Intermediate commit to save work

Author: pavelvelikhov

ydb/core/kqp/opt/logical/kqp_opt_log_join.cpp

@@ -859,6 +859,16 @@ TMaybeNode<TExprBase> KqpJoinToIndexLookupImpl(const TDqJoin& join, TExprContext
 
 } // anonymous namespace
 
+bool KqpLookupJoinIsApplicable(const NYql::NNodes::TExprBase& node, TVector<TString> joinColumns, NYql::TExprContext& ctx,
+    const TKqpOptimizeContext& kqpCtx) {
+    if (kqpCtx.IsScanQuery() && !kqpCtx.Config->EnableKqpScanQueryStreamIdxLookupJoin) {
+        return false;
+    }
+
+    
+
+}
+
 TExprBase KqpJoinToIndexLookup(const TExprBase& node, TExprContext& ctx, const TKqpOptimizeContext& kqpCtx)
 {
     if ((kqpCtx.IsScanQuery() && !kqpCtx.Config->EnableKqpScanQueryStreamIdxLookupJoin) || !node.Maybe<TDqJoin>()) {

ydb/core/kqp/opt/logical/kqp_opt_log_rules.h

@@ -38,6 +38,9 @@ NYql::NNodes::TExprBase KqpRewriteIndexRead(const NYql::NNodes::TExprBase& node,
 NYql::NNodes::TExprBase KqpRewriteLookupIndex(const NYql::NNodes::TExprBase& node, NYql::TExprContext& ctx,
     const TKqpOptimizeContext& kqpCtx);
 
+bool KqpLookupJoinIsApplicable(const NYql::NNodes::TExprBase& node, TVector<TString> joinColumns, NYql::TExprContext& ctx,
+    const TKqpOptimizeContext& kqpCtx);
+
 NYql::NNodes::TExprBase KqpRewriteStreamLookupIndex(const NYql::NNodes::TExprBase& node, NYql::TExprContext& ctx,
     const TKqpOptimizeContext& kqpCtx);
 

ydb/library/yql/core/cbo/cbo_optimizer_new.h

@@ -49,8 +49,13 @@ struct IBaseOptimizerNode {
 struct TRelOptimizerNode : public IBaseOptimizerNode {
     TString Label;
 
+    // Temporary solution to check if a LookupJoin is possible in KQP
+    TExprNode::TPtr Expr;
+
     TRelOptimizerNode(TString label, std::shared_ptr<TOptimizerStatistics> stats) : 
         IBaseOptimizerNode(RelNodeType, stats), Label(label) { }
+    TRelOptimizerNode(TString label, std::shared_ptr<TOptimizerStatistics> stats, const TExprNode::TPtr expr) : 
+        IBaseOptimizerNode(RelNodeType, stats), Label(label), Expr(expr) { }
     virtual ~TRelOptimizerNode() {}
 
     virtual TVector<TString> Labels();

ydb/library/yql/core/yql_cost_function.h

@@ -44,11 +44,12 @@ bool operator < (const TJoinColumn& c1, const TJoinColumn& c2);
 }
 
 enum EJoinImplType {
-    DictJoin,
-    MapJoin,
-    GraceJoin
+    HashJoin,
+    LookupJoin
 };
 
+static const EJoinImplType AllJoinTypes[] = { HashJoin, LookupJoin };
+
 TOptimizerStatistics ComputeJoinStats(const TOptimizerStatistics& leftStats, const TOptimizerStatistics& rightStats, 
     const std::set<std::pair<NDq::TJoinColumn, NDq::TJoinColumn>>& joinConditions, EJoinImplType joinType);
 

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

@@ -92,6 +92,66 @@ void ComputeJoinConditions(const TCoEquiJoinTuple& joinTuple,
     }
 }
 
+bool IsLookupJoinApplicable(std::shared_ptr<IBaseOptimizerNode> left, 
+    std::shared_ptr<IBaseOptimizerNode> right, 
+    const std::set<std::pair<TJoinColumn, TJoinColumn>>& joinCondition) {
+
+    
+}
+
+bool IsJoinApplicable(std::shared_ptr<IBaseOptimizerNode> left, 
+    std::shared_ptr<IBaseOptimizerNode> right, 
+    const std::set<std::pair<TJoinColumn, TJoinColumn>>& joinConditions,
+    EJoinImplType joinImpl) {
+
+        swtich(joinImpl) {
+            case EJoinImplType::LookupJoin:
+                return IsLookupJoinApplicable(left,right,joinConditions);
+            default:
+                return true;
+        }
+}
+
+std::shared_ptr<TJoinOptimizerNode> PickBestJoin(std::shared_ptr<IBaseOptimizerNode> left, 
+    std::shared_ptr<IBaseOptimizerNode> right, 
+    const std::set<std::pair<TJoinColumn, TJoinColumn>>& leftJoinConditions,
+    const std::set<std::pair<TJoinColumn, TJoinColumn>>& rightJoinConditions,
+    const std::shared_ptr<TOptimizerStatistics> leftStats,
+    const std::shared_ptr<TOptimizerStatistics> rightStats) {
+
+    auto res = std::shared_ptr<TJoinOptimizerNode>();
+
+    for ( auto joinType : AllJoinTypes ) {
+        auto p1 = IsJoinApplicable(left, right, leftJoinConditions, joinType) ? 
+            MakeJoin(left, right, leftJoinConditions, joinType ) :
+            std::shared_ptr<TJoinOptimizerNode>();
+        auto p2 = IsJoinApplicable(right, left, rightJoinConditions, joinType) ? 
+            MakeJoin(right, left, rightJoinConditions, joinType ) :
+            std::shared_ptr<TJoinOptimizerNode>();
+            
+        if (p1) {
+            if (res) {
+                if (p1->Cost < res->Cost) {
+                    res = p1;
+                }
+            } else {
+                res = p1;
+            }
+        }
+        if (p2) {
+            if (res) {
+                if (p2->Cost < res->Cost) {
+                    res = p2;
+                }
+            } else {
+                res = p2;
+            }
+        }
+    }
+
+    return res;
+}
+
 /**
  * Create a new join and compute its statistics and cost
 */