Fixed a problem with cardinality estimation for PK joins (#907)

Author: pavelvelikhov

ydb/core/kqp/opt/kqp_statistics_transformer.cpp

@@ -55,7 +55,7 @@ void InferStatisticsForKqpTable(const TExprNode::TPtr& input, TTypeAnnotationCon
     const auto& tableData = kqpCtx.Tables->ExistingTable(kqpCtx.Cluster, path.Value());
     double nRows = tableData.Metadata->RecordsCount;
     int nAttrs = tableData.Metadata->Columns.size();
-    YQL_CLOG(TRACE, CoreDq) << "Infer statistics for table: " << path.Value() << ", nrows: " << nRows << ", nattrs: " << nAttrs;
+    YQL_CLOG(TRACE, CoreDq) << "Infer statistics for table: " << path.Value() << ", nrows: " << nRows << ", nattrs: " << nAttrs << ", nKeyColumns: " << tableData.Metadata->KeyColumnNames.size();
 
     auto outputStats = TOptimizerStatistics(EStatisticsType::BaseTable, nRows, nAttrs, 0.0, tableData.Metadata->KeyColumnNames);
     typeCtx->SetStats(input.Get(), std::make_shared<TOptimizerStatistics>(outputStats));

ydb/library/yql/core/yql_cost_function.cpp

@@ -41,9 +41,11 @@ TOptimizerStatistics NYql::ComputeJoinStats(const TOptimizerStatistics& leftStat
 
     double newCard;
     EStatisticsType outputType;
+    TVector<TString> joinedTableKeys;
 
     if (IsPKJoin(rightStats,rightJoinKeys)) {
         newCard = std::max(leftStats.Nrows,rightStats.Nrows);
+        joinedTableKeys = leftStats.KeyColumns;
         if (leftStats.Type == EStatisticsType::BaseTable){
             outputType = EStatisticsType::FilteredFactTable;
         } else {
@@ -52,6 +54,7 @@ TOptimizerStatistics NYql::ComputeJoinStats(const TOptimizerStatistics& leftStat
     }
     else if (IsPKJoin(leftStats,leftJoinKeys)) {
         newCard = std::max(leftStats.Nrows,rightStats.Nrows);
+        joinedTableKeys = rightStats.KeyColumns;
         if (rightStats.Type == EStatisticsType::BaseTable){
             outputType = EStatisticsType::FilteredFactTable;
         } else {
@@ -69,7 +72,7 @@ TOptimizerStatistics NYql::ComputeJoinStats(const TOptimizerStatistics& leftStat
         + newCard 
         + leftStats.Cost + rightStats.Cost;
 
-    return TOptimizerStatistics(outputType, newCard, newNCols, cost);
+    return TOptimizerStatistics(outputType, newCard, newNCols, cost, joinedTableKeys);
 }
 
 TOptimizerStatistics NYql::ComputeJoinStats(const TOptimizerStatistics& leftStats, const TOptimizerStatistics& rightStats, 

ydb/library/yql/dq/opt/dq_opt_stat.cpp

@@ -194,7 +194,7 @@ void InferStatisticsForFlatMap(const TExprNode::TPtr& input, TTypeAnnotationCont
 
         double selectivity = ComputePredicateSelectivity(flatmap.Lambda().Body(), inputStats);
 
-        auto outputStats = TOptimizerStatistics(inputStats->Type, inputStats->Nrows * selectivity, inputStats->Ncols, inputStats->Cost );
+        auto outputStats = TOptimizerStatistics(inputStats->Type, inputStats->Nrows * selectivity, inputStats->Ncols, inputStats->Cost, inputStats->KeyColumns );
 
         typeCtx->SetStats(input.Get(), std::make_shared<TOptimizerStatistics>(outputStats) );
     }
@@ -235,7 +235,7 @@ void InferStatisticsForFilter(const TExprNode::TPtr& input, TTypeAnnotationConte
 
     double selectivity = ComputePredicateSelectivity(filterBody, inputStats);
 
-    auto outputStats = TOptimizerStatistics(inputStats->Type, inputStats->Nrows * selectivity, inputStats->Ncols, inputStats->Cost);
+    auto outputStats = TOptimizerStatistics(inputStats->Type, inputStats->Nrows * selectivity, inputStats->Ncols, inputStats->Cost, inputStats->KeyColumns);
 
     typeCtx->SetStats(input.Get(), std::make_shared<TOptimizerStatistics>(outputStats) );
 }