kuzudb · andyfengHKU · Jun 14, 2023 · Jun 13, 2023
diff --git a/src/binder/bind_expression/bind_function_expression.cpp b/src/binder/bind_expression/bind_function_expression.cpp
@@ -69,7 +69,7 @@ std::shared_ptr<Expression> ExpressionBinder::bindScalarFunctionExpression(
         ScalarFunctionExpression::getUniqueName(function->name, childrenAfterCast);
     return make_shared<ScalarFunctionExpression>(functionName, FUNCTION, std::move(bindData),
         std::move(childrenAfterCast), function->execFunc, function->selectFunc,
-        uniqueExpressionName);
+        function->compileFunc, uniqueExpressionName);
 }
 
 std::shared_ptr<Expression> ExpressionBinder::bindAggregateFunctionExpression(

diff --git a/src/expression_evaluator/function_evaluator.cpp b/src/expression_evaluator/function_evaluator.cpp
@@ -1,7 +1,6 @@
 #include "expression_evaluator/function_evaluator.h"
 
 #include "binder/expression/function_expression.h"
-#include "function/struct/vector_struct_operations.h"
 
 using namespace kuzu::common;
 using namespace kuzu::processor;
@@ -16,17 +15,13 @@ void FunctionExpressionEvaluator::init(const ResultSet& resultSet, MemoryManager
     if (expression->dataType.getLogicalTypeID() == LogicalTypeID::BOOL) {
         selectFunc = ((binder::ScalarFunctionExpression&)*expression).selectFunc;
     }
-    for (auto& child : children) {
-        parameters.push_back(child->resultVector);
-    }
 }
 
 void FunctionExpressionEvaluator::evaluate() {
     for (auto& child : children) {
         child->evaluate();
     }
     if (execFunc != nullptr) {
-        // Some functions are evaluated at compile time (e.g. struct_extract).
         execFunc(parameters, *resultVector);
     }
 }
@@ -63,41 +58,19 @@ std::unique_ptr<BaseExpressionEvaluator> FunctionExpressionEvaluator::clone() {
 
 void FunctionExpressionEvaluator::resolveResultVector(
     const ResultSet& resultSet, MemoryManager* memoryManager) {
-    auto& functionExpression = (binder::ScalarFunctionExpression&)*expression;
-    auto functionName = functionExpression.getFunctionName();
-    if (functionName == STRUCT_EXTRACT_FUNC_NAME || functionName == UNION_EXTRACT_FUNC_NAME) {
-        auto& bindData = (function::StructExtractBindData&)*functionExpression.getBindData();
-        resultVector =
-            StructVector::getFieldVector(children[0]->resultVector.get(), bindData.childIdx);
-    } else {
-        resultVector = std::make_shared<ValueVector>(expression->dataType, memoryManager);
+    for (auto& child : children) {
+        parameters.push_back(child->resultVector);
     }
+    resultVector = std::make_shared<ValueVector>(expression->dataType, memoryManager);
     std::vector<BaseExpressionEvaluator*> inputEvaluators;
     inputEvaluators.reserve(children.size());
     for (auto& child : children) {
         inputEvaluators.push_back(child.get());
     }
     resolveResultStateFromChildren(inputEvaluators);
-    // TODO(Ziyi): We should move result valueVector state resolution to each function.
-    if (functionExpression.getFunctionName() == STRUCT_PACK_FUNC_NAME) {
-        // Our goal is to make the state of the resultVector consistent with its children vectors.
-        // If the resultVector and inputVector are in different dataChunks, we should create a new
-        // child valueVector, which shares the state with the resultVector, instead of reusing the
-        // inputVector.
-        for (auto i = 0u; i < inputEvaluators.size(); i++) {
-            auto inputEvaluator = inputEvaluators[i];
-            if (inputEvaluator->resultVector->state == resultVector->state) {
-                common::StructVector::referenceVector(
-                    resultVector.get(), i, inputEvaluator->resultVector);
-            }
-        }
-    } else if (functionExpression.getFunctionName() == UNION_VALUE_FUNC_NAME) {
-        assert(inputEvaluators.size() == 1);
-        resultVector->setState(inputEvaluators[0]->resultVector->state);
-        common::UnionVector::getTagVector(resultVector.get())
-            ->setState(inputEvaluators[0]->resultVector->state);
-        common::UnionVector::referenceVector(
-            resultVector.get(), 0 /* fieldIdx */, inputEvaluators[0]->resultVector);
+    auto& functionExpression = (binder::ScalarFunctionExpression&)*expression;
+    if (functionExpression.compileFunc != nullptr) {
+        functionExpression.compileFunc(functionExpression.getBindData(), parameters, resultVector);
     }
 }
 

diff --git a/src/function/CMakeLists.txt b/src/function/CMakeLists.txt
@@ -12,6 +12,7 @@ add_library(kuzu_function
         vector_hash_operations.cpp
         vector_list_operation.cpp
         vector_null_operations.cpp
+        vector_node_rel_operations.cpp
         vector_string_operations.cpp
         vector_timestamp_operations.cpp
         vector_struct_operations.cpp

diff --git a/src/function/built_in_vector_operations.cpp b/src/function/built_in_vector_operations.cpp
@@ -7,7 +7,7 @@
 #include "function/interval/vector_interval_operations.h"
 #include "function/list/vector_list_operations.h"
 #include "function/map/vector_map_operations.h"
-#include "function/schema/vector_offset_operations.h"
+#include "function/schema/vector_node_rel_operations.h"
 #include "function/string/vector_string_operations.h"
 #include "function/struct/vector_struct_operations.h"
 #include "function/timestamp/vector_timestamp_operations.h"
@@ -30,8 +30,7 @@ void BuiltInVectorOperations::registerVectorOperations() {
     registerStructOperations();
     registerMapOperations();
     registerUnionOperations();
-    // register internal offset operation
-    vectorOperations.insert({OFFSET_FUNC_NAME, OffsetVectorOperation::getDefinitions()});
+    registerNodeRelOperations();
 }
 
 bool BuiltInVectorOperations::canApplyStaticEvaluation(
@@ -505,5 +504,11 @@ void BuiltInVectorOperations::registerUnionOperations() {
         {UNION_EXTRACT_FUNC_NAME, UnionExtractVectorOperations::getDefinitions()});
 }
 
+void BuiltInVectorOperations::registerNodeRelOperations() {
+    vectorOperations.insert({OFFSET_FUNC_NAME, OffsetVectorOperation::getDefinitions()});
+    vectorOperations.insert({NODES_FUNC_NAME, NodesVectorOperation::getDefinitions()});
+    vectorOperations.insert({RELS_FUNC_NAME, RelsVectorOperation::getDefinitions()});
+}
+
 } // namespace function
 } // namespace kuzu
diff --git a/src/function/vector_list_operation.cpp b/src/function/vector_list_operation.cpp
@@ -399,8 +399,7 @@
 void ListSortVectorOperation::getExecFunction(
     const binder::expression_vector& arguments, scalar_exec_func& func) {
     if (arguments.size() == 1) {
-        func = UnaryExecListStructFunctionWithVectors<list_entry_t, list_entry_t,
-            operation::ListSort<T>>;
+        func = UnaryExecListStructFunction<list_entry_t, list_entry_t, operation::ListSort<T>>;
         return;
     } else if (arguments.size() == 2) {
         func = BinaryExecListStructFunction<list_entry_t, ku_string_t, list_entry_t,
@@ -472,8 +471,8 @@
 void ListReverseSortVectorOperation::getExecFunction(
     const binder::expression_vector& arguments, scalar_exec_func& func) {
     if (arguments.size() == 1) {
-        func = UnaryExecListStructFunctionWithVectors<list_entry_t, list_entry_t,
-            operation::ListReverseSort<T>>;
+        func =
+            UnaryExecListStructFunction<list_entry_t, list_entry_t, operation::ListReverseSort<T>>;
         return;
     } else if (arguments.size() == 2) {
         func = BinaryExecListStructFunction<list_entry_t, ku_string_t, list_entry_t,
@@ -500,23 +499,23 @@
     case LogicalTypeID::SERIAL:
     case LogicalTypeID::INT64: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, int64_t, operation::ListSum>;
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListSum>;
     } break;
     case LogicalTypeID::INT32: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, int32_t, operation::ListSum>;
+            UnaryExecListStructFunction<list_entry_t, int32_t, operation::ListSum>;
     } break;
     case LogicalTypeID::INT16: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, int16_t, operation::ListSum>;
+            UnaryExecListStructFunction<list_entry_t, int16_t, operation::ListSum>;
     } break;
     case LogicalTypeID::DOUBLE: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, double_t, operation::ListSum>;
+            UnaryExecListStructFunction<list_entry_t, double_t, operation::ListSum>;
     } break;
     case LogicalTypeID::FLOAT: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, float_t, operation::ListSum>;
+            UnaryExecListStructFunction<list_entry_t, float_t, operation::ListSum>;
     } break;
     default: {
         throw common::NotImplementedException("ListSumVectorOperation::bindFunc");
@@ -539,47 +538,47 @@
     switch (VarListType::getChildType(&arguments[0]->dataType)->getLogicalTypeID()) {
     case LogicalTypeID::SERIAL:
     case LogicalTypeID::INT64: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<int64_t>>;
     } break;
     case LogicalTypeID::INT32: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<int32_t>>;
     } break;
     case LogicalTypeID::INT16: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<int16_t>>;
     } break;
     case LogicalTypeID::DOUBLE: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<double_t>>;
     } break;
     case LogicalTypeID::FLOAT: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<float_t>>;
     } break;
     case LogicalTypeID::BOOL: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<uint8_t>>;
     } break;
     case LogicalTypeID::STRING: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<ku_string_t>>;
     } break;
     case LogicalTypeID::DATE: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<date_t>>;
     } break;
     case LogicalTypeID::TIMESTAMP: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<timestamp_t>>;
     } break;
     case LogicalTypeID::INTERVAL: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<interval_t>>;
     } break;
     case LogicalTypeID::INTERNAL_ID: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t,
             list_entry_t, operation::ListDistinct<internalID_t>>;
     } break;
     default: {
@@ -603,48 +602,48 @@
     switch (common::VarListType::getChildType(&arguments[0]->dataType)->getLogicalTypeID()) {
     case LogicalTypeID::SERIAL:
     case LogicalTypeID::INT64: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<int64_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<int64_t>>;
     } break;
     case LogicalTypeID::INT32: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<int32_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<int32_t>>;
     } break;
     case LogicalTypeID::INT16: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<int16_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<int16_t>>;
     } break;
     case LogicalTypeID::DOUBLE: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<double_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<double_t>>;
     } break;
     case LogicalTypeID::FLOAT: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<float_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<float_t>>;
     } break;
     case LogicalTypeID::BOOL: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<uint8_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<uint8_t>>;
     } break;
     case LogicalTypeID::STRING: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<ku_string_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<ku_string_t>>;
     } break;
     case LogicalTypeID::DATE: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<date_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<date_t>>;
     } break;
     case LogicalTypeID::TIMESTAMP: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<timestamp_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<timestamp_t>>;
     } break;
     case LogicalTypeID::INTERVAL: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<interval_t>>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListUnique<interval_t>>;
     } break;
     case LogicalTypeID::INTERNAL_ID: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            int64_t, operation::ListUnique<common::internalID_t>>;
+        vectorOperationDefinition->execFunc = UnaryExecListStructFunction<list_entry_t, int64_t,
+            operation::ListUnique<common::internalID_t>>;
     } break;
     default: {
         throw common::NotImplementedException("ListUniqueVectorOperation::bindFunc");
@@ -669,51 +668,51 @@
     case LogicalTypeID::SERIAL:
     case LogicalTypeID::INT64: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, int64_t, operation::ListAnyValue>;
+            UnaryExecListStructFunction<list_entry_t, int64_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::INT32: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, int32_t, operation::ListAnyValue>;
+            UnaryExecListStructFunction<list_entry_t, int32_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::INT16: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, int16_t, operation::ListAnyValue>;
+            UnaryExecListStructFunction<list_entry_t, int16_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::DOUBLE: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, double_t, operation::ListAnyValue>;
+            UnaryExecListStructFunction<list_entry_t, double_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::FLOAT: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, float_t, operation::ListAnyValue>;
+            UnaryExecListStructFunction<list_entry_t, float_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::BOOL: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, uint8_t, operation::ListAnyValue>;
+            UnaryExecListStructFunction<list_entry_t, uint8_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::STRING: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            ku_string_t, operation::ListAnyValue>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, ku_string_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::DATE: {
         vectorOperationDefinition->execFunc =
-            UnaryExecListStructFunctionWithVectors<list_entry_t, date_t, operation::ListAnyValue>;
+            UnaryExecListStructFunction<list_entry_t, date_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::TIMESTAMP: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            timestamp_t, operation::ListAnyValue>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, timestamp_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::INTERVAL: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            interval_t, operation::ListAnyValue>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, interval_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::VAR_LIST: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            list_entry_t, operation::ListAnyValue>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, list_entry_t, operation::ListAnyValue>;
     } break;
     case LogicalTypeID::INTERNAL_ID: {
-        vectorOperationDefinition->execFunc = UnaryExecListStructFunctionWithVectors<list_entry_t,
-            internalID_t, operation::ListAnyValue>;
+        vectorOperationDefinition->execFunc =
+            UnaryExecListStructFunction<list_entry_t, internalID_t, operation::ListAnyValue>;
     } break;
     default: {
         throw common::NotImplementedException("ListAnyValueVectorOperation::bindFunc");