fixes

Author: Shamzik

runtime-light/stdlib/string/regex-functions.cpp

@@ -118,7 +118,8 @@ bool compile_regex(RegexInfo& regex_info) noexcept {
   }
 
   // check runtime cache
-  if (const auto it{regex_state.regex_pcre2_code_cache.find(regex_info.regex)}; it != regex_state.regex_pcre2_code_cache.end()) {
+  if (const auto it{regex_state.regex_pcre2_code_cache.find({regex_info.regex.data(), static_cast<string::size_type>(regex_info.regex.size())})};
+      it != regex_state.regex_pcre2_code_cache.end()) {
     auto& [compile_options, regex_code] = it->second;
     regex_info.compile_options = compile_options;
     regex_info.regex_code = regex_code;
@@ -695,7 +696,7 @@ std::optional<array<mixed>> split_regex(RegexInfo& regex_info, int64_t limit_val
     auto match_start{offsets[0]};
     auto match_end{offsets[1]};
 
-    make_output_val(regex_info, no_empty, offset_capture, offset, match_start).transform([&output, &limit_val](auto&& output_val)noexcept {
+    make_output_val(regex_info, no_empty, offset_capture, offset, match_start).transform([&output, &limit_val](auto&& output_val) noexcept {
       output.push_back(std::move(output_val));
       if (limit_val != kphp::regex::PREG_NOLIMIT) {
         limit_val--;
@@ -706,7 +707,7 @@ std::optional<array<mixed>> split_regex(RegexInfo& regex_info, int64_t limit_val
     if (delim_capture) {
       for (auto i{1uz}; i < regex_info.match_count; i++) {
         auto j{2 * i};
-        make_output_val(regex_info, no_empty, offset_capture, offsets[j], offsets[j + 1]).transform([&output](auto&& output_val)noexcept {
+        make_output_val(regex_info, no_empty, offset_capture, offsets[j], offsets[j + 1]).transform([&output](auto&& output_val) noexcept {
           output.push_back(std::move(output_val));
           return 0;
         });

runtime-light/stdlib/string/regex-state.h

@@ -9,14 +9,16 @@
 
 #include "common/mixin/not_copyable.h"
 #include "runtime-common/core/allocator/script-allocator.h"
+#include "runtime-common/core/runtime-core.h"
 #include "runtime-common/core/std/containers.h"
 #include "runtime-light/stdlib/string/regex-include.h"
 #include "runtime-light/utils/concepts.h"
 
 struct RegexInstanceState final : private vk::not_copyable {
 private:
-  template<hashable Key, typename Value>
-  using unordered_map = kphp::stl::unordered_map<Key, Value, kphp::memory::script_allocator>;
+  template<typename Value>
+  using string_unordered_map =
+      kphp::stl::unordered_map<string, Value, kphp::memory::script_allocator, decltype([](const string& s) noexcept { return static_cast<size_t>(s.hash()); })>;
 
   static constexpr size_t MAX_SUBPATTERNS_COUNT = 512;
 
@@ -34,12 +36,13 @@ struct RegexInstanceState final : private vk::not_copyable {
   const regex_pcre2_compile_context_t compile_context;
   const regex_pcre2_match_context_t match_context;
   regex_pcre2_match_data_t regex_pcre2_match_data;
-  unordered_map<std::string_view, compiled_regex_cache_entry> regex_pcre2_code_cache;
+  string_unordered_map<compiled_regex_cache_entry> regex_pcre2_code_cache;
 
   RegexInstanceState() noexcept;
 
   void add_compiled_code(std::string_view regex, uint32_t compile_options, regex_pcre2_code_t regex_code) noexcept {
-    regex_pcre2_code_cache.emplace(regex, compiled_regex_cache_entry{.compile_options = compile_options, .regex_code = regex_code});
+    regex_pcre2_code_cache.emplace(string{regex.data(), static_cast<string::size_type>(regex.size())},
+                                   compiled_regex_cache_entry{.compile_options = compile_options, .regex_code = regex_code});
   }
 
   static RegexInstanceState& get() noexcept;