draft

kuzudb · Feb 27, 2024 · 14dc603 · 14dc603
1 parent 29f2534
commit 14dc603
Show file tree

Hide file tree

Showing 45 changed files with 3,515 additions and 3,171 deletions.
diff --git a/examples/cpp/main.cpp b/examples/cpp/main.cpp
@@ -4,17 +4,19 @@
 using namespace kuzu::main;
 
 int main() {
-    auto database = std::make_unique<Database>("" /* fill db path */);
+    auto database = std::make_unique<Database>("test" /* fill db path */);
     auto connection = std::make_unique<Connection>(database.get());
 
     // Create schema.
-    connection->query("CREATE NODE TABLE Person(name STRING, age INT64, PRIMARY KEY(name));");
-    // Create nodes.
-    connection->query("CREATE (:Person {name: 'Alice', age: 25});");
-    connection->query("CREATE (:Person {name: 'Bob', age: 30});");
+//    connection->query("CREATE NODE TABLE Person(name STRING, age INT64, PRIMARY KEY(name));");
+//    // Create nodes.
+//    connection->query("CREATE (:Person {name: 'Alice', age: 25});");
+//    connection->query("CREATE (:Person {name: 'Bob', age: 30});");
 
     // Execute a simple query.
-    auto result = connection->query("MATCH (a:Person) RETURN a.name AS NAME, a.age AS AGE;");
+//    auto result = connection->query("CREATE NODE TABLE Person(name STRING,age INT64, PRIMARY KEY(name)); COPY Person FROM \"db_name3/Person.csv\" (escape = '\\\\' , delim = ',' , quote = '\\\"', header=false);");
+    auto result = connection->query("import database 'db_name3';");
+//    auto result = connection->query("match (n) return n;");
     // Print query result.
     std::cout << result->toString();
 }
diff --git a/scripts/antlr4/Cypher.g4.copy b/scripts/antlr4/Cypher.g4.copy
@@ -30,7 +30,8 @@ oC_Statement
         | kU_CommentOn
         | kU_Transaction
         | kU_Extension
-        | kU_ExportDatabase;
+        | kU_ExportDatabase
+        | kU_ImportDatabase;
 
 kU_CopyFrom
     : COPY SP oC_SchemaName ( ( SP? '(' SP? kU_ColumnNames SP? ')' SP? ) | SP ) FROM SP (kU_FilePaths | oC_Variable) ( SP? '(' SP? kU_ParsingOptions SP? ')' )? ;
@@ -47,6 +48,10 @@ kU_CopyTO
 kU_ExportDatabase
     : EXPORT SP DATABASE SP StringLiteral ( SP? '(' SP? kU_ParsingOptions SP? ')' )? ;
 
+kU_ImportDatabase
+    : IMPORT SP DATABASE SP StringLiteral;
+
+
 kU_StandaloneCall
     : CALL SP oC_SymbolicName SP? '=' SP? oC_Literal ;
 
@@ -89,6 +94,8 @@ COLUMN : ( 'C' | 'c' ) ( 'O' | 'o' ) ( 'L' | 'l' ) ( 'U' | 'u' ) ( 'M' | 'm' ) (
 
 EXPORT: ( 'E' | 'e') ( 'X' | 'x') ( 'P' | 'p') ( 'O' | 'o') ( 'R' | 'r') ( 'T' | 't');
 
+IMPORT: ( 'I' | 'i') ( 'M' | 'm') ( 'P' | 'p') ( 'O' | 'o') ( 'R' | 'r') ( 'T' | 't');
+
 DATABASE: ( 'D' | 'd') ( 'A' | 'a') ( 'T' | 't') ( 'A' | 'a') ( 'B' | 'b') ( 'A' | 'a') ( 'S' | 's')( 'E' | 'e');
 
 kU_DDL
@@ -754,6 +761,7 @@ kU_NonReservedKeywords
         | BEGIN
         | END
         | IN
+        | IMPORT
         | EXPORT
         | DATABASE
         ;

diff --git a/src/antlr4/Cypher.g4 b/src/antlr4/Cypher.g4
@@ -30,7 +30,8 @@ oC_Statement
         | kU_CommentOn
         | kU_Transaction
         | kU_Extension
-        | kU_ExportDatabase;
+        | kU_ExportDatabase
+        | kU_ImportDatabase;
 
 kU_CopyFrom
     : COPY SP oC_SchemaName ( ( SP? '(' SP? kU_ColumnNames SP? ')' SP? ) | SP ) FROM SP (kU_FilePaths | oC_Variable) ( SP? '(' SP? kU_ParsingOptions SP? ')' )? ;
@@ -47,6 +48,10 @@ kU_CopyTO
 kU_ExportDatabase
     : EXPORT SP DATABASE SP StringLiteral ( SP? '(' SP? kU_ParsingOptions SP? ')' )? ;
 
+kU_ImportDatabase
+    : IMPORT SP DATABASE SP StringLiteral;
+
+
 kU_StandaloneCall
     : CALL SP oC_SymbolicName SP? '=' SP? oC_Literal ;
 
@@ -89,6 +94,8 @@ COLUMN : ( 'C' | 'c' ) ( 'O' | 'o' ) ( 'L' | 'l' ) ( 'U' | 'u' ) ( 'M' | 'm' ) (
 
 EXPORT: ( 'E' | 'e') ( 'X' | 'x') ( 'P' | 'p') ( 'O' | 'o') ( 'R' | 'r') ( 'T' | 't');
 
+IMPORT: ( 'I' | 'i') ( 'M' | 'm') ( 'P' | 'p') ( 'O' | 'o') ( 'R' | 'r') ( 'T' | 't');
+
 DATABASE: ( 'D' | 'd') ( 'A' | 'a') ( 'T' | 't') ( 'A' | 'a') ( 'B' | 'b') ( 'A' | 'a') ( 'S' | 's')( 'E' | 'e');
 
 kU_DDL
@@ -754,6 +761,7 @@ kU_NonReservedKeywords
         | BEGIN
         | END
         | IN
+        | IMPORT
         | EXPORT
         | DATABASE
         ;

diff --git a/src/binder/bind/CMakeLists.txt b/src/binder/bind/CMakeLists.txt
@@ -18,7 +18,8 @@ add_library(
         bind_transaction.cpp
         bind_updating_clause.cpp
         bind_extension.cpp
-        bind_export_database.cpp)
+        bind_export_database.cpp
+        bind_import_database.cpp)
 
 set(ALL_OBJECT_FILES
         ${ALL_OBJECT_FILES} $<TARGET_OBJECTS:kuzu_binder_bind>

diff --git a/src/binder/bind/bind_import_database.cpp b/src/binder/bind/bind_import_database.cpp
@@ -0,0 +1,39 @@
+#include <fstream>
+#include <iostream>
+
+#include "binder/copy/bound_import_database.h"
+#include "binder/query/bound_regular_query.h"
+#include "common/exception/binder.h"
+#include "parser/parser.h"
+#include "parser/port_db.h"
+
+using namespace kuzu::binder;
+using namespace kuzu::common;
+using namespace kuzu::parser;
+using namespace kuzu::catalog;
+using namespace kuzu::transaction;
+using namespace kuzu::storage;
+
+namespace kuzu {
+namespace binder {
+
+std::unique_ptr<BoundStatement> Binder::bindImportDatabaseClause(const Statement& statement) {
+    auto& importDatabaseStatement = ku_dynamic_cast<const Statement&, const ImportDB&>(statement);
+    auto boundFilePath = importDatabaseStatement.getFilePath();
+    std::vector<std::string> files = {"schema.cypher", "copy.cypher", "macro.cypher"};
+    std::string finalQueryStatements;
+    for (auto& file : files) {
+        auto filePath = vfs->joinPath(boundFilePath, file);
+        std::ifstream cypherFile(filePath);
+        if (!cypherFile.is_open()) {
+            throw Exception(stringFormat("Error opening file: {}, errno: {}.", filePath, errno));
+        }
+        std::string line;
+        while (getline(cypherFile, line)) {
+            finalQueryStatements += line;
+        }
+    }
+    return std::make_unique<BoundImportDatabase>(boundFilePath, finalQueryStatements);
+}
+} // namespace binder
+} // namespace kuzu
diff --git a/src/binder/binder.cpp b/src/binder/binder.cpp
@@ -58,6 +58,9 @@ std::unique_ptr<BoundStatement> Binder::bind(const Statement& statement) {
     case StatementType::EXPORT_DATABASE: {
         boundStatement = bindExportDatabaseClause(statement);
     } break;
+    case StatementType::IMPORT_DATABASE: {
+        boundStatement = bindImportDatabaseClause(statement);
+    } break;
     default: {
         KU_UNREACHABLE;
     }

diff --git a/src/binder/bound_statement_visitor.cpp b/src/binder/bound_statement_visitor.cpp
@@ -50,6 +50,9 @@ void BoundStatementVisitor::visit(const BoundStatement& statement) {
     case StatementType::EXPORT_DATABASE: {
         visitExportDatabase(statement);
     } break;
+    case StatementType::IMPORT_DATABASE: {
+        visitImportDatabase(statement);
+    } break;
     default:
         KU_UNREACHABLE;
     }

diff --git a/src/include/binder/binder.h b/src/include/binder/binder.h
@@ -146,6 +146,7 @@ class Binder {
     std::unique_ptr<BoundStatement> bindCopyToClause(const parser::Statement& statement);
 
     std::unique_ptr<BoundStatement> bindExportDatabaseClause(const parser::Statement& statement);
+    std::unique_ptr<BoundStatement> bindImportDatabaseClause(const parser::Statement& statement);
 
     /*** bind file scan ***/
     std::unordered_map<std::string, common::Value> bindParsingOptions(

diff --git a/src/include/binder/bound_statement_visitor.h b/src/include/binder/bound_statement_visitor.h
@@ -23,6 +23,7 @@ class BoundStatementVisitor {
     virtual void visitCopyFrom(const BoundStatement&) {}
     virtual void visitCopyTo(const BoundStatement&) {}
     virtual void visitExportDatabase(const BoundStatement&) {}
+    virtual void visitImportDatabase(const BoundStatement&) {}
     virtual void visitStandaloneCall(const BoundStatement&) {}
     virtual void visitCommentOn(const BoundStatement&) {}
     virtual void visitExplain(const BoundStatement&);

diff --git a/src/include/binder/copy/bound_import_database.h b/src/include/binder/copy/bound_import_database.h
@@ -0,0 +1,26 @@
+#pragma once
+#include "binder/binder.h"
+#include "binder/bound_statement.h"
+#include "common/copier_config/csv_reader_config.h"
+#include "common/copier_config/reader_config.h"
+
+namespace kuzu {
+namespace binder {
+
+class BoundImportDatabase : public BoundStatement {
+public:
+    BoundImportDatabase(std::string filePath, std::string query)
+        : BoundStatement{common::StatementType::IMPORT_DATABASE,
+              BoundStatementResult::createEmptyResult()},
+          filePath{std::move(filePath)}, query{query} {}
+
+    inline std::string getFilePath() const { return filePath; }
+    inline std::string getQuery() const { return query; }
+
+private:
+    std::string filePath;
+    std::string query;
+};
+
+} // namespace binder
+} // namespace kuzu
diff --git a/src/include/common/enums/statement_type.h b/src/include/common/enums/statement_type.h
@@ -19,6 +19,7 @@ enum class StatementType : uint8_t {
     TRANSACTION = 30,
     EXTENSION = 31,
     EXPORT_DATABASE = 32,
+    IMPORT_DATABASE = 33,
 };
 
 struct StatementTypeUtils {

diff --git a/src/include/main/client_context.h b/src/include/main/client_context.h
@@ -17,6 +17,10 @@
 
 namespace kuzu {
 
+namespace parser {
+class Statement;
+}
+
 namespace binder {
 class Binder;
 class ExpressionBinder;
@@ -113,6 +117,8 @@ class ClientContext {
 
     std::unique_ptr<QueryResult> query(std::string_view queryStatement);
 
+    void runQuery(std::string query);
+
 private:
     inline void resetActiveQuery() { activeQuery.reset(); }
 

diff --git a/src/include/parser/parsed_statement_visitor.h b/src/include/parser/parsed_statement_visitor.h
@@ -49,6 +49,7 @@ class StatementVisitor {
     virtual void visitTransaction(const Statement& /*statement*/) {}
     virtual void visitExtension(const Statement& /*statement*/) {}
     virtual void visitExportDatabase(const Statement& /*statement*/) {}
+    virtual void visitImportDatabase(const Statement& /*statement*/) {}
     // LCOV_EXCL_STOP
 };
 

diff --git a/src/include/parser/port_db.h b/src/include/parser/port_db.h
@@ -20,5 +20,16 @@ class ExportDB : public Statement {
     std::string filePath;
 };
 
+class ImportDB : public Statement {
+public:
+    explicit ImportDB(std::string filePath)
+        : Statement{common::StatementType::IMPORT_DATABASE}, filePath{std::move(filePath)} {}
+
+    inline std::string getFilePath() const { return filePath; }
+
+private:
+    std::string filePath;
+};
+
 } // namespace parser
 } // namespace kuzu
diff --git a/src/include/parser/transformer.h b/src/include/parser/transformer.h
@@ -53,6 +53,7 @@ class Transformer {
     parsing_option_t transformParsingOptions(CypherParser::KU_ParsingOptionsContext& ctx);
 
     std::unique_ptr<Statement> transformExportDatabase(CypherParser::KU_ExportDatabaseContext& ctx);
+    std::unique_ptr<Statement> transformImportDatabase(CypherParser::KU_ImportDatabaseContext& ctx);
 
     // Transform query statement.
     std::unique_ptr<Statement> transformQuery(CypherParser::OC_QueryContext& ctx);

diff --git a/src/include/planner/operator/logical_operator.h b/src/include/planner/operator/logical_operator.h
@@ -52,7 +52,8 @@ enum class LogicalOperatorType : uint8_t {
     UNION_ALL,
     UNWIND,
     EXTENSION,
-    EXPORT_DATABASE
+    EXPORT_DATABASE,
+    IMPORT_DATABASE,
 };
 
 class LogicalOperatorUtils {

diff --git a/src/include/planner/operator/persistent/logical_import_db.h b/src/include/planner/operator/persistent/logical_import_db.h
@@ -0,0 +1,29 @@
+#pragma once
+
+#include "planner/operator/logical_operator.h"
+
+namespace kuzu {
+namespace planner {
+
+class LogicalImportDatabase : public LogicalOperator {
+public:
+    explicit LogicalImportDatabase(std::string query)
+        : LogicalOperator{LogicalOperatorType::IMPORT_DATABASE}, query{query} {}
+
+    inline std::string getExpressionsForPrinting() const override { return std::string{}; }
+
+    void computeFactorizedSchema() override;
+    void computeFlatSchema() override;
+
+    inline std::string getQuery() const { return query; }
+
+    inline std::unique_ptr<LogicalOperator> copy() override {
+        return make_unique<LogicalImportDatabase>(query);
+    }
+
+private:
+    std::string query;
+};
+
+} // namespace planner
+} // namespace kuzu
diff --git a/src/include/planner/planner.h b/src/include/planner/planner.h
@@ -65,6 +65,7 @@ class Planner {
 
     // Plan export database
     std::unique_ptr<LogicalPlan> planExportDatabase(const binder::BoundStatement& statement);
+    std::unique_ptr<LogicalPlan> planImportDatabase(const binder::BoundStatement& statement);
 
     // Plan query.
     std::vector<std::unique_ptr<LogicalPlan>> planQuery(

diff --git a/src/include/processor/operator/persistent/import_db.h b/src/include/processor/operator/persistent/import_db.h
@@ -0,0 +1,28 @@
+#pragma once
+
+#include "common/copier_config/reader_config.h"
+#include "processor/operator/physical_operator.h"
+
+namespace kuzu {
+namespace processor {
+
+class ImportDB : public PhysicalOperator {
+public:
+    ImportDB(std::string query, uint32_t id, const std::string& paramsString)
+        : PhysicalOperator{PhysicalOperatorType::IMPORT_DATABASE, id, paramsString}, query{query} {}
+
+    bool canParallel() const override { return false; }
+
+    bool isSource() const override { return true; }
+
+    bool getNextTuplesInternal(ExecutionContext* context) override;
+
+    inline std::unique_ptr<PhysicalOperator> clone() override {
+        return std::make_unique<ImportDB>(query, id, paramsString);
+    }
+
+private:
+    std::string query;
+};
+} // namespace processor
+} // namespace kuzu
diff --git a/src/include/processor/operator/physical_operator.h b/src/include/processor/operator/physical_operator.h
@@ -33,6 +33,7 @@ enum class PhysicalOperatorType : uint8_t {
     FLATTEN,
     HASH_JOIN_BUILD,
     HASH_JOIN_PROBE,
+    IMPORT_DATABASE,
     INDEX_LOOKUP,
     INDEX_SCAN,
     INSERT,

diff --git a/src/include/processor/plan_mapper.h b/src/include/processor/plan_mapper.h
@@ -105,6 +105,7 @@ class PlanMapper {
     std::unique_ptr<PhysicalOperator> mapTransaction(planner::LogicalOperator* logicalOperator);
     std::unique_ptr<PhysicalOperator> mapExtension(planner::LogicalOperator* logicalOperator);
     std::unique_ptr<PhysicalOperator> mapExportDatabase(planner::LogicalOperator* logicalOperator);
+    std::unique_ptr<PhysicalOperator> mapImportDatabase(planner::LogicalOperator* logicalOperator);
 
     std::unique_ptr<PhysicalOperator> createCopyRel(
         std::shared_ptr<PartitionerSharedState> partitionerSharedState,