From 6cba5ba7e6565fa06cc093ccd177a1b2954063fe Mon Sep 17 00:00:00 2001
From: NicknineTheEagle <nik18.1995@gmail.com>
Date: Sun, 22 Sep 2024 20:24:37 +0300
Subject: [PATCH] Convert STFS filenames from Win-1252 to UTF-8

---
 src/xenia/base/string.cc                      | 77 +++++++++++++++++++
 src/xenia/base/string.h                       |  2 +
 src/xenia/vfs/devices/disc_image_device.cc    |  4 +-
 .../xcontent_devices/stfs_container_device.cc |  6 +-
 .../xcontent_devices/svod_container_device.cc |  4 +-
 5 files changed, 89 insertions(+), 4 deletions(-)

diff --git a/src/xenia/base/string.cc b/src/xenia/base/string.cc
index 38b0ccf5e..9d13b7d86 100644
--- a/src/xenia/base/string.cc
+++ b/src/xenia/base/string.cc
@@ -14,6 +14,9 @@
 #include <locale>
 
 #include "xenia/base/platform.h"
+#if XE_PLATFORM_WIN32
+#include "xenia/base/platform_win.h"
+#endif
 
 #if !XE_PLATFORM_WIN32
 #include <strings.h>
@@ -58,4 +61,78 @@ std::u16string to_utf16(const std::string_view source) {
   return utfcpp::utf8to16(source);
 }
 
+std::string utf8_to_win1252(const std::string_view source) {
+#if XE_PLATFORM_WIN32
+  std::string input_str(source);
+  int srclen = static_cast<int>(input_str.size());
+
+  int wlen =
+      MultiByteToWideChar(CP_UTF8, 0, input_str.c_str(), srclen, NULL, 0);
+  if (!wlen) {
+    return "";
+  }
+  std::vector<WCHAR> wbuf(wlen);
+  int result = MultiByteToWideChar(CP_UTF8, 0, input_str.c_str(), srclen,
+                                   wbuf.data(), wlen);
+  if (!result) {
+    return "";
+  }
+
+  int len = WideCharToMultiByte(1252, 0, wbuf.data(), wlen, NULL, 0, "_", NULL);
+  if (!len) {
+    return "";
+  }
+  std::vector<CHAR> buf(len);
+  result = WideCharToMultiByte(1252, 0, wbuf.data(), wlen, buf.data(), len, "_",
+                               NULL);
+  if (!result) {
+    return "";
+  }
+
+  std::string output_str(buf.begin(), buf.end());
+  return output_str;
+#else
+  // TODO: Use iconv on POSIX.
+  std::string output_str(source);
+  return output_str;
+#endif
+}
+
+std::string win1252_to_utf8(const std::string_view source) {
+#if XE_PLATFORM_WIN32
+  std::string input_str(source);
+  int srclen = static_cast<int>(input_str.size());
+
+  int wlen = MultiByteToWideChar(1252, 0, input_str.c_str(), srclen, NULL, 0);
+  if (!wlen) {
+    return "";
+  }
+  std::vector<WCHAR> wbuf(wlen);
+  int result = MultiByteToWideChar(1252, 0, input_str.c_str(), srclen,
+                                   wbuf.data(), wlen);
+  if (!result) {
+    return "";
+  }
+
+  int len =
+      WideCharToMultiByte(CP_UTF8, 0, wbuf.data(), wlen, NULL, 0, "_", NULL);
+  if (!len) {
+    return "";
+  }
+  std::vector<CHAR> buf(len);
+  result = WideCharToMultiByte(CP_UTF8, 0, wbuf.data(), wlen, buf.data(), len,
+                               "_", NULL);
+  if (!result) {
+    return "";
+  }
+
+  std::string output_str(buf.begin(), buf.end());
+  return output_str;
+#else
+  // TODO: Use iconv on POSIX.
+  std::string output_str(source);
+  return output_str;
+#endif
+}
+
 }  // namespace xe
diff --git a/src/xenia/base/string.h b/src/xenia/base/string.h
index b768d22e6..70f39d3be 100644
--- a/src/xenia/base/string.h
+++ b/src/xenia/base/string.h
@@ -22,6 +22,8 @@ char* xe_strdup(const char* source);
 
 std::string to_utf8(const std::u16string_view source);
 std::u16string to_utf16(const std::string_view source);
+std::string utf8_to_win1252(const std::string_view source);
+std::string win1252_to_utf8(const std::string_view source);
 
 }  // namespace xe
 
diff --git a/src/xenia/vfs/devices/disc_image_device.cc b/src/xenia/vfs/devices/disc_image_device.cc
index ef8800a03..4ab72d1fb 100644
--- a/src/xenia/vfs/devices/disc_image_device.cc
+++ b/src/xenia/vfs/devices/disc_image_device.cc
@@ -140,7 +140,9 @@ bool DiscImageDevice::ReadEntry(ParseState* state, const uint8_t* buffer,
     return false;
   }
 
-  auto name = std::string(name_buffer, name_length);
+  // Filename is stored as Windows-1252, convert it to UTF-8.
+  auto ansi_name = std::string(name_buffer, name_length);
+  auto name = xe::win1252_to_utf8(ansi_name);
 
   auto entry = DiscImageEntry::Create(this, parent, name, mmap_.get());
   entry->attributes_ = attributes | kFileAttributeReadOnly;
diff --git a/src/xenia/vfs/devices/xcontent_devices/stfs_container_device.cc b/src/xenia/vfs/devices/xcontent_devices/stfs_container_device.cc
index 417f05414..a8d9e62de 100644
--- a/src/xenia/vfs/devices/xcontent_devices/stfs_container_device.cc
+++ b/src/xenia/vfs/devices/xcontent_devices/stfs_container_device.cc
@@ -113,8 +113,10 @@ StfsContainerDevice::Result StfsContainerDevice::Read() {
 std::unique_ptr<XContentContainerEntry> StfsContainerDevice::ReadEntry(
     Entry* parent, MultiFileHandles* files,
     const StfsDirectoryEntry* dir_entry) {
-  std::string name(reinterpret_cast<const char*>(dir_entry->name),
-                   dir_entry->flags.name_length & 0x3F);
+  // Filename is stored as Windows-1252, convert it to UTF-8.
+  std::string ansi_name(reinterpret_cast<const char*>(dir_entry->name),
+                        dir_entry->flags.name_length & 0x3F);
+  std::string name = xe::win1252_to_utf8(ansi_name);
 
   auto entry = XContentContainerEntry::Create(this, parent, name, &files_);
 
diff --git a/src/xenia/vfs/devices/xcontent_devices/svod_container_device.cc b/src/xenia/vfs/devices/xcontent_devices/svod_container_device.cc
index 21b367cc8..8ec80ebf2 100644
--- a/src/xenia/vfs/devices/xcontent_devices/svod_container_device.cc
+++ b/src/xenia/vfs/devices/xcontent_devices/svod_container_device.cc
@@ -153,7 +153,9 @@ SvodContainerDevice::Result SvodContainerDevice::ReadEntry(
     return Result::kReadError;
   }
 
-  auto name = std::string(name_buffer.get(), dir_entry.name_length);
+  // Filename is stored as Windows-1252, convert it to UTF-8.
+  auto ansi_name = std::string(name_buffer.get(), dir_entry.name_length);
+  auto name = xe::win1252_to_utf8(ansi_name);
 
   // Read the left node
   if (dir_entry.node_l) {