refactor: support c++20

CMakeLists.txt

@@ -171,6 +171,7 @@ if(MSVC)
   set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /Zi")
   set(CMAKE_SHARED_LINKER_FLAGS_RELEASE "${CMAKE_SHARED_LINKER_FLAGS_RELEASE} /DEBUG /OPT:REF /OPT:ICF")
   add_definitions("/wd4244 /wd4996")
+  add_compile_options("/Zc:__cplusplus")
 
   # large address aware
   set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} /LARGEADDRESSAWARE")

include/darts.h

@@ -1429,9 +1429,9 @@ class DoubleArrayBuilder {
   void clear();
 
  private:
-  enum { BLOCK_SIZE = 256 };
-  enum { NUM_EXTRA_BLOCKS = 16 };
-  enum { NUM_EXTRAS = BLOCK_SIZE * NUM_EXTRA_BLOCKS };
+  static const std::size_t BLOCK_SIZE = 256;
+  static const std::size_t NUM_EXTRA_BLOCKS = 16;
+  static const std::size_t NUM_EXTRAS = BLOCK_SIZE * NUM_EXTRA_BLOCKS;
 
   enum { UPPER_MASK = 0xFF << 21 };
   enum { LOWER_MASK = 0xFF };

src/rime/common.h

@@ -104,6 +104,27 @@ class path : public std::filesystem::path {
   path& operator/=(const std::string& p) { return *this /= path(p); }
   path& operator/=(const char* p) { return *this /= path(p); }
 
+  path stem() const { return path(fs_path::stem()); }
+  path filename() const { return path(fs_path::filename()); }
+  path extension() const { return path(fs_path::extension()); }
+  // return UTF-8 encoded std::string
+  std::string string_utf8() const {
+#if __cplusplus >= 202002L
+    const auto u8s = this->u8string();
+    return std::string(u8s.begin(), u8s.end());
+#else
+    return this->u8string();
+#endif
+  }
+  // return UTF-8 encoded std::string of generic format
+  std::string generic_string_utf8() const {
+#if __cplusplus >= 202002L
+    const auto u8s = this->generic_u8string();
+    return std::string(u8s.begin(), u8s.end());
+#else
+    return this->generic_u8string();
+#endif
+  }
   friend path operator/(const path& lhs, const path& rhs) {
     return path(lhs) /= rhs;
   }
@@ -128,7 +149,7 @@ class path : public std::filesystem::path {
   }
 #ifdef RIME_ENABLE_LOGGING
   friend std::ostream& operator<<(std::ostream& os, const path& p) {
-    return os << p.u8string();
+    return os << p.string_utf8();
   }
 #endif
 };

src/rime/config/config_data.cc

@@ -66,7 +66,7 @@ bool ConfigData::LoadFromFile(const path& file_path, ConfigCompiler* compiler) {
   modified_ = false;
   root.reset();
   if (!std::filesystem::exists(file_path)) {
-    if (!boost::ends_with(file_path.u8string(), ".custom.yaml"))
+    if (!boost::ends_with(file_path.string_utf8(), ".custom.yaml"))
       LOG(WARNING) << "nonexistent config file '" << file_path << "'.";
     return false;
   }

src/rime/dict/dict_compiler.cc

@@ -209,7 +209,7 @@ bool DictCompiler::Compile(const path& schema_file) {
 
 static path relocate_target(const path& source_path,
                             ResourceResolver* target_resolver) {
-  auto resource_id = source_path.filename().u8string();
+  auto resource_id = source_path.filename().string_utf8();
   return target_resolver->ResolvePath(resource_id);
 }
 

src/rime/dict/entry_collector.cc

@@ -56,7 +56,7 @@ void EntryCollector::LoadPresetVocabulary(DictSettings* settings) {
 
 void EntryCollector::Collect(const path& dict_file) {
   LOG(INFO) << "collecting entries from " << dict_file;
-  current_dict_file = dict_file.u8string();
+  current_dict_file = dict_file.string_utf8();
   line_number = 0;
   // read table
   std::ifstream fin(dict_file.c_str());

src/rime/dict/user_db.cc

@@ -108,7 +108,7 @@ bool UserDbHelper::UpdateUserInfo() {
 }
 
 bool UserDbHelper::IsUniformFormat(const path& file_path) {
-  return boost::ends_with(file_path.filename().u8string(),
+  return boost::ends_with(file_path.filename().string_utf8(),
                           plain_userdb_extension);
 }
 

src/rime/dict/vocabulary.h

@@ -40,6 +40,8 @@ struct ShortDictEntry {
   double weight = 0.0;
 
   ShortDictEntry() = default;
+  ShortDictEntry(const string& t, const Code& c, double w)
+      : text(t), code(c), weight(w) {}
   bool operator<(const ShortDictEntry& other) const;
 };
 

src/rime/gear/simplifier.cc

@@ -41,7 +41,7 @@ class Opencc {
     opencc::Config config;
     try {
       // opencc accepts file path encoded in UTF-8.
-      converter_ = config.NewFromFile(config_path_.u8string());
+      converter_ = config.NewFromFile(config_path_.string_utf8());
 
       const list<opencc::ConversionPtr> conversions =
           converter_->GetConversionChain()->GetConversions();
@@ -315,7 +315,7 @@ Simplifier* SimplifierComponent::Create(const Ticket& ticket) {
     return new Simplifier(ticket, opencc);
   }
   path opencc_config_path = path(opencc_config);
-  if (opencc_config_path.extension().u8string() == ".ini") {
+  if (opencc_config_path.extension().string_utf8() == ".ini") {
     LOG(ERROR) << "please upgrade opencc_config to an opencc 1.0 config file.";
     return nullptr;
   }

src/rime/lever/deployment_tasks.cc

@@ -52,8 +52,8 @@ bool DetectModifications::Run(Deployer* deployer) {
         for (fs::directory_iterator iter(p), end; iter != end; ++iter) {
           path entry(iter->path());
           if (fs::is_regular_file(fs::canonical(entry)) &&
-              entry.extension().u8string() == ".yaml" &&
-              entry.filename().u8string() != "user.yaml") {
+              entry.extension().string_utf8() == ".yaml" &&
+              entry.filename().string_utf8() != "user.yaml") {
             last_modified =
                 (std::max)(last_modified,
                            filesystem::to_time_t(fs::last_write_time(entry)));
@@ -460,7 +460,7 @@ bool PrebuildAllSchemas::Run(Deployer* deployer) {
   for (fs::directory_iterator iter(shared_data_path), end; iter != end;
        ++iter) {
     path entry(iter->path());
-    if (boost::ends_with(entry.filename().u8string(), ".schema.yaml")) {
+    if (boost::ends_with(entry.filename().string_utf8(), ".schema.yaml")) {
       the<DeploymentTask> t(new SchemaUpdate(entry));
       if (!t->Run(deployer))
         success = false;
@@ -525,7 +525,7 @@ bool UserDictSync::Run(Deployer* deployer) {
 }
 
 static bool IsCustomizedCopy(const path& file_path) {
-  auto file_name = file_path.filename().u8string();
+  auto file_name = file_path.filename().string_utf8();
   if (boost::ends_with(file_name, ".yaml") &&
       !boost::ends_with(file_name, ".custom.yaml")) {
     Config config;
@@ -556,7 +556,7 @@ bool BackupConfigFiles::Run(Deployer* deployer) {
     path entry(iter->path());
     if (!fs::is_regular_file(entry))
       continue;
-    auto file_extension = entry.extension().u8string();
+    auto file_extension = entry.extension().string_utf8();
     bool is_yaml_file = file_extension == ".yaml";
     bool is_text_file = file_extension == ".txt";
     if (!is_yaml_file && !is_text_file)
@@ -596,7 +596,7 @@ bool CleanupTrash::Run(Deployer* deployer) {
     path entry(iter->path());
     if (!fs::is_regular_file(entry))
       continue;
-    auto file_name = entry.filename().u8string();
+    auto file_name = entry.filename().string_utf8();
     if (file_name == "rime.log" || boost::ends_with(file_name, ".bin") ||
         boost::ends_with(file_name, ".reverse.kct") ||
         boost::ends_with(file_name, ".userdb.kct.old") ||
@@ -650,15 +650,15 @@ bool CleanOldLogFiles::Run(Deployer* deployer) {
     try {
       // preparing files
       for (const auto& entry : fs::directory_iterator(dir)) {
-        const string& file_name(entry.path().filename().u8string());
+        const string& file_name(path(entry.path().filename()).string_utf8());
         if (entry.is_regular_file() && !entry.is_symlink() &&
             boost::starts_with(file_name, app_name) &&
             boost::ends_with(file_name, ".log") &&
             !boost::contains(file_name, today)) {
           files_to_remove.push_back(entry.path());
         } else if (entry.is_symlink()) {
           auto target = fs::read_symlink(entry.path());
-          const string& target_file_name(target.filename().u8string());
+          const string& target_file_name(path(target.filename()).string_utf8());
           if (boost::starts_with(target_file_name, app_name) &&
               boost::ends_with(target_file_name, ".log")) {
             files_in_use.insert(target);

src/rime/lever/switcher_settings.cc

@@ -53,7 +53,7 @@ void SwitcherSettings::GetAvailableSchemasFromDirectory(const path& dir) {
   }
   for (fs::directory_iterator it(dir), end; it != end; ++it) {
     path file_path(it->path());
-    if (boost::ends_with(file_path.u8string(), ".schema.yaml")) {
+    if (boost::ends_with(file_path.string_utf8(), ".schema.yaml")) {
       Config config;
       if (config.LoadFromFile(file_path)) {
         SchemaInfo info;

src/rime/lever/user_dict_manager.cc

@@ -40,7 +40,7 @@ void UserDictManager::GetUserDictList(UserDictList* user_dict_list,
     return;
   }
   for (fs::directory_iterator it(path_), end; it != end; ++it) {
-    string name = it->path().filename().u8string();
+    string name = path(it->path().filename()).string_utf8();
     if (boost::ends_with(name, component->extension())) {
       boost::erase_last(name, component->extension());
       user_dict_list->push_back(name);

src/rime/resource.cc

@@ -10,7 +10,7 @@
 namespace rime {
 
 string ResourceResolver::ToResourceId(const string& file_path) const {
-  string string_path = path(file_path).generic_u8string();
+  string string_path = path(file_path).generic_string_utf8();
   bool has_prefix = boost::starts_with(string_path, type_.prefix);
   bool has_suffix = boost::ends_with(string_path, type_.suffix);
   size_t start = (has_prefix ? type_.prefix.length() : 0);

tools/rime_table_decompiler.cc

@@ -122,7 +122,7 @@ int main(int argc, char* argv[]) {
   fout << "# Rime dictionary\n\n";
   fout << "---\n"
           "name: "
-       << file_path.stem().u8string()
+       << rime::path(file_path).stem().string_utf8()
        << "\n"
           "version: \"1.0\"\n"
           "...\n\n";