dbms: add tests for sequenceCount and sequenceMatch, fix infinite loop for sequenceCount of empty pattern, fix decltype deducing reference type [#METR-17427]

dbms/include/DB/AggregateFunctions/AggregateFunctionSequenceMatch.h

@@ -346,17 +346,18 @@ private:
 	}
 
 protected:
-	template <typename T1, typename T2>
+	template <typename T>
-	bool match(T1 & events_it, const T2 events_end) const
+	bool match(T & events_it, const T events_end) const
 	{
 		const auto action_begin = std::begin(actions);
 		const auto action_end = std::end(actions);
 		auto action_it = action_begin;
 
+		const auto events_begin = events_it;
 		auto base_it = events_it;
 
 		/// an iterator to action plus an iterator to row in events list plus timestamp at the start of sequence
-		using backtrack_info = std::tuple<decltype(action_it), decltype(events_it), decltype(base_it)>;
+		using backtrack_info = std::tuple<decltype(action_it), T, T>;
 		std::stack<backtrack_info> back_stack;
 
 		/// backtrack if possible
@@ -473,6 +474,9 @@ protected:
 				++action_it;
 		}
 
+		if (events_it == events_begin)
+			++events_it;
+
 		return action_it == action_end;
 	}
 

dbms/tests/queries/0_stateless/00222_sequence_aggregate_function_family.reference

@@ -0,0 +1,42 @@
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1

dbms/tests/queries/0_stateless/00222_sequence_aggregate_function_family.sql

@@ -0,0 +1,51 @@
+drop table if exists sequence_test;
+
+create table sequence_test (time UInt32, data UInt8) engine=Memory;
+
+insert into sequence_test values (0,0),(1,0),(2,0),(3,0),(4,1),(5,2),(6,0),(7,0),(8,0),(9,0),(10,1),(11,1);
+
+select 1 = sequenceMatch('')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('.')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('.*')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?3)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceMatch('(?4)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)(?1)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)(?1)(?1)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceMatch('(?1)(?1)(?1)(?1)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)(?1)(?1)(?1)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)(?t>10)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceMatch('(?1)(?t>11)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)(?t<11)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)(?t<3)(?3)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?1)(?t<=2)(?3)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceMatch('(?1)(?t<2)(?3)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?2)(?t>=7)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceMatch('(?2)(?t>7)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceMatch('(?2)(?3)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+
+select count() = sequenceCount('')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select count() = sequenceCount('.')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select count() = sequenceCount('.*')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 8 = sequenceCount('(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 3 = sequenceCount('(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceCount('(?3)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceCount('(?4)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 4 = sequenceCount('(?1)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 2 = sequenceCount('(?1)(?1)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 2 = sequenceCount('(?1)(?1)(?1)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceCount('(?1)(?1)(?1)(?1)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 2 = sequenceCount('(?1)(?1)(?1)(?1)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceCount('(?1)(?t>10)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceCount('(?1)(?t>11)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 2 = sequenceCount('(?1)(?t<11)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceCount('(?1)(?t<3)(?3)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceCount('(?1)(?t<=2)(?3)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceCount('(?1)(?t<2)(?3)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceCount('(?2)(?t>=7)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 0 = sequenceCount('(?2)(?t>7)(?2)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+select 1 = sequenceCount('(?2)(?3)(?1)')(toDateTime(time), data = 0, data = 1, data = 2, data = 3) from sequence_test;
+
+drop table sequence_test;