Merge pull request #2627 from kuzudb/table-schema-refactor

Table schema refactor

src/binder/CMakeLists.txt

@@ -1,6 +1,5 @@
 add_subdirectory(bind)
 add_subdirectory(bind_expression)
-add_subdirectory(ddl)
 add_subdirectory(expression)
 add_subdirectory(query)
 add_subdirectory(rewriter)

src/binder/bind/bind_copy.cpp

@@ -216,11 +216,11 @@ static void bindExpectedColumns(TableSchema* tableSchema,
     } else {
         // No column specified. Fall back to schema columns.
         for (auto& property : tableSchema->properties) {
-            if (skipPropertyInFile(*property)) {
+            if (skipPropertyInFile(property)) {
                 continue;
             }
-            columnNames.push_back(property->getName());
-            columnTypes.push_back(property->getDataType()->copy());
+            columnNames.push_back(property.getName());
+            columnTypes.push_back(property.getDataType()->copy());
         }
     }
 }

src/binder/bind/bind_ddl.cpp

@@ -7,7 +7,6 @@
 #include "catalog/rel_table_schema.h"
 #include "common/exception/binder.h"
 #include "common/string_format.h"
-#include "common/string_utils.h"
 #include "main/client_context.h"
 #include "parser/ddl/alter.h"
 #include "parser/ddl/create_table.h"
@@ -20,52 +19,61 @@ using namespace kuzu::catalog;
 namespace kuzu {
 namespace binder {
 
-std::vector<Property> Binder::bindProperties(
-    const std::vector<std::pair<std::string, std::string>>& propertyNameDataTypes) {
-    std::vector<Property> boundPropertyNameDataTypes;
-    std::unordered_set<std::string> boundPropertyNames;
-    boundPropertyNames.reserve(propertyNameDataTypes.size());
-    boundPropertyNameDataTypes.reserve(propertyNameDataTypes.size());
-    for (auto& propertyNameDataType : propertyNameDataTypes) {
-        if (boundPropertyNames.contains(propertyNameDataType.first)) {
+static void validateUniquePropertyName(const std::vector<PropertyInfo>& infos) {
+    std::unordered_set<std::string> nameSet;
+    for (auto& info : infos) {
+        if (nameSet.contains(info.name)) {
             throw BinderException(
-                stringFormat("Duplicated column name: {}, column name must be unique.",
-                    propertyNameDataType.first));
-        } else if (TableSchema::isReservedPropertyName(propertyNameDataType.first)) {
+                stringFormat("Duplicated column name: {}, column name must be unique.", info.name));
+        }
+        nameSet.insert(info.name);
+    }
+}
+
+static void validateReservedPropertyName(const std::vector<PropertyInfo>& infos) {
+    for (auto& info : infos) {
+        if (TableSchema::isReservedPropertyName(info.name)) {
             throw BinderException(
-                stringFormat("PropertyName: {} is an internal reserved propertyName.",
-                    propertyNameDataType.first));
+                stringFormat("PropertyName: {} is an internal reserved propertyName.", info.name));
         }
-        boundPropertyNameDataTypes.emplace_back(
-            propertyNameDataType.first, bindDataType(propertyNameDataType.second));
-        boundPropertyNames.emplace(propertyNameDataType.first);
     }
-    return boundPropertyNameDataTypes;
 }
 
-static uint32_t bindPrimaryKey(const std::string& pkColName,
-    std::vector<std::pair<std::string, std::string>> propertyNameDataTypes) {
+std::vector<PropertyInfo> Binder::bindPropertyInfo(
+    const std::vector<std::pair<std::string, std::string>>& propertyNameDataTypes) {
+    std::vector<PropertyInfo> propertyInfos;
+    propertyInfos.reserve(propertyNameDataTypes.size());
+    for (auto& propertyNameDataType : propertyNameDataTypes) {
+        propertyInfos.emplace_back(
+            propertyNameDataType.first, *bindDataType(propertyNameDataType.second));
+    }
+    validateUniquePropertyName(propertyInfos);
+    validateReservedPropertyName(propertyInfos);
+    return propertyInfos;
+}
+
+static uint32_t bindPrimaryKey(
+    const std::string& pkColName, const std::vector<PropertyInfo>& infos) {
     uint32_t primaryKeyIdx = UINT32_MAX;
-    for (auto i = 0u; i < propertyNameDataTypes.size(); i++) {
-        if (propertyNameDataTypes[i].first == pkColName) {
+    for (auto i = 0u; i < infos.size(); i++) {
+        if (infos[i].name == pkColName) {
             primaryKeyIdx = i;
         }
     }
     if (primaryKeyIdx == UINT32_MAX) {
         throw BinderException(
             "Primary key " + pkColName + " does not match any of the predefined node properties.");
     }
-    auto primaryKey = propertyNameDataTypes[primaryKeyIdx];
-    StringUtils::toUpper(primaryKey.second);
+    auto pkType = infos[primaryKeyIdx].type;
     // We only support INT64, STRING and SERIAL column as the primary key.
-    switch (LogicalTypeUtils::dataTypeFromString(primaryKey.second).getLogicalTypeID()) {
+    switch (pkType.getLogicalTypeID()) {
     case LogicalTypeID::INT64:
     case LogicalTypeID::STRING:
     case LogicalTypeID::SERIAL:
         break;
     default:
         throw BinderException(
-            "Invalid primary key type: " + primaryKey.second + ". Expected STRING or INT64.");
+            "Invalid primary key type: " + pkType.toString() + ". Expected STRING or INT64.");
     }
     return primaryKeyIdx;
 }
@@ -91,27 +99,31 @@ BoundCreateTableInfo Binder::bindCreateTableInfo(const parser::CreateTableInfo*
 }
 
 BoundCreateTableInfo Binder::bindCreateNodeTableInfo(const CreateTableInfo* info) {
-    auto boundProperties = bindProperties(info->propertyNameDataTypes);
-    auto extraInfo = (ExtraCreateNodeTableInfo*)info->extraInfo.get();
-    auto primaryKeyIdx = bindPrimaryKey(extraInfo->pKName, info->propertyNameDataTypes);
-    for (auto i = 0u; i < boundProperties.size(); ++i) {
-        if (boundProperties[i].getDataType()->getLogicalTypeID() == LogicalTypeID::SERIAL &&
-            primaryKeyIdx != i) {
+    auto propertyInfos = bindPropertyInfo(info->propertyNameDataTypes);
+    auto extraInfo = ku_dynamic_cast<const ExtraCreateTableInfo*, const ExtraCreateNodeTableInfo*>(
+        info->extraInfo.get());
+    auto primaryKeyIdx = bindPrimaryKey(extraInfo->pKName, propertyInfos);
+    for (auto i = 0u; i < propertyInfos.size(); ++i) {
+        if (propertyInfos[i].type == *LogicalType::SERIAL() && primaryKeyIdx != i) {
             throw BinderException("Serial property in node table must be the primary key.");
         }
     }
     auto boundExtraInfo =
-        std::make_unique<BoundExtraCreateNodeTableInfo>(primaryKeyIdx, std::move(boundProperties));
+        std::make_unique<BoundExtraCreateNodeTableInfo>(primaryKeyIdx, std::move(propertyInfos));
     return BoundCreateTableInfo(TableType::NODE, info->tableName, std::move(boundExtraInfo));
 }
 
 BoundCreateTableInfo Binder::bindCreateRelTableInfo(const CreateTableInfo* info) {
-    auto boundProperties = bindProperties(info->propertyNameDataTypes);
-    for (auto& boundProperty : boundProperties) {
-        if (boundProperty.getDataType()->getLogicalTypeID() == LogicalTypeID::SERIAL ||
-            boundProperty.getDataType()->getLogicalTypeID() == LogicalTypeID::MAP) {
-            throw BinderException(stringFormat("{} property is not supported in rel table.",
-                boundProperty.getDataType()->toString()));
+    std::vector<PropertyInfo> propertyInfos;
+    propertyInfos.emplace_back(InternalKeyword::ID, *LogicalType::INTERNAL_ID());
+    for (auto& propertyInfo : bindPropertyInfo(info->propertyNameDataTypes)) {
+        propertyInfos.push_back(propertyInfo.copy());
+    }
+    for (auto& propertyInfo : propertyInfos) {
+        if (propertyInfo.type == *LogicalType::SERIAL() ||
+            propertyInfo.type.getLogicalTypeID() == LogicalTypeID::MAP) {
+            throw BinderException(stringFormat(
+                "{} property is not supported in rel table.", propertyInfo.type.toString()));
         }
     }
     auto extraInfo = (ExtraCreateRelTableInfo*)info->extraInfo.get();
@@ -122,7 +134,7 @@ BoundCreateTableInfo Binder::bindCreateRelTableInfo(const CreateTableInfo* info)
     auto dstTableID = bindTableID(extraInfo->dstTableName);
     validateTableType(dstTableID, TableType::NODE);
     auto boundExtraInfo = std::make_unique<BoundExtraCreateRelTableInfo>(
-        srcMultiplicity, dstMultiplicity, srcTableID, dstTableID, std::move(boundProperties));
+        srcMultiplicity, dstMultiplicity, srcTableID, dstTableID, std::move(propertyInfos));
     return BoundCreateTableInfo(TableType::REL, info->tableName, std::move(boundExtraInfo));
 }
 

src/binder/bind/bind_graph_pattern.cpp

@@ -130,11 +130,11 @@ static std::vector<std::string> getPropertyNames(const std::vector<TableSchema*>
     std::unordered_set<std::string> propertyNamesSet;
     for (auto& tableSchema : tableSchemas) {
         for (auto& property : tableSchema->properties) {
-            if (propertyNamesSet.contains(property->getName())) {
+            if (propertyNamesSet.contains(property.getName())) {
                 continue;
             }
-            propertyNamesSet.insert(property->getName());
-            result.push_back(property->getName());
+            propertyNamesSet.insert(property.getName());
+            result.push_back(property.getName());
         }
     }
     return result;
@@ -150,7 +150,7 @@ static std::unique_ptr<Expression> createPropertyExpression(const std::string& p
                        nodeTableSchema->getPropertyID(propertyName);
     }
     std::unordered_map<common::table_id_t, common::property_id_t> tableIDToPropertyID;
-    std::vector<LogicalType*> propertyDataTypes;
+    std::vector<const LogicalType*> propertyDataTypes;
     for (auto& tableSchema : tableSchemas) {
         if (!tableSchema->containProperty(propertyName)) {
             continue;

src/binder/bind/bind_updating_clause.cpp

@@ -177,13 +177,13 @@ BoundInsertInfo Binder::bindInsertRelInfo(
 std::vector<expression_pair> Binder::bindSetItems(const PropertyKeyValCollection& collection,
     TableSchema* tableSchema, const std::shared_ptr<Expression>& nodeOrRel) {
     std::vector<expression_pair> setItems;
-    for (auto& property : tableSchema->getProperties()) {
-        if (collection.hasKeyVal(nodeOrRel, property->getName())) { // input specifies rhs.
-            setItems.emplace_back(collection.getKeyVal(nodeOrRel, property->getName()));
+    for (auto& property : tableSchema->getPropertiesRef()) {
+        if (collection.hasKeyVal(nodeOrRel, property.getName())) { // input specifies rhs.
+            setItems.emplace_back(collection.getKeyVal(nodeOrRel, property.getName()));
             continue;
         }
         auto propertyExpression =
-            expressionBinder.bindNodeOrRelPropertyExpression(*nodeOrRel, property->getName());
+            expressionBinder.bindNodeOrRelPropertyExpression(*nodeOrRel, property.getName());
         auto nullExpression = expressionBinder.createNullLiteralExpression();
         nullExpression =
             ExpressionBinder::implicitCastIfNecessary(nullExpression, propertyExpression->dataType);

src/binder/bind/ddl/bind_create_rdf_graph.cpp

@@ -29,38 +29,36 @@ static std::string getRdfLiteralTripleTableName(const std::string& rdfName) {
 
 BoundCreateTableInfo Binder::bindCreateRdfGraphInfo(const CreateTableInfo* info) {
     auto rdfGraphName = info->tableName;
-    auto stringType = LogicalType(LogicalTypeID::STRING);
-    auto serialType = LogicalType(LogicalTypeID::SERIAL);
     // Resource table.
     auto resourceTableName = getRdfResourceTableName(rdfGraphName);
-    std::vector<Property> resourceProperties;
-    resourceProperties.emplace_back(std::string(rdf::IRI), stringType.copy());
+    std::vector<PropertyInfo> resourceProperties;
+    resourceProperties.emplace_back(std::string(rdf::IRI), *LogicalType::STRING());
     auto resourceExtraInfo = std::make_unique<BoundExtraCreateNodeTableInfo>(
         0 /* primaryKeyIdx */, std::move(resourceProperties));
     auto resourceCreateInfo =
         BoundCreateTableInfo(TableType::NODE, resourceTableName, std::move(resourceExtraInfo));
     // Literal table.
     auto literalTableName = getRdfLiteralTableName(rdfGraphName);
-    std::vector<Property> literalProperties;
-    literalProperties.emplace_back(std::string(rdf::ID), serialType.copy());
-    literalProperties.emplace_back(std::string(rdf::VAL), RdfVariantType::getType());
+    std::vector<PropertyInfo> literalProperties;
+    literalProperties.emplace_back(std::string(rdf::ID), *LogicalType::SERIAL());
+    literalProperties.emplace_back(std::string(rdf::VAL), *RdfVariantType::getType());
     auto literalExtraInfo = std::make_unique<BoundExtraCreateNodeTableInfo>(
         0 /* primaryKeyIdx */, std::move(literalProperties));
     auto literalCreateInfo =
         BoundCreateTableInfo(TableType::NODE, literalTableName, std::move(literalExtraInfo));
     // Resource triple table.
     auto resourceTripleTableName = getRdfResourceTripleTableName(rdfGraphName);
-    std::vector<Property> resourceTripleProperties;
-    resourceTripleProperties.emplace_back(std::string(rdf::PID), LogicalType::INTERNAL_ID());
+    std::vector<PropertyInfo> resourceTripleProperties;
+    resourceTripleProperties.emplace_back(std::string(rdf::PID), *LogicalType::INTERNAL_ID());
     auto boundResourceTripleExtraInfo =
         std::make_unique<BoundExtraCreateRelTableInfo>(RelMultiplicity::MANY, RelMultiplicity::MANY,
             INVALID_TABLE_ID, INVALID_TABLE_ID, std::move(resourceTripleProperties));
     auto boundResourceTripleCreateInfo = BoundCreateTableInfo(
         TableType::REL, resourceTripleTableName, std::move(boundResourceTripleExtraInfo));
     // Literal triple table.
     auto literalTripleTableName = getRdfLiteralTripleTableName(rdfGraphName);
-    std::vector<Property> literalTripleProperties;
-    literalTripleProperties.emplace_back(std::string(rdf::PID), LogicalType::INTERNAL_ID());
+    std::vector<PropertyInfo> literalTripleProperties;
+    literalTripleProperties.emplace_back(std::string(rdf::PID), *LogicalType::INTERNAL_ID());
     auto boundLiteralTripleExtraInfo =
         std::make_unique<BoundExtraCreateRelTableInfo>(RelMultiplicity::MANY, RelMultiplicity::MANY,
             INVALID_TABLE_ID, INVALID_TABLE_ID, std::move(literalTripleProperties));

src/catalog/catalog_content.cpp

@@ -2,6 +2,7 @@
 
 #include <fcntl.h>
 
+#include "binder/ddl/bound_create_table_info.h"
 #include "catalog/node_table_schema.h"
 #include "catalog/rdf_graph_schema.h"
 #include "catalog/rel_table_group_schema.h"
@@ -34,57 +35,36 @@ CatalogContent::CatalogContent(const std::string& directory, VirtualFileSystem*
     registerBuiltInFunctions();
 }
 
-static void assignPropertyIDAndTableID(std::vector<Property>& properties, table_id_t tableID) {
-    for (auto i = 0u; i < properties.size(); ++i) {
-        properties[i].setPropertyID(i);
-        properties[i].setTableID(tableID);
-    }
-}
-
-static std::vector<std::unique_ptr<Property>> getPropertiesUniquePtr(
-    const std::vector<Property>& properties) {
-    std::vector<std::unique_ptr<Property>> result;
-    for (auto& property : properties) {
-        result.push_back(std::make_unique<Property>(property.copy()));
-    }
-    return result;
-}
-
 table_id_t CatalogContent::addNodeTableSchema(const BoundCreateTableInfo& info) {
     table_id_t tableID = assignNextTableID();
     auto extraInfo = ku_dynamic_cast<BoundExtraCreateTableInfo*, BoundExtraCreateNodeTableInfo*>(
         info.extraInfo.get());
-    auto properties = Property::copy(extraInfo->properties);
-    assignPropertyIDAndTableID(properties, tableID);
-    auto nodeTableSchema = std::make_unique<NodeTableSchema>(
-        info.tableName, tableID, extraInfo->primaryKeyIdx, getPropertiesUniquePtr(properties));
+    auto nodeTableSchema =
+        std::make_unique<NodeTableSchema>(info.tableName, tableID, extraInfo->primaryKeyIdx);
+    for (auto& propertyInfo : extraInfo->propertyInfos) {
+        nodeTableSchema->addProperty(propertyInfo.name, propertyInfo.type.copy());
+    }
     tableNameToIDMap.emplace(nodeTableSchema->tableName, tableID);
     tableSchemas.emplace(tableID, std::move(nodeTableSchema));
     return tableID;
 }
 
-// TODO(Xiyang): move this to binding stage
-static void addRelInternalIDProperty(std::vector<Property>& properties) {
-    auto relInternalIDProperty = Property(InternalKeyword::ID, LogicalType::INTERNAL_ID());
-    properties.insert(properties.begin(), std::move(relInternalIDProperty));
-}
-
 table_id_t CatalogContent::addRelTableSchema(const BoundCreateTableInfo& info) {
     table_id_t tableID = assignNextTableID();
     auto extraInfo = ku_dynamic_cast<BoundExtraCreateTableInfo*, BoundExtraCreateRelTableInfo*>(
         info.extraInfo.get());
-    auto properties = Property::copy(extraInfo->properties);
-    addRelInternalIDProperty(properties);
-    assignPropertyIDAndTableID(properties, tableID);
     auto srcNodeTableSchema =
         ku_dynamic_cast<TableSchema*, NodeTableSchema*>(getTableSchema(extraInfo->srcTableID));
     auto dstNodeTableSchema =
         ku_dynamic_cast<TableSchema*, NodeTableSchema*>(getTableSchema(extraInfo->dstTableID));
     srcNodeTableSchema->addFwdRelTableID(tableID);
     dstNodeTableSchema->addBwdRelTableID(tableID);
-    auto relTableSchema = std::make_unique<RelTableSchema>(info.tableName, tableID,
-        getPropertiesUniquePtr(properties), extraInfo->srcMultiplicity, extraInfo->dstMultiplicity,
-        extraInfo->srcTableID, extraInfo->dstTableID);
+    auto relTableSchema =
+        std::make_unique<RelTableSchema>(info.tableName, tableID, extraInfo->srcMultiplicity,
+            extraInfo->dstMultiplicity, extraInfo->srcTableID, extraInfo->dstTableID);
+    for (auto& propertyInfo : extraInfo->propertyInfos) {
+        relTableSchema->addProperty(propertyInfo.name, propertyInfo.type.copy());
+    }
     tableNameToIDMap.emplace(relTableSchema->tableName, tableID);
     tableSchemas.emplace(tableID, std::move(relTableSchema));
     return tableID;

src/catalog/node_table_schema.cpp

@@ -9,26 +9,26 @@ namespace kuzu {
 namespace catalog {
 
 NodeTableSchema::NodeTableSchema(const NodeTableSchema& other) : TableSchema{other} {
-    primaryKeyPropertyID = other.primaryKeyPropertyID;
+    primaryKeyPID = other.primaryKeyPID;
     fwdRelTableIDSet = other.fwdRelTableIDSet;
     bwdRelTableIDSet = other.bwdRelTableIDSet;
 }
 
 void NodeTableSchema::serializeInternal(Serializer& serializer) {
-    serializer.serializeValue(primaryKeyPropertyID);
+    serializer.serializeValue(primaryKeyPID);
     serializer.serializeUnorderedSet(fwdRelTableIDSet);
     serializer.serializeUnorderedSet(bwdRelTableIDSet);
 }
 
 std::unique_ptr<NodeTableSchema> NodeTableSchema::deserialize(Deserializer& deserializer) {
-    property_id_t primaryKeyPropertyID;
+    property_id_t primaryKeyPID;
     std::unordered_set<table_id_t> fwdRelTableIDSet;
     std::unordered_set<table_id_t> bwdRelTableIDSet;
-    deserializer.deserializeValue(primaryKeyPropertyID);
+    deserializer.deserializeValue(primaryKeyPID);
     deserializer.deserializeUnorderedSet(fwdRelTableIDSet);
     deserializer.deserializeUnorderedSet(bwdRelTableIDSet);
     auto schema = std::make_unique<NodeTableSchema>();
-    schema->primaryKeyPropertyID = primaryKeyPropertyID;
+    schema->primaryKeyPID = primaryKeyPID;
     schema->fwdRelTableIDSet = std::move(fwdRelTableIDSet);
     schema->bwdRelTableIDSet = std::move(bwdRelTableIDSet);
     return schema;