kuzudb · andyfengHKU · Sep 12, 2023 · Sep 11, 2023
diff --git a/dataset/rel-group/copy.cypher b/dataset/rel-group/copy.cypher
@@ -0,0 +1,8 @@
+COPY personA FROM "dataset/rel-group/node.csv" (HeaDER=true, deLim=',');
+COPY personB FROM "dataset/rel-group/node.csv" (HeaDER=true, deLim=',');
+COPY personC FROM "dataset/rel-group/node.csv" (HeaDER=true, deLim=',');
+COPY knows_personA_personB FROM "dataset/rel-group/edge.csv";
+COPY knows_personA_personC FROM "dataset/rel-group/edge.csv";
+COPY knows_personB_personC FROM "dataset/rel-group/edge.csv";
+COPY likes_personA_personB FROM "dataset/rel-group/edge.csv";
+COPY likes_personB_personA FROM "dataset/rel-group/edge.csv";
diff --git a/dataset/rel-group/edge.csv b/dataset/rel-group/edge.csv
@@ -0,0 +1,14 @@
+0,2,2021-06-30
+0,3,2021-06-30
+0,5,2021-06-30
+2,0,2021-06-30
+2,3,1950-05-14
+2,5,1950-05-14
+3,0,2021-06-30
+3,2,1950-05-14
+3,5,2000-01-01
+5,0,2021-06-30
+5,2,1950-05-14
+5,3,2000-01-01
+7,8,1905-12-12
+7,9,1905-12-12
diff --git a/dataset/rel-group/node.csv b/dataset/rel-group/node.csv
@@ -0,0 +1,9 @@
+id,fname
+0,Alice
+2,Bob
+3,Carol
+5,Dan
+7,Elizabeth
+8,Farooq
+9,Greg
+10,Hubert Blaine Wolfeschlegelsteinhausenbergerdorff
diff --git a/dataset/rel-group/schema.cypher b/dataset/rel-group/schema.cypher
@@ -0,0 +1,5 @@
+create node table personA (ID INt64, fName StRING, PRIMARY KEY (ID));
+create node table personB (ID INt64, fName StRING, PRIMARY KEY (ID));
+create node table personC (ID INt64, fName StRING, PRIMARY KEY (ID));
+create rel table group knows (FROM personA TO personB, FROM personA To personC, FROM personB TO personC, date DATE);
+create rel table group likes (FROM personA TO personB, FROM personB To personA, date DATE);
diff --git a/src/binder/bind/bind_copy.cpp b/src/binder/bind/bind_copy.cpp
@@ -88,7 +88,7 @@ expression_vector Binder::bindColumnExpressions(
     return columnExpressions;
 }
 
-bool Binder::bindContainsSerial(TableSchema* tableSchema, CopyDescription::FileType fileType) {
+static bool bindContainsSerial(TableSchema* tableSchema, CopyDescription::FileType fileType) {
     bool containsSerial = false;
     for (auto& property : tableSchema->properties) {
         if (property->getDataType()->getLogicalTypeID() == LogicalTypeID::SERIAL) {
@@ -104,7 +104,7 @@ std::unique_ptr<BoundStatement> Binder::bindCopyFromClause(const Statement& stat
     auto catalogContent = catalog.getReadOnlyVersion();
     auto tableName = copyStatement.getTableName();
     // Bind to table schema.
-    validateTableExist(catalog, tableName);
+    validateNodeRelTableExist(tableName);
     auto tableID = catalogContent->getTableID(tableName);
     auto tableSchema = catalogContent->getTableSchema(tableID);
     // Bind csv reader configuration
@@ -127,17 +127,17 @@ std::unique_ptr<BoundStatement> Binder::bindCopyFromClause(const Statement& stat
         actualFileType, boundFilePaths, std::move(csvReaderConfig));
     auto nodeOffsetExpression =
         createVariable(std::string(Property::OFFSET_NAME), LogicalType{LogicalTypeID::INT64});
-    auto boundOffsetExpression = tableSchema->tableType == TableType::REL ?
-                                     createVariable(std::string(Property::REL_BOUND_OFFSET_NAME),
-                                         LogicalType{LogicalTypeID::ARROW_COLUMN}) :
-                                     nullptr;
-    auto nbrOffsetExpression = tableSchema->tableType == TableType::REL ?
-                                   createVariable(std::string(Property::REL_NBR_OFFSET_NAME),
-                                       LogicalType{LogicalTypeID::ARROW_COLUMN}) :
-                                   nullptr;
+    std::shared_ptr<Expression> boundOffset = nullptr;
+    std::shared_ptr<Expression> nbrOffset = nullptr;
+    if (tableSchema->tableType == TableType::REL) {
+        boundOffset = createVariable(
+            std::string(Property::REL_BOUND_OFFSET_NAME), LogicalType{LogicalTypeID::ARROW_COLUMN});
+        nbrOffset = createVariable(
+            std::string(Property::REL_NBR_OFFSET_NAME), LogicalType{LogicalTypeID::ARROW_COLUMN});
+    }
     auto boundCopyFromInfo = std::make_unique<BoundCopyFromInfo>(std::move(copyDescription),
         tableSchema, std::move(columnExpressions), std::move(nodeOffsetExpression),
-        std::move(boundOffsetExpression), std::move(nbrOffsetExpression), containsSerial);
+        std::move(boundOffset), std::move(nbrOffset), containsSerial);
     return std::make_unique<BoundCopyFrom>(std::move(boundCopyFromInfo));
 }
 

diff --git a/src/binder/bind/bind_ddl.cpp b/src/binder/bind/bind_ddl.cpp
@@ -234,7 +234,7 @@ std::unique_ptr<BoundStatement> Binder::bindCreateTable(const parser::Statement&
 std::unique_ptr<BoundStatement> Binder::bindDropTable(const Statement& statement) {
     auto& dropTable = (DropTable&)statement;
     auto tableName = dropTable.getTableName();
-    validateTableExist(catalog, tableName);
+    validateNodeRelTableExist(tableName);
     auto catalogContent = catalog.getReadOnlyVersion();
     auto tableID = catalogContent->getTableID(tableName);
     if (catalogContent->containNodeTable(tableName)) {
@@ -247,7 +247,7 @@ std::unique_ptr<BoundStatement> Binder::bindRenameTable(const Statement& stateme
     auto renameTable = (RenameTable&)statement;
     auto tableName = renameTable.getTableName();
     auto catalogContent = catalog.getReadOnlyVersion();
-    validateTableExist(catalog, tableName);
+    validateNodeRelTableExist(tableName);
     if (catalogContent->containTable(renameTable.getNewName())) {
         throw BinderException("Table: " + renameTable.getNewName() + " already exists.");
     }
@@ -258,7 +258,7 @@ std::unique_ptr<BoundStatement> Binder::bindRenameTable(const Statement& stateme
 std::unique_ptr<BoundStatement> Binder::bindAddProperty(const Statement& statement) {
     auto& addProperty = (AddProperty&)statement;
     auto tableName = addProperty.getTableName();
-    validateTableExist(catalog, tableName);
+    validateNodeRelTableExist(tableName);
     auto catalogContent = catalog.getReadOnlyVersion();
     auto tableID = catalogContent->getTableID(tableName);
     auto dataType = bindDataType(addProperty.getDataType());
@@ -277,7 +277,7 @@ std::unique_ptr<BoundStatement> Binder::bindAddProperty(const Statement& stateme
 std::unique_ptr<BoundStatement> Binder::bindDropProperty(const Statement& statement) {
     auto& dropProperty = (DropProperty&)statement;
     auto tableName = dropProperty.getTableName();
-    validateTableExist(catalog, tableName);
+    validateNodeRelTableExist(tableName);
     auto catalogContent = catalog.getReadOnlyVersion();
     auto tableID = catalogContent->getTableID(tableName);
     auto tableSchema = catalogContent->getTableSchema(tableID);
@@ -292,7 +292,7 @@ std::unique_ptr<BoundStatement> Binder::bindDropProperty(const Statement& statem
 std::unique_ptr<BoundStatement> Binder::bindRenameProperty(const Statement& statement) {
     auto& renameProperty = (RenameProperty&)statement;
     auto tableName = renameProperty.getTableName();
-    validateTableExist(catalog, tableName);
+    validateNodeRelTableExist(tableName);
     auto catalogContent = catalog.getReadOnlyVersion();
     auto tableID = catalogContent->getTableID(tableName);
     auto tableSchema = catalogContent->getTableSchema(tableID);

diff --git a/src/binder/bind/bind_graph_pattern.cpp b/src/binder/bind/bind_graph_pattern.cpp
@@ -4,6 +4,7 @@
 #include "binder/expression/path_expression.h"
 #include "binder/expression/property_expression.h"
 #include "catalog/node_table_schema.h"
+#include "catalog/rel_table_group_schema.h"
 #include "catalog/rel_table_schema.h"
 
 using namespace kuzu::common;
@@ -457,17 +458,34 @@
 }
 
 std::vector<table_id_t> Binder::bindRelTableIDs(const std::vector<std::string>& tableNames) {
-    if (catalog.getReadOnlyVersion()->getRelTableIDs().empty()) {
+    auto catalogContent = catalog.getReadOnlyVersion();
+    if (catalogContent->getRelTableIDs().empty()) {
         throw BinderException("No rel table exists in database.");
     }
+    // Rewrite empty rel pattern "-[]-" as all rel tables
     std::unordered_set<table_id_t> tableIDs;
     if (tableNames.empty()) {
-        for (auto tableID : catalog.getReadOnlyVersion()->getRelTableIDs()) {
+        for (auto tableID : catalogContent->getRelTableIDs()) {
             tableIDs.insert(tableID);
         }
     }
     for (auto& tableName : tableNames) {
-        tableIDs.insert(bindRelTableID(tableName));
+        validateTableExist(tableName);
+        auto tableID = catalogContent->getTableID(tableName);
+        auto tableSchema = catalogContent->getTableSchema(tableID);
+        switch (tableSchema->getTableType()) {
+        case TableType::REL: {
+            tableIDs.insert(tableID);
+        } break;
+        case TableType::REL_GROUP: {
+            auto relGroupSchema = reinterpret_cast<RelTableGroupSchema*>(tableSchema);
+            for (auto& relTableID : relGroupSchema->getRelTableIDs()) {
+                tableIDs.insert(relTableID);
+            }
+        } break;
+        default:
+            throw BinderException("Rel table " + tableName + " does not exist.");
+        }
     }
     auto result = std::vector<table_id_t>{tableIDs.begin(), tableIDs.end()};
     std::sort(result.begin(), result.end());

diff --git a/src/binder/binder.cpp b/src/binder/binder.cpp
@@ -67,18 +67,12 @@ std::shared_ptr<Expression> Binder::bindWhereExpression(const ParsedExpression&
     return whereExpression;
 }
 
-table_id_t Binder::bindRelTableID(const std::string& tableName) const {
-    if (!catalog.getReadOnlyVersion()->containRelTable(tableName)) {
-        throw BinderException("Rel table " + tableName + " does not exist.");
-    }
-    return catalog.getReadOnlyVersion()->getTableID(tableName);
-}
-
 table_id_t Binder::bindNodeTableID(const std::string& tableName) const {
-    if (!catalog.getReadOnlyVersion()->containNodeTable(tableName)) {
+    auto catalogContent = catalog.getReadOnlyVersion();
+    if (!catalogContent->containNodeTable(tableName)) {
         throw BinderException("Node table " + tableName + " does not exist.");
     }
-    return catalog.getReadOnlyVersion()->getTableID(tableName);
+    return catalogContent->getTableID(tableName);
 }
 
 std::shared_ptr<Expression> Binder::createVariable(
@@ -164,10 +158,16 @@ void Binder::validateReadNotFollowUpdate(const NormalizedSingleQuery& singleQuer
     }
 }
 
-void Binder::validateTableExist(const Catalog& _catalog, std::string& tableName) {
-    if (!_catalog.getReadOnlyVersion()->containNodeTable(tableName) &&
-        !_catalog.getReadOnlyVersion()->containRelTable(tableName)) {
-        throw BinderException("Node/Rel " + tableName + " does not exist.");
+void Binder::validateTableExist(const std::string& tableName) {
+    if (!catalog.getReadOnlyVersion()->containTable(tableName)) {
+        throw BinderException("Table " + tableName + " does not exist.");
+    }
+}
+
+void Binder::validateNodeRelTableExist(const std::string& tableName) {
+    if (!catalog.getReadOnlyVersion()->containNodeTable(tableName) &&
+        !catalog.getReadOnlyVersion()->containRelTable(tableName)) {
+        throw BinderException("Table " + tableName + " does not exist.");
     }
 }
 

diff --git a/src/catalog/catalog_content.cpp b/src/catalog/catalog_content.cpp
@@ -109,22 +109,6 @@ table_id_t CatalogContent::addRdfGraphSchema(const BoundCreateTableInfo& info) {
     return rdfGraphID;
 }
 
-bool CatalogContent::containNodeTable(const std::string& tableName) const {
-    if (!tableNameToIDMap.contains(tableName)) {
-        return false;
-    }
-    auto tableID = getTableID(tableName);
-    return tableSchemas.at(tableID)->tableType == TableType::NODE;
-}
-
-bool CatalogContent::containRelTable(const std::string& tableName) const {
-    if (!tableNameToIDMap.contains(tableName)) {
-        return false;
-    }
-    auto tableID = getTableID(tableName);
-    return tableSchemas.at(tableID)->tableType == TableType::REL;
-}
-
 Property* CatalogContent::getNodeProperty(
     table_id_t tableID, const std::string& propertyName) const {
     for (auto& property : tableSchemas.at(tableID)->properties) {
@@ -268,7 +252,15 @@ void CatalogContent::registerBuiltInFunctions() {
     builtInTableFunctions = std::make_unique<function::BuiltInTableFunctions>();
 }
 
-std::vector<TableSchema*> CatalogContent::getTableSchemas(common::TableType tableType) const {
+bool CatalogContent::containTable(const std::string& tableName, TableType tableType) const {
+    if (!tableNameToIDMap.contains(tableName)) {
+        return false;
+    }
+    auto tableID = getTableID(tableName);
+    return tableSchemas.at(tableID)->tableType == tableType;
+}
+
+std::vector<TableSchema*> CatalogContent::getTableSchemas(TableType tableType) const {
     std::vector<TableSchema*> result;
     for (auto& [id, schema] : tableSchemas) {
         if (schema->getTableType() == tableType) {

diff --git a/src/include/binder/binder.h b/src/include/binder/binder.h
@@ -77,7 +77,6 @@ class Binder {
     std::shared_ptr<Expression> bindWhereExpression(
         const parser::ParsedExpression& parsedExpression);
 
-    common::table_id_t bindRelTableID(const std::string& tableName) const;
     common::table_id_t bindNodeTableID(const std::string& tableName) const;
 
     std::shared_ptr<Expression> createVariable(
@@ -105,8 +104,6 @@ class Binder {
         catalog::TableSchema* tableSchema, const std::string& propertyName);
 
     /*** bind copy from/to ***/
-    static bool bindContainsSerial(
-        catalog::TableSchema* tableSchema, common::CopyDescription::FileType fileType);
     expression_vector bindColumnExpressions(
         catalog::TableSchema* tableSchema, common::CopyDescription::FileType fileType);
     std::unique_ptr<BoundStatement> bindCopyFromClause(const parser::Statement& statement);
@@ -243,7 +240,9 @@ class Binder {
     // multiple statement.
     static void validateReadNotFollowUpdate(const NormalizedSingleQuery& singleQuery);
 
-    static void validateTableExist(const catalog::Catalog& _catalog, std::string& tableName);
+    void validateTableExist(const std::string& tableName);
+    // TODO(Xiyang): remove this validation once we refactor DDL.
+    void validateNodeRelTableExist(const std::string& tableName);
 
     static bool validateStringParsingOptionName(std::string& parsingOptionName);
 

diff --git a/src/include/catalog/catalog_content.h b/src/include/catalog/catalog_content.h
@@ -32,6 +32,15 @@ class CatalogContent {
     inline bool containTable(const std::string& name) const {
         return tableNameToIDMap.contains(name);
     }
+    inline bool containNodeTable(const std::string& tableName) const {
+        return containTable(tableName, common::TableType::NODE);
+    }
+    inline bool containRelTable(const std::string& tableName) const {
+        return containTable(tableName, common::TableType::REL);
+    }
+    inline bool containRelGroup(const std::string& tableName) const {
+        return containTable(tableName, common::TableType::REL_GROUP);
+    }
     inline std::string getTableName(common::table_id_t tableID) const {
         assert(tableSchemas.contains(tableID));
         return getTableSchema(tableID)->tableName;
@@ -53,10 +62,6 @@ class CatalogContent {
     common::table_id_t addRelTableGroupSchema(const binder::BoundCreateTableInfo& info);
     common::table_id_t addRdfGraphSchema(const binder::BoundCreateTableInfo& info);
 
-    bool containNodeTable(const std::string& tableName) const;
-
-    bool containRelTable(const std::string& tableName) const;
-
     /**
      * Node and Rel property functions.
      */
@@ -113,6 +118,7 @@ class CatalogContent {
 
     void registerBuiltInFunctions();
 
+    bool containTable(const std::string& tableName, common::TableType tableType) const;
     std::vector<TableSchema*> getTableSchemas(common::TableType tableType) const;
     std::vector<common::table_id_t> getTableIDs(common::TableType tableType) const;
 

diff --git a/test/test_files/ddl/ddl.test b/test/test_files/ddl/ddl.test
@@ -8,12 +8,12 @@
 ---- ok
 -STATEMENT MATCH ()-[e:studyAt]->() RETURN count(*)
 ---- error
-Binder exception: Rel table studyAt does not exist.
+Binder exception: Table studyAt does not exist.
 -STATEMENT DROP TABLE workAt
 ---- ok
 -STATEMENT MATCH ()-[e:workAt]->() RETURN count(*)
 ---- error
-Binder exception: Rel table workAt does not exist.
+Binder exception: Table workAt does not exist.
 -STATEMENT DROP TABLE organisation
 ---- ok
 -STATEMENT MATCH (n:organisation) RETURN count(*)

diff --git a/test/test_files/exceptions/binder/binder_error.test b/test/test_files/exceptions/binder/binder_error.test
@@ -275,7 +275,7 @@ Binder exception: Unrecognized parsing csv option: PK.
 -LOG CopyCSVInvalidSchemaName
 -STATEMENT COPY university FROM "person_0_0.csv" (pk=",")
 ---- error
-Binder exception: Node/Rel university does not exist.
+Binder exception: Table university does not exist.
 
 -LOG CopyCSVInvalidEscapeChar
 -STATEMENT COPY person FROM "person_0_0.csv" (ESCAPE="..")
@@ -305,7 +305,7 @@ Binder exception: PropertyName: _id is an internal reserved propertyName.
 -LOG DropNotExistsTable
 -STATEMENT DROP TABLE person1;
 ---- error
-Binder exception: Node/Rel person1 does not exist.
+Binder exception: Table person1 does not exist.
 
 -LOG InvalidLimitNumberType
 -STATEMENT MATCH (a:person) RETURN a.age LIMIT "abc"
@@ -360,7 +360,7 @@ DISTINCT (SERIAL) -> SERIAL
 -LOG DropColumnFromNonExistedTable
 -STATEMENT alter table person1 drop k
 ---- error
-Binder exception: Node/Rel person1 does not exist.
+Binder exception: Table person1 does not exist.
 
 -LOG DropNonExistedColumn
 -STATEMENT alter table person drop random
@@ -385,7 +385,7 @@ Binder exception: Expression 3.2 has data type STRING but expect INT64. Implicit
 -LOG RenameNonExistedTable
 -STATEMENT alter table person1 rename to person2
 ---- error
-Binder exception: Node/Rel person1 does not exist.
+Binder exception: Table person1 does not exist.
 
 -LOG RenameTableDuplicateName
 -STATEMENT alter table person rename to organisation
@@ -395,7 +395,7 @@ Binder exception: Table: organisation already exists.
 -LOG RenamePropertyOfNonExistedTable
 -STATEMENT alter table person1 rename col1 to col2
 ---- error
-Binder exception: Node/Rel person1 does not exist.
+Binder exception: Table person1 does not exist.
 
 -LOG RenamePropertyDuplicateName
 -STATEMENT alter table person rename fName to gender