From fb63e1fe5c2683d004bd4f204ef8baffd60b7749 Mon Sep 17 00:00:00 2001
From: "Zezheng.Li" <Zezheng.Li@linux.alibaba.com>
Date: Mon, 27 Nov 2023 16:03:21 +0800
Subject: [PATCH] [struct_pack] improve performance when deserialize
 vector<pod>/string

---
 include/ylt/struct_pack/unpacker.hpp | 10 ++++----
 include/ylt/struct_pack/util.h       | 34 ++++++++++++++++++++++------
 2 files changed, 32 insertions(+), 12 deletions(-)

diff --git a/include/ylt/struct_pack/unpacker.hpp b/include/ylt/struct_pack/unpacker.hpp
index 112655af5..786c655c8 100644
--- a/include/ylt/struct_pack/unpacker.hpp
+++ b/include/ylt/struct_pack/unpacker.hpp
@@ -995,15 +995,15 @@ class unpacker {
               }
               else if constexpr (is_little_endian_copyable<sizeof(
                                      value_type)>) {
-                item.resize(size64);
-                if SP_UNLIKELY (!read_bytes_array(
-                                    reader_, (char *)item.data(),
-                                    size64 * sizeof(value_type))) {
+                resize(item, size64);
+                // item.resize(size64);
+                if SP_UNLIKELY (!read_bytes_array(reader_, (char *)item.data(),
+                                                  mem_sz)) {
                   return struct_pack::errc::no_buffer_space;
                 }
               }
               else {
-                item.resize(size64);
+                resize(item, size64);
                 for (auto &i : item) {
                   code = deserialize_one<size_type, version, NotSkip>(i);
                   if SP_UNLIKELY (code != struct_pack::errc{}) {
diff --git a/include/ylt/struct_pack/util.h b/include/ylt/struct_pack/util.h
index 988899928..87d86321e 100644
--- a/include/ylt/struct_pack/util.h
+++ b/include/ylt/struct_pack/util.h
@@ -21,8 +21,15 @@
 #include <vector>
 
 #include "marco.h"
+
 namespace struct_pack::detail {
 
+#if __cpp_concepts >= 201907L
+constexpr bool is_string_reserve_shrink = requires { std::string{}.reserve(); };
+#else
+constexpr bool is_string_reserve_shrink = true;
+#endif
+
 template <typename T>
 using remove_cvref_t = std::remove_cv_t<std::remove_reference_t<T>>;
 
@@ -134,18 +141,27 @@ template class string_thief<decltype(&std::string::_Mypair),
 
 void string_set_length_hacker(std::string &, std::size_t);
 
+#ifndef __clang__
+#define __has_feature(X) false
+#endif
+
 template <typename ch>
 inline void resize(std::basic_string<ch> &raw_str, std::size_t sz) {
   std::string &str = *reinterpret_cast<std::string *>(&raw_str);
-#if defined(__SANITIZE_ADDRESS__)
+#if defined(__SANITIZE_ADDRESS__) || __has_feature(address_sanitizer)
   raw_str.resize(sz);
 #elif defined(__GLIBCXX__) || defined(_LIBCPP_VERSION) || \
     defined(_MSVC_STL_VERSION)
-  if (sz > str.capacity()) {
-    str.reserve(sz);
+  if constexpr (is_string_reserve_shrink) {
+    if (sz > raw_str.capacity()) {
+      str.reserve(sz * sizeof(ch));
+    }
+  }
+  else {
+    str.reserve(sz * sizeof(ch));
   }
   string_set_length_hacker(str, sz);
-  str[sz] = '\0';
+  for (auto i = sz; i < sz + sizeof(ch); ++i) str[i] = '\0';
 #else
   raw_str.resize(sz);
 #endif
@@ -195,20 +211,24 @@ template class vector_thief<decltype(&std::vector<char>::_Mypair),
 
 template <typename ch>
 inline void resize(std::vector<ch> &raw_vec, std::size_t sz) {
-#if defined(__SANITIZE_ADDRESS__)
+#if defined(__SANITIZE_ADDRESS__) || __has_feature(address_sanitizer)
   raw_vec.resize(sz);
 #elif defined(__GLIBCXX__) ||                                     \
     (defined(_LIBCPP_VERSION) && defined(_LIBCPP_HAS_NO_ASAN)) || \
     defined(_MSVC_STL_VERSION)
   std::vector<char> &vec = *reinterpret_cast<std::vector<char> *>(&raw_vec);
-  vec.reserve(sz);
-  vector_set_length_hacker(vec, sz);
+  vec.reserve(sz * sizeof(ch));
+  vector_set_length_hacker(vec, sz * sizeof(ch));
 #else
   raw_vec.resize(sz);
 #endif
 }
 #endif
 
+#ifndef __clang__
+#undef __has_feature
+#endif
+
 template <typename T>
 inline void resize(T &str, std::size_t sz) {
   str.resize(sz);