Add storage version info

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.11)
 
-project(Kuzu VERSION 0.0.3.1 LANGUAGES CXX)
+project(Kuzu VERSION 0.0.3.3 LANGUAGES CXX)
 
 find_package(Threads REQUIRED)
 
@@ -78,6 +78,7 @@ function(add_kuzu_test TEST_NAME)
 endfunction()
 
 add_definitions(-DKUZU_ROOT_DIRECTORY="${PROJECT_SOURCE_DIR}")
+add_definitions(-DKUZU_STORAGE_VERSION="${CMAKE_PROJECT_VERSION}")
 
 set(ARROW_INSTALL ${CMAKE_CURRENT_SOURCE_DIR}/external/build/arrow/install)
 find_library(ARROW_DEPS_PATH arrow_bundled_dependencies HINTS ${ARROW_INSTALL}/lib ${ARROW_INSTALL}/lib64)

src/catalog/catalog.cpp

@@ -296,6 +296,9 @@ void CatalogContent::saveToFile(const std::string& directory, DBFileType dbFileT
     auto catalogPath = StorageUtils::getCatalogFilePath(directory, dbFileType);
     auto fileInfo = FileUtils::openFile(catalogPath, O_WRONLY | O_CREAT);
     uint64_t offset = 0;
+    writeMagicBytes(fileInfo.get(), offset);
+    offset = SerDeser::serializeValue<uint64_t>(
+        StorageVersionInfo::getStorageVersion(), fileInfo.get(), offset);
     offset = SerDeser::serializeValue<uint64_t>(nodeTableSchemas.size(), fileInfo.get(), offset);
     offset = SerDeser::serializeValue<uint64_t>(relTableSchemas.size(), fileInfo.get(), offset);
     for (auto& nodeTableSchema : nodeTableSchemas) {
@@ -316,6 +319,10 @@ void CatalogContent::readFromFile(const std::string& directory, DBFileType dbFil
     logger->debug("Reading from {}.", catalogPath);
     auto fileInfo = FileUtils::openFile(catalogPath, O_RDONLY);
     uint64_t offset = 0;
+    validateMagicBytes(fileInfo.get(), offset);
+    storage::storage_version_t savedStorageVersion;
+    offset = SerDeser::deserializeValue<uint64_t>(savedStorageVersion, fileInfo.get(), offset);
+    validateStorageVersion(savedStorageVersion);
     uint64_t numNodeTables, numRelTables;
     offset = SerDeser::deserializeValue<uint64_t>(numNodeTables, fileInfo.get(), offset);
     offset = SerDeser::deserializeValue<uint64_t>(numRelTables, fileInfo.get(), offset);
@@ -342,6 +349,36 @@ void CatalogContent::readFromFile(const std::string& directory, DBFileType dbFil
     SerDeser::deserializeValue<table_id_t>(nextTableID, fileInfo.get(), offset);
 }
 
+void CatalogContent::validateStorageVersion(storage_version_t savedStorageVersion) const {
+    auto storageVersion = StorageVersionInfo::getStorageVersion();
+    if (savedStorageVersion != storageVersion) {
+        throw common::RuntimeException(StringUtils::string_format(
+            "Trying to read a database file with a different version. "
+            "Database file version: {}, Current build storage version: {}",
+            savedStorageVersion, storageVersion));
+    }
+}
+
+void CatalogContent::validateMagicBytes(FileInfo* fileInfo, offset_t& offset) const {
+    auto numMagicBytes = strlen(StorageVersionInfo::MAGIC_BYTES);
+    uint8_t magicBytes[4];
+    for (auto i = 0u; i < numMagicBytes; i++) {
+        offset = SerDeser::deserializeValue<uint8_t>(magicBytes[i], fileInfo, offset);
+    }
+    if (memcmp(magicBytes, StorageVersionInfo::MAGIC_BYTES, numMagicBytes) != 0) {
+        throw common::RuntimeException(
+            "This is not a valid Kuzu database directory for the current version of Kuzu.");
+    }
+}
+
+void CatalogContent::writeMagicBytes(FileInfo* fileInfo, offset_t& offset) const {
+    auto numMagicBytes = strlen(StorageVersionInfo::MAGIC_BYTES);
+    for (auto i = 0u; i < numMagicBytes; i++) {
+        offset =
+            SerDeser::serializeValue<uint8_t>(StorageVersionInfo::MAGIC_BYTES[i], fileInfo, offset);
+    }
+}
+
 Catalog::Catalog() : wal{nullptr} {
     catalogContentForReadOnlyTrx = std::make_unique<CatalogContent>();
     builtInVectorOperations = std::make_unique<function::BuiltInVectorOperations>();

src/include/catalog/catalog.h

@@ -11,6 +11,7 @@
 #include "common/utils.h"
 #include "function/aggregate/built_in_aggregate_functions.h"
 #include "function/built_in_vector_operations.h"
+#include "storage/storage_info.h"
 #include "storage/wal/wal.h"
 
 namespace spdlog {
@@ -133,6 +134,12 @@ class CatalogContent {
 private:
     inline common::table_id_t assignNextTableID() { return nextTableID++; }
 
+    void validateStorageVersion(storage::storage_version_t savedStorageVersion) const;
+
+    void validateMagicBytes(common::FileInfo* fileInfo, common::offset_t& offset) const;
+
+    void writeMagicBytes(common::FileInfo* fileInfo, common::offset_t& offset) const;
+
 private:
     std::shared_ptr<spdlog::logger> logger;
     std::unordered_map<common::table_id_t, std::unique_ptr<NodeTableSchema>> nodeTableSchemas;

src/include/storage/storage_info.h

@@ -0,0 +1,31 @@
+#pragma once
+
+#include "exception"
+#include <string>
+
+#include "common/utils.h"
+
+namespace kuzu {
+namespace storage {
+
+using storage_version_t = uint64_t;
+
+struct StorageVersionInfo {
+    static std::unordered_map<std::string, storage_version_t> getStorageVersionInfo() {
+        return {{"0.0.3.2", 2}, {"0.0.3.1", 1}};
+    }
+
+    static storage_version_t getStorageVersion() {
+        auto storageVersionInfo = getStorageVersionInfo();
+        if (!storageVersionInfo.contains(KUZU_STORAGE_VERSION)) {
+            throw common::RuntimeException(common::StringUtils::string_format(
+                "Invalid storage version name: {}", KUZU_STORAGE_VERSION));
+        }
+        return storageVersionInfo.at(KUZU_STORAGE_VERSION);
+    }
+
+    static constexpr const char* MAGIC_BYTES = "KUZU";
+};
+
+} // namespace storage
+} // namespace kuzu