duckdb 0.8.1-dev31.0 → 0.8.1-dev327.0

package/src/duckdb/src/include/duckdb/common/vector_operations/aggregate_executor.hpp

@@ -11,6 +11,7 @@
 #include "duckdb/common/exception.hpp"
 #include "duckdb/common/types/vector.hpp"
 #include "duckdb/common/vector_operations/vector_operations.hpp"
+#include "duckdb/function/aggregate_state.hpp"
 
 namespace duckdb {
 
@@ -24,7 +25,7 @@ private:
 	static inline void NullaryFlatLoop(STATE_TYPE **__restrict states, AggregateInputData &aggr_input_data,
 	                                   idx_t count) {
 		for (idx_t i = 0; i < count; i++) {
-			OP::template Operation<STATE_TYPE, OP>(states[i], aggr_input_data, i);
+			OP::template Operation<STATE_TYPE, OP>(*states[i], aggr_input_data, i);
 		}
 	}
 
@@ -34,24 +35,25 @@ private:
 
 		for (idx_t i = 0; i < count; i++) {
 			auto sidx = ssel.get_index(i);
-			OP::template Operation<STATE_TYPE, OP>(states[sidx], aggr_input_data, sidx);
+			OP::template Operation<STATE_TYPE, OP>(*states[sidx], aggr_input_data, sidx);
 		}
 	}
 
 	template <class STATE_TYPE, class INPUT_TYPE, class OP>
-	static inline void UnaryFlatLoop(INPUT_TYPE *__restrict idata, AggregateInputData &aggr_input_data,
+	static inline void UnaryFlatLoop(const INPUT_TYPE *__restrict idata, AggregateInputData &aggr_input_data,
 	                                 STATE_TYPE **__restrict states, ValidityMask &mask, idx_t count) {
-		if (!mask.AllValid()) {
-			idx_t base_idx = 0;
+		if (OP::IgnoreNull() && !mask.AllValid()) {
+			AggregateUnaryInput input(aggr_input_data, mask);
+			auto &base_idx = input.input_idx;
+			base_idx = 0;
 			auto entry_count = ValidityMask::EntryCount(count);
 			for (idx_t entry_idx = 0; entry_idx < entry_count; entry_idx++) {
 				auto validity_entry = mask.GetValidityEntry(entry_idx);
 				idx_t next = MinValue<idx_t>(base_idx + ValidityMask::BITS_PER_VALUE, count);
-				if (!OP::IgnoreNull() || ValidityMask::AllValid(validity_entry)) {
+				if (ValidityMask::AllValid(validity_entry)) {
 					// all valid: perform operation
 					for (; base_idx < next; base_idx++) {
-						OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(states[base_idx], aggr_input_data, idata,
-						                                                   mask, base_idx);
+						OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*states[base_idx], idata[base_idx], input);
 					}
 				} else if (ValidityMask::NoneValid(validity_entry)) {
 					// nothing valid: skip all
@@ -62,46 +64,52 @@ private:
 					idx_t start = base_idx;
 					for (; base_idx < next; base_idx++) {
 						if (ValidityMask::RowIsValid(validity_entry, base_idx - start)) {
-							OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(states[base_idx], aggr_input_data, idata,
-							                                                   mask, base_idx);
+							OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*states[base_idx], idata[base_idx],
+							                                                   input);
 						}
 					}
 				}
 			}
 		} else {
-			for (idx_t i = 0; i < count; i++) {
-				OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(states[i], aggr_input_data, idata, mask, i);
+			AggregateUnaryInput input(aggr_input_data, mask);
+			auto &i = input.input_idx;
+			for (i = 0; i < count; i++) {
+				OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*states[i], idata[i], input);
 			}
 		}
 	}
 
 	template <class STATE_TYPE, class INPUT_TYPE, class OP>
-	static inline void UnaryScatterLoop(INPUT_TYPE *__restrict idata, AggregateInputData &aggr_input_data,
+	static inline void UnaryScatterLoop(const INPUT_TYPE *__restrict idata, AggregateInputData &aggr_input_data,
 	                                    STATE_TYPE **__restrict states, const SelectionVector &isel,
 	                                    const SelectionVector &ssel, ValidityMask &mask, idx_t count) {
 		if (OP::IgnoreNull() && !mask.AllValid()) {
 			// potential NULL values and NULL values are ignored
+			AggregateUnaryInput input(aggr_input_data, mask);
 			for (idx_t i = 0; i < count; i++) {
-				auto idx = isel.get_index(i);
+				input.input_idx = isel.get_index(i);
 				auto sidx = ssel.get_index(i);
-				if (mask.RowIsValid(idx)) {
-					OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(states[sidx], aggr_input_data, idata, mask, idx);
+				if (mask.RowIsValid(input.input_idx)) {
+					OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*states[sidx], idata[input.input_idx], input);
 				}
 			}
 		} else {
 			// quick path: no NULL values or NULL values are not ignored
+			AggregateUnaryInput input(aggr_input_data, mask);
 			for (idx_t i = 0; i < count; i++) {
-				auto idx = isel.get_index(i);
+				input.input_idx = isel.get_index(i);
 				auto sidx = ssel.get_index(i);
-				OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(states[sidx], aggr_input_data, idata, mask, idx);
+				OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*states[sidx], idata[input.input_idx], input);
 			}
 		}
 	}
 
 	template <class STATE_TYPE, class INPUT_TYPE, class OP>
-	static inline void UnaryFlatUpdateLoop(INPUT_TYPE *__restrict idata, AggregateInputData &aggr_input_data,
+	static inline void UnaryFlatUpdateLoop(const INPUT_TYPE *__restrict idata, AggregateInputData &aggr_input_data,
 	                                       STATE_TYPE *__restrict state, idx_t count, ValidityMask &mask) {
-		idx_t base_idx = 0;
+		AggregateUnaryInput input(aggr_input_data, mask);
+		auto &base_idx = input.input_idx;
+		base_idx = 0;
 		auto entry_count = ValidityMask::EntryCount(count);
 		for (idx_t entry_idx = 0; entry_idx < entry_count; entry_idx++) {
 			auto validity_entry = mask.GetValidityEntry(entry_idx);
@@ -109,7 +117,7 @@ private:
 			if (!OP::IgnoreNull() || ValidityMask::AllValid(validity_entry)) {
 				// all valid: perform operation
 				for (; base_idx < next; base_idx++) {
-					OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(state, aggr_input_data, idata, mask, base_idx);
+					OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*state, idata[base_idx], input);
 				}
 			} else if (ValidityMask::NoneValid(validity_entry)) {
 				// nothing valid: skip all
@@ -120,8 +128,7 @@ private:
 				idx_t start = base_idx;
 				for (; base_idx < next; base_idx++) {
 					if (ValidityMask::RowIsValid(validity_entry, base_idx - start)) {
-						OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(state, aggr_input_data, idata, mask,
-						                                                   base_idx);
+						OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*state, idata[base_idx], input);
 					}
 				}
 			}
@@ -129,77 +136,80 @@ private:
 	}
 
 	template <class STATE_TYPE, class INPUT_TYPE, class OP>
-	static inline void UnaryUpdateLoop(INPUT_TYPE *__restrict idata, AggregateInputData &aggr_input_data,
+	static inline void UnaryUpdateLoop(const INPUT_TYPE *__restrict idata, AggregateInputData &aggr_input_data,
 	                                   STATE_TYPE *__restrict state, idx_t count, ValidityMask &mask,
 	                                   const SelectionVector &__restrict sel_vector) {
+		AggregateUnaryInput input(aggr_input_data, mask);
 		if (OP::IgnoreNull() && !mask.AllValid()) {
 			// potential NULL values and NULL values are ignored
 			for (idx_t i = 0; i < count; i++) {
-				auto idx = sel_vector.get_index(i);
-				if (mask.RowIsValid(idx)) {
-					OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(state, aggr_input_data, idata, mask, idx);
+				input.input_idx = sel_vector.get_index(i);
+				if (mask.RowIsValid(input.input_idx)) {
+					OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*state, idata[input.input_idx], input);
 				}
 			}
 		} else {
 			// quick path: no NULL values or NULL values are not ignored
 			for (idx_t i = 0; i < count; i++) {
-				auto idx = sel_vector.get_index(i);
-				OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(state, aggr_input_data, idata, mask, idx);
+				input.input_idx = sel_vector.get_index(i);
+				OP::template Operation<INPUT_TYPE, STATE_TYPE, OP>(*state, idata[input.input_idx], input);
 			}
 		}
 	}
 
 	template <class STATE_TYPE, class A_TYPE, class B_TYPE, class OP>
-	static inline void BinaryScatterLoop(A_TYPE *__restrict adata, AggregateInputData &aggr_input_data,
-	                                     B_TYPE *__restrict bdata, STATE_TYPE **__restrict states, idx_t count,
+	static inline void BinaryScatterLoop(const A_TYPE *__restrict adata, AggregateInputData &aggr_input_data,
+	                                     const B_TYPE *__restrict bdata, STATE_TYPE **__restrict states, idx_t count,
 	                                     const SelectionVector &asel, const SelectionVector &bsel,
 	                                     const SelectionVector &ssel, ValidityMask &avalidity,
 	                                     ValidityMask &bvalidity) {
+		AggregateBinaryInput input(aggr_input_data, avalidity, bvalidity);
 		if (OP::IgnoreNull() && (!avalidity.AllValid() || !bvalidity.AllValid())) {
 			// potential NULL values and NULL values are ignored
 			for (idx_t i = 0; i < count; i++) {
-				auto aidx = asel.get_index(i);
-				auto bidx = bsel.get_index(i);
+				input.lidx = asel.get_index(i);
+				input.ridx = bsel.get_index(i);
 				auto sidx = ssel.get_index(i);
-				if (avalidity.RowIsValid(aidx) && bvalidity.RowIsValid(bidx)) {
-					OP::template Operation<A_TYPE, B_TYPE, STATE_TYPE, OP>(states[sidx], aggr_input_data, adata, bdata,
-					                                                       avalidity, bvalidity, aidx, bidx);
+				if (avalidity.RowIsValid(input.lidx) && bvalidity.RowIsValid(input.ridx)) {
+					OP::template Operation<A_TYPE, B_TYPE, STATE_TYPE, OP>(*states[sidx], adata[input.lidx],
+					                                                       bdata[input.ridx], input);
 				}
 			}
 		} else {
 			// quick path: no NULL values or NULL values are not ignored
 			for (idx_t i = 0; i < count; i++) {
-				auto aidx = asel.get_index(i);
-				auto bidx = bsel.get_index(i);
+				input.lidx = asel.get_index(i);
+				input.ridx = bsel.get_index(i);
 				auto sidx = ssel.get_index(i);
-				OP::template Operation<A_TYPE, B_TYPE, STATE_TYPE, OP>(states[sidx], aggr_input_data, adata, bdata,
-				                                                       avalidity, bvalidity, aidx, bidx);
+				OP::template Operation<A_TYPE, B_TYPE, STATE_TYPE, OP>(*states[sidx], adata[input.lidx],
+				                                                       bdata[input.ridx], input);
 			}
 		}
 	}
 
 	template <class STATE_TYPE, class A_TYPE, class B_TYPE, class OP>
-	static inline void BinaryUpdateLoop(A_TYPE *__restrict adata, AggregateInputData &aggr_input_data,
-	                                    B_TYPE *__restrict bdata, STATE_TYPE *__restrict state, idx_t count,
+	static inline void BinaryUpdateLoop(const A_TYPE *__restrict adata, AggregateInputData &aggr_input_data,
+	                                    const B_TYPE *__restrict bdata, STATE_TYPE *__restrict state, idx_t count,
 	                                    const SelectionVector &asel, const SelectionVector &bsel,
 	                                    ValidityMask &avalidity, ValidityMask &bvalidity) {
+		AggregateBinaryInput input(aggr_input_data, avalidity, bvalidity);
 		if (OP::IgnoreNull() && (!avalidity.AllValid() || !bvalidity.AllValid())) {
 			// potential NULL values and NULL values are ignored
 			for (idx_t i = 0; i < count; i++) {
-				auto aidx = asel.get_index(i);
-				auto bidx = bsel.get_index(i);
-				if (avalidity.RowIsValid(aidx) && bvalidity.RowIsValid(bidx)) {
-					OP::template Operation<A_TYPE, B_TYPE, STATE_TYPE, OP>(state, aggr_input_data, adata, bdata,
-					                                                       avalidity, bvalidity, aidx, bidx);
+				input.lidx = asel.get_index(i);
+				input.ridx = bsel.get_index(i);
+				if (avalidity.RowIsValid(input.lidx) && bvalidity.RowIsValid(input.ridx)) {
+					OP::template Operation<A_TYPE, B_TYPE, STATE_TYPE, OP>(*state, adata[input.lidx], bdata[input.ridx],
+					                                                       input);
 				}
 			}
 		} else {
 			// quick path: no NULL values or NULL values are not ignored
 			for (idx_t i = 0; i < count; i++) {
-				auto aidx = asel.get_index(i);
-				auto bidx = bsel.get_index(i);
-				OP::template Operation<A_TYPE, B_TYPE, STATE_TYPE, OP>(state, aggr_input_data, adata, bdata, avalidity,
-				                                                       bvalidity, aidx, bidx);
+				input.lidx = asel.get_index(i);
+				input.ridx = bsel.get_index(i);
+				OP::template Operation<A_TYPE, B_TYPE, STATE_TYPE, OP>(*state, adata[input.lidx], bdata[input.ridx],
+				                                                       input);
 			}
 		}
 	}
@@ -209,7 +219,7 @@ public:
 	static void NullaryScatter(Vector &states, AggregateInputData &aggr_input_data, idx_t count) {
 		if (states.GetVectorType() == VectorType::CONSTANT_VECTOR) {
 			auto sdata = ConstantVector::GetData<STATE_TYPE *>(states);
-			OP::template ConstantOperation<STATE_TYPE, OP>(*sdata, aggr_input_data, count);
+			OP::template ConstantOperation<STATE_TYPE, OP>(**sdata, aggr_input_data, count);
 		} else if (states.GetVectorType() == VectorType::FLAT_VECTOR) {
 			auto sdata = FlatVector::GetData<STATE_TYPE *>(states);
 			NullaryFlatLoop<STATE_TYPE, OP>(sdata, aggr_input_data, count);
@@ -222,7 +232,7 @@ public:
 
 	template <class STATE_TYPE, class OP>
 	static void NullaryUpdate(data_ptr_t state, AggregateInputData &aggr_input_data, idx_t count) {
-		OP::template ConstantOperation<STATE_TYPE, OP>((STATE_TYPE *)state, aggr_input_data, count);
+		OP::template ConstantOperation<STATE_TYPE, OP>(*reinterpret_cast<STATE_TYPE *>(state), aggr_input_data, count);
 	}
 
 	template <class STATE_TYPE, class INPUT_TYPE, class OP>
@@ -236,8 +246,8 @@ public:
 			// regular constant: get first state
 			auto idata = ConstantVector::GetData<INPUT_TYPE>(input);
 			auto sdata = ConstantVector::GetData<STATE_TYPE *>(states);
-			OP::template ConstantOperation<INPUT_TYPE, STATE_TYPE, OP>(*sdata, aggr_input_data, idata,
-			                                                           ConstantVector::Validity(input), count);
+			AggregateUnaryInput input_data(aggr_input_data, ConstantVector::Validity(input));
+			OP::template ConstantOperation<INPUT_TYPE, STATE_TYPE, OP>(**sdata, *idata, input_data, count);
 		} else if (input.GetVectorType() == VectorType::FLAT_VECTOR &&
 		           states.GetVectorType() == VectorType::FLAT_VECTOR) {
 			auto idata = FlatVector::GetData<INPUT_TYPE>(input);
@@ -248,9 +258,9 @@ public:
 			UnifiedVectorFormat idata, sdata;
 			input.ToUnifiedFormat(count, idata);
 			states.ToUnifiedFormat(count, sdata);
-			UnaryScatterLoop<STATE_TYPE, INPUT_TYPE, OP>((INPUT_TYPE *)idata.data, aggr_input_data,
-			                                             (STATE_TYPE **)sdata.data, *idata.sel, *sdata.sel,
-			                                             idata.validity, count);
+			UnaryScatterLoop<STATE_TYPE, INPUT_TYPE, OP>(UnifiedVectorFormat::GetData<INPUT_TYPE>(idata),
+			                                             aggr_input_data, (STATE_TYPE **)sdata.data, *idata.sel,
+			                                             *sdata.sel, idata.validity, count);
 		}
 	}
 
@@ -262,8 +272,9 @@ public:
 				return;
 			}
 			auto idata = ConstantVector::GetData<INPUT_TYPE>(input);
-			OP::template ConstantOperation<INPUT_TYPE, STATE_TYPE, OP>((STATE_TYPE *)state, aggr_input_data, idata,
-			                                                           ConstantVector::Validity(input), count);
+			AggregateUnaryInput input_data(aggr_input_data, ConstantVector::Validity(input));
+			OP::template ConstantOperation<INPUT_TYPE, STATE_TYPE, OP>(*reinterpret_cast<STATE_TYPE *>(state), *idata,
+			                                                           input_data, count);
 			break;
 		}
 		case VectorType::FLAT_VECTOR: {
@@ -275,8 +286,9 @@ public:
 		default: {
 			UnifiedVectorFormat idata;
 			input.ToUnifiedFormat(count, idata);
-			UnaryUpdateLoop<STATE_TYPE, INPUT_TYPE, OP>((INPUT_TYPE *)idata.data, aggr_input_data, (STATE_TYPE *)state,
-			                                            count, idata.validity, *idata.sel);
+			UnaryUpdateLoop<STATE_TYPE, INPUT_TYPE, OP>(UnifiedVectorFormat::GetData<INPUT_TYPE>(idata),
+			                                            aggr_input_data, (STATE_TYPE *)state, count, idata.validity,
+			                                            *idata.sel);
 			break;
 		}
 		}
@@ -290,9 +302,9 @@ public:
 		b.ToUnifiedFormat(count, bdata);
 		states.ToUnifiedFormat(count, sdata);
 
-		BinaryScatterLoop<STATE_TYPE, A_TYPE, B_TYPE, OP>((A_TYPE *)adata.data, aggr_input_data, (B_TYPE *)bdata.data,
-		                                                  (STATE_TYPE **)sdata.data, count, *adata.sel, *bdata.sel,
-		                                                  *sdata.sel, adata.validity, bdata.validity);
+		BinaryScatterLoop<STATE_TYPE, A_TYPE, B_TYPE, OP>(
+		    UnifiedVectorFormat::GetData<A_TYPE>(adata), aggr_input_data, UnifiedVectorFormat::GetData<B_TYPE>(bdata),
+		    (STATE_TYPE **)sdata.data, count, *adata.sel, *bdata.sel, *sdata.sel, adata.validity, bdata.validity);
 	}
 
 	template <class STATE_TYPE, class A_TYPE, class B_TYPE, class OP>
@@ -302,9 +314,9 @@ public:
 		a.ToUnifiedFormat(count, adata);
 		b.ToUnifiedFormat(count, bdata);
 
-		BinaryUpdateLoop<STATE_TYPE, A_TYPE, B_TYPE, OP>((A_TYPE *)adata.data, aggr_input_data, (B_TYPE *)bdata.data,
-		                                                 (STATE_TYPE *)state, count, *adata.sel, *bdata.sel,
-		                                                 adata.validity, bdata.validity);
+		BinaryUpdateLoop<STATE_TYPE, A_TYPE, B_TYPE, OP>(
+		    UnifiedVectorFormat::GetData<A_TYPE>(adata), aggr_input_data, UnifiedVectorFormat::GetData<B_TYPE>(bdata),
+		    (STATE_TYPE *)state, count, *adata.sel, *bdata.sel, adata.validity, bdata.validity);
 	}
 
 	template <class STATE_TYPE, class OP>
@@ -314,7 +326,7 @@ public:
 		auto tdata = FlatVector::GetData<STATE_TYPE *>(target);
 
 		for (idx_t i = 0; i < count; i++) {
-			OP::template Combine<STATE_TYPE, OP>(*sdata[i], tdata[i], aggr_input_data);
+			OP::template Combine<STATE_TYPE, OP>(*sdata[i], *tdata[i], aggr_input_data);
 		}
 	}
 
@@ -326,17 +338,41 @@ public:
 
 			auto sdata = ConstantVector::GetData<STATE_TYPE *>(states);
 			auto rdata = ConstantVector::GetData<RESULT_TYPE>(result);
-			OP::template Finalize<RESULT_TYPE, STATE_TYPE>(result, aggr_input_data, *sdata, rdata,
-			                                               ConstantVector::Validity(result), 0);
+			AggregateFinalizeData finalize_data(result, aggr_input_data);
+			OP::template Finalize<RESULT_TYPE, STATE_TYPE>(**sdata, *rdata, finalize_data);
 		} else {
 			D_ASSERT(states.GetVectorType() == VectorType::FLAT_VECTOR);
 			result.SetVectorType(VectorType::FLAT_VECTOR);
 
 			auto sdata = FlatVector::GetData<STATE_TYPE *>(states);
 			auto rdata = FlatVector::GetData<RESULT_TYPE>(result);
+			AggregateFinalizeData finalize_data(result, aggr_input_data);
+			for (idx_t i = 0; i < count; i++) {
+				finalize_data.result_idx = i + offset;
+				OP::template Finalize<RESULT_TYPE, STATE_TYPE>(*sdata[i], rdata[finalize_data.result_idx],
+				                                               finalize_data);
+			}
+		}
+	}
+
+	template <class STATE_TYPE, class OP>
+	static void VoidFinalize(Vector &states, AggregateInputData &aggr_input_data, Vector &result, idx_t count,
+	                         idx_t offset) {
+		if (states.GetVectorType() == VectorType::CONSTANT_VECTOR) {
+			result.SetVectorType(VectorType::CONSTANT_VECTOR);
+
+			auto sdata = ConstantVector::GetData<STATE_TYPE *>(states);
+			AggregateFinalizeData finalize_data(result, aggr_input_data);
+			OP::template Finalize<STATE_TYPE>(**sdata, finalize_data);
+		} else {
+			D_ASSERT(states.GetVectorType() == VectorType::FLAT_VECTOR);
+			result.SetVectorType(VectorType::FLAT_VECTOR);
+
+			auto sdata = FlatVector::GetData<STATE_TYPE *>(states);
+			AggregateFinalizeData finalize_data(result, aggr_input_data);
 			for (idx_t i = 0; i < count; i++) {
-				OP::template Finalize<RESULT_TYPE, STATE_TYPE>(result, aggr_input_data, sdata[i], rdata,
-				                                               FlatVector::Validity(result), i + offset);
+				finalize_data.result_idx = i + offset;
+				OP::template Finalize<STATE_TYPE>(*sdata[i], finalize_data);
 			}
 		}
 	}
@@ -348,15 +384,15 @@ public:
 
 		auto idata = FlatVector::GetData<const INPUT_TYPE>(input) - bias;
 		const auto &ivalid = FlatVector::Validity(input);
-		OP::template Window<STATE, INPUT_TYPE, RESULT_TYPE>(idata, ifilter, ivalid, aggr_input_data, (STATE *)state,
-		                                                    frame, prev, result, rid, bias);
+		OP::template Window<STATE, INPUT_TYPE, RESULT_TYPE>(
+		    idata, ifilter, ivalid, aggr_input_data, *reinterpret_cast<STATE *>(state), frame, prev, result, rid, bias);
 	}
 
 	template <class STATE_TYPE, class OP>
 	static void Destroy(Vector &states, AggregateInputData &aggr_input_data, idx_t count) {
 		auto sdata = FlatVector::GetData<STATE_TYPE *>(states);
 		for (idx_t i = 0; i < count; i++) {
-			OP::template Destroy<STATE_TYPE>(aggr_input_data, sdata[i]);
+			OP::template Destroy<STATE_TYPE>(*sdata[i], aggr_input_data);
 		}
 	}
 };

package/src/duckdb/src/include/duckdb/common/vector_operations/binary_executor.hpp

@@ -70,7 +70,7 @@ struct BinaryLambdaWrapperWithNulls {
 struct BinaryExecutor {
 	template <class LEFT_TYPE, class RIGHT_TYPE, class RESULT_TYPE, class OPWRAPPER, class OP, class FUNC,
 	          bool LEFT_CONSTANT, bool RIGHT_CONSTANT>
-	static void ExecuteFlatLoop(LEFT_TYPE *__restrict ldata, RIGHT_TYPE *__restrict rdata,
+	static void ExecuteFlatLoop(const LEFT_TYPE *__restrict ldata, const RIGHT_TYPE *__restrict rdata,
 	                            RESULT_TYPE *__restrict result_data, idx_t count, ValidityMask &mask, FUNC fun) {
 		if (!LEFT_CONSTANT) {
 			ASSERT_RESTRICT(ldata, ldata + count, result_data, result_data + count);
@@ -184,7 +184,7 @@ struct BinaryExecutor {
 	}
 
 	template <class LEFT_TYPE, class RIGHT_TYPE, class RESULT_TYPE, class OPWRAPPER, class OP, class FUNC>
-	static void ExecuteGenericLoop(LEFT_TYPE *__restrict ldata, RIGHT_TYPE *__restrict rdata,
+	static void ExecuteGenericLoop(const LEFT_TYPE *__restrict ldata, const RIGHT_TYPE *__restrict rdata,
 	                               RESULT_TYPE *__restrict result_data, const SelectionVector *__restrict lsel,
 	                               const SelectionVector *__restrict rsel, idx_t count, ValidityMask &lvalidity,
 	                               ValidityMask &rvalidity, ValidityMask &result_validity, FUNC fun) {
@@ -221,8 +221,9 @@ struct BinaryExecutor {
 		result.SetVectorType(VectorType::FLAT_VECTOR);
 		auto result_data = FlatVector::GetData<RESULT_TYPE>(result);
 		ExecuteGenericLoop<LEFT_TYPE, RIGHT_TYPE, RESULT_TYPE, OPWRAPPER, OP, FUNC>(
-		    (LEFT_TYPE *)ldata.data, (RIGHT_TYPE *)rdata.data, result_data, ldata.sel, rdata.sel, count, ldata.validity,
-		    rdata.validity, FlatVector::Validity(result), fun);
+		    UnifiedVectorFormat::GetData<LEFT_TYPE>(ldata), UnifiedVectorFormat::GetData<RIGHT_TYPE>(rdata),
+		    result_data, ldata.sel, rdata.sel, count, ldata.validity, rdata.validity, FlatVector::Validity(result),
+		    fun);
 	}
 
 	template <class LEFT_TYPE, class RIGHT_TYPE, class RESULT_TYPE, class OPWRAPPER, class OP, class FUNC>
@@ -300,7 +301,7 @@ public:
 
 	template <class LEFT_TYPE, class RIGHT_TYPE, class OP, bool LEFT_CONSTANT, bool RIGHT_CONSTANT, bool HAS_TRUE_SEL,
 	          bool HAS_FALSE_SEL>
-	static inline idx_t SelectFlatLoop(LEFT_TYPE *__restrict ldata, RIGHT_TYPE *__restrict rdata,
+	static inline idx_t SelectFlatLoop(const LEFT_TYPE *__restrict ldata, const RIGHT_TYPE *__restrict rdata,
 	                                   const SelectionVector *sel, idx_t count, ValidityMask &validity_mask,
 	                                   SelectionVector *true_sel, SelectionVector *false_sel) {
 		idx_t true_count = 0, false_count = 0;
@@ -364,7 +365,7 @@ public:
 	}
 
 	template <class LEFT_TYPE, class RIGHT_TYPE, class OP, bool LEFT_CONSTANT, bool RIGHT_CONSTANT>
-	static inline idx_t SelectFlatLoopSwitch(LEFT_TYPE *__restrict ldata, RIGHT_TYPE *__restrict rdata,
+	static inline idx_t SelectFlatLoopSwitch(const LEFT_TYPE *__restrict ldata, const RIGHT_TYPE *__restrict rdata,
 	                                         const SelectionVector *sel, idx_t count, ValidityMask &mask,
 	                                         SelectionVector *true_sel, SelectionVector *false_sel) {
 		if (true_sel && false_sel) {
@@ -419,10 +420,10 @@ public:
 
 	template <class LEFT_TYPE, class RIGHT_TYPE, class OP, bool NO_NULL, bool HAS_TRUE_SEL, bool HAS_FALSE_SEL>
 	static inline idx_t
-	SelectGenericLoop(LEFT_TYPE *__restrict ldata, RIGHT_TYPE *__restrict rdata, const SelectionVector *__restrict lsel,
-	                  const SelectionVector *__restrict rsel, const SelectionVector *__restrict result_sel, idx_t count,
-	                  ValidityMask &lvalidity, ValidityMask &rvalidity, SelectionVector *true_sel,
-	                  SelectionVector *false_sel) {
+	SelectGenericLoop(const LEFT_TYPE *__restrict ldata, const RIGHT_TYPE *__restrict rdata,
+	                  const SelectionVector *__restrict lsel, const SelectionVector *__restrict rsel,
+	                  const SelectionVector *__restrict result_sel, idx_t count, ValidityMask &lvalidity,
+	                  ValidityMask &rvalidity, SelectionVector *true_sel, SelectionVector *false_sel) {
 		idx_t true_count = 0, false_count = 0;
 		for (idx_t i = 0; i < count; i++) {
 			auto result_idx = result_sel->get_index(i);
@@ -447,7 +448,7 @@ public:
 	}
 	template <class LEFT_TYPE, class RIGHT_TYPE, class OP, bool NO_NULL>
 	static inline idx_t
-	SelectGenericLoopSelSwitch(LEFT_TYPE *__restrict ldata, RIGHT_TYPE *__restrict rdata,
+	SelectGenericLoopSelSwitch(const LEFT_TYPE *__restrict ldata, const RIGHT_TYPE *__restrict rdata,
 	                           const SelectionVector *__restrict lsel, const SelectionVector *__restrict rsel,
 	                           const SelectionVector *__restrict result_sel, idx_t count, ValidityMask &lvalidity,
 	                           ValidityMask &rvalidity, SelectionVector *true_sel, SelectionVector *false_sel) {
@@ -466,7 +467,7 @@ public:
 
 	template <class LEFT_TYPE, class RIGHT_TYPE, class OP>
 	static inline idx_t
-	SelectGenericLoopSwitch(LEFT_TYPE *__restrict ldata, RIGHT_TYPE *__restrict rdata,
+	SelectGenericLoopSwitch(const LEFT_TYPE *__restrict ldata, const RIGHT_TYPE *__restrict rdata,
 	                        const SelectionVector *__restrict lsel, const SelectionVector *__restrict rsel,
 	                        const SelectionVector *__restrict result_sel, idx_t count, ValidityMask &lvalidity,
 	                        ValidityMask &rvalidity, SelectionVector *true_sel, SelectionVector *false_sel) {
@@ -487,9 +488,9 @@ public:
 		left.ToUnifiedFormat(count, ldata);
 		right.ToUnifiedFormat(count, rdata);
 
-		return SelectGenericLoopSwitch<LEFT_TYPE, RIGHT_TYPE, OP>((LEFT_TYPE *)ldata.data, (RIGHT_TYPE *)rdata.data,
-		                                                          ldata.sel, rdata.sel, sel, count, ldata.validity,
-		                                                          rdata.validity, true_sel, false_sel);
+		return SelectGenericLoopSwitch<LEFT_TYPE, RIGHT_TYPE, OP>(
+		    UnifiedVectorFormat::GetData<LEFT_TYPE>(ldata), UnifiedVectorFormat::GetData<RIGHT_TYPE>(rdata), ldata.sel,
+		    rdata.sel, sel, count, ldata.validity, rdata.validity, true_sel, false_sel);
 	}
 
 	template <class LEFT_TYPE, class RIGHT_TYPE, class OP>

package/src/duckdb/src/include/duckdb/common/vector_operations/generic_executor.hpp

@@ -38,7 +38,7 @@ struct PrimitiveType {
 	static bool ConstructType(STRUCT_STATE &state, idx_t i, PrimitiveType<INPUT_TYPE> &result) {
 		auto &vdata = state.main_data;
 		auto idx = vdata.sel->get_index(i);
-		auto ptr = (INPUT_TYPE *)vdata.data;
+		auto ptr = UnifiedVectorFormat::GetData<INPUT_TYPE>(vdata);
 		result.val = ptr[idx];
 		return true;
 	}
@@ -77,7 +77,7 @@ struct StructTypeUnary {
 		if (!a_data.validity.RowIsValid(a_idx)) {
 			return false;
 		}
-		auto a_ptr = (A_TYPE *)a_data.data;
+		auto a_ptr = UnifiedVectorFormat::GetData<A_TYPE>(a_data);
 		result.a_val = a_ptr[a_idx];
 		return true;
 	}
@@ -106,8 +106,8 @@ struct StructTypeBinary {
 		if (!a_data.validity.RowIsValid(a_idx) || !b_data.validity.RowIsValid(b_idx)) {
 			return false;
 		}
-		auto a_ptr = (A_TYPE *)a_data.data;
-		auto b_ptr = (B_TYPE *)b_data.data;
+		auto a_ptr = UnifiedVectorFormat::GetData<A_TYPE>(a_data);
+		auto b_ptr = UnifiedVectorFormat::GetData<B_TYPE>(b_data);
 		result.a_val = a_ptr[a_idx];
 		result.b_val = b_ptr[b_idx];
 		return true;
@@ -143,9 +143,9 @@ struct StructTypeTernary {
 		    !c_data.validity.RowIsValid(c_idx)) {
 			return false;
 		}
-		auto a_ptr = (A_TYPE *)a_data.data;
-		auto b_ptr = (B_TYPE *)b_data.data;
-		auto c_ptr = (C_TYPE *)c_data.data;
+		auto a_ptr = UnifiedVectorFormat::GetData<A_TYPE>(a_data);
+		auto b_ptr = UnifiedVectorFormat::GetData<B_TYPE>(b_data);
+		auto c_ptr = UnifiedVectorFormat::GetData<C_TYPE>(c_data);
 		result.a_val = a_ptr[a_idx];
 		result.b_val = b_ptr[b_idx];
 		result.c_val = c_ptr[c_idx];
@@ -188,10 +188,10 @@ struct StructTypeQuaternary {
 		    !c_data.validity.RowIsValid(c_idx) || !d_data.validity.RowIsValid(d_idx)) {
 			return false;
 		}
-		auto a_ptr = (A_TYPE *)a_data.data;
-		auto b_ptr = (B_TYPE *)b_data.data;
-		auto c_ptr = (C_TYPE *)c_data.data;
-		auto d_ptr = (D_TYPE *)d_data.data;
+		auto a_ptr = UnifiedVectorFormat::GetData<A_TYPE>(a_data);
+		auto b_ptr = UnifiedVectorFormat::GetData<B_TYPE>(b_data);
+		auto c_ptr = UnifiedVectorFormat::GetData<C_TYPE>(c_data);
+		auto d_ptr = UnifiedVectorFormat::GetData<D_TYPE>(d_data);
 		result.a_val = a_ptr[a_idx];
 		result.b_val = b_ptr[b_idx];
 		result.c_val = c_ptr[c_idx];

package/src/duckdb/src/include/duckdb/common/vector_operations/ternary_executor.hpp

@@ -41,11 +41,11 @@ struct TernaryLambdaWrapperWithNulls {
 struct TernaryExecutor {
 private:
 	template <class A_TYPE, class B_TYPE, class C_TYPE, class RESULT_TYPE, class OPWRAPPER, class FUN>
-	static inline void ExecuteLoop(A_TYPE *__restrict adata, B_TYPE *__restrict bdata, C_TYPE *__restrict cdata,
-	                               RESULT_TYPE *__restrict result_data, idx_t count, const SelectionVector &asel,
-	                               const SelectionVector &bsel, const SelectionVector &csel, ValidityMask &avalidity,
-	                               ValidityMask &bvalidity, ValidityMask &cvalidity, ValidityMask &result_validity,
-	                               FUN fun) {
+	static inline void ExecuteLoop(const A_TYPE *__restrict adata, const B_TYPE *__restrict bdata,
+	                               const C_TYPE *__restrict cdata, RESULT_TYPE *__restrict result_data, idx_t count,
+	                               const SelectionVector &asel, const SelectionVector &bsel,
+	                               const SelectionVector &csel, ValidityMask &avalidity, ValidityMask &bvalidity,
+	                               ValidityMask &cvalidity, ValidityMask &result_validity, FUN fun) {
 		if (!avalidity.AllValid() || !bvalidity.AllValid() || !cvalidity.AllValid()) {
 			for (idx_t i = 0; i < count; i++) {
 				auto aidx = asel.get_index(i);
@@ -95,9 +95,10 @@ public:
 			c.ToUnifiedFormat(count, cdata);
 
 			ExecuteLoop<A_TYPE, B_TYPE, C_TYPE, RESULT_TYPE, OPWRAPPER>(
-			    (A_TYPE *)adata.data, (B_TYPE *)bdata.data, (C_TYPE *)cdata.data,
-			    FlatVector::GetData<RESULT_TYPE>(result), count, *adata.sel, *bdata.sel, *cdata.sel, adata.validity,
-			    bdata.validity, cdata.validity, FlatVector::Validity(result), fun);
+			    UnifiedVectorFormat::GetData<A_TYPE>(adata), UnifiedVectorFormat::GetData<B_TYPE>(bdata),
+			    UnifiedVectorFormat::GetData<C_TYPE>(cdata), FlatVector::GetData<RESULT_TYPE>(result), count,
+			    *adata.sel, *bdata.sel, *cdata.sel, adata.validity, bdata.validity, cdata.validity,
+			    FlatVector::Validity(result), fun);
 		}
 	}
 
@@ -122,11 +123,11 @@ public:
 
 private:
 	template <class A_TYPE, class B_TYPE, class C_TYPE, class OP, bool NO_NULL, bool HAS_TRUE_SEL, bool HAS_FALSE_SEL>
-	static inline idx_t SelectLoop(A_TYPE *__restrict adata, B_TYPE *__restrict bdata, C_TYPE *__restrict cdata,
-	                               const SelectionVector *result_sel, idx_t count, const SelectionVector &asel,
-	                               const SelectionVector &bsel, const SelectionVector &csel, ValidityMask &avalidity,
-	                               ValidityMask &bvalidity, ValidityMask &cvalidity, SelectionVector *true_sel,
-	                               SelectionVector *false_sel) {
+	static inline idx_t SelectLoop(const A_TYPE *__restrict adata, const B_TYPE *__restrict bdata,
+	                               const C_TYPE *__restrict cdata, const SelectionVector *result_sel, idx_t count,
+	                               const SelectionVector &asel, const SelectionVector &bsel,
+	                               const SelectionVector &csel, ValidityMask &avalidity, ValidityMask &bvalidity,
+	                               ValidityMask &cvalidity, SelectionVector *true_sel, SelectionVector *false_sel) {
 		idx_t true_count = 0, false_count = 0;
 		for (idx_t i = 0; i < count; i++) {
 			auto result_idx = result_sel->get_index(i);
@@ -158,17 +159,20 @@ private:
 	                                        SelectionVector *true_sel, SelectionVector *false_sel) {
 		if (true_sel && false_sel) {
 			return SelectLoop<A_TYPE, B_TYPE, C_TYPE, OP, NO_NULL, true, true>(
-			    (A_TYPE *)adata.data, (B_TYPE *)bdata.data, (C_TYPE *)cdata.data, sel, count, *adata.sel, *bdata.sel,
-			    *cdata.sel, adata.validity, bdata.validity, cdata.validity, true_sel, false_sel);
+			    UnifiedVectorFormat::GetData<A_TYPE>(adata), UnifiedVectorFormat::GetData<B_TYPE>(bdata),
+			    UnifiedVectorFormat::GetData<C_TYPE>(cdata), sel, count, *adata.sel, *bdata.sel, *cdata.sel,
+			    adata.validity, bdata.validity, cdata.validity, true_sel, false_sel);
 		} else if (true_sel) {
 			return SelectLoop<A_TYPE, B_TYPE, C_TYPE, OP, NO_NULL, true, false>(
-			    (A_TYPE *)adata.data, (B_TYPE *)bdata.data, (C_TYPE *)cdata.data, sel, count, *adata.sel, *bdata.sel,
-			    *cdata.sel, adata.validity, bdata.validity, cdata.validity, true_sel, false_sel);
+			    UnifiedVectorFormat::GetData<A_TYPE>(adata), UnifiedVectorFormat::GetData<B_TYPE>(bdata),
+			    UnifiedVectorFormat::GetData<C_TYPE>(cdata), sel, count, *adata.sel, *bdata.sel, *cdata.sel,
+			    adata.validity, bdata.validity, cdata.validity, true_sel, false_sel);
 		} else {
 			D_ASSERT(false_sel);
 			return SelectLoop<A_TYPE, B_TYPE, C_TYPE, OP, NO_NULL, false, true>(
-			    (A_TYPE *)adata.data, (B_TYPE *)bdata.data, (C_TYPE *)cdata.data, sel, count, *adata.sel, *bdata.sel,
-			    *cdata.sel, adata.validity, bdata.validity, cdata.validity, true_sel, false_sel);
+			    UnifiedVectorFormat::GetData<A_TYPE>(adata), UnifiedVectorFormat::GetData<B_TYPE>(bdata),
+			    UnifiedVectorFormat::GetData<C_TYPE>(cdata), sel, count, *adata.sel, *bdata.sel, *cdata.sel,
+			    adata.validity, bdata.validity, cdata.validity, true_sel, false_sel);
 		}
 	}