Merge pull request #2009 from kuzudb/issue-1888

Support defining multiple FROM TO in CREATE REL TABLE GROUP
kuzudb · Sep 8, 2023 · 46f4039 · 46f4039
2 parents d03d306 + 8fd6d3f
commit 46f4039
Show file tree

Hide file tree

Showing 55 changed files with 4,790 additions and 4,101 deletions.
diff --git a/src/antlr4/Cypher.g4 b/src/antlr4/Cypher.g4
@@ -58,26 +58,33 @@ COPY : ( 'C' | 'c' ) ( 'O' | 'o' ) ( 'P' | 'p') ( 'Y' | 'y' ) ;
 
 FROM : ( 'F' | 'f' ) ( 'R' | 'r' ) ( 'O' | 'o' ) ( 'M' | 'm' ) ;
 
-NPY : ( 'N' | 'n' ) ( 'P' | 'p' ) ( 'Y' | 'y' ) ;
-
 COLUMN : ( 'C' | 'c' ) ( 'O' | 'o' ) ( 'L' | 'l' ) ( 'U' | 'u' ) ( 'M' | 'm' ) ( 'N' | 'n' ) ;
 
 kU_DDL
-    : kU_CreateNode
-        | kU_CreateRel
+    : kU_CreateNodeTable
+        | kU_CreateRelTable
+        | kU_CreateRelTableGroup
         | kU_CreateRdfGraph
         | kU_DropTable
         | kU_AlterTable;
 
-kU_CreateNode
+kU_CreateNodeTable
     : CREATE SP NODE SP TABLE SP oC_SchemaName SP? '(' SP? kU_PropertyDefinitions SP? ( ',' SP? kU_CreateNodeConstraint ) SP? ')' ;
 
 NODE : ( 'N' | 'n' ) ( 'O' | 'o' ) ( 'D' | 'd' ) ( 'E' | 'e' ) ;
 
 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_CreateRelTable
+    : CREATE SP REL SP TABLE SP oC_SchemaName SP? '(' SP? kU_RelTableConnection SP? ( ',' SP? kU_PropertyDefinitions SP? )? ( ',' SP? oC_SymbolicName SP? )?  ')' ;
+
+kU_CreateRelTableGroup
+   : CREATE SP REL SP TABLE SP GROUP SP oC_SchemaName SP? '(' SP? kU_RelTableConnection SP ? (',' SP? kU_RelTableConnection)+ SP? ( ',' SP? kU_PropertyDefinitions SP? )? ( ',' SP? oC_SymbolicName SP? )?  ')' ;
+
+GROUP : ( 'G' | 'g' ) ( 'R' | 'r' ) ( 'O' | 'o' ) ( 'U' | 'u' ) ( 'P' | 'p' ) ;
+
+kU_RelTableConnection
+    : FROM SP oC_SchemaName SP TO SP oC_SchemaName ;
 
 kU_CreateRdfGraph
     : CREATE SP RDF SP GRAPH SP oC_SchemaName ;

diff --git a/src/binder/CMakeLists.txt b/src/binder/CMakeLists.txt
@@ -1,6 +1,7 @@
 add_subdirectory(bind)
 add_subdirectory(bind_expression)
 add_subdirectory(copy)
+add_subdirectory(ddl)
 add_subdirectory(expression)
 add_subdirectory(query)
 add_subdirectory(rewriter)

diff --git a/src/binder/bind/bind_ddl.cpp b/src/binder/bind/bind_ddl.cpp
diff --git a/src/binder/binder.cpp b/src/binder/binder.cpp
@@ -14,14 +14,8 @@ namespace binder {
 std::unique_ptr<BoundStatement> Binder::bind(const Statement& statement) {
     std::unique_ptr<BoundStatement> boundStatement;
     switch (statement.getStatementType()) {
-    case StatementType::CREATE_NODE_TABLE: {
-        boundStatement = bindCreateNodeTableClause(statement);
-    } break;
-    case StatementType::CREATE_REL_TABLE: {
-        boundStatement = bindCreateRelTableClause(statement);
-    } break;
-    case StatementType::CREATE_RDF_GRAPH: {
-        boundStatement = bindCreateRdfGraphClause(statement);
+    case StatementType::CREATE_TABLE: {
+        boundStatement = bindCreateTable(statement);
     } break;
     case StatementType::COPY_FROM: {
         boundStatement = bindCopyFromClause(statement);
@@ -30,19 +24,19 @@ std::unique_ptr<BoundStatement> Binder::bind(const Statement& statement) {
         boundStatement = bindCopyToClause(statement);
     } break;
     case StatementType::DROP_TABLE: {
-        boundStatement = bindDropTableClause(statement);
+        boundStatement = bindDropTable(statement);
     } break;
     case StatementType::RENAME_TABLE: {
-        boundStatement = bindRenameTableClause(statement);
+        boundStatement = bindRenameTable(statement);
     } break;
     case StatementType::ADD_PROPERTY: {
-        boundStatement = bindAddPropertyClause(statement);
+        boundStatement = bindAddProperty(statement);
     } break;
     case StatementType::DROP_PROPERTY: {
-        boundStatement = bindDropPropertyClause(statement);
+        boundStatement = bindDropProperty(statement);
     } break;
     case StatementType::RENAME_PROPERTY: {
-        boundStatement = bindRenamePropertyClause(statement);
+        boundStatement = bindRenameProperty(statement);
     } break;
     case StatementType::QUERY: {
         boundStatement = bindQuery((const RegularQuery&)statement);

diff --git a/src/binder/bound_statement_visitor.cpp b/src/binder/bound_statement_visitor.cpp
@@ -12,14 +12,8 @@ void BoundStatementVisitor::visit(const kuzu::binder::BoundStatement& statement)
     case StatementType::QUERY: {
         visitRegularQuery((BoundRegularQuery&)statement);
     } break;
-    case StatementType::CREATE_NODE_TABLE: {
-        visitCreateNodeTable(statement);
-    } break;
-    case StatementType::CREATE_REL_TABLE: {
-        visitCreateRelTable(statement);
-    } break;
-    case StatementType::CREATE_RDF_GRAPH: {
-        visitCreateRdfGraph(statement);
+    case StatementType::CREATE_TABLE: {
+        visitCreateTable(statement);
     } break;
     case StatementType::DROP_TABLE: {
         visitDropTable(statement);

diff --git a/src/binder/ddl/CMakeLists.txt b/src/binder/ddl/CMakeLists.txt
@@ -0,0 +1,8 @@
+add_library(
+        kuzu_binder_ddl
+        OBJECT
+        bound_create_table_info.cpp)
+
+set(ALL_OBJECT_FILES
+        ${ALL_OBJECT_FILES} $<TARGET_OBJECTS:kuzu_binder_ddl>
+        PARENT_SCOPE)
diff --git a/src/binder/ddl/bound_create_table_info.cpp b/src/binder/ddl/bound_create_table_info.cpp
@@ -0,0 +1,17 @@
+#include "binder/ddl/bound_create_table_info.h"
+
+namespace kuzu {
+namespace binder {
+
+std::vector<std::unique_ptr<BoundCreateTableInfo>> BoundCreateTableInfo::copy(
+    const std::vector<std::unique_ptr<BoundCreateTableInfo>>& infos) {
+    std::vector<std::unique_ptr<BoundCreateTableInfo>> result;
+    result.reserve(infos.size());
+    for (auto& info : infos) {
+        result.push_back(info->copy());
+    }
+    return result;
+}
+
+} // namespace binder
+} // namespace kuzu
diff --git a/src/catalog/CMakeLists.txt b/src/catalog/CMakeLists.txt
@@ -3,7 +3,8 @@ add_library(kuzu_catalog
         catalog.cpp
         catalog_content.cpp
         table_schema.cpp
-        property.cpp)
+        property.cpp
+        rel_table_group_schema.cpp)
 
 set(ALL_OBJECT_FILES
         ${ALL_OBJECT_FILES} $<TARGET_OBJECTS:kuzu_catalog>

diff --git a/src/catalog/catalog.cpp b/src/catalog/catalog.cpp
@@ -1,5 +1,6 @@
 #include "catalog/catalog.h"
 
+#include "catalog/rel_table_group_schema.h"
 #include "common/ser_deser.h"
 #include "common/string_utils.h"
 #include "storage/storage_utils.h"
@@ -52,6 +53,19 @@ table_id_t Catalog::addRelTableSchema(const binder::BoundCreateTableInfo& info)
     return tableID;
 }
 
+common::table_id_t Catalog::addRelTableGroupSchema(const binder::BoundCreateTableInfo& info) {
+    initCatalogContentForWriteTrxIfNecessary();
+    auto tableID = catalogContentForWriteTrx->addRelTableGroupSchema(info);
+    auto relTableGroupSchema =
+        (RelTableGroupSchema*)catalogContentForWriteTrx->getTableSchema(tableID);
+    // TODO(Ziyi): remove this when we can log variable size record. See also wal_record.h
+    for (auto relTableID : relTableGroupSchema->getRelTableIDs()) {
+        wal->logRelTableRecord(relTableID);
+    }
+    wal->logRelTableGroupRecord(tableID, relTableGroupSchema->getRelTableIDs());
+    return tableID;
+}
+
 common::table_id_t Catalog::addRdfGraphSchema(const binder::BoundCreateTableInfo& info) {
     initCatalogContentForWriteTrxIfNecessary();
     auto tableID = catalogContentForWriteTrx->addRdfGraphSchema(info);

diff --git a/src/catalog/catalog_content.cpp b/src/catalog/catalog_content.cpp
@@ -1,5 +1,6 @@
 #include "catalog/catalog_content.h"
 
+#include "catalog/rel_table_group_schema.h"
 #include "common/ser_deser.h"
 #include "common/string_utils.h"
 #include "storage/storage_utils.h"
@@ -31,9 +32,9 @@ static void assignPropertyIDAndTableID(
 
 table_id_t CatalogContent::addNodeTableSchema(const BoundCreateTableInfo& info) {
     table_id_t tableID = assignNextTableID();
-    auto properties = Property::copy(info.properties);
+    auto extraInfo = reinterpret_cast<BoundExtraCreateNodeTableInfo*>(info.extraInfo.get());
+    auto properties = Property::copy(extraInfo->properties);
     assignPropertyIDAndTableID(properties, tableID);
-    auto extraInfo = (BoundNodeExtraCreateTableInfo*)info.extraInfo.get();
     auto nodeTableSchema = std::make_unique<NodeTableSchema>(
         info.tableName, tableID, extraInfo->primaryKeyIdx, std::move(properties));
     tableNameToIDMap.emplace(nodeTableSchema->tableName, tableID);
@@ -50,10 +51,10 @@ static void addRelInternalIDProperty(std::vector<std::unique_ptr<Property>>& pro
 
 table_id_t CatalogContent::addRelTableSchema(const BoundCreateTableInfo& info) {
     table_id_t tableID = assignNextTableID();
-    auto properties = Property::copy(info.properties);
+    auto extraInfo = reinterpret_cast<BoundExtraCreateRelTableInfo*>(info.extraInfo.get());
+    auto properties = Property::copy(extraInfo->properties);
     addRelInternalIDProperty(properties);
     assignPropertyIDAndTableID(properties, tableID);
-    auto extraInfo = (BoundRelExtraCreateTableInfo*)info.extraInfo.get();
     getNodeTableSchema(extraInfo->srcTableID)->addFwdRelTableID(tableID);
     getNodeTableSchema(extraInfo->dstTableID)->addBwdRelTableID(tableID);
     auto relTableSchema = std::make_unique<RelTableSchema>(info.tableName, tableID,
@@ -64,13 +65,28 @@ table_id_t CatalogContent::addRelTableSchema(const BoundCreateTableInfo& info) {
     return tableID;
 }
 
-common::table_id_t CatalogContent::addRdfGraphSchema(const BoundCreateTableInfo& info) {
+table_id_t CatalogContent::addRelTableGroupSchema(const binder::BoundCreateTableInfo& info) {
+    table_id_t relTableGroupID = assignNextTableID();
+    auto extraInfo = (BoundExtraCreateRelTableGroupInfo*)info.extraInfo.get();
+    std::vector<table_id_t> relTableIDs;
+    relTableIDs.reserve(extraInfo->infos.size());
+    for (auto& childInfo : extraInfo->infos) {
+        relTableIDs.push_back(addRelTableSchema(*childInfo));
+    }
+    auto relTableGroupName = info.tableName;
+    auto relTableGroupSchema = std::make_unique<RelTableGroupSchema>(
+        relTableGroupName, relTableGroupID, std::move(relTableIDs));
+    tableNameToIDMap.emplace(relTableGroupName, relTableGroupID);
+    tableSchemas.emplace(relTableGroupID, std::move(relTableGroupSchema));
+    return relTableGroupID;
+}
+
+table_id_t CatalogContent::addRdfGraphSchema(const BoundCreateTableInfo& info) {
     table_id_t rdfGraphID = assignNextTableID();
-    auto extraInfo = (BoundRdfExtraCreateTableInfo*)info.extraInfo.get();
+    auto extraInfo = reinterpret_cast<BoundExtraCreateRdfGraphInfo*>(info.extraInfo.get());
     auto nodeInfo = extraInfo->nodeInfo.get();
     auto relInfo = extraInfo->relInfo.get();
-    auto nodeExtraInfo = (BoundNodeExtraCreateTableInfo*)nodeInfo->extraInfo.get();
-    auto relExtraInfo = (BoundRelExtraCreateTableInfo*)relInfo->extraInfo.get();
+    auto relExtraInfo = (BoundExtraCreateRelTableInfo*)relInfo->extraInfo.get();
     // Node table schema
     auto nodeTableID = addNodeTableSchema(*nodeInfo);
     // Rel table schema

diff --git a/src/catalog/rel_table_group_schema.cpp b/src/catalog/rel_table_group_schema.cpp
@@ -0,0 +1,22 @@
+#include "catalog/rel_table_group_schema.h"
+
+#include "common/ser_deser.h"
+
+using namespace kuzu::common;
+
+namespace kuzu {
+namespace catalog {
+
+void RelTableGroupSchema::serializeInternal(FileInfo* fileInfo, uint64_t& offset) {
+    SerDeser::serializeVector(relTableIDs, fileInfo, offset);
+}
+
+std::unique_ptr<RelTableGroupSchema> RelTableGroupSchema::deserialize(
+    FileInfo* fileInfo, uint64_t& offset) {
+    std::vector<table_id_t> relTableIDs;
+    SerDeser::deserializeVector(relTableIDs, fileInfo, offset);
+    return std::make_unique<RelTableGroupSchema>(std::move(relTableIDs));
+}
+
+} // namespace catalog
+} // namespace kuzu
diff --git a/src/catalog/table_schema.cpp b/src/catalog/table_schema.cpp
@@ -1,5 +1,6 @@
 #include "catalog/table_schema.h"
 
+#include "catalog/rel_table_group_schema.h"
 #include "common/exception.h"
 #include "common/ser_deser.h"
 #include "common/string_utils.h"
@@ -124,6 +125,9 @@ std::unique_ptr<TableSchema> TableSchema::deserialize(FileInfo* fileInfo, uint64
     case TableType::REL: {
         result = RelTableSchema::deserialize(fileInfo, offset);
     } break;
+    case TableType::REL_GROUP: {
+        result = RelTableGroupSchema::deserialize(fileInfo, offset);
+    } break;
     case TableType::RDF: {
         result = RdfGraphSchema::deserialize(fileInfo, offset);
     } break;

diff --git a/src/include/binder/binder.h b/src/include/binder/binder.h
@@ -7,6 +7,10 @@
 #include "parser/query/regular_query.h"
 
 namespace kuzu {
+namespace parser {
+struct CreateTableInfo;
+}
+
 namespace main {
 class ClientContext;
 }
@@ -80,22 +84,25 @@ class Binder {
         const std::string& name, const common::LogicalType& dataType);
 
     /*** 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);
-    std::unique_ptr<BoundStatement> bindDropPropertyClause(const parser::Statement& statement);
-    std::unique_ptr<BoundStatement> bindRenamePropertyClause(const parser::Statement& statement);
-
-    std::vector<std::unique_ptr<catalog::Property>> bindProperties(
-        std::vector<std::pair<std::string, std::string>> propertyNameDataTypes);
-    uint32_t bindPrimaryKey(const std::string& pkColName,
-        std::vector<std::pair<std::string, std::string>> propertyNameDataTypes);
+    std::unique_ptr<BoundCreateTableInfo> bindCreateTableInfo(const parser::CreateTableInfo* info);
+    std::unique_ptr<BoundCreateTableInfo> bindCreateNodeTableInfo(
+        const parser::CreateTableInfo* info);
+    std::unique_ptr<BoundCreateTableInfo> bindCreateRelTableInfo(
+        const parser::CreateTableInfo* info);
+    std::unique_ptr<BoundCreateTableInfo> bindCreateRelTableGroupInfo(
+        const parser::CreateTableInfo* info);
+    std::unique_ptr<BoundCreateTableInfo> bindCreateRdfGraphInfo(
+        const parser::CreateTableInfo* info);
+    std::unique_ptr<BoundStatement> bindCreateTable(const parser::Statement& statement);
+
+    std::unique_ptr<BoundStatement> bindDropTable(const parser::Statement& statement);
+    std::unique_ptr<BoundStatement> bindRenameTable(const parser::Statement& statement);
+    std::unique_ptr<BoundStatement> bindAddProperty(const parser::Statement& statement);
+    std::unique_ptr<BoundStatement> bindDropProperty(const parser::Statement& statement);
+    std::unique_ptr<BoundStatement> bindRenameProperty(const parser::Statement& statement);
+
     common::property_id_t bindPropertyName(
         catalog::NodeTableSchema::TableSchema* tableSchema, const std::string& propertyName);
-    std::unique_ptr<common::LogicalType> bindDataType(const std::string& dataType);
 
     /*** bind copy from/to ***/
     static bool bindContainsSerial(

diff --git a/src/include/binder/bound_statement_visitor.h b/src/include/binder/bound_statement_visitor.h
@@ -18,9 +18,7 @@ class BoundStatementVisitor {
     virtual void visitQueryPart(const NormalizedQueryPart& queryPart);
 
 protected:
-    virtual void visitCreateNodeTable(const BoundStatement& statement) {}
-    virtual void visitCreateRelTable(const BoundStatement& statement) {}
-    virtual void visitCreateRdfGraph(const BoundStatement& statement) {}
+    virtual void visitCreateTable(const BoundStatement& statement) {}
     virtual void visitDropTable(const BoundStatement& statement) {}
     virtual void visitRenameTable(const BoundStatement& statement) {}
     virtual void visitAddProperty(const BoundStatement& statement) {}

diff --git a/src/include/binder/ddl/bound_create_table.h b/src/include/binder/ddl/bound_create_table.h
@@ -8,9 +8,9 @@ namespace binder {
 
 class BoundCreateTable : public BoundDDL {
 public:
-    explicit BoundCreateTable(common::StatementType statementType, std::string tableName,
-        std::unique_ptr<BoundCreateTableInfo> info)
-        : BoundDDL{statementType, std::move(tableName)}, info{std::move(info)} {}
+    explicit BoundCreateTable(std::string tableName, std::unique_ptr<BoundCreateTableInfo> info)
+        : BoundDDL{common::StatementType::CREATE_TABLE, std::move(tableName)}, info{std::move(
+                                                                                   info)} {}
 
     inline BoundCreateTableInfo* getBoundCreateTableInfo() const { return info.get(); }