support alter table add property

src/CMakeLists.txt

@@ -11,7 +11,7 @@ add_subdirectory(storage)
 add_subdirectory(transaction)
 
 add_library(kuzu STATIC ${ALL_OBJECT_FILES})
-target_link_libraries(kuzu PUBLIC antlr4_cypher antlr4_runtime utf8proc 
+target_link_libraries(kuzu PUBLIC antlr4_cypher antlr4_runtime utf8proc
         parquet_lib arrow_lib arrow_deps Threads::Threads)
 target_include_directories(
         kuzu PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>

src/antlr4/Cypher.g4

@@ -51,10 +51,26 @@ kU_DropTable
 DROP : ( 'D' | 'd' ) ( 'R' | 'r' ) ( 'O' | 'o' ) ( 'P' | 'p' ) ;
 
 kU_AlterTable
-    : ALTER SP TABLE SP oC_SchemaName SP DROP SP oC_PropertyKeyName ;
+    : ALTER SP TABLE SP oC_SchemaName SP kU_AlterOptions ;
 
 ALTER: ( 'A' | 'a' ) ( 'L' | 'l' ) ( 'T' | 't' ) ( 'E' | 'e' ) ( 'R' | 'r' ) ;
 
+kU_AlterOptions
+    : kU_AddProperty
+        | kU_DropProperty;
+
+kU_AddProperty
+    : ADD SP ( COLUMN SP )? oC_PropertyKeyName SP kU_DataType ( SP DEFAULT SP oC_Expression )? ;
+
+DEFAULT : ( 'D' | 'd' ) ( 'E' | 'e' ) ( 'F' | 'f' ) ( 'A' | 'a' ) ( 'U' | 'u' ) ( 'L' | 'l' ) ( 'T' | 't' ) ;
+
+kU_DropProperty
+    : DROP SP ( COLUMN SP )? oC_PropertyKeyName ;
+
+ADD: ( 'A' | 'a' ) ( 'D' | 'd' ) ( 'D' | 'd' ) ;
+
+COLUMN: ( 'C' | 'c' ) ( 'O' | 'o' ) ( 'L' | 'l' ) ( 'U' | 'u'  ) ( 'M' | 'm' ) ( 'N' | 'n' ) ;
+
 kU_RelConnections : kU_RelConnection ( SP? ',' SP? kU_RelConnection )* ;
 
 kU_RelConnection: FROM SP kU_NodeLabels SP TO SP kU_NodeLabels ;

src/binder/bind/bind_copy.cpp

@@ -12,8 +12,7 @@ unique_ptr<BoundStatement> Binder::bindCopy(const Statement& statement) {
     auto tableName = copyCSV.getTableName();
     validateTableExist(catalog, tableName);
     auto isNodeTable = catalogContent->containNodeTable(tableName);
-    auto tableID = isNodeTable ? catalogContent->getNodeTableIDFromName(tableName) :
-                                 catalogContent->getRelTableIDFromName(tableName);
+    auto tableID = catalogContent->getTableID(tableName);
     auto filePath = copyCSV.getCSVFileName();
     auto csvReaderConfig = bindParsingOptions(copyCSV.getParsingOptions());
     return make_unique<BoundCopy>(CopyDescription(filePath, csvReaderConfig), tableID, tableName);

src/binder/bind/bind_ddl.cpp

@@ -1,8 +1,10 @@
 #include "binder/binder.h"
+#include "binder/ddl/bound_add_property.h"
 #include "binder/ddl/bound_create_node_clause.h"
 #include "binder/ddl/bound_create_rel_clause.h"
 #include "binder/ddl/bound_drop_property.h"
 #include "binder/ddl/bound_drop_table.h"
+#include "parser/ddl/add_property.h"
 #include "parser/ddl/create_node_clause.h"
 #include "parser/ddl/create_rel_clause.h"
 #include "parser/ddl/drop_property.h"
@@ -49,8 +51,7 @@ unique_ptr<BoundStatement> Binder::bindDropTable(const Statement& statement) {
     validateTableExist(catalog, tableName);
     auto catalogContent = catalog.getReadOnlyVersion();
     auto isNodeTable = catalogContent->containNodeTable(tableName);
-    auto tableID = isNodeTable ? catalogContent->getNodeTableIDFromName(tableName) :
-                                 catalogContent->getRelTableIDFromName(tableName);
+    auto tableID = catalogContent->getTableID(tableName);
     if (isNodeTable) {
         validateNodeTableHasNoEdge(catalog, tableID);
     }
@@ -63,8 +64,7 @@ unique_ptr<BoundStatement> Binder::bindDropProperty(const Statement& statement)
     validateTableExist(catalog, tableName);
     auto catalogContent = catalog.getReadOnlyVersion();
     auto isNodeTable = catalogContent->containNodeTable(tableName);
-    auto tableID = isNodeTable ? catalogContent->getNodeTableIDFromName(tableName) :
-                                 catalogContent->getRelTableIDFromName(tableName);
+    auto tableID = catalogContent->getTableID(tableName);
     auto propertyID =
         bindPropertyName(catalogContent->getTableSchema(tableID), dropProperty.getPropertyName());
     if (isNodeTable &&
@@ -75,6 +75,22 @@ unique_ptr<BoundStatement> Binder::bindDropProperty(const Statement& statement)
     return make_unique<BoundDropProperty>(tableID, propertyID, tableName);
 }
 
+unique_ptr<BoundStatement> Binder::bindAddProperty(const Statement& statement) {
+    auto& addProperty = (AddProperty&)statement;
+    auto tableName = addProperty.getTableName();
+    validateTableExist(catalog, tableName);
+    auto catalogContent = catalog.getReadOnlyVersion();
+    auto tableID = catalogContent->getTableID(tableName);
+    auto dataType = Types::dataTypeFromString(addProperty.getDataType());
+    if (catalogContent->getTableSchema(tableID)->containProperty(addProperty.getPropertyName())) {
+        throw BinderException("Property: " + addProperty.getPropertyName() + " already exists.");
+    }
+    return make_unique<BoundAddProperty>(tableID, addProperty.getPropertyName(), dataType,
+        ExpressionBinder::implicitCastIfNecessary(
+            expressionBinder.bindExpression(*addProperty.getDefaultValue()), dataType),
+        tableName);
+}
+
 vector<PropertyNameDataType> Binder::bindPropertyNameDataTypes(
     vector<pair<string, string>> propertyNameDataTypes) {
     vector<PropertyNameDataType> boundPropertyNameDataTypes;

src/binder/binder.cpp

@@ -20,6 +20,9 @@ unique_ptr<BoundStatement> Binder::bind(const Statement& statement) {
     case StatementType::DROP_PROPERTY: {
         return bindDropProperty(statement);
     }
+    case StatementType::ADD_PROPERTY: {
+        return bindAddProperty(statement);
+    }
     case StatementType::QUERY: {
         return bindQuery((const RegularQuery&)statement);
     }
@@ -38,14 +41,14 @@ table_id_t Binder::bindRelTableID(const string& tableName) const {
     if (!catalog.getReadOnlyVersion()->containRelTable(tableName)) {
         throw BinderException("Rel table " + tableName + " does not exist.");
     }
-    return catalog.getReadOnlyVersion()->getRelTableIDFromName(tableName);
+    return catalog.getReadOnlyVersion()->getTableID(tableName);
 }
 
 table_id_t Binder::bindNodeTableID(const string& tableName) const {
     if (!catalog.getReadOnlyVersion()->containNodeTable(tableName)) {
         throw BinderException("Node table " + tableName + " does not exist.");
     }
-    return catalog.getReadOnlyVersion()->getNodeTableIDFromName(tableName);
+    return catalog.getReadOnlyVersion()->getTableID(tableName);
 }
 
 shared_ptr<Expression> Binder::createVariable(const string& name, const DataType& dataType) {

src/catalog/catalog.cpp

@@ -380,9 +380,15 @@ void Catalog::dropTableSchema(table_id_t tableID) {
 
 void Catalog::dropProperty(table_id_t tableID, property_id_t propertyID) {
     initCatalogContentForWriteTrxIfNecessary();
-    catalogContentForWriteTrx->dropProperty(tableID, propertyID);
+    catalogContentForWriteTrx->getTableSchema(tableID)->dropProperty(propertyID);
     wal->logDropPropertyRecord(tableID, propertyID);
 }
 
+void Catalog::addProperty(table_id_t tableID, string propertyName, DataType dataType) {
+    initCatalogContentForWriteTrxIfNecessary();
+    catalogContentForWriteTrx->getTableSchema(tableID)->addProperty(
+        std::move(propertyName), std::move(dataType));
+}
+
 } // namespace catalog
 } // namespace kuzu

src/catalog/catalog_structs.cpp

@@ -48,6 +48,27 @@ string TableSchema::getPropertyName(property_id_t propertyID) const {
         "Table: %s doesn't have a property with propertyID=%d.", tableName.c_str(), propertyID));
 }
 
+property_id_t TableSchema::getPropertyID(string propertyName) const {
+    for (auto& property : properties) {
+        if (property.name == propertyName) {
+            return property.propertyID;
+        }
+    }
+    throw RuntimeException(
+        StringUtils::string_format("Table: %s doesn't have a property with propertyName=%s.",
+            tableName.c_str(), propertyName.c_str()));
+}
+
+Property TableSchema::getProperty(property_id_t propertyID) const {
+    for (auto& property : properties) {
+        if (property.propertyID == propertyID) {
+            return property;
+        }
+    }
+    throw RuntimeException(StringUtils::string_format(
+        "Table: %s doesn't have a property with propertyID=%d.", tableName.c_str(), propertyID));
+}
+
 unordered_set<table_id_t> RelTableSchema::getAllNodeTableIDs() const {
     unordered_set<table_id_t> allNodeTableIDs;
     for (auto& [srcTableID, dstTableID] : srcDstTableIDs) {

src/include/binder/binder.h

@@ -45,6 +45,7 @@ class Binder {
     unique_ptr<BoundStatement> bindCreateRelClause(const Statement& statement);
     unique_ptr<BoundStatement> bindDropTable(const Statement& statement);
     unique_ptr<BoundStatement> bindDropProperty(const Statement& statement);
+    unique_ptr<BoundStatement> bindAddProperty(const Statement& statement);
 
     vector<PropertyNameDataType> bindPropertyNameDataTypes(
         vector<pair<string, string>> propertyNameDataTypes);

src/include/binder/ddl/bound_add_property.h

@@ -0,0 +1,32 @@
+#pragma once
+
+#include "bound_ddl.h"
+
+namespace kuzu {
+namespace binder {
+
+class BoundAddProperty : public BoundDDL {
+public:
+    explicit BoundAddProperty(table_id_t tableID, string propertyName, DataType dataType,
+        shared_ptr<Expression> defaultValue, string tableName)
+        : BoundDDL{StatementType::ADD_PROPERTY, std::move(tableName)}, tableID{tableID},
+          propertyName{std::move(propertyName)}, dataType{std::move(dataType)}, defaultValue{
+                                                                                    defaultValue} {}
+
+    inline table_id_t getTableID() const { return tableID; }
+
+    inline string getPropertyName() const { return propertyName; }
+
+    inline DataType getDataType() const { return dataType; }
+
+    inline shared_ptr<Expression> getDefaultValue() const { return defaultValue; }
+
+private:
+    table_id_t tableID;
+    string propertyName;
+    DataType dataType;
+    shared_ptr<Expression> defaultValue;
+};
+
+} // namespace binder
+} // namespace kuzu

src/include/catalog/catalog.h

@@ -77,13 +77,10 @@ class CatalogContent {
         return relTableNameToIDMap.contains(tableName);
     }
 
-    inline table_id_t getNodeTableIDFromName(const string& tableName) const {
-        return nodeTableNameToIDMap.at(tableName);
+    inline table_id_t getTableID(const string& tableName) const {
+        return nodeTableNameToIDMap.contains(tableName) ? nodeTableNameToIDMap.at(tableName) :
+                                                          relTableNameToIDMap.at(tableName);
     }
-    inline table_id_t getRelTableIDFromName(const string& tableName) const {
-        return relTableNameToIDMap.at(tableName);
-    }
-
     inline bool isSingleMultiplicityInDirection(table_id_t tableID, RelDirection direction) const {
         return relTableSchemas.at(tableID)->isSingleMultiplicityInDirection(direction);
     }
@@ -126,9 +123,6 @@ class CatalogContent {
         return direction == FWD ? relTableSchema->getUniqueSrcTableIDs() :
                                   relTableSchema->getUniqueDstTableIDs();
     }
-    inline void dropProperty(table_id_t tableID, property_id_t propertyID) {
-        getTableSchema(tableID)->dropProperty(propertyID);
-    }
 
     void dropTableSchema(table_id_t tableID);
 
@@ -199,6 +193,8 @@ class Catalog {
 
     void dropProperty(table_id_t tableID, property_id_t propertyID);
 
+    void addProperty(table_id_t tableID, string propertyName, DataType dataType);
+
 protected:
     unique_ptr<BuiltInVectorOperations> builtInVectorOperations;
     unique_ptr<BuiltInAggregateFunctions> builtInAggregateFunctions;

src/include/catalog/catalog_structs.h

@@ -33,23 +33,22 @@ struct PropertyNameDataType {
 };
 
 struct Property : PropertyNameDataType {
-private:
-    Property(string name, DataType dataType, uint32_t propertyID, table_id_t tableID)
-        : PropertyNameDataType{std::move(name), std::move(dataType)},
-          propertyID{propertyID}, tableID{tableID} {}
-
 public:
     // This constructor is needed for ser/deser functions
-    Property() {}
+    Property() = default;
+
+    Property(string name, DataType dataType, property_id_t propertyID, table_id_t tableID)
+        : PropertyNameDataType{std::move(name), std::move(dataType)},
+          propertyID{propertyID}, tableID{tableID} {}
 
     static Property constructNodeProperty(
-        const PropertyNameDataType& nameDataType, uint32_t propertyID, table_id_t tableID) {
-        return Property(nameDataType.name, nameDataType.dataType, propertyID, tableID);
+        const PropertyNameDataType& nameDataType, property_id_t propertyID, table_id_t tableID) {
+        return {nameDataType.name, nameDataType.dataType, propertyID, tableID};
     }
 
     static inline Property constructRelProperty(
-        const PropertyNameDataType& nameDataType, uint32_t propertyID, table_id_t tableID) {
-        return Property(nameDataType.name, nameDataType.dataType, propertyID, tableID);
+        const PropertyNameDataType& nameDataType, property_id_t propertyID, table_id_t tableID) {
+        return {nameDataType.name, nameDataType.dataType, propertyID, tableID};
     }
 
     property_id_t propertyID;
@@ -60,7 +59,8 @@ struct TableSchema {
 public:
     TableSchema(string tableName, table_id_t tableID, bool isNodeTable, vector<Property> properties)
         : tableName{std::move(tableName)}, tableID{tableID}, isNodeTable{isNodeTable},
-          properties{std::move(properties)} {}
+          properties{std::move(properties)}, nextPropertyID{
+                                                 (property_id_t)this->properties.size()} {}
 
     virtual ~TableSchema() = default;
 
@@ -83,13 +83,28 @@ struct TableSchema {
             [&propertyName](const Property& property) { return property.name == propertyName; });
     }
 
+    inline void addProperty(string propertyName, DataType dataType) {
+        properties.emplace_back(
+            std::move(propertyName), std::move(dataType), increaseNextPropertyID(), tableID);
+    }
+
     string getPropertyName(property_id_t propertyID) const;
 
+    property_id_t getPropertyID(string propertyName) const;
+
+    Property getProperty(property_id_t propertyID) const;
+
+private:
+    inline property_id_t increaseNextPropertyID() { return nextPropertyID++; }
+
 public:
     string tableName;
     table_id_t tableID;
     bool isNodeTable;
     vector<Property> properties;
+
+private:
+    property_id_t nextPropertyID;
 };
 
 struct NodeTableSchema : TableSchema {

src/include/common/statement_type.h

@@ -12,6 +12,7 @@ enum class StatementType : uint8_t {
     COPY_CSV = 3,
     DROP_TABLE = 4,
     DROP_PROPERTY = 5,
+    ADD_PROPERTY = 6,
 };
 
 class StatementTypeUtils {

src/include/parser/ddl/add_property.h

@@ -0,0 +1,33 @@
+#pragma once
+
+#include "common/types/types.h"
+#include "parser/ddl/ddl.h"
+#include "parser/expression/parsed_expression.h"
+
+namespace kuzu {
+namespace parser {
+
+using namespace std;
+
+class AddProperty : public DDL {
+public:
+    explicit AddProperty(string tableName, string propertyName, string dataType,
+        unique_ptr<ParsedExpression> defaultValue)
+        : DDL{StatementType::ADD_PROPERTY, std::move(tableName)}, propertyName{std::move(
+                                                                      propertyName)},
+          dataType{std::move(dataType)}, defaultValue{std::move(defaultValue)} {}
+
+    inline string getPropertyName() const { return propertyName; }
+
+    inline string getDataType() const { return dataType; }
+
+    inline ParsedExpression* getDefaultValue() const { return defaultValue.get(); }
+
+private:
+    string propertyName;
+    string dataType;
+    unique_ptr<ParsedExpression> defaultValue;
+};
+
+} // namespace parser
+} // namespace kuzu

src/include/parser/query/updating_clause/create_clause.h

@@ -10,7 +10,6 @@ class CreateClause : public UpdatingClause {
 public:
     CreateClause(vector<unique_ptr<PatternElement>> patternElements)
         : UpdatingClause{ClauseType::CREATE}, patternElements{std::move(patternElements)} {};
-    ~CreateClause() override = default;
 
     inline const vector<unique_ptr<PatternElement>>& getPatternElements() const {
         return patternElements;

src/include/parser/transformer.h

@@ -208,6 +208,10 @@ class Transformer {
 
     unique_ptr<Statement> transformAlterTable(CypherParser::KU_AlterTableContext& ctx);
 
+    unique_ptr<Statement> transformAddProperty(CypherParser::KU_AlterTableContext& ctx);
+
+    unique_ptr<Statement> transformDropProperty(CypherParser::KU_AlterTableContext& ctx);
+
     string transformDataType(CypherParser::KU_DataTypeContext& ctx);
 
     string transformListIdentifiers(CypherParser::KU_ListIdentifiersContext& ctx);

src/include/planner/logical_plan/logical_operator/base_logical_operator.h

@@ -9,6 +9,7 @@ namespace planner {
 
 enum class LogicalOperatorType : uint8_t {
     ACCUMULATE,
+    ADD_PROPERTY,
     AGGREGATE,
     COPY_CSV,
     CREATE_NODE,