Merge 870b68e into bfbba84

Author: lll-phill-lll

ydb/library/yql/dq/tasks/dq_task_program.cpp

@@ -2,13 +2,37 @@
 
 #include <ydb/library/yql/core/yql_expr_optimize.h>
 #include <ydb/library/yql/minikql/mkql_node_serialization.h>
+#include <ydb/library/yql/minikql/mkql_runtime_version.h>
 #include <ydb/library/yql/providers/common/mkql/yql_type_mkql.h>
 
+#include <util/generic/xrange.h>
+
 namespace NYql::NDq {
 
 using namespace NKikimr::NMiniKQL;
 using namespace NYql::NNodes;
 
+
+class TSpillingTransformProvider {
+public:
+    TCallableVisitFunc operator()(TInternName name) {
+        if (name == "GraceJoin" || name == "GraceSelfJoin") {
+            return [name](NKikimr::NMiniKQL::TCallable& callable, const TTypeEnvironment& env) {
+                TCallableBuilder callableBuilder(env,
+                    TStringBuilder() << callable.GetType()->GetName() << "WithSpilling",
+                    callable.GetType()->GetReturnType(), false);
+                for (ui32 i: xrange(callable.GetInputsCount())) {
+                    callableBuilder.Add(callable.GetInput(i));
+                }
+                return TRuntimeNode(callableBuilder.Build(), false);
+            };
+        }
+        
+
+        return TCallableVisitFunc();
+    }
+};
+
 const TStructExprType* CollectParameters(NNodes::TCoLambda program, TExprContext& ctx) {
     TVector<const TItemExprType*> memberTypes;
 
@@ -49,6 +73,15 @@ TString BuildProgram(NNodes::TCoLambda program, const TStructExprType& paramsTyp
 
     TRuntimeNode rootNode = MkqlBuildExpr(program.Body().Ref(), ctx);
 
+    if (RuntimeVersion >= 50U) {
+        TExploringNodeVisitor explorer;
+        explorer.Walk(rootNode.GetNode(), typeEnv);
+        bool wereChanges = false;
+        rootNode = SinglePassVisitCallables(rootNode, explorer, TSpillingTransformProvider(), typeEnv, true, wereChanges);
+
+        std::cerr << "MISHA " << wereChanges << std::endl; 
+    }
+
     TStructLiteralBuilder structBuilder(typeEnv);
     structBuilder.Add("Program", rootNode);
     structBuilder.Add("Inputs", pgmBuilder.NewTuple(inputNodes));

ydb/library/yql/minikql/comp_nodes/mkql_factory.cpp

@@ -234,8 +234,8 @@ struct TCallableComputationNodeBuilderFuncMapFiller {
         {"JoinDict", &WrapJoinDict},
         {"GraceJoin", &WrapGraceJoin},
         {"GraceSelfJoin", &WrapGraceSelfJoin},
-        {"GraceJoinWithSpilling", &WrapGraceJoinWithSpilling},
-        {"GraceSelfJoinWithSpilling", &WrapGraceSelfJoinWithSpilling},
+        {"GraceJoinWithSpilling", &WrapGraceJoin},
+        {"GraceSelfJoinWithSpilling", &WrapGraceSelfJoin},
         {"MapJoinCore", &WrapMapJoinCore},
         {"CommonJoinCore", &WrapCommonJoinCore},
         {"CombineCore", &WrapCombineCore},

ydb/library/yql/minikql/comp_nodes/mkql_grace_join.cpp

@@ -1170,29 +1170,27 @@ IComputationNode* WrapGraceJoinCommon(TCallable& callable, const TComputationNod
 
 IComputationNode* WrapGraceJoin(TCallable& callable, const TComputationNodeFactoryContext& ctx) {
     MKQL_ENSURE(callable.GetInputsCount() == 8, "Expected 8 args");
+    bool isSpillingAllowed = false;
+    if (callable.GetType()->GetName() == "GraceJoinWithSpilling") {
+        isSpillingAllowed = true;
+    }
+
+    std::cerr << "MISHA " << callable.GetType()->GetName() << std::endl;
 
-    return WrapGraceJoinCommon(callable, ctx, false, false);
+    return WrapGraceJoinCommon(callable, ctx, false, isSpillingAllowed);
 }
 
 IComputationNode* WrapGraceSelfJoin(TCallable& callable, const TComputationNodeFactoryContext& ctx) {
     MKQL_ENSURE(callable.GetInputsCount() == 7, "Expected 7 args");
-    
-    return WrapGraceJoinCommon(callable, ctx, true, false);
-}
-
-IComputationNode* WrapGraceJoinWithSpilling(TCallable& callable, const TComputationNodeFactoryContext& ctx) {
-    MKQL_ENSURE(callable.GetInputsCount() == 8, "Expected 8 args");
 
-    return WrapGraceJoinCommon(callable, ctx, false, true);
-}
-
-IComputationNode* WrapGraceSelfJoinWithSpilling(TCallable& callable, const TComputationNodeFactoryContext& ctx) {
-    MKQL_ENSURE(callable.GetInputsCount() == 7, "Expected 7 args");
+    bool isSpillingAllowed = false;
+    if (callable.GetType()->GetName() == "GraceSelfJoinWithSpilling") {
+        isSpillingAllowed = true;
+    }
 
-    return WrapGraceJoinCommon(callable, ctx, true, true);
+    return WrapGraceJoinCommon(callable, ctx, true, isSpillingAllowed);
 }
 
-
 }
 
 }

ydb/library/yql/minikql/mkql_program_builder.cpp

@@ -2160,28 +2160,6 @@ TRuntimeNode TProgramBuilder::GraceSelfJoin(TRuntimeNode flowLeft,  EJoinKind jo
     return GraceJoinCommon(__func__, flowLeft, {}, joinKind, leftKeyColumns, rightKeyColumns, leftRenames, rightRenames, returnType, anyJoinSettings);
 }
 
-TRuntimeNode TProgramBuilder::GraceJoinWithSpilling(TRuntimeNode flowLeft, TRuntimeNode flowRight, EJoinKind joinKind,
-        const TArrayRef<const ui32>& leftKeyColumns, const TArrayRef<const ui32>& rightKeyColumns,
-        const TArrayRef<const ui32>& leftRenames, const TArrayRef<const ui32>& rightRenames, TType* returnType, EAnyJoinSettings anyJoinSettings ) {
-
-    if constexpr (RuntimeVersion < 50U) {
-        THROW yexception() << "Runtime version (" << RuntimeVersion << ") too old for " << __func__;
-    }
-
-    return GraceJoinCommon(__func__, flowLeft, flowRight, joinKind, leftKeyColumns, rightKeyColumns, leftRenames, rightRenames, returnType, anyJoinSettings);
-}
-
-TRuntimeNode TProgramBuilder::GraceSelfJoinWithSpilling(TRuntimeNode flowLeft, EJoinKind joinKind,
-        const TArrayRef<const ui32>& leftKeyColumns, const TArrayRef<const ui32>& rightKeyColumns,
-        const TArrayRef<const ui32>& leftRenames, const TArrayRef<const ui32>& rightRenames, TType* returnType, EAnyJoinSettings anyJoinSettings ) {
-
-    if constexpr (RuntimeVersion < 50U) {
-        THROW yexception() << "Runtime version (" << RuntimeVersion << ") too old for " << __func__;
-    }
-    
-    return GraceJoinCommon(__func__, flowLeft, {}, joinKind, leftKeyColumns, rightKeyColumns, leftRenames, rightRenames, returnType, anyJoinSettings);
-}
-
 TRuntimeNode TProgramBuilder::ToSortedDict(TRuntimeNode list, bool all, const TUnaryLambda& keySelector,
     const TUnaryLambda& payloadSelector, bool isCompact, ui64 itemsCountHint) {
     return ToDict(list, all, keySelector, payloadSelector, __func__, isCompact, itemsCountHint);

ydb/library/yql/providers/common/mkql/yql_provider_mkql.cpp

@@ -1704,13 +1704,6 @@ TMkqlCommonCallableCompiler::TShared::TShared() {
 
         const auto returnType = BuildType(node, *node.GetTypeAnn(), ctx.ProgramBuilder);
 
-        // TODO: use PRAGMA
-        bool IsSpillingAllowed = false;
-        if (RuntimeVersion >= 50U && IsSpillingAllowed) {
-            return selfJoin
-                ? ctx.ProgramBuilder.GraceSelfJoinWithSpilling(flowLeft, joinKind, leftKeyColumns, rightKeyColumns, leftRenames, rightRenames, returnType, anyJoinSettings)
-                : ctx.ProgramBuilder.GraceJoinWithSpilling(flowLeft, flowRight, joinKind, leftKeyColumns, rightKeyColumns, leftRenames, rightRenames, returnType, anyJoinSettings);
-        }
         return selfJoin
             ? ctx.ProgramBuilder.GraceSelfJoin(flowLeft, joinKind, leftKeyColumns, rightKeyColumns, leftRenames, rightRenames, returnType, anyJoinSettings)
             : ctx.ProgramBuilder.GraceJoin(flowLeft, flowRight, joinKind, leftKeyColumns, rightKeyColumns, leftRenames, rightRenames, returnType, anyJoinSettings);

ydb/library/yql/providers/dq/common/yql_dq_settings.cpp

@@ -98,6 +98,7 @@ TDqConfiguration::TDqConfiguration() {
 
     REGISTER_SETTING(*this, _MaxAttachmentsSize);
     REGISTER_SETTING(*this, DisableCheckpoints);
+    REGISTER_SETTING(*this, EnableSpillingInGraceJoin);
 }
 
 } // namespace NYql

ydb/library/yql/providers/dq/common/yql_dq_settings.h

@@ -60,6 +60,7 @@ struct TDqSettings {
         static constexpr ui32 MaxDPccpDPTableSize = 16400U;
         static constexpr ui64 MaxAttachmentsSize = 2_GB;
         static constexpr bool SplitStageOnDqReplicate = true;
+        static constexpr bool EnableSpillingInGraceJoin = false;
     };
 
     using TPtr = std::shared_ptr<TDqSettings>;
@@ -131,6 +132,8 @@ struct TDqSettings {
     NCommon::TConfSetting<bool, false> DisableLLVMForBlockStages;
     NCommon::TConfSetting<bool, false> SplitStageOnDqReplicate;
 
+    NCommon::TConfSetting<bool, false> EnableSpillingInGraceJoin;
+
     NCommon::TConfSetting<ui64, false> _MaxAttachmentsSize;
     NCommon::TConfSetting<bool, false> DisableCheckpoints;