fs: convert to u8 string for filesystem path

PR-URL: #54653
Fixes: #54476
Reviewed-By: Yagiz Nizipli <[email protected]>
Reviewed-By: Antoine du Hamel <[email protected]>
Reviewed-By: Joyee Cheung <[email protected]>
Reviewed-By: James M Snell <[email protected]>

Author: jazelly

src/node_file.cc

@@ -3130,22 +3130,24 @@ static void CpSyncCheckPaths(const FunctionCallbackInfo<Value>& args) {
   ToNamespacedPath(env, &src);
   THROW_IF_INSUFFICIENT_PERMISSIONS(
       env, permission::PermissionScope::kFileSystemRead, src.ToStringView());
-  auto src_path = std::filesystem::path(src.ToStringView());
+
+  auto src_path = std::filesystem::path(src.ToU8StringView());
 
   BufferValue dest(isolate, args[1]);
   CHECK_NOT_NULL(*dest);
   ToNamespacedPath(env, &dest);
   THROW_IF_INSUFFICIENT_PERMISSIONS(
       env, permission::PermissionScope::kFileSystemWrite, dest.ToStringView());
-  auto dest_path = std::filesystem::path(dest.ToStringView());
 
+  auto dest_path = std::filesystem::path(dest.ToU8StringView());
   bool dereference = args[2]->IsTrue();
   bool recursive = args[3]->IsTrue();
 
   std::error_code error_code;
   auto src_status = dereference
                         ? std::filesystem::symlink_status(src_path, error_code)
                         : std::filesystem::status(src_path, error_code);
+
   if (error_code) {
 #ifdef _WIN32
     int errorno = uv_translate_sys_error(error_code.value());
@@ -3169,34 +3171,41 @@ static void CpSyncCheckPaths(const FunctionCallbackInfo<Value>& args) {
   if (!error_code) {
     // Check if src and dest are identical.
     if (std::filesystem::equivalent(src_path, dest_path)) {
-      std::string message =
-          "src and dest cannot be the same " + dest_path.string();
-      return THROW_ERR_FS_CP_EINVAL(env, message.c_str());
+      std::u8string message =
+          u8"src and dest cannot be the same " + dest_path.u8string();
+      return THROW_ERR_FS_CP_EINVAL(
+          env, reinterpret_cast<const char*>(message.c_str()));
     }
 
     const bool dest_is_dir =
         dest_status.type() == std::filesystem::file_type::directory;
 
     if (src_is_dir && !dest_is_dir) {
-      std::string message = "Cannot overwrite non-directory " +
-                            src_path.string() + " with directory " +
-                            dest_path.string();
-      return THROW_ERR_FS_CP_DIR_TO_NON_DIR(env, message.c_str());
+      std::u8string message = u8"Cannot overwrite non-directory " +
+                              src_path.u8string() + u8" with directory " +
+                              dest_path.u8string();
+      return THROW_ERR_FS_CP_DIR_TO_NON_DIR(
+          env, reinterpret_cast<const char*>(message.c_str()));
     }
 
     if (!src_is_dir && dest_is_dir) {
-      std::string message = "Cannot overwrite directory " + dest_path.string() +
-                            " with non-directory " + src_path.string();
-      return THROW_ERR_FS_CP_NON_DIR_TO_DIR(env, message.c_str());
+      std::u8string message = u8"Cannot overwrite directory " +
+                              dest_path.u8string() + u8" with non-directory " +
+                              src_path.u8string();
+      return THROW_ERR_FS_CP_NON_DIR_TO_DIR(
+          env, reinterpret_cast<const char*>(message.c_str()));
     }
   }
 
-  std::string dest_path_str = dest_path.string();
+  std::u8string dest_path_str = dest_path.u8string();
+
   // Check if dest_path is a subdirectory of src_path.
-  if (src_is_dir && dest_path_str.starts_with(src_path.string())) {
-    std::string message = "Cannot copy " + src_path.string() +
-                          " to a subdirectory of self " + dest_path.string();
-    return THROW_ERR_FS_CP_EINVAL(env, message.c_str());
+  if (src_is_dir && dest_path_str.starts_with(src_path.u8string())) {
+    std::u8string message = u8"Cannot copy " + src_path.u8string() +
+                            u8" to a subdirectory of self " +
+                            dest_path.u8string();
+    return THROW_ERR_FS_CP_EINVAL(
+        env, reinterpret_cast<const char*>(message.c_str()));
   }
 
   auto dest_parent = dest_path.parent_path();
@@ -3207,9 +3216,11 @@ static void CpSyncCheckPaths(const FunctionCallbackInfo<Value>& args) {
          dest_parent.parent_path() != dest_parent) {
     if (std::filesystem::equivalent(
             src_path, dest_path.parent_path(), error_code)) {
-      std::string message = "Cannot copy " + src_path.string() +
-                            " to a subdirectory of self " + dest_path.string();
-      return THROW_ERR_FS_CP_EINVAL(env, message.c_str());
+      std::u8string message = u8"Cannot copy " + src_path.u8string() +
+                              u8" to a subdirectory of self " +
+                              dest_path.u8string();
+      return THROW_ERR_FS_CP_EINVAL(
+          env, reinterpret_cast<const char*>(message.c_str()));
     }
 
     // If equivalent fails, it's highly likely that dest_parent does not exist
@@ -3221,25 +3232,29 @@ static void CpSyncCheckPaths(const FunctionCallbackInfo<Value>& args) {
   }
 
   if (src_is_dir && !recursive) {
-    std::string message =
-        "Recursive option not enabled, cannot copy a directory: " +
-        src_path.string();
-    return THROW_ERR_FS_EISDIR(env, message.c_str());
+    std::u8string message =
+        u8"Recursive option not enabled, cannot copy a directory: " +
+        src_path.u8string();
+    return THROW_ERR_FS_EISDIR(env,
+                               reinterpret_cast<const char*>(message.c_str()));
   }
 
   switch (src_status.type()) {
     case std::filesystem::file_type::socket: {
-      std::string message = "Cannot copy a socket file: " + dest_path.string();
-      return THROW_ERR_FS_CP_SOCKET(env, message.c_str());
+      std::u8string message = u8"Cannot copy a socket file: " + dest_path_str;
+      return THROW_ERR_FS_CP_SOCKET(
+          env, reinterpret_cast<const char*>(message.c_str()));
     }
     case std::filesystem::file_type::fifo: {
-      std::string message = "Cannot copy a FIFO pipe: " + dest_path.string();
-      return THROW_ERR_FS_CP_FIFO_PIPE(env, message.c_str());
+      std::u8string message = u8"Cannot copy a FIFO pipe: " + dest_path_str;
+      return THROW_ERR_FS_CP_FIFO_PIPE(
+          env, reinterpret_cast<const char*>(message.c_str()));
     }
     case std::filesystem::file_type::unknown: {
-      std::string message =
-          "Cannot copy an unknown file type: " + dest_path.string();
-      return THROW_ERR_FS_CP_UNKNOWN(env, message.c_str());
+      std::u8string message =
+          u8"Cannot copy an unknown file type: " + dest_path_str;
+      return THROW_ERR_FS_CP_UNKNOWN(
+          env, reinterpret_cast<const char*>(message.c_str()));
     }
     default:
       break;

src/util.h

@@ -562,6 +562,10 @@ class BufferValue : public MaybeStackBuffer<char> {
   inline std::string_view ToStringView() const {
     return std::string_view(out(), length());
   }
+  inline std::u8string_view ToU8StringView() const {
+    return std::u8string_view(reinterpret_cast<const char8_t*>(out()),
+                              length());
+  }
 };
 
 #define SPREAD_BUFFER_ARG(val, name)                                           \

test/fixtures/copy/utf/新建文件夹/index.js

@@ -0,0 +1,3 @@
+module.exports = {
+  purpose: 'testing copy'
+};

test/parallel/test-fs-cp.mjs

@@ -30,6 +30,14 @@ function nextdir() {
 
 // Synchronous implementation of copy.
 
+// It copies a nested folder containing UTF characters.
+{
+  const src = './test/fixtures/copy/utf/新建文件夹';
+  const dest = nextdir();
+  cpSync(src, dest, mustNotMutateObjectDeep({ recursive: true }));
+  assertDirEquivalent(src, dest);
+}
+
 // It copies a nested folder structure with files and folders.
 {
   const src = './test/fixtures/copy/kitchen-sink';