add rdf ddl.

Co-authored-by: Gaurav Sehgal <gaurav.sehgal8297@gmail.com>
kuzudb · Sep 1, 2023 · 50bd660 · 50bd660
1 parent 9aaff49
commit 50bd660
Show file tree

Hide file tree

Showing 63 changed files with 4,243 additions and 3,862 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.11)
 
-project(Kuzu VERSION 0.0.8.1 LANGUAGES CXX)
+project(Kuzu VERSION 0.0.8.2 LANGUAGES CXX)
 
 find_package(Threads REQUIRED)
 

diff --git a/src/antlr4/Cypher.g4 b/src/antlr4/Cypher.g4
@@ -65,6 +65,7 @@ COLUMN : ( 'C' | 'c' ) ( 'O' | 'o' ) ( 'L' | 'l' ) ( 'U' | 'u' ) ( 'M' | 'm' ) (
 kU_DDL
     : kU_CreateNode
         | kU_CreateRel
+        | kU_CreateRdfGraph
         | kU_DropTable
         | kU_AlterTable;
 
@@ -78,6 +79,13 @@ TABLE: ( 'T' | 't' ) ( 'A' | 'a' ) ( 'B' | 'b' ) ( 'L' | 'l' ) ( 'E' | 'e' ) ;
 kU_CreateRel
     : CREATE SP REL SP TABLE SP oC_SchemaName SP? '(' SP? FROM SP oC_SchemaName SP TO SP oC_SchemaName SP? ( ',' SP? kU_PropertyDefinitions SP? )? ( ',' SP? oC_SymbolicName SP? )?  ')' ;
 
+kU_CreateRdfGraph
+    : CREATE SP RDF SP GRAPH SP oC_SchemaName ;
+
+RDF : ('R' | 'r') ('D' | 'd') ('F' | 'f') ;
+
+GRAPH : ('G' | 'g') ('R' | 'r') ('A' | 'a') ('P' | 'p') ('H' | 'h') ;
+
 kU_DropTable
     : DROP SP TABLE SP oC_SchemaName ;
 

diff --git a/src/binder/bind/bind_ddl.cpp b/src/binder/bind/bind_ddl.cpp
@@ -1,15 +1,13 @@
 #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_create_table.h"
 #include "binder/ddl/bound_drop_property.h"
 #include "binder/ddl/bound_drop_table.h"
 #include "binder/ddl/bound_rename_property.h"
 #include "binder/ddl/bound_rename_table.h"
 #include "common/string_utils.h"
 #include "parser/ddl/add_property.h"
-#include "parser/ddl/create_node_clause.h"
-#include "parser/ddl/create_rel_clause.h"
+#include "parser/ddl/create_table_clause.h"
 #include "parser/ddl/drop_property.h"
 #include "parser/ddl/drop_table.h"
 #include "parser/ddl/rename_property.h"
@@ -24,40 +22,102 @@ namespace binder {
 
 std::unique_ptr<BoundStatement> Binder::bindCreateNodeTableClause(
     const parser::Statement& statement) {
-    auto& createNodeTableClause = (parser::CreateNodeTableClause&)statement;
-    auto tableName = createNodeTableClause.getTableName();
+    auto& createTableClause = reinterpret_cast<const CreateTableClause&>(statement);
+    auto tableName = createTableClause.getTableName();
     if (catalog.getReadOnlyVersion()->containTable(tableName)) {
         throw BinderException("Node " + tableName + " already exists.");
     }
-    auto boundProperties = bindProperties(createNodeTableClause.getPropertyNameDataTypes());
-    auto primaryKeyIdx = bindPrimaryKey(
-        createNodeTableClause.getPKColName(), createNodeTableClause.getPropertyNameDataTypes());
+    auto createInfo = createTableClause.getInfo();
+    auto boundProperties = bindProperties(createInfo->propertyNameDataTypes);
+    auto extraInfo = (NodeExtraCreateTableInfo*)createInfo->extraInfo.get();
+    auto primaryKeyIdx = bindPrimaryKey(extraInfo->pKName, createInfo->propertyNameDataTypes);
     for (auto i = 0u; i < boundProperties.size(); ++i) {
         if (boundProperties[i]->getDataType()->getLogicalTypeID() == LogicalTypeID::SERIAL &&
             primaryKeyIdx != i) {
             throw BinderException("Serial property in node table must be the primary key.");
         }
     }
-    return make_unique<BoundCreateNodeClause>(tableName, std::move(boundProperties), primaryKeyIdx);
+    auto boundExtraInfo = std::make_unique<BoundNodeExtraCreateTableInfo>(primaryKeyIdx);
+    auto boundCreateInfo = std::make_unique<BoundCreateTableInfo>(
+        tableName, std::move(boundProperties), std::move(boundExtraInfo));
+    return make_unique<BoundCreateTable>(
+        StatementType::CREATE_NODE_TABLE, tableName, std::move(boundCreateInfo));
 }
 
 std::unique_ptr<BoundStatement> Binder::bindCreateRelTableClause(
     const parser::Statement& statement) {
-    auto& createRelClause = (CreateRelClause&)statement;
-    auto tableName = createRelClause.getTableName();
+    auto& createTableClause = reinterpret_cast<const CreateTableClause&>(statement);
+    auto tableName = createTableClause.getTableName();
     if (catalog.getReadOnlyVersion()->containTable(tableName)) {
         throw BinderException("Rel " + tableName + " already exists.");
     }
-    auto boundProperties = bindProperties(createRelClause.getPropertyNameDataTypes());
+    auto createInfo = createTableClause.getInfo();
+    auto boundProperties = bindProperties(createInfo->propertyNameDataTypes);
     for (auto& boundProperty : boundProperties) {
         if (boundProperty->getDataType()->getLogicalTypeID() == LogicalTypeID::SERIAL) {
             throw BinderException("Serial property is not supported in rel table.");
         }
     }
-    auto relMultiplicity = getRelMultiplicityFromString(createRelClause.getRelMultiplicity());
-    return make_unique<BoundCreateRelClause>(tableName, std::move(boundProperties), relMultiplicity,
-        bindNodeTableID(createRelClause.getSrcTableName()),
-        bindNodeTableID(createRelClause.getDstTableName()));
+    auto extraInfo = (RelExtraCreateTableInfo*)createInfo->extraInfo.get();
+    auto srcTableID = bindNodeTableID(extraInfo->srcTableName);
+    auto srcTableSchema =
+        (NodeTableSchema*)catalog.getReadOnlyVersion()->getTableSchema(srcTableID);
+    auto srcPkDataType = srcTableSchema->getPrimaryKey()->getDataType();
+    auto dstTableID = bindNodeTableID(extraInfo->dstTableName);
+    auto dstTableSchema =
+        (NodeTableSchema*)catalog.getReadOnlyVersion()->getTableSchema(dstTableID);
+    auto dstPkDataType = dstTableSchema->getPrimaryKey()->getDataType();
+    auto relMultiplicity = getRelMultiplicityFromString(extraInfo->relMultiplicity);
+    auto boundExtraInfo = std::make_unique<BoundRelExtraCreateTableInfo>(
+        relMultiplicity, srcTableID, dstTableID, srcPkDataType->copy(), dstPkDataType->copy());
+    auto boundCreateInfo = std::make_unique<BoundCreateTableInfo>(
+        tableName, std::move(boundProperties), std::move(boundExtraInfo));
+    return make_unique<BoundCreateTable>(
+        StatementType::CREATE_REL_TABLE, tableName, std::move(boundCreateInfo));
+}
+
+static constexpr char RDF_IRI[] = "_IRI";
+static constexpr char RDF_PREDICT_ID[] = "_PREDICT_ID";
+static constexpr char RDF_NODE_TABLE_SUFFIX[] = "_RESOURCE";
+static constexpr char RDF_REL_TABLE_SUFFIX[] = "_TRIPLES";
+
+static inline std::string getRdfNodeTableName(const std::string& rdfName) {
+    return rdfName + RDF_NODE_TABLE_SUFFIX;
+}
+
+static inline std::string getRdfRelTableName(const std::string& rdfName) {
+    return rdfName + RDF_REL_TABLE_SUFFIX;
+}
+
+std::unique_ptr<BoundStatement> Binder::bindCreateRdfGraphClause(
+    const parser::Statement& statement) {
+    auto& createTableClause = reinterpret_cast<const CreateTableClause&>(statement);
+    auto rdfGraphName = createTableClause.getTableName();
+    auto stringType = std::make_unique<LogicalType>(LogicalTypeID::STRING);
+    // RDF node (resource) table
+    auto nodeTableName = getRdfNodeTableName(rdfGraphName);
+    std::vector<std::unique_ptr<Property>> nodeProperties;
+    nodeProperties.push_back(std::make_unique<Property>(RDF_IRI, stringType->copy()));
+    auto boundNodeExtraInfo =
+        std::make_unique<BoundNodeExtraCreateTableInfo>(0 /* primaryKeyIdx */);
+    auto boundNodeCreateInfo = std::make_unique<BoundCreateTableInfo>(
+        nodeTableName, std::move(nodeProperties), std::move(boundNodeExtraInfo));
+    // RDF rel (triples) table
+    auto relTableName = getRdfRelTableName(rdfGraphName);
+    std::vector<std::unique_ptr<Property>> relProperties;
+    relProperties.push_back(std::make_unique<Property>(
+        RDF_PREDICT_ID, std::make_unique<LogicalType>(LogicalTypeID::INTERNAL_ID)));
+    auto boundRelExtraInfo =
+        std::make_unique<BoundRelExtraCreateTableInfo>(RelMultiplicity::MANY_MANY, INVALID_TABLE_ID,
+            INVALID_TABLE_ID, stringType->copy(), stringType->copy());
+    auto boundRelCreateInfo = std::make_unique<BoundCreateTableInfo>(
+        relTableName, std::move(relProperties), std::move(boundRelExtraInfo));
+    auto boundExtraInfo = std::make_unique<BoundRdfExtraCreateTableInfo>(
+        std::move(boundNodeCreateInfo), std::move(boundRelCreateInfo));
+    auto boundCreateInfo =
+        std::make_unique<BoundCreateTableInfo>(rdfGraphName, std::move(boundExtraInfo));
+    return std::make_unique<BoundCreateTable>(
+        StatementType::CREATE_RDF_GRAPH, rdfGraphName, std::move(boundCreateInfo));
 }
 
 std::unique_ptr<BoundStatement> Binder::bindDropTableClause(const parser::Statement& statement) {

diff --git a/src/binder/binder.cpp b/src/binder/binder.cpp
@@ -20,6 +20,9 @@ std::unique_ptr<BoundStatement> Binder::bind(const Statement& statement) {
     case StatementType::CREATE_REL_TABLE: {
         boundStatement = bindCreateRelTableClause(statement);
     } break;
+    case StatementType::CREATE_RDF_GRAPH: {
+        boundStatement = bindCreateRdfGraphClause(statement);
+    } break;
     case StatementType::COPY_FROM: {
         boundStatement = bindCopyFromClause(statement);
     } break;

diff --git a/src/binder/bound_statement_visitor.cpp b/src/binder/bound_statement_visitor.cpp
@@ -18,6 +18,9 @@ void BoundStatementVisitor::visit(const kuzu::binder::BoundStatement& statement)
     case StatementType::CREATE_REL_TABLE: {
         visitCreateRelTable(statement);
     } break;
+    case StatementType::CREATE_RDF_GRAPH: {
+        visitCreateRdfGraph(statement);
+    } break;
     case StatementType::DROP_TABLE: {
         visitDropTable(statement);
     } break;

diff --git a/src/catalog/catalog.cpp b/src/catalog/catalog.cpp
@@ -38,27 +38,29 @@ ExpressionType Catalog::getFunctionType(const std::string& name) const {
     return catalogContentForReadOnlyTrx->getFunctionType(name);
 }
 
-table_id_t Catalog::addNodeTableSchema(std::string tableName, property_id_t primaryKeyId,
-    std::vector<std::unique_ptr<Property>> propertyDefinitions) {
+table_id_t Catalog::addNodeTableSchema(const binder::BoundCreateTableInfo& info) {
     initCatalogContentForWriteTrxIfNecessary();
-    auto tableID = catalogContentForWriteTrx->addNodeTableSchema(
-        std::move(tableName), primaryKeyId, std::move(propertyDefinitions));
+    auto tableID = catalogContentForWriteTrx->addNodeTableSchema(info);
     wal->logNodeTableRecord(tableID);
     return tableID;
 }
 
-table_id_t Catalog::addRelTableSchema(std::string tableName, RelMultiplicity relMultiplicity,
-    std::vector<std::unique_ptr<Property>> propertyDefinitions, table_id_t srcTableID,
-    table_id_t dstTableID, std::unique_ptr<LogicalType> srcPKDataType,
-    std::unique_ptr<LogicalType> dstPKDataType) {
+table_id_t Catalog::addRelTableSchema(const binder::BoundCreateTableInfo& info) {
     initCatalogContentForWriteTrxIfNecessary();
-    auto tableID = catalogContentForWriteTrx->addRelTableSchema(std::move(tableName),
-        relMultiplicity, std::move(propertyDefinitions), srcTableID, dstTableID,
-        std::move(srcPKDataType), std::move(dstPKDataType));
+    auto tableID = catalogContentForWriteTrx->addRelTableSchema(info);
     wal->logRelTableRecord(tableID);
     return tableID;
 }
 
+common::table_id_t Catalog::addRdfGraphSchema(const binder::BoundCreateTableInfo& info) {
+    initCatalogContentForWriteTrxIfNecessary();
+    auto tableID = catalogContentForWriteTrx->addRdfGraphSchema(info);
+    auto rdfGraphSchema = (RdfGraphSchema*)catalogContentForWriteTrx->getTableSchema(tableID);
+    wal->logRdfGraphRecord(
+        tableID, rdfGraphSchema->getNodeTableID(), rdfGraphSchema->getRelTableID());
+    return tableID;
+}
+
 void Catalog::dropTableSchema(table_id_t tableID) {
     initCatalogContentForWriteTrxIfNecessary();
     catalogContentForWriteTrx->dropTableSchema(tableID);

diff --git a/src/catalog/catalog_content.cpp b/src/catalog/catalog_content.cpp
@@ -20,41 +20,71 @@ CatalogContent::CatalogContent(const std::string& directory) {
     registerBuiltInFunctions();
 }
 
-table_id_t CatalogContent::addNodeTableSchema(std::string tableName, property_id_t primaryKeyId,
-    std::vector<std::unique_ptr<Property>> properties) {
-    table_id_t tableID = assignNextTableID();
+static void assignPropertyIDAndTableID(
+    std::vector<std::unique_ptr<Property>>& properties, table_id_t tableID) {
     for (auto i = 0u; i < properties.size(); ++i) {
         properties[i]->setPropertyID(i);
         properties[i]->setTableID(tableID);
     }
+}
+
+table_id_t CatalogContent::addNodeTableSchema(const binder::BoundCreateTableInfo& info) {
+    table_id_t tableID = assignNextTableID();
+    auto properties = Property::copy(info.properties);
+    assignPropertyIDAndTableID(properties, tableID);
+    auto extraInfo = (binder::BoundNodeExtraCreateTableInfo*)info.extraInfo.get();
     auto nodeTableSchema = std::make_unique<NodeTableSchema>(
-        std::move(tableName), tableID, primaryKeyId, std::move(properties));
+        info.tableName, tableID, extraInfo->primaryKeyIdx, std::move(properties));
     tableNameToIDMap.emplace(nodeTableSchema->tableName, tableID);
     tableSchemas.emplace(tableID, std::move(nodeTableSchema));
     return tableID;
 }
 
-table_id_t CatalogContent::addRelTableSchema(std::string tableName, RelMultiplicity relMultiplicity,
-    std::vector<std::unique_ptr<Property>> properties, table_id_t srcTableID, table_id_t dstTableID,
-    std::unique_ptr<LogicalType> srcPKDataType, std::unique_ptr<LogicalType> dstPKDataType) {
-    table_id_t tableID = assignNextTableID();
-    getNodeTableSchema(srcTableID)->addFwdRelTableID(tableID);
-    getNodeTableSchema(dstTableID)->addBwdRelTableID(tableID);
+// TODO(Xiyang): move this to binding stage
+static void addRelInternalIDProperty(std::vector<std::unique_ptr<Property>>& properties) {
     auto relInternalIDProperty = std::make_unique<Property>(
         InternalKeyword::ID, std::make_unique<LogicalType>(LogicalTypeID::INTERNAL_ID));
     properties.insert(properties.begin(), std::move(relInternalIDProperty));
-    for (auto i = 0u; i < properties.size(); ++i) {
-        properties[i]->setPropertyID(i);
-        properties[i]->setTableID(tableID);
-    }
-    auto relTableSchema = std::make_unique<RelTableSchema>(std::move(tableName), tableID,
-        relMultiplicity, std::move(properties), srcTableID, dstTableID, std::move(srcPKDataType),
-        std::move(dstPKDataType));
+}
+
+table_id_t CatalogContent::addRelTableSchema(const binder::BoundCreateTableInfo& info) {
+    table_id_t tableID = assignNextTableID();
+    auto properties = Property::copy(info.properties);
+    addRelInternalIDProperty(properties);
+    assignPropertyIDAndTableID(properties, tableID);
+    auto extraInfo = (binder::BoundRelExtraCreateTableInfo*)info.extraInfo.get();
+    getNodeTableSchema(extraInfo->srcTableID)->addFwdRelTableID(tableID);
+    getNodeTableSchema(extraInfo->dstTableID)->addBwdRelTableID(tableID);
+    auto relTableSchema = std::make_unique<RelTableSchema>(info.tableName, tableID,
+        extraInfo->relMultiplicity, std::move(properties), extraInfo->srcTableID,
+        extraInfo->dstTableID, extraInfo->srcPkDataType->copy(), extraInfo->dstPkDataType->copy());
     tableNameToIDMap.emplace(relTableSchema->tableName, tableID);
     tableSchemas.emplace(tableID, std::move(relTableSchema));
     return tableID;
 }
 
+common::table_id_t CatalogContent::addRdfGraphSchema(const binder::BoundCreateTableInfo& info) {
+    table_id_t rdfGraphID = assignNextTableID();
+    auto extraInfo = (binder::BoundRdfExtraCreateTableInfo*)info.extraInfo.get();
+    auto nodeInfo = extraInfo->nodeInfo.get();
+    auto relInfo = extraInfo->relInfo.get();
+    auto nodeExtraInfo = (binder::BoundNodeExtraCreateTableInfo*)nodeInfo->extraInfo.get();
+    auto relExtraInfo = (binder::BoundRelExtraCreateTableInfo*)relInfo->extraInfo.get();
+    // Node table schema
+    auto nodeTableID = addNodeTableSchema(*nodeInfo);
+    // Rel table schema
+    relExtraInfo->srcTableID = nodeTableID;
+    relExtraInfo->dstTableID = nodeTableID;
+    auto relTableID = addRelTableSchema(*relInfo);
+    // Rdf table schema
+    auto rdfGraphName = info.tableName;
+    auto rdfGraphSchema =
+        std::make_unique<RdfGraphSchema>(rdfGraphName, rdfGraphID, nodeTableID, relTableID);
+    tableNameToIDMap.emplace(rdfGraphName, rdfGraphID);
+    tableSchemas.emplace(rdfGraphID, std::move(rdfGraphSchema));
+    return rdfGraphID;
+}
+
 bool CatalogContent::containNodeTable(const std::string& tableName) const {
     if (!tableNameToIDMap.contains(tableName)) {
         return false;

diff --git a/src/catalog/property.cpp b/src/catalog/property.cpp
@@ -43,14 +43,13 @@ std::unique_ptr<Property> Property::deserialize(FileInfo* fileInfo, uint64_t& of
         name, std::move(dataType), propertyID, tableID, std::move(metadataDAHInfo));
 }
 
-std::vector<std::unique_ptr<catalog::Property>> Property::copyProperties(
-    const std::vector<std::unique_ptr<catalog::Property>>& propertiesToCopy) {
-    std::vector<std::unique_ptr<catalog::Property>> propertiesToReturn;
-    propertiesToReturn.reserve(propertiesToCopy.size());
-    for (const auto& property : propertiesToCopy) {
-        propertiesToReturn.push_back(property->copy());
+std::vector<std::unique_ptr<Property>> Property::copy(
+    const std::vector<std::unique_ptr<Property>>& properties) {
+    std::vector<std::unique_ptr<Property>> result;
+    for (auto& property : properties) {
+        result.push_back(property->copy());
     }
-    return propertiesToReturn;
+    return result;
 }
 
 } // namespace catalog

diff --git a/src/catalog/table_schema.cpp b/src/catalog/table_schema.cpp
@@ -124,6 +124,9 @@ std::unique_ptr<TableSchema> TableSchema::deserialize(FileInfo* fileInfo, uint64
     case TableType::REL: {
         result = RelTableSchema::deserialize(fileInfo, offset);
     } break;
+    case TableType::RDF: {
+        result = RdfGraphSchema::deserialize(fileInfo, offset);
+    } break;
     default: {
         throw NotImplementedException{"TableSchema::deserialize"};
     }
@@ -136,15 +139,6 @@ std::unique_ptr<TableSchema> TableSchema::deserialize(FileInfo* fileInfo, uint64
     return result;
 }
 
-std::vector<std::unique_ptr<Property>> TableSchema::copyProperties() const {
-    std::vector<std::unique_ptr<Property>> propertiesCopy;
-    propertiesCopy.reserve(properties.size());
-    for (auto& property : properties) {
-        propertiesCopy.push_back(property->copy());
-    }
-    return propertiesCopy;
-}
-
 void NodeTableSchema::serializeInternal(FileInfo* fileInfo, uint64_t& offset) {
     SerDeser::serializeValue(primaryKeyPropertyID, fileInfo, offset);
     SerDeser::serializeUnorderedSet(fwdRelTableIDSet, fileInfo, offset);
@@ -184,5 +178,18 @@ std::unique_ptr<RelTableSchema> RelTableSchema::deserialize(FileInfo* fileInfo,
         std::move(srcPKDataType), std::move(dstPKDataType));
 }
 
+void RdfGraphSchema::serializeInternal(FileInfo* fileInfo, uint64_t& offset) {
+    SerDeser::serializeValue(nodeTableID, fileInfo, offset);
+    SerDeser::serializeValue(relTableID, fileInfo, offset);
+}
+
+std::unique_ptr<RdfGraphSchema> RdfGraphSchema::deserialize(FileInfo* fileInfo, uint64_t& offset) {
+    table_id_t nodeTableID;
+    table_id_t relTableID;
+    SerDeser::deserializeValue(nodeTableID, fileInfo, offset);
+    SerDeser::deserializeValue(relTableID, fileInfo, offset);
+    return std::make_unique<RdfGraphSchema>(nodeTableID, relTableID);
+}
+
 } // namespace catalog
 } // namespace kuzu
diff --git a/src/include/binder/binder.h b/src/include/binder/binder.h
@@ -80,6 +80,7 @@ class Binder {
     /*** bind DDL ***/
     std::unique_ptr<BoundStatement> bindCreateNodeTableClause(const parser::Statement& statement);
     std::unique_ptr<BoundStatement> bindCreateRelTableClause(const parser::Statement& statement);
+    std::unique_ptr<BoundStatement> bindCreateRdfGraphClause(const parser::Statement& statement);
     std::unique_ptr<BoundStatement> bindDropTableClause(const parser::Statement& statement);
     std::unique_ptr<BoundStatement> bindRenameTableClause(const parser::Statement& statement);
     std::unique_ptr<BoundStatement> bindAddPropertyClause(const parser::Statement& statement);

diff --git a/src/include/binder/bound_statement_visitor.h b/src/include/binder/bound_statement_visitor.h
@@ -20,6 +20,7 @@ class BoundStatementVisitor {
 protected:
     virtual void visitCreateNodeTable(const BoundStatement& statement) {}
     virtual void visitCreateRelTable(const BoundStatement& statement) {}
+    virtual void visitCreateRdfGraph(const BoundStatement& statement) {}
     virtual void visitDropTable(const BoundStatement& statement) {}
     virtual void visitRenameTable(const BoundStatement& statement) {}
     virtual void visitAddProperty(const BoundStatement& statement) {}