MS-176 Add create table parameter check

Former-commit-id: 28500a443d0a55339ac987f1f13417d23a2edf46

cpp/src/server/RequestTask.cpp

@@ -8,6 +8,7 @@
 #include "utils/CommonUtil.h"
 #include "utils/Log.h"
 #include "utils/TimeRecorder.h"
+#include "utils/ValidationUtil.h"
 #include "DBWrapper.h"
 #include "version.h"
 
@@ -133,19 +134,23 @@ BaseTaskPtr CreateTableTask::Create(const thrift::TableSchema& schema) {
 
 ServerError CreateTableTask::OnExecute() {
     TimeRecorder rc("CreateTableTask");
-    
+
     try {
         //step 1: check arguments
-        if(schema_.table_name.empty()) {
-            return SetError(SERVER_INVALID_TABLE_NAME, "Empty table name");
+        ServerError res = SERVER_SUCCESS;
+        res = ValidateTableName(schema_.table_name);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
-        if(schema_.dimension <= 0) {
-            return SetError(SERVER_INVALID_TABLE_DIMENSION, "Invalid table dimension: " + std::to_string(schema_.dimension));
+
+        res = ValidateTableDimension(schema_.dimension);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
 
-        engine::EngineType engine_type = EngineType(schema_.index_type);
-        if(engine_type == engine::EngineType::INVALID) {
-            return SetError(SERVER_INVALID_INDEX_TYPE, "Invalid index type: " + std::to_string(schema_.index_type));
+        res = ValidateTableIndexType(schema_.index_type);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
 
         //step 2: construct table schema
@@ -187,8 +192,10 @@ ServerError DescribeTableTask::OnExecute() {
 
     try {
         //step 1: check arguments
-        if(table_name_.empty()) {
-            return SetError(SERVER_INVALID_TABLE_NAME, "Empty table name");
+        ServerError res = SERVER_SUCCESS;
+        res = ValidateTableName(table_name_);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
 
         //step 2: get table info
@@ -230,10 +237,11 @@ ServerError HasTableTask::OnExecute() {
         TimeRecorder rc("HasTableTask");
 
         //step 1: check arguments
-        if(table_name_.empty()) {
-            return SetError(SERVER_INVALID_TABLE_NAME, "Empty table name");
+        ServerError res = SERVER_SUCCESS;
+        res = ValidateTableName(table_name_);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
-
         //step 2: check table existence
         engine::Status stat = DBWrapper::DB()->HasTable(table_name_, has_table_);
         if(!stat.ok()) {
@@ -264,8 +272,10 @@ ServerError DeleteTableTask::OnExecute() {
         TimeRecorder rc("DeleteTableTask");
 
         //step 1: check arguments
-        if (table_name_.empty()) {
-            return SetError(SERVER_INVALID_TABLE_NAME, "Empty table name");
+        ServerError res = SERVER_SUCCESS;
+        res = ValidateTableName(table_name_);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
 
         //step 2: check table existence
@@ -346,8 +356,10 @@ ServerError AddVectorTask::OnExecute() {
         TimeRecorder rc("AddVectorTask");
 
         //step 1: check arguments
-        if (table_name_.empty()) {
-            return SetError(SERVER_INVALID_TABLE_NAME, "Empty table name");
+        ServerError res = SERVER_SUCCESS;
+        res = ValidateTableName(table_name_);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
 
         if(record_array_.empty()) {
@@ -435,8 +447,10 @@ ServerError SearchVectorTask::OnExecute() {
         TimeRecorder rc("SearchVectorTask");
 
         //step 1: check arguments
-        if (table_name_.empty()) {
-            return SetError(SERVER_INVALID_TABLE_NAME, "Empty table name");
+        ServerError res = SERVER_SUCCESS;
+        res = ValidateTableName(table_name_);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
 
         if(top_k_ <= 0) {
@@ -548,8 +562,10 @@ ServerError GetTableRowCountTask::OnExecute() {
         TimeRecorder rc("GetTableRowCountTask");
 
         //step 1: check arguments
-        if (table_name_.empty()) {
-            return SetError(SERVER_INVALID_TABLE_NAME, "Empty table name");
+        ServerError res = SERVER_SUCCESS;
+        res = ValidateTableName(table_name_);
+        if(res != SERVER_SUCCESS) {
+            return res;
         }
 
         //step 2: get row count

cpp/src/utils/ValidationUtil.cpp

+#include <src/db/ExecutionEngine.h>
+#include "ValidationUtil.h"
+#include "Log.h"
+
+
+namespace zilliz {
+namespace milvus {
+namespace server {
+
+constexpr size_t table_name_size_limit = 16384;
+constexpr int64_t table_dimension_limit = 16384;
+
+ServerError
+ValidateTableName(const std::string &table_name) {
+
+    // Table name shouldn't be empty.
+    if (table_name.empty()) {
+        SERVER_LOG_ERROR << "Empty table name";
+        return SERVER_INVALID_TABLE_NAME;
+    }
+
+    // Table name size shouldn't exceed 16384.
+    if (table_name.size() > table_name_size_limit) {
+        SERVER_LOG_ERROR << "Table name size exceed the limitation";
+        return SERVER_INVALID_TABLE_NAME;
+    }
+
+    // Table name first character should be underscore or character.
+    char first_char = table_name[0];
+    if (first_char != '_' && std::isalpha(first_char) == 0) {
+        SERVER_LOG_ERROR << "Table name first character isn't underscore or character: " << first_char;
+        return SERVER_INVALID_TABLE_NAME;
+    }
+
+    int64_t table_name_size = table_name.size();
+    for (int64_t i = 1; i < table_name_size; ++i) {
+        char name_char = table_name[i];
+        if (name_char != '_' && std::isalnum(name_char) == 0) {
+            SERVER_LOG_ERROR << "Table name character isn't underscore or alphanumber: " << name_char;
+            return SERVER_INVALID_TABLE_NAME;
+        }
+    }
+
+    return SERVER_SUCCESS;
+}
+
+ServerError
+ValidateTableDimension(int64_t dimension) {
+    if (dimension <= 0 || dimension > table_dimension_limit) {
+        SERVER_LOG_ERROR << "Table dimension excceed the limitation: " << table_dimension_limit;
+        return SERVER_INVALID_VECTOR_DIMENSION;
+    } else {
+        return SERVER_SUCCESS;
+    }
+}
+
+ServerError
+ValidateTableIndexType(int32_t index_type) {
+    auto engine_type = engine::EngineType(index_type);
+    switch (engine_type) {
+        case engine::EngineType::FAISS_IDMAP:
+        case engine::EngineType::FAISS_IVFFLAT: {
+            SERVER_LOG_DEBUG << "Index type: " << index_type;
+            return SERVER_SUCCESS;
+        }
+        default: {
+            return SERVER_INVALID_INDEX_TYPE;
+        }
+    }
+}
+
+}
+}
+}
\ No newline at end of file

cpp/src/utils/ValidationUtil.h

+#pragma once
+
+#include "Error.h"
+
+namespace zilliz {
+namespace milvus {
+namespace server {
+
+ServerError
+ValidateTableName(const std::string& table_name);
+
+ServerError
+ValidateTableDimension(int64_t dimension);
+
+ServerError
+ValidateTableIndexType(int32_t index_type);
+
+}
+}
+}
\ No newline at end of file

cpp/unittest/CMakeLists.txt

@@ -12,7 +12,6 @@ aux_source_directory(${MILVUS_ENGINE_SRC}/config config_files)
 
 set(unittest_srcs
         ${CMAKE_CURRENT_SOURCE_DIR}/main.cpp)
-        #${EASYLOGGINGPP_INCLUDE_DIR}/easylogging++.cc)
 
 set(require_files
         ${MILVUS_ENGINE_SRC}/server/ServerConfig.cpp
@@ -44,4 +43,5 @@ add_subdirectory(index_wrapper)
 #add_subdirectory(faiss_wrapper)
 #add_subdirectory(license)
 add_subdirectory(metrics)
-add_subdirectory(storage)
+add_subdirectory(storage)
\ No newline at end of file
+add_subdirectory(utils)
\ No newline at end of file

cpp/unittest/db/db_tests.cpp

@@ -3,17 +3,20 @@
 // Unauthorized copying of this file, via any medium is strictly prohibited.
 // Proprietary and confidential.
 ////////////////////////////////////////////////////////////////////////////////
-#include <gtest/gtest.h>
-#include <thread>
-#include <easylogging++.h>
-#include <boost/filesystem.hpp>
-
 #include "utils.h"
 #include "db/DB.h"
 #include "db/DBImpl.h"
 #include "db/MetaConsts.h"
 #include "db/Factories.h"
 
+#include <gtest/gtest.h>
+#include <easylogging++.h>
+
+#include <boost/filesystem.hpp>
+
+#include <thread>
+#include <random>
+
 using namespace zilliz::milvus;
 
 namespace {

cpp/unittest/db/mysql_db_test.cpp

@@ -3,17 +3,19 @@
 // Unauthorized copying of this file, via any medium is strictly prohibited.
 // Proprietary and confidential.
 ////////////////////////////////////////////////////////////////////////////////
-#include <gtest/gtest.h>
-#include <thread>
-#include <easylogging++.h>
-#include <boost/filesystem.hpp>
-
 #include "utils.h"
 #include "db/DB.h"
 #include "db/DBImpl.h"
 #include "db/MetaConsts.h"
 #include "db/Factories.h"
 
+#include <gtest/gtest.h>
+#include <easylogging++.h>
+#include <boost/filesystem.hpp>
+
+#include <thread>
+#include <random>
+
 using namespace zilliz::milvus;
 
 namespace {

cpp/unittest/db/search_test.cpp

@@ -3,10 +3,11 @@
 // Unauthorized copying of this file, via any medium is strictly prohibited.
 // Proprietary and confidential.
 ////////////////////////////////////////////////////////////////////////////////
-#include <gtest/gtest.h>
-
 #include "db/scheduler/task/SearchTask.h"
 
+#include <gtest/gtest.h>
+
+#include <cmath>
 #include <vector>
 
 using namespace zilliz::milvus;

cpp/unittest/faiss_wrapper/wrapper_test.cpp

@@ -4,12 +4,15 @@
 // Proprietary and confidential.
 ////////////////////////////////////////////////////////////////////////////////
 
-#include <gtest/gtest.h>
+
 
 #include "wrapper/Operand.h"
 #include "wrapper/Index.h"
 #include "wrapper/IndexBuilder.h"
 
+#include <gtest/gtest.h>
+#include <random>
+
 using namespace zilliz::milvus::engine;
 
 

cpp/unittest/utils/CMakeLists.txt

+#-------------------------------------------------------------------------------
+# Copyright 上海赜睿信息科技有限公司(Zilliz) - All Rights Reserved
+# Unauthorized copying of this file, via any medium is strictly prohibited.
+# Proprietary and confidential.
+#-------------------------------------------------------------------------------
+
+# Make sure that your call to link_directories takes place before your call to the relevant add_executable.
+include_directories("${CUDA_TOOLKIT_ROOT_DIR}/include")
+link_directories("${CUDA_TOOLKIT_ROOT_DIR}/lib64")
+
+set(validation_util_src
+        ${MILVUS_ENGINE_SRC}/utils/ValidationUtil.cpp
+        ${MILVUS_ENGINE_SRC}/utils/ValidationUtil.h)
+
+set(validation_util_test_src
+        ${unittest_srcs}
+        ${validation_util_src}
+        ${require_files}
+        ValidationUtilTest.cpp
+        )
+
+add_executable(valication_util_test
+        ${validation_util_test_src}
+        ${config_files})
+
+target_link_libraries(valication_util_test
+        ${unittest_libs}
+        boost_filesystem)
+
+install(TARGETS valication_util_test DESTINATION bin)
\ No newline at end of file

cpp/unittest/utils/ValidationUtilTest.cpp

+////////////////////////////////////////////////////////////////////////////////
+// Copyright 上海赜睿信息科技有限公司(Zilliz) - All Rights Reserved
+// Unauthorized copying of this file, via any medium is strictly prohibited.
+// Proprietary and confidential.
+////////////////////////////////////////////////////////////////////////////////
+#include <gtest/gtest.h>
+
+#include "utils/ValidationUtil.h"
+#include "utils/Error.h"
+
+#include <string>
+
+using namespace zilliz::milvus::server;
+
+TEST(ValidationUtilTest, TableNameTest) {
+    std::string table_name = "Normal123_";
+    ServerError res = ValidateTableName(table_name);
+    ASSERT_EQ(res, SERVER_SUCCESS);
+
+    table_name = "12sds";
+    res = ValidateTableName(table_name);
+    ASSERT_EQ(res, SERVER_INVALID_TABLE_NAME);
+
+    table_name = "";
+    res = ValidateTableName(table_name);
+    ASSERT_EQ(res, SERVER_INVALID_TABLE_NAME);
+
+    table_name = "_asdasd";
+    res = ValidateTableName(table_name);
+    ASSERT_EQ(res, SERVER_SUCCESS);
+
+    table_name = "!@#!@";
+    res = ValidateTableName(table_name);
+    ASSERT_EQ(res, SERVER_INVALID_TABLE_NAME);
+
+    table_name = "中文";
+    res = ValidateTableName(table_name);
+    ASSERT_EQ(res, SERVER_INVALID_TABLE_NAME);
+
+
+    table_name = std::string('a', 32768);
+    res = ValidateTableName(table_name);
+    ASSERT_EQ(res, SERVER_INVALID_TABLE_NAME);
+}
+
+
+TEST(ValidationUtilTest, TableDimensionTest) {
+    ASSERT_EQ(ValidateTableDimension(-1), SERVER_INVALID_VECTOR_DIMENSION);
+    ASSERT_EQ(ValidateTableDimension(0), SERVER_INVALID_VECTOR_DIMENSION);
+    ASSERT_EQ(ValidateTableDimension(16385), SERVER_INVALID_VECTOR_DIMENSION);
+    ASSERT_EQ(ValidateTableDimension(16384), SERVER_SUCCESS);
+    ASSERT_EQ(ValidateTableDimension(1), SERVER_SUCCESS);
+}
+
+TEST(ValidationUtilTest, TableIndexTypeTest) {
+    ASSERT_EQ(ValidateTableIndexType(0), SERVER_INVALID_INDEX_TYPE);
+    ASSERT_EQ(ValidateTableIndexType(1), SERVER_SUCCESS);
+    ASSERT_EQ(ValidateTableIndexType(2), SERVER_SUCCESS);
+    ASSERT_EQ(ValidateTableIndexType(3), SERVER_INVALID_INDEX_TYPE);
+    ASSERT_EQ(ValidateTableIndexType(4), SERVER_INVALID_INDEX_TYPE);
+}