remove ValueVectorUtils

kuzudb · Jun 17, 2023 · 06a35df · 06a35df
1 parent b069813
commit 06a35df
Show file tree

Hide file tree

Showing 26 changed files with 106 additions and 139 deletions.
diff --git a/src/common/vector/CMakeLists.txt b/src/common/vector/CMakeLists.txt
@@ -1,7 +1,6 @@
 add_library(kuzu_common_vector
         OBJECT
         value_vector.cpp
-        value_vector_utils.cpp
         auxiliary_buffer.cpp)
 
 set(ALL_OBJECT_FILES

diff --git a/src/common/vector/value_vector.cpp b/src/common/vector/value_vector.cpp
@@ -70,6 +70,33 @@ void ValueVector::copyFromRowData(uint32_t pos, const uint8_t* rowData) {
     }
 }
 
+void ValueVector::copyToRowData(
+    uint32_t pos, uint8_t* rowData, InMemOverflowBuffer* rowOverflowBuffer) const {
+    switch (dataType.getPhysicalType()) {
+    case PhysicalTypeID::STRUCT: {
+        StructVector::copyToRowData(this, pos, rowData, rowOverflowBuffer);
+    } break;
+    case PhysicalTypeID::VAR_LIST: {
+        ListVector::copyToRowData(this, pos, rowData, rowOverflowBuffer);
+    } break;
+    case PhysicalTypeID::STRING: {
+        auto& srcStr = getValue<ku_string_t>(pos);
+        auto& dstStr = *(ku_string_t*)rowData;
+        if (ku_string_t::isShortString(srcStr.len)) {
+            dstStr.setShortString(srcStr);
+        } else {
+            dstStr.overflowPtr =
+                reinterpret_cast<uint64_t>(rowOverflowBuffer->allocateSpace(srcStr.len));
+            dstStr.setLongString(srcStr);
+        }
+    } break;
+    default: {
+        auto dataTypeSize = LogicalTypeUtils::getRowLayoutSize(dataType);
+        memcpy(rowData, getData() + pos * dataTypeSize, dataTypeSize);
+    }
+    }
+}
+
 void ValueVector::copyFromVectorData(
     uint8_t* dstData, const ValueVector* srcVector, const uint8_t* srcVectorData) {
     assert(srcVector->dataType.getPhysicalType() == dataType.getPhysicalType());
@@ -238,6 +265,30 @@ void ListVector::copyFromRowData(ValueVector* vector, uint32_t pos, const uint8_
     }
 }
 
+void ListVector::copyToRowData(const ValueVector* vector, uint32_t pos, uint8_t* rowData,
+    InMemOverflowBuffer* rowOverflowBuffer) {
+    auto& srcListEntry = vector->getValue<list_entry_t>(pos);
+    auto srcListDataVector = common::ListVector::getDataVector(vector);
+    auto& dstListEntry = *(ku_list_t*)rowData;
+    dstListEntry.size = srcListEntry.size;
+    auto nullBytesSize = NullBuffer::getNumBytesForNullValues(dstListEntry.size);
+    auto dataRowLayoutSize = LogicalTypeUtils::getRowLayoutSize(srcListDataVector->dataType);
+    auto dstListOverflowSize = dataRowLayoutSize * dstListEntry.size + nullBytesSize;
+    auto dstListOverflow = rowOverflowBuffer->allocateSpace(dstListOverflowSize);
+    dstListEntry.overflowPtr = reinterpret_cast<uint64_t>(dstListOverflow);
+    NullBuffer::initNullBytes(dstListOverflow, dstListEntry.size);
+    auto dstListValues = dstListOverflow + nullBytesSize;
+    for (auto i = 0u; i < srcListEntry.size; i++) {
+        if (srcListDataVector->isNull(srcListEntry.offset + i)) {
+            NullBuffer::setNull(dstListOverflow, i);
+        } else {
+            srcListDataVector->copyToRowData(
+                srcListEntry.offset + i, dstListValues, rowOverflowBuffer);
+        }
+        dstListValues += dataRowLayoutSize;
+    }
+}
+
 void ListVector::copyFromVectorData(ValueVector* dstVector, uint8_t* dstData,
     const ValueVector* srcVector, const uint8_t* srcData) {
     auto& srcListEntry = *(common::list_entry_t*)(srcData);
@@ -275,6 +326,25 @@ void StructVector::copyFromRowData(ValueVector* vector, uint32_t pos, const uint
     }
 }
 
+void StructVector::copyToRowData(const ValueVector* vector, uint32_t pos, uint8_t* rowData,
+    InMemOverflowBuffer* rowOverflowBuffer) {
+    // The storage structure of STRUCT type in factorizedTable is:
+    // [NULLBYTES, FIELD1, FIELD2, ...]
+    auto& structFields = StructVector::getFieldVectors(vector);
+    NullBuffer::initNullBytes(rowData, structFields.size());
+    auto structNullBytes = rowData;
+    auto structValues = structNullBytes + NullBuffer::getNumBytesForNullValues(structFields.size());
+    for (auto i = 0u; i < structFields.size(); i++) {
+        auto structField = structFields[i];
+        if (structField->isNull(pos)) {
+            NullBuffer::setNull(structNullBytes, i);
+        } else {
+            structField->copyToRowData(pos, structValues, rowOverflowBuffer);
+        }
+        structValues += LogicalTypeUtils::getRowLayoutSize(structField->dataType);
+    }
+}
+
 void StructVector::copyFromVectorData(ValueVector* dstVector, const uint8_t* dstData,
     const ValueVector* srcVector, const uint8_t* srcData) {
     auto& srcPos = *(int64_t*)srcData;

diff --git a/src/common/vector/value_vector_utils.cpp b/src/common/vector/value_vector_utils.cpp
diff --git a/src/expression_evaluator/case_evaluator.cpp b/src/expression_evaluator/case_evaluator.cpp
@@ -1,7 +1,5 @@
 #include "expression_evaluator/case_evaluator.h"
 
-#include "common/vector/value_vector_utils.h"
-
 using namespace kuzu::common;
 using namespace kuzu::processor;
 using namespace kuzu::storage;

diff --git a/src/expression_evaluator/literal_evaluator.cpp b/src/expression_evaluator/literal_evaluator.cpp
@@ -1,7 +1,5 @@
 #include "expression_evaluator/literal_evaluator.h"
 
-#include "common/vector/value_vector_utils.h"
-
 using namespace kuzu::common;
 using namespace kuzu::storage;
 

diff --git a/src/function/vector_cast_operations.cpp b/src/function/vector_cast_operations.cpp
@@ -1,6 +1,5 @@
 #include "function/cast/vector_cast_operations.h"
 
-#include "common/vector/value_vector_utils.h"
 #include "function/cast/cast_operations.h"
 
 using namespace kuzu::common;

diff --git a/src/function/vector_list_operation.cpp b/src/function/vector_list_operation.cpp
@@ -1,6 +1,5 @@
 #include "binder/expression_binder.h"
 #include "common/types/ku_list.h"
-#include "common/vector/value_vector_utils.h"
 #include "function/list/operations/list_any_value_operation.h"
 #include "function/list/operations/list_append_operation.h"
 #include "function/list/operations/list_concat_operation.h"

diff --git a/src/include/common/vector/value_vector.h b/src/include/common/vector/value_vector.h
@@ -56,6 +56,8 @@ class ValueVector {
     void setValue(uint32_t pos, T val);
     // This function assumes rowData is non-NULL.
     void copyFromRowData(uint32_t pos, const uint8_t* rowData);
+    void copyToRowData(
+        uint32_t pos, uint8_t* rowData, InMemOverflowBuffer* rowOverflowBuffer) const;
     void copyFromVectorData(
         uint8_t* dstData, const ValueVector* srcVector, const uint8_t* srcVectorData);
 
@@ -128,6 +130,8 @@ class ListVector {
     }
 
     static void copyFromRowData(ValueVector* vector, uint32_t pos, const uint8_t* rowData);
+    static void copyToRowData(const ValueVector* vector, uint32_t pos, uint8_t* rowData,
+        InMemOverflowBuffer* rowOverflowBuffer);
     static void copyFromVectorData(ValueVector* dstVector, uint8_t* dstData,
         const ValueVector* srcVector, const uint8_t* srcData);
 };
@@ -160,6 +164,8 @@ class StructVector {
     }
 
     static void copyFromRowData(ValueVector* vector, uint32_t pos, const uint8_t* rowData);
+    static void copyToRowData(const ValueVector* vector, uint32_t pos, uint8_t* rowData,
+        InMemOverflowBuffer* rowOverflowBuffer);
     static void copyFromVectorData(ValueVector* dstVector, const uint8_t* dstData,
         const ValueVector* srcVector, const uint8_t* srcData);
 };

diff --git a/src/include/common/vector/value_vector_utils.h b/src/include/common/vector/value_vector_utils.h
@@ -1,15 +0,0 @@
-#pragma once
-
-#include "common/vector/value_vector.h"
-
-namespace kuzu {
-namespace common {
-
-class ValueVectorUtils {
-public:
-    static void copyNonNullDataWithSameTypeOutFromPos(const ValueVector& srcVector, uint64_t pos,
-        uint8_t* dstData, InMemOverflowBuffer& dstOverflowBuffer);
-};
-
-} // namespace common
-} // namespace kuzu

diff --git a/src/include/function/aggregate/collect.h b/src/include/function/aggregate/collect.h
@@ -1,6 +1,5 @@
 #pragma once
 
-#include "common/vector/value_vector_utils.h"
 #include "processor/result/factorized_table.h"
 
 namespace kuzu {
@@ -74,8 +73,7 @@ struct CollectFunction {
         for (auto i = 0u; i < multiplicity; ++i) {
             auto tuple = state->factorizedTable->appendEmptyTuple();
             state->isNull = false;
-            common::ValueVectorUtils::copyNonNullDataWithSameTypeOutFromPos(
-                *input, pos, tuple, *state->factorizedTable->getInMemOverflowBuffer());
+            input->copyToRowData(pos, tuple, state->factorizedTable->getInMemOverflowBuffer());
         }
     }
 

diff --git a/src/include/function/list/operations/list_any_value_operation.h b/src/include/function/list/operations/list_any_value_operation.h
@@ -1,6 +1,6 @@
 #pragma once
 
-#include "common/vector/value_vector_utils.h"
+#include "common/vector/value_vector.h"
 
 namespace kuzu {
 namespace function {

diff --git a/src/include/function/list/operations/list_distinct_operation.h b/src/include/function/list/operations/list_distinct_operation.h
@@ -2,7 +2,7 @@
 
 #include <set>
 
-#include "common/vector/value_vector_utils.h"
+#include "common/vector/value_vector.h"
 
 namespace kuzu {
 namespace function {

diff --git a/src/include/function/list/operations/list_extract_operation.h b/src/include/function/list/operations/list_extract_operation.h
@@ -4,7 +4,7 @@
 #include <cstring>
 
 #include "common/types/ku_string.h"
-#include "common/vector/value_vector_utils.h"
+#include "common/vector/value_vector.h"
 #include "function/string/operations/array_extract_operation.h"
 
 namespace kuzu {

diff --git a/src/include/function/list/operations/list_unique_operation.h b/src/include/function/list/operations/list_unique_operation.h
@@ -2,7 +2,7 @@
 
 #include <set>
 
-#include "common/vector/value_vector_utils.h"
+#include "common/vector/value_vector.h"
 
 namespace kuzu {
 namespace function {

diff --git a/src/include/function/map/operations/base_map_extract_operation.h b/src/include/function/map/operations/base_map_extract_operation.h
@@ -1,7 +1,6 @@
 #pragma once
 
 #include "common/vector/value_vector.h"
-#include "common/vector/value_vector_utils.h"
 
 namespace kuzu {
 namespace function {

diff --git a/src/include/function/map/operations/map_creation_operation.h b/src/include/function/map/operations/map_creation_operation.h
@@ -1,7 +1,6 @@
 #pragma once
 
 #include "common/vector/value_vector.h"
-#include "common/vector/value_vector_utils.h"
 
 namespace kuzu {
 namespace function {

diff --git a/src/include/function/map/operations/map_extract_operation.h b/src/include/function/map/operations/map_extract_operation.h
@@ -1,7 +1,6 @@
 #pragma once
 
 #include "common/vector/value_vector.h"
-#include "common/vector/value_vector_utils.h"
 
 namespace kuzu {
 namespace function {

diff --git a/src/include/function/map/operations/map_values_operation.h b/src/include/function/map/operations/map_values_operation.h
@@ -1,7 +1,6 @@
 #pragma once
 
 #include "common/vector/value_vector.h"
-#include "common/vector/value_vector_utils.h"
 
 namespace kuzu {
 namespace function {

diff --git a/src/include/function/map/vector_map_operations.h b/src/include/function/map/vector_map_operations.h
@@ -1,6 +1,6 @@
 #pragma once
 
-#include "common/vector/value_vector_utils.h"
+#include "common/vector/value_vector.h"
 #include "function/vector_operations.h"
 
 namespace kuzu {

diff --git a/src/include/function/string/operations/regexp_extract_all_operation.h b/src/include/function/string/operations/regexp_extract_all_operation.h
@@ -1,7 +1,7 @@
 #pragma once
 
 #include "base_regexp_operation.h"
-#include "common/vector/value_vector_utils.h"
+#include "common/vector/value_vector.h"
 #include "re2.h"
 
 namespace kuzu {

diff --git a/src/include/function/struct/vector_struct_operations.h b/src/include/function/struct/vector_struct_operations.h
@@ -1,6 +1,6 @@
 #pragma once
 
-#include "common/vector/value_vector_utils.h"
+#include "common/vector/value_vector.h"
 #include "function/vector_operations.h"
 
 namespace kuzu {

diff --git a/src/include/function/union/vector_union_operations.h b/src/include/function/union/vector_union_operations.h
@@ -1,6 +1,5 @@
 #pragma once
 
-#include "common/vector/value_vector_utils.h"
 #include "function/vector_operations.h"
 
 namespace kuzu {

diff --git a/src/include/processor/operator/unwind.h b/src/include/processor/operator/unwind.h
@@ -1,7 +1,6 @@
 #pragma once
 
 #include "binder/expression/expression.h"
-#include "common/vector/value_vector_utils.h"
 #include "expression_evaluator/base_evaluator.h"
 #include "processor/operator/physical_operator.h"
 #include "processor/result/result_set.h"

diff --git a/src/processor/mapper/map_expressions_scan.cpp b/src/processor/mapper/map_expressions_scan.cpp
@@ -1,4 +1,3 @@
-#include "common/vector/value_vector_utils.h"
 #include "planner/logical_plan/logical_operator/logical_expressions_scan.h"
 #include "processor/mapper/plan_mapper.h"
 #include "processor/operator/table_scan/factorized_table_scan.h"