Refactor node table to take in columnID instead of propertyID

src/catalog/table_schema.cpp

@@ -5,7 +5,6 @@
 #include "catalog/rel_table_group_schema.h"
 #include "catalog/rel_table_schema.h"
 #include "common/constants.h"
-#include "common/exception/catalog.h"
 #include "common/exception/internal.h"
 #include "common/exception/not_implemented.h"
 #include "common/exception/runtime.h"
@@ -50,6 +49,16 @@ property_id_t TableSchema::getPropertyID(const std::string& propertyName) const
         "Table: {} doesn't have a property with propertyName={}.", tableName, propertyName));
 }
 
+// TODO(Guodong): Instead of looping over properties, cache a map between propertyID and columnID.
+column_id_t TableSchema::getColumnID(const property_id_t propertyID) const {
+    for (auto i = 0u; i < properties.size(); i++) {
+        if (properties[i]->getPropertyID() == propertyID) {
+            return i;
+        }
+    }
+    return INVALID_COLUMN_ID;
+}
+
 Property* TableSchema::getProperty(property_id_t propertyID) const {
     for (auto& property : properties) {
         if (property->getPropertyID() == propertyID) {

src/include/catalog/table_schema.h

@@ -66,6 +66,7 @@ class TableSchema {
     std::string getPropertyName(common::property_id_t propertyID) const;
 
     common::property_id_t getPropertyID(const std::string& propertyName) const;
+    common::column_id_t getColumnID(common::property_id_t propertyID) const;
 
     Property* getProperty(common::property_id_t propertyID) const;
 

src/include/processor/operator/mask.h

@@ -23,8 +23,10 @@ struct MaskData {
         std::fill(data, data + size, 0);
     }
 
-    inline void setMask(uint64_t pos, uint8_t maskValue) { data[pos] = maskValue; }
-    inline bool isMasked(uint64_t pos, uint8_t trueMaskVal) { return data[pos] == trueMaskVal; }
+    inline void setMask(uint64_t pos, uint8_t maskValue) const { data[pos] = maskValue; }
+    inline bool isMasked(uint64_t pos, uint8_t trueMaskVal) const {
+        return data[pos] == trueMaskVal;
+    }
 
 private:
     std::unique_ptr<uint8_t[]> dataBuffer;
@@ -63,7 +65,7 @@ class MaskCollection {
 
 class NodeSemiMask {
 public:
-    NodeSemiMask(storage::NodeTable* nodeTable) : nodeTable{nodeTable} {}
+    explicit NodeSemiMask(storage::NodeTable* nodeTable) : nodeTable{nodeTable} {}
 
     virtual void init(transaction::Transaction* trx) = 0;
 
@@ -72,7 +74,7 @@ class NodeSemiMask {
     virtual uint8_t getNumMasks() const = 0;
     virtual void incrementNumMasks() = 0;
 
-    inline bool isEnabled() { return getNumMasks() > 0; }
+    inline bool isEnabled() const { return getNumMasks() > 0; }
     inline storage::NodeTable* getNodeTable() const { return nodeTable; }
 
 protected:
@@ -81,7 +83,7 @@ class NodeSemiMask {
 
 class NodeOffsetSemiMask : public NodeSemiMask {
 public:
-    NodeOffsetSemiMask(storage::NodeTable* nodeTable) : NodeSemiMask{nodeTable} {
+    explicit NodeOffsetSemiMask(storage::NodeTable* nodeTable) : NodeSemiMask{nodeTable} {
         offsetMask = std::make_unique<MaskCollection>();
     }
 
@@ -110,7 +112,7 @@ class NodeOffsetSemiMask : public NodeSemiMask {
 
 class NodeOffsetAndMorselSemiMask : public NodeSemiMask {
 public:
-    NodeOffsetAndMorselSemiMask(storage::NodeTable* nodeTable) : NodeSemiMask{nodeTable} {
+    explicit NodeOffsetAndMorselSemiMask(storage::NodeTable* nodeTable) : NodeSemiMask{nodeTable} {
         offsetMask = std::make_unique<MaskCollection>();
         morselMask = std::make_unique<MaskCollection>();
     }

src/include/processor/operator/persistent/insert_executor.h

@@ -12,12 +12,10 @@ class NodeInsertExecutor {
 public:
     NodeInsertExecutor(storage::NodeTable* table, std::vector<storage::RelTable*> relTablesToInit,
         const DataPos& nodeIDVectorPos, std::vector<DataPos> propertyLhsPositions,
-        std::vector<std::unique_ptr<evaluator::ExpressionEvaluator>> propertyRhsEvaluators,
-        std::unordered_map<common::property_id_t, common::vector_idx_t> propertyIDToVectorIdx)
+        std::vector<std::unique_ptr<evaluator::ExpressionEvaluator>> propertyRhsEvaluators)
         : table{table}, relTablesToInit{std::move(relTablesToInit)},
           nodeIDVectorPos{nodeIDVectorPos}, propertyLhsPositions{std::move(propertyLhsPositions)},
-          propertyRhsEvaluators{std::move(propertyRhsEvaluators)}, propertyIDToVectorIdx{std::move(
-                                                                       propertyIDToVectorIdx)} {}
+          propertyRhsEvaluators{std::move(propertyRhsEvaluators)}, nodeIDVector{nullptr} {}
     NodeInsertExecutor(const NodeInsertExecutor& other);
 
     void init(ResultSet* resultSet, ExecutionContext* context);
@@ -37,8 +35,6 @@ class NodeInsertExecutor {
     DataPos nodeIDVectorPos;
     std::vector<DataPos> propertyLhsPositions;
     std::vector<std::unique_ptr<evaluator::ExpressionEvaluator>> propertyRhsEvaluators;
-    // TODO(Guodong): remove this.
-    std::unordered_map<common::property_id_t, common::vector_idx_t> propertyIDToVectorIdx;
 
     common::ValueVector* nodeIDVector;
     std::vector<common::ValueVector*> propertyLhsVectors;
@@ -53,7 +49,8 @@ class RelInsertExecutor {
         std::vector<std::unique_ptr<evaluator::ExpressionEvaluator>> propertyRhsEvaluators)
         : relsStatistics{relsStatistics}, table{table}, srcNodePos{srcNodePos},
           dstNodePos{dstNodePos}, propertyLhsPositions{std::move(propertyLhsPositions)},
-          propertyRhsEvaluators{std::move(propertyRhsEvaluators)} {}
+          propertyRhsEvaluators{std::move(propertyRhsEvaluators)}, srcNodeIDVector{nullptr},
+          dstNodeIDVector{nullptr} {}
     RelInsertExecutor(const RelInsertExecutor& other);
 
     void init(ResultSet* resultSet, ExecutionContext* context);
@@ -75,8 +72,8 @@ class RelInsertExecutor {
     std::vector<DataPos> propertyLhsPositions;
     std::vector<std::unique_ptr<evaluator::ExpressionEvaluator>> propertyRhsEvaluators;
 
-    common::ValueVector* srcNodeIDVector = nullptr;
-    common::ValueVector* dstNodeIDVector = nullptr;
+    common::ValueVector* srcNodeIDVector;
+    common::ValueVector* dstNodeIDVector;
     std::vector<common::ValueVector*> propertyLhsVectors;
     std::vector<common::ValueVector*> propertyRhsVectors;
 };

src/include/processor/operator/persistent/set_executor.h

@@ -39,7 +39,7 @@ class NodeSetExecutor {
 
 struct NodeSetInfo {
     storage::NodeTable* table;
-    common::property_id_t propertyID;
+    common::column_id_t columnID;
 };
 
 class SingleLabelNodeSetExecutor : public NodeSetExecutor {

src/include/processor/operator/scan/scan_node_table.h

@@ -9,30 +9,31 @@ namespace processor {
 class ScanSingleNodeTable : public ScanColumns {
 public:
     ScanSingleNodeTable(const DataPos& inVectorPos, std::vector<DataPos> outVectorsPos,
-        storage::NodeTable* table, std::vector<uint32_t> propertyColumnIds,
+        storage::NodeTable* table, std::vector<common::column_id_t> columnIDs,
         std::unique_ptr<PhysicalOperator> prevOperator, uint32_t id,
         const std::string& paramsString)
         : ScanColumns{inVectorPos, std::move(outVectorsPos), std::move(prevOperator), id,
               paramsString},
-          table{table}, propertyColumnIds{std::move(propertyColumnIds)} {}
+          table{table}, columnIDs{std::move(columnIDs)} {}
 
     bool getNextTuplesInternal(ExecutionContext* context) override;
 
     inline std::unique_ptr<PhysicalOperator> clone() override {
         return make_unique<ScanSingleNodeTable>(inputNodeIDVectorPos, outPropertyVectorsPos, table,
-            propertyColumnIds, children[0]->clone(), id, paramsString);
+            columnIDs, children[0]->clone(), id, paramsString);
     }
 
 private:
     storage::NodeTable* table;
-    std::vector<uint32_t> propertyColumnIds;
+    std::vector<common::column_id_t> columnIDs;
 };
 
 class ScanMultiNodeTables : public ScanColumns {
 public:
     ScanMultiNodeTables(const DataPos& inVectorPos, std::vector<DataPos> outVectorsPos,
         std::unordered_map<common::table_id_t, storage::NodeTable*> tables,
-        std::unordered_map<common::table_id_t, std::vector<uint32_t>> tableIDToScanColumnIds,
+        std::unordered_map<common::table_id_t, std::vector<common::column_id_t>>
+            tableIDToScanColumnIds,
         std::unique_ptr<PhysicalOperator> prevOperator, uint32_t id,
         const std::string& paramsString)
         : ScanColumns{inVectorPos, std::move(outVectorsPos), std::move(prevOperator), id,
@@ -48,7 +49,7 @@ class ScanMultiNodeTables : public ScanColumns {
 
 private:
     std::unordered_map<common::table_id_t, storage::NodeTable*> tables;
-    std::unordered_map<common::table_id_t, std::vector<uint32_t>> tableIDToScanColumnIds;
+    std::unordered_map<common::table_id_t, std::vector<common::column_id_t>> tableIDToScanColumnIds;
 };
 
 } // namespace processor

src/include/processor/plan_mapper.h

@@ -130,7 +130,7 @@ class PlanMapper {
 
     std::unique_ptr<NodeInsertExecutor> getNodeInsertExecutor(storage::NodesStore* nodesStore,
         storage::RelsStore* relsStore, planner::LogicalCreateNodeInfo* info,
-        const planner::Schema& inSchema, const planner::Schema& outSchema);
+        const planner::Schema& inSchema, const planner::Schema& outSchema) const;
     std::unique_ptr<RelInsertExecutor> getRelInsertExecutor(storage::RelsStore* relsStore,
         planner::LogicalCreateRelInfo* info, const planner::Schema& inSchema,
         const planner::Schema& outSchema);

src/include/storage/copier/node_group.h

@@ -18,8 +18,9 @@ class NodeGroup {
     inline void setNodeGroupIdx(uint64_t nodeGroupIdx_) { this->nodeGroupIdx = nodeGroupIdx_; }
     inline uint64_t getNodeGroupIdx() const { return nodeGroupIdx; }
     inline common::offset_t getNumNodes() const { return numNodes; }
-    inline ColumnChunk* getColumnChunk(common::property_id_t propertyID) {
-        return chunks.contains(propertyID) ? chunks.at(propertyID).get() : nullptr;
+    inline ColumnChunk* getColumnChunk(common::column_id_t columnID) {
+        assert(columnID < chunks.size());
+        return chunks[columnID].get();
     }
     inline bool isFull() const { return numNodes == common::StorageConstants::NODE_GROUP_SIZE; }
 
@@ -33,7 +34,7 @@ class NodeGroup {
 private:
     uint64_t nodeGroupIdx;
     common::offset_t numNodes;
-    std::unordered_map<common::property_id_t, std::unique_ptr<ColumnChunk>> chunks;
+    std::vector<std::unique_ptr<ColumnChunk>> chunks;
 };
 
 } // namespace storage

src/include/storage/local_storage.h

@@ -20,9 +20,9 @@ class LocalStorage {
     void lookup(common::table_id_t tableID, common::ValueVector* nodeIDVector,
         const std::vector<common::column_id_t>& columnIDs,
         const std::vector<common::ValueVector*>& outputVectors);
-    void update(common::table_id_t tableID, common::property_id_t propertyID,
+    void update(common::table_id_t tableID, common::column_id_t columnID,
         common::ValueVector* nodeIDVector, common::ValueVector* propertyVector);
-    void update(common::table_id_t tableID, common::property_id_t propertyID,
+    void update(common::table_id_t tableID, common::column_id_t columnID,
         common::offset_t nodeOffset, common::ValueVector* propertyVector,
         common::sel_t posInPropertyVector);
 

src/include/storage/local_table.h

@@ -145,15 +145,15 @@ class LocalTable {
     void lookup(common::ValueVector* nodeIDVector,
         const std::vector<common::column_id_t>& columnIDs,
         const std::vector<common::ValueVector*>& outputVectors);
-    void update(common::property_id_t propertyID, common::ValueVector* nodeIDVector,
+    void update(common::column_id_t columnID, common::ValueVector* nodeIDVector,
         common::ValueVector* propertyVector, MemoryManager* mm);
-    void update(common::property_id_t propertyID, common::offset_t nodeOffset,
+    void update(common::column_id_t columnID, common::offset_t nodeOffset,
         common::ValueVector* propertyVector, common::sel_t posInPropertyVector, MemoryManager* mm);
 
     void prepareCommit();
 
 private:
-    std::map<common::property_id_t, std::unique_ptr<LocalColumn>> columns;
+    std::map<common::column_id_t, std::unique_ptr<LocalColumn>> columns;
     NodeTable* table;
 };
 

src/include/storage/store/node_table.h

@@ -39,35 +39,34 @@ class NodeTable {
     inline BMFileHandle* getDataFH() const { return dataFH; }
 
     void read(transaction::Transaction* transaction, common::ValueVector* nodeIDVector,
-        const std::vector<common::column_id_t>& columnIds,
+        const std::vector<common::column_id_t>& columnIDs,
         const std::vector<common::ValueVector*>& outputVectors);
     void insert(transaction::Transaction* transaction, common::ValueVector* nodeIDVector,
-        const std::vector<common::ValueVector*>& propertyVectors,
-        const std::unordered_map<common::property_id_t, common::vector_idx_t>&
-            propertyIDToVectorIdx);
-    void update(transaction::Transaction* transaction, common::property_id_t propertyID,
+        const std::vector<common::ValueVector*>& propertyVectors);
+    void update(transaction::Transaction* transaction, common::column_id_t columnID,
         common::ValueVector* nodeIDVector, common::ValueVector* propertyVector);
-    void update(transaction::Transaction* transaction, common::property_id_t propertyID,
+    void update(transaction::Transaction* transaction, common::column_id_t columnID,
         common::offset_t nodeOffset, common::ValueVector* propertyVector,
-        common::sel_t posInPropertyVector);
+        common::sel_t posInPropertyVector) const;
     void delete_(transaction::Transaction* transaction, common::ValueVector* nodeIDVector,
         DeleteState* deleteState);
     void append(NodeGroup* nodeGroup);
 
-    std::unordered_set<common::property_id_t> getPropertyIDs() const;
-
-    inline NodeColumn* getPropertyColumn(common::property_id_t propertyIdx) {
-        assert(propertyColumns.contains(propertyIdx));
-        return propertyColumns.at(propertyIdx).get();
+    inline common::column_id_t getNumColumns() const { return columns.size(); }
+    inline NodeColumn* getColumn(common::column_id_t columnID) {
+        assert(columnID < columns.size());
+        return columns[columnID].get();
     }
+    inline common::column_id_t getPKColumnID() const { return pkColumnID; }
     inline PrimaryKeyIndex* getPKIndex() const { return pkIndex.get(); }
     inline NodesStatisticsAndDeletedIDs* getNodeStatisticsAndDeletedIDs() const {
         return nodesStatisticsAndDeletedIDs;
     }
-    inline common::property_id_t getPKPropertyID() const { return pkPropertyID; }
     inline common::table_id_t getTableID() const { return tableID; }
 
-    inline void dropColumn(common::property_id_t propertyID) { propertyColumns.erase(propertyID); }
+    inline void dropColumn(common::column_id_t columnID) {
+        columns.erase(columns.begin() + columnID);
+    }
     void addColumn(const catalog::Property& property, common::ValueVector* defaultValueVector,
         transaction::Transaction* transaction);
 
@@ -89,16 +88,16 @@ class NodeTable {
 
     void insertPK(common::ValueVector* nodeIDVector, common::ValueVector* primaryKeyVector);
     inline uint64_t getNumNodeGroups(transaction::Transaction* transaction) const {
-        return propertyColumns.begin()->second->getNumNodeGroups(transaction);
+        assert(!columns.empty());
+        return columns[0]->getNumNodeGroups(transaction);
     }
 
 private:
     NodesStatisticsAndDeletedIDs* nodesStatisticsAndDeletedIDs;
-    // TODO(Guodong): use vector here
-    std::map<common::property_id_t, std::unique_ptr<NodeColumn>> propertyColumns;
+    std::vector<std::unique_ptr<NodeColumn>> columns;
     BMFileHandle* dataFH;
     BMFileHandle* metadataFH;
-    common::property_id_t pkPropertyID;
+    common::column_id_t pkColumnID;
     std::unique_ptr<PrimaryKeyIndex> pkIndex;
     common::table_id_t tableID;
     BufferManager& bufferManager;

src/include/storage/store/nodes_store.h

@@ -16,7 +16,7 @@ class NodesStore {
 
     inline NodeColumn* getNodePropertyColumn(
         common::table_id_t tableID, uint64_t propertyIdx) const {
-        return nodeTables.at(tableID)->getPropertyColumn(propertyIdx);
+        return nodeTables.at(tableID)->getColumn(propertyIdx);
     }
     inline PrimaryKeyIndex* getPKIndex(common::table_id_t tableID) {
         return nodeTables[tableID]->getPKIndex();

src/main/storage_driver.cpp

@@ -20,7 +20,7 @@ void StorageDriver::scan(const std::string& nodeName, const std::string& propert
     auto nodeTableID = catalogContent->getTableID(nodeName);
     auto propertyID = catalogContent->getTableSchema(nodeTableID)->getPropertyID(propertyName);
     auto nodeTable = storageManager->getNodesStore().getNodeTable(nodeTableID);
-    auto column = nodeTable->getPropertyColumn(propertyID);
+    auto column = nodeTable->getColumn(propertyID);
     auto current_buffer = result;
     std::vector<std::thread> threads;
     auto numElementsPerThread = size / numThreads + 1;

src/processor/map/map_create.cpp

@@ -27,7 +27,7 @@ static std::vector<DataPos> populateLhsVectorPositions(
 std::unique_ptr<NodeInsertExecutor> PlanMapper::getNodeInsertExecutor(
     storage::NodesStore* nodesStore, storage::RelsStore* relsStore,
     planner::LogicalCreateNodeInfo* info, const planner::Schema& inSchema,
-    const planner::Schema& outSchema) {
+    const planner::Schema& outSchema) const {
     auto node = info->node;
     auto nodeTableID = node->getSingleTableID();
     auto table = nodesStore->getNodeTable(nodeTableID);
@@ -41,15 +41,11 @@ std::unique_ptr<NodeInsertExecutor> PlanMapper::getNodeInsertExecutor(
     auto nodeIDPos = DataPos(outSchema.getExpressionPos(*node->getInternalIDProperty()));
     std::vector<DataPos> lhsVectorPositions = populateLhsVectorPositions(info->setItems, outSchema);
     std::vector<std::unique_ptr<ExpressionEvaluator>> evaluators;
-    std::unordered_map<common::property_id_t, common::vector_idx_t> propertyIDToVectorIdx;
-    for (auto i = 0u; i < info->setItems.size(); ++i) {
-        auto& [lhs, rhs] = info->setItems[i];
-        auto propertyExpression = (binder::PropertyExpression*)lhs.get();
+    for (auto& [_, rhs] : info->setItems) {
         evaluators.push_back(ExpressionMapper::getEvaluator(rhs, &inSchema));
-        propertyIDToVectorIdx.insert({propertyExpression->getPropertyID(nodeTableID), i});
     }
     return std::make_unique<NodeInsertExecutor>(table, std::move(relTablesToInit), nodeIDPos,
-        std::move(lhsVectorPositions), std::move(evaluators), std::move(propertyIDToVectorIdx));
+        std::move(lhsVectorPositions), std::move(evaluators));
 }
 
 std::unique_ptr<PhysicalOperator> PlanMapper::mapCreateNode(LogicalOperator* logicalOperator) {

src/processor/map/map_scan_node_property.cpp

@@ -23,7 +23,7 @@ std::unique_ptr<PhysicalOperator> PlanMapper::mapScanNodeProperty(
         outVectorsPos.emplace_back(outSchema->getExpressionPos(*expression));
     }
     if (node->isMultiLabeled()) {
-        std::unordered_map<table_id_t, std::vector<uint32_t>> tableIDToColumns;
+        std::unordered_map<table_id_t, std::vector<column_id_t>> tableIDToColumns;
         std::unordered_map<table_id_t, storage::NodeTable*> tables;
         for (auto& tableID : node->getTableIDs()) {
             tables.insert({tableID, nodeStore.getNodeTable(tableID)});
@@ -33,7 +33,9 @@ std::unique_ptr<PhysicalOperator> PlanMapper::mapScanNodeProperty(
                 if (!property->hasPropertyID(tableID)) {
                     columns.push_back(UINT32_MAX);
                 } else {
-                    columns.push_back(property->getPropertyID(tableID));
+                    columns.push_back(
+                        catalog->getReadOnlyVersion()->getTableSchema(tableID)->getColumnID(
+                            property->getPropertyID(tableID)));
                 }
             }
             tableIDToColumns.insert({tableID, std::move(columns)});
@@ -43,13 +45,14 @@ std::unique_ptr<PhysicalOperator> PlanMapper::mapScanNodeProperty(
             getOperatorID(), scanProperty.getExpressionsForPrinting());
     } else {
         auto tableID = node->getSingleTableID();
-        std::vector<uint32_t> columnIds;
+        auto tableSchema = catalog->getReadOnlyVersion()->getTableSchema(tableID);
+        std::vector<column_id_t> columnIDs;
         for (auto& expression : scanProperty.getProperties()) {
             auto property = static_pointer_cast<PropertyExpression>(expression);
-            columnIds.push_back(property->getPropertyID(tableID));
+            columnIDs.push_back(tableSchema->getColumnID(property->getPropertyID(tableID)));
         }
         return std::make_unique<ScanSingleNodeTable>(inputNodeIDVectorPos, std::move(outVectorsPos),
-            nodeStore.getNodeTable(tableID), std::move(columnIds), std::move(prevOperator),
+            nodeStore.getNodeTable(tableID), std::move(columnIDs), std::move(prevOperator),
             getOperatorID(), scanProperty.getExpressionsForPrinting());
     }
 }