fixed join vs distinct facets; fixed join vs implicit groupby

src/joinsorter.cpp

@@ -559,7 +559,8 @@ JoinSorter_c::JoinSorter_c ( const CSphIndex * pIndex, const CSphIndex * pJoined
 		bHaveAggregates |= tSorterSchema.GetAttr(i).m_eAggrFunc!=SPH_AGGR_NONE;
 
 	CSphVector<std::pair<int,bool>> dRightFilters = FetchJoinRightTableFilters ( m_tQuery.m_dFilters, tSorterSchema, m_tQuery.m_sJoinIdx.cstr() );
-	m_bFinalCalcOnly = !bJoinedGroupSort && !bHaveAggregates && !dRightFilters.GetLength() && !NeedToMoveMixedJoinFilters ( m_tQuery, tSorterSchema );
+	bool bDisableByImplicitGrouping = HasImplicitGrouping(m_tQuery) && m_tQuery.m_eJoinType!=JoinType_e::LEFT;
+	m_bFinalCalcOnly = !bJoinedGroupSort && !bHaveAggregates && !dRightFilters.GetLength() && !NeedToMoveMixedJoinFilters ( m_tQuery, tSorterSchema ) && !pSorter->IsPrecalc() && !bDisableByImplicitGrouping;
 	m_bErrorFlag = !SetupJoinQuery ( m_pSorter->GetSchema()->GetDynamicSize(), m_sErrorMessage );
 }
 
@@ -1306,7 +1307,7 @@ void JoinSorter_c::SetupJoinSelectList()
 class JoinMultiSorter_c : public JoinSorter_c
 {
 public:
-			JoinMultiSorter_c ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> dSorters, bool bJoinedGroupSort );
+			JoinMultiSorter_c ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> dSorters );
 
 	bool	Push ( const CSphMatch & tEntry ) override;
 	bool	PushGrouped ( const CSphMatch & tEntry, bool bNewSet ) override;
@@ -1321,8 +1322,8 @@ class JoinMultiSorter_c : public JoinSorter_c
 };
 
 
-JoinMultiSorter_c::JoinMultiSorter_c ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> dSorters, bool bJoinedGroupSort )
-	: JoinSorter_c ( pIndex, pJoinedIndex, dQueries, dSorters[0], bJoinedGroupSort )
+JoinMultiSorter_c::JoinMultiSorter_c ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> dSorters )
+	: JoinSorter_c ( pIndex, pJoinedIndex, dQueries, dSorters[0], true )
 {
 	m_dSorters.Resize ( dSorters.GetLength() );
 	memcpy ( m_dSorters.Begin(), dSorters.Begin(), dSorters.GetLengthBytes() );
@@ -1476,7 +1477,7 @@ ISphMatchSorter * CreateJoinSorter ( const CSphIndex * pIndex, const CSphIndex *
 }
 
 
-bool CreateJoinMultiSorter ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const SphQueueSettings_t & tSettings, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> & dSorters, bool bJoinedGroupSort, CSphString & sError )
+bool CreateJoinMultiSorter ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const SphQueueSettings_t & tSettings, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> & dSorters, CSphString & sError )
 {
 	if ( !tSettings.m_pJoinArgs )
 		return true;
@@ -1486,7 +1487,7 @@ bool CreateJoinMultiSorter ( const CSphIndex * pIndex, const CSphIndex * pJoined
 
 	// the idea is that 1st sorter does the join AND it also pushes joined matches to all other sorters
 	// to avoid double push to 1..N sorters they are wrapped in a class that prevents pushing matches
-	std::unique_ptr<JoinMultiSorter_c> pJoinSorter = std::make_unique<JoinMultiSorter_c> ( pIndex, pJoinedIndex, dQueries, dSorters, bJoinedGroupSort );
+	std::unique_ptr<JoinMultiSorter_c> pJoinSorter = std::make_unique<JoinMultiSorter_c> ( pIndex, pJoinedIndex, dQueries, dSorters );
 	if ( pJoinSorter->GetErrorFlag() )
 	{
 		sError = pJoinSorter->GetErrorMessage();

src/joinsorter.h

@@ -19,4 +19,4 @@ bool				NeedToMoveMixedJoinFilters ( const CSphQuery & tQuery, const ISphSchema
 std::unique_ptr<ISphFilter> CreateJoinNullFilter ( const CSphFilterSettings & tSettings, const CSphAttrLocator & tNullMapLocator );
 
 ISphMatchSorter *	CreateJoinSorter ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const SphQueueSettings_t & tSettings, const CSphQuery & tQuery, ISphMatchSorter * pSorter, bool bJoinedGroupSort, CSphString & sError );
-bool				CreateJoinMultiSorter ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const SphQueueSettings_t & tSettings, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> & dSorters, bool bJoinedGroupSort, CSphString & sError );
+bool				CreateJoinMultiSorter ( const CSphIndex * pIndex, const CSphIndex * pJoinedIndex, const SphQueueSettings_t & tSettings, const VecTraits_T<CSphQuery> & dQueries, VecTraits_T<ISphMatchSorter *> & dSorters, CSphString & sError );

src/queuecreator.cpp

@@ -621,6 +621,9 @@ bool QueueCreator_c::SetupGroupbySettings ( bool bHasImplicitGrouping )
 		return Err ( "group-by attribute '%s' not found", m_tQuery.m_sGroupBy.cstr() );
 
 	const CSphColumnInfo & tGroupByAttr = tSchema.GetAttr(iGroupBy);
+	if ( tGroupByAttr.m_pExpr && tGroupByAttr.m_pExpr->UsesDocstore() )
+		return Err ( "unable to group by stored field '%s'", m_tQuery.m_sGroupBy.cstr() );
+
 	ESphAttr eType = tGroupByAttr.m_eAttrType;
 	CSphAttrLocator tLoc = tGroupByAttr.m_tLocator;
 	m_bJoinedGroupSort |= IsJoinAttr ( tGroupByAttr.m_sName );
@@ -2231,21 +2234,21 @@ int QueueCreator_c::AdjustMaxMatches ( int iMaxMatches ) const
 bool QueueCreator_c::CanCalcFastCountDistinct() const
 {
 	bool bHasAggregates = PredictAggregates();
-	return !bHasAggregates && m_tGroupSorterSettings.m_bImplicit && m_tGroupSorterSettings.m_bDistinct && m_tQuery.m_dFilters.IsEmpty() && m_tQuery.m_sQuery.IsEmpty() && m_tQuery.m_sKNNAttr.IsEmpty();
+	return !bHasAggregates && m_tGroupSorterSettings.m_bImplicit && m_tGroupSorterSettings.m_bDistinct && m_tQuery.m_dFilters.IsEmpty() && m_tQuery.m_sQuery.IsEmpty() && m_tQuery.m_sKNNAttr.IsEmpty() && m_tQuery.m_eJoinType!=JoinType_e::INNER;
 }
 
 
 bool QueueCreator_c::CanCalcFastCountFilter() const
 {
 	bool bHasAggregates = PredictAggregates();
-	return !bHasAggregates && m_tGroupSorterSettings.m_bImplicit && !m_tGroupSorterSettings.m_bDistinct && m_tQuery.m_dFilters.GetLength()==1 && m_tQuery.m_sQuery.IsEmpty() && m_tQuery.m_sKNNAttr.IsEmpty();
+	return !bHasAggregates && m_tGroupSorterSettings.m_bImplicit && !m_tGroupSorterSettings.m_bDistinct && m_tQuery.m_dFilters.GetLength()==1 && m_tQuery.m_sQuery.IsEmpty() && m_tQuery.m_sKNNAttr.IsEmpty() && m_tQuery.m_eJoinType!=JoinType_e::INNER;
 }
 
 
 bool QueueCreator_c::CanCalcFastCount() const
 {
 	bool bHasAggregates = PredictAggregates();
-	return !bHasAggregates && m_tGroupSorterSettings.m_bImplicit && !m_tGroupSorterSettings.m_bDistinct && m_tQuery.m_dFilters.IsEmpty() && m_tQuery.m_sQuery.IsEmpty() && m_tQuery.m_sKNNAttr.IsEmpty();
+	return !bHasAggregates && m_tGroupSorterSettings.m_bImplicit && !m_tGroupSorterSettings.m_bDistinct && m_tQuery.m_dFilters.IsEmpty() && m_tQuery.m_sQuery.IsEmpty() && m_tQuery.m_sKNNAttr.IsEmpty() && m_tQuery.m_eJoinType!=JoinType_e::INNER;
 }
 
 

src/searchd.cpp

@@ -5507,7 +5507,7 @@ int SearchHandler_c::CreateMultiQueryOrFacetSorters ( const CSphIndex * pIndex,
 		return 0;
 	}
 
-	if ( m_bFacetQueue && !CreateJoinMultiSorter ( pIndex, dJoinedIndexes[0], tQueueSettings, m_dNQueries, dSorters, tQueueRes.m_bJoinedGroupSort, dErrors[0] ) )
+	if ( m_bFacetQueue && !CreateJoinMultiSorter ( pIndex, dJoinedIndexes[0], tQueueSettings, m_dNQueries, dSorters, dErrors[0] ) )
 	{
 		dSorters.Apply ( [] ( ISphMatchSorter *& pSorter ) { SafeDelete (pSorter); } );
 		return 0;

test/test_278/test.xml

@@ -317,10 +317,47 @@ select * from join1 inner join join2 on join1.string_id = join2.id where id=1 or
 <!-- left join with a single right filter -->
 select * from join1 left join join2 on join1.id = join2.id where join2.name = 'name1' order by id asc;
 
+<!-- multiple facet -->
+select join1.j.table, j.table from join2 join join1 on join1.id = join2.id facet join1.j.table distinct facet j.table;;
+
+<!-- precalc (fast count) sorter -->
+select id, count(*) from join1 left join join2 on join1.id=join2.id;
+
+<!-- count(*) sorter vs left join -->
+select id, * from join1 left join join2 on join1.id=join2.id;
+select id, count(*) from join1 left join join2 on join1.id=join2.id;
+
+<!-- count(*) sorter vs inner join -->
+select id, * from join1 join join2 on join1.id=join2.id;
+select id, count(*) from join1 join join2 on join1.id=join2.id;
+
 <!-- cleanup -->
 drop table join1;
 drop table join2;
 
+<!-- populate tables -->
+CREATE TABLE customers (name text, email string attribute, address text );
+
+INSERT INTO customers (id, name, email, address) VALUES
+(1, 'Alice Johnson', 'alice@example.com', '123 Maple St'),
+(2, 'Bob Smith', 'bob@example.com', '456 Oak St'),
+(3, 'Carol White', 'carol@example.com', '789 Pine St');
+
+CREATE TABLE orders (customer_id bigint, product text, quantity integer, order_date string, tags multi, details json );
+
+INSERT INTO orders (id, customer_id, product, quantity, order_date, tags, details) VALUES 
+(1, 1, 'Laptop', 1, '2023-01-01', (101, 102), '{"price":1200, "warranty":"2 years"}'),
+(2, 2, 'Phone', 2, '2023-01-02', (103), '{"price":800, "warranty":"1 year"}'),
+(3, 1, 'Tablet', 1, '2023-01-03', (101, 104), '{"price":450, "warranty":"1 year"}'),
+(4, 3, 'Monitor', 1, '2023-01-04', (105), '{"price":300, "warranty":"1 year"}');
+
+<!-- group by fulltext (error) -->
+SELECT name, COUNT(*) FROM customers INNER JOIN orders ON customers.id = orders.customer_id GROUP BY name;
+
+<!-- cleanup -->
+DROP TABLE customers;
+DROP TABLE orders;
+
 </sphinxql>
 </queries>