeyFun = fun(Ele) -> list_to_tuple([field(Ele, Key) || Key <- KeyL]) end,
-    group_by(KeyFun, ValFun, AggFun);
-group_by(Key, ValFun, AggFun) when is_atom(Key) ->
-    group_by(curry_field(Key), ValFun, AggFun);
-group_by(KeyFun, Val, AggFun) when is_atom(Val) ->
-    group_by(KeyFun, curry_field(Val), AggFun);
-group_by(KeyFun, ValFun, AggFun) ->
+    group_by(KeyFun, ValFun, AggFun, Limit);
+group_by(Key, ValFun, AggFun, Limit) when is_atom(Key) ->
+    group_by(curry_field(Key), ValFun, AggFun, Limit);
+group_by(KeyFun, Val, AggFun, Limit) when is_atom(Val) ->
+    group_by(KeyFun, curry_field(Val), AggFun, Limit);
+group_by(KeyFun, ValFun, AggFun, Limit) ->
     FoldFun = fun(Ele, Acc) ->
-        Key = KeyFun(Ele),
-        Val = ValFun(Ele),
-        CurrVal = maps:get(Key, Acc, 0),
-        case AggFun(CurrVal, Val) of
-            0 ->
-                Acc;
-            NewVal ->
-                maps:put(Key, NewVal, Acc)
-        end
+        case maps:size(Acc) =< Limit of
+            true ->
+                Key = KeyFun(Ele),
+                Val = ValFun(Ele),
+                CurrVal = maps:get(Key, Acc, 0),
+                case AggFun(CurrVal, Val) of
+                    0 ->
+                        Acc;
+                    NewVal ->
+                        maps:put(Key, NewVal, Acc)
+                end;
+            false ->
+                throw({limit, Acc})
+            end
     end,
-    ets:foldl(FoldFun, #{}, ?CSRT_ETS).
+    try
+        {ok, ets:foldl(FoldFun, #{}, ?CSRT_ETS)}
+    catch throw:{limit, Acc} ->
+        {limit, Acc}
+    end.
 
 %% Sorts largest first
 sorted(Map) when is_map(Map) ->