add std regex search wrapper to improve script performance.

Author: GeekNeo

CHANGELOG.md

@@ -12,6 +12,7 @@ Bug fixes:
 Improvements:
  * icpp: automatically generate the standard module pcm file in user local system;
  * icpp: add qemu tb buffer to 128MB per thread;
+ * icpp: add std regex search wrapper to improve script performance;
  * imod: add assets configuration to support icpp module extra payloads;
  * imod: make an os independent module if there's no extra include paths;
 

runtime/include/icpp.hpp

@@ -37,6 +37,7 @@
 // c style
 #include <filesystem>
 #include <format>
+#include <regex>
 #include <string>
 #include <string_view>
 #else
@@ -153,4 +154,21 @@ template <size_t N> std::string rand_string() {
   return {tstr.data(), N};
 }
 
+struct regex {
+  regex(std::string_view pattern, int flags = std::regex_constants::ECMAScript |
+                                              std::regex_constants::icase) {
+    init(pattern, flags);
+  }
+  ~regex() { deinit(); }
+  regex() = delete;
+
+  // return true if str matches the initial pattern
+  bool search(std::string_view str) const;
+
+private:
+  void init(std::string_view pattern, int flags);
+  void deinit();
+  void *context_;
+};
+
 } // namespace icpp

src/loader.cpp

@@ -296,6 +296,20 @@ struct ModuleLoader {
         {apisym(
              __ZN4icpp10result_setERKNSt3__117basic_string_viewIcNS0_11char_traitsIcEEEE),
          reinterpret_cast<const void *>(&api::result_sets)});
+
+    auto regexInit = &api::regex::init;
+    auto regexDeinit = &api::regex::deinit;
+    auto regexSearch = &api::regex::search;
+    syms_.insert(
+        {apisym(
+             __ZN4icpp5regex4initENSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEi),
+         *(const void **)(&regexInit)});
+    syms_.insert(
+        {apisym(__ZN4icpp5regex6deinitEv), *(const void **)(&regexDeinit)});
+    syms_.insert(
+        {apisym(
+             __ZNK4icpp5regex6searchENSt3__117basic_string_viewIcNS1_11char_traitsIcEEEE),
+         *(const void **)(&regexSearch)});
 #endif
   }
 

src/object-llvm.cpp

@@ -1774,9 +1774,13 @@ void Object::parseSections() {
                       reinterpret_cast<uint64_t>(expContent->data())});
       // file rva relative to text[0] section
       news.frva = news.vm - textsects_[0].vm;
+      if (ofile_->isELF())
+        news.vrva = news.frva - textsects_[0].frva;
       if (0) {
-        log_print(Develop, "Section {} frva={:x}, vmrva={:x} vm={:x} size={}.",
-                  name.data(), news.frva, news.vrva, news.vm, news.size);
+        log_print(Develop,
+                  "Section index={} frva={:x} vmrva={:x} vm={:x} size={} {}.",
+                  news.index, news.frva, news.vrva, news.vm, news.size,
+                  name.data());
       }
     } else if (s.isBSS() || name.ends_with("bss") || name.ends_with("common")) {
       dynsects_.push_back(

src/runtime.cpp

@@ -202,6 +202,22 @@ std::string_view rand_string(char *buff, int length) {
   return {buff, static_cast<size_t>(length)};
 }
 
+void regex::init(std::string_view pattern, int flags) {
+  auto rflags = static_cast<std::regex_constants::syntax_option_type>(flags);
+  context_ = new std::regex(pattern.data(), rflags);
+}
+
+void regex::deinit() {
+  auto preg = static_cast<std::regex *>(context_);
+  delete preg;
+}
+
+// return true if str matches the initial pattern
+bool regex::search(std::string_view str) const {
+  auto preg = static_cast<std::regex *>(context_);
+  return std::regex_search(str.data(), str.data() + str.size(), *preg);
+}
+
 } // namespace api
 
 } // namespace icpp

src/runtime.h

@@ -8,6 +8,7 @@
 
 #include "utils.h"
 #include <map>
+#include <regex>
 #include <vector>
 
 namespace com {
@@ -155,10 +156,27 @@ will be built by clang-icpp, so the std::string may be defined in a different
 way, to avoid the type mismatch, herein gives it an old C style one.
 
 As of this, if you want to extend icpp runtime with native modules, the type
-mismatch situation must be considered  on Windows.
+mismatch situation must be considered on Windows.
 */
 std::string_view rand_string(char *buff, int length);
 
+struct regex {
+  regex(std::string_view pattern, int flags = std::regex_constants::ECMAScript |
+                                              std::regex_constants::icase) {
+    init(pattern, flags);
+  }
+  ~regex() { deinit(); }
+  regex() = delete;
+
+  // return true if str matches the initial pattern
+  bool search(std::string_view str) const;
+
+  // private:
+  void init(std::string_view pattern, int flags);
+  void deinit();
+  void *context_;
+};
+
 } // namespace api
 
 } // namespace icpp