From 9faf9a6145367043cce0123fb41e624a992f58d4 Mon Sep 17 00:00:00 2001
From: Megamouse <studienricky89@googlemail.com>
Date: Sat, 10 Jan 2026 12:36:33 +0100
Subject: [PATCH] ISO: optimize byteswap and utf16 string conversion

---
 rpcs3/Loader/ISO.cpp | 23 ++++++++---------------
 1 file changed, 8 insertions(+), 15 deletions(-)

diff --git a/rpcs3/Loader/ISO.cpp b/rpcs3/Loader/ISO.cpp
index b3f9ccb8cf..e46a1a6b82 100644
--- a/rpcs3/Loader/ISO.cpp
+++ b/rpcs3/Loader/ISO.cpp
@@ -102,24 +102,17 @@ std::optional<iso_fs_metadata> iso_read_directory_entry(fs::file& file, bool nam
 	}
 	else if (names_in_ucs2) // for strings in joliet descriptor
 	{
-		std::string new_file_name;
-		int read = 0;
-		const u8* raw_str = reinterpret_cast<const u8*>(file_name.c_str());
-		while (read < file_name_length)
+		// characters are stored in big endian format.
+		std::u16string utf16;
+		utf16.resize(file_name_length / 2);
+
+		const u16* raw = reinterpret_cast<const u16*>(file_name.data());
+		for (size_t i = 0; i < utf16.size(); ++i, raw++)
 		{
-			// characters are stored in big endian format.
-			const u16 upper = raw_str[read];
-			const u8 lower = raw_str[read + 1];
-
-			const u16 code_point = (upper << 8) + lower;
-
-			std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> convert;
-			new_file_name += convert.to_bytes(code_point);
-
-			read += 2;
+			utf16[i] = *reinterpret_cast<const be_t<u16>*>(raw);
 		}
 
-		file_name = new_file_name;
+		file_name = utf16_to_utf8(utf16);
 	}
 
 	if (file_name.ends_with(";1"))