From 7c63dddd8c88de5347e7f7cbcd12078a71e8e7ce Mon Sep 17 00:00:00 2001
From: DrakkLord <DrakkLord@users.noreply.github.com>
Date: Thu, 25 Jul 2024 19:18:04 +0200
Subject: [PATCH] added support for windows UTF-8 file name handling
---
CMakeLists.txt | 4 +-
source/cppfs/CMakeLists.txt | 9 +++
.../cppfs/include/cppfs/windows/UTF8Handler.h | 29 +++++++
.../cppfs/source/windows/LocalFileHandle.cpp | 24 +++---
.../source/windows/LocalFileIterator.cpp | 7 +-
.../cppfs/source/windows/LocalFileWatcher.cpp | 78 ++++++++-----------
source/cppfs/source/windows/UTF8Handler.cpp | 50 ++++++++++++
7 files changed, 139 insertions(+), 62 deletions(-)
create mode 100644 source/cppfs/include/cppfs/windows/UTF8Handler.h
create mode 100644 source/cppfs/source/windows/UTF8Handler.cpp
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 34d80c0..ad3ab72 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -70,7 +70,9 @@ option(OPTION_BUILD_DOCS "Build documentation."
option(OPTION_BUILD_EXAMPLES "Build examples." OFF)
option(OPTION_BUILD_SSH_BACKEND "Build SSH backend" OFF)
option(OPTION_FORCE_SYSTEM_DIR_INSTALL "Force system dir install" OFF)
-
+if(WIN32)
+ option(OPTION_WINDOWS_UTF8 "Build Windows UTF8 support" ON)
+endif()
#
# Declare project
diff --git a/source/cppfs/CMakeLists.txt b/source/cppfs/CMakeLists.txt
index 520512c..379fc41 100644
--- a/source/cppfs/CMakeLists.txt
+++ b/source/cppfs/CMakeLists.txt
@@ -119,10 +119,12 @@ endif()
if("${CMAKE_SYSTEM_NAME}" MATCHES "Windows")
set(headers ${headers}
${include_path}/windows/LocalFileWatcher.h
+ ${include_path}/windows/UTF8Handler.h
)
set(sources ${sources}
${source_path}/windows/LocalFileWatcher.cpp
+ ${source_path}/windows/UTF8Handler.cpp
)
endif()
@@ -188,6 +190,13 @@ set_target_properties(${target}
SOVERSION ${META_VERSION_MAJOR}
)
+if(WIN32 AND OPTION_WINDOWS_UTF8)
+ target_compile_options(${target}
+ PRIVATE
+ -DUNICODE
+ -D_UNICODE
+ )
+endif()
#
# Include directories
diff --git a/source/cppfs/include/cppfs/windows/UTF8Handler.h b/source/cppfs/include/cppfs/windows/UTF8Handler.h
new file mode 100644
index 0000000..0c6bbcf
--- /dev/null
+++ b/source/cppfs/include/cppfs/windows/UTF8Handler.h
@@ -0,0 +1,29 @@
+
+#pragma once
+
+
+#include <memory>
+
+#include <cppfs/AbstractFileSystem.h>
+
+#if defined(_UNICODE)
+ #define UTF8Convert_UTF8toW(s) UTF8Handler::utf8_to_wstring(s)
+ #define UTF8Convert_WtoUTF8(s) UTF8Handler::wstring_to_utf8(s)
+#else
+ #define UTF8Convert_UTF8toW(s) s
+ #define UTF8Convert_WtoUTF8(s) s
+#endif
+
+namespace cppfs
+{
+ /**
+ * @brief
+ * Windows specific UTF-8 string conversions
+ */
+ namespace UTF8Handler
+ {
+ std::wstring utf8_to_wstring(const std::string& utf8_str);
+ std::string wstring_to_utf8(const std::wstring& wstr);
+ } // namespace UTF8Handler
+
+} // namespace cppfs
diff --git a/source/cppfs/source/windows/LocalFileHandle.cpp b/source/cppfs/source/windows/LocalFileHandle.cpp
index c8fda29..0d0fbc3 100644
--- a/source/cppfs/source/windows/LocalFileHandle.cpp
+++ b/source/cppfs/source/windows/LocalFileHandle.cpp
@@ -9,7 +9,7 @@
#include <cppfs/FilePath.h>
#include <cppfs/windows/LocalFileSystem.h>
#include <cppfs/windows/LocalFileIterator.h>
-
+#include <cppfs/windows/UTF8Handler.h>
namespace cppfs
{
@@ -105,7 +105,7 @@ std::vector<std::string> LocalFileHandle::listFiles() const
// Open directory
WIN32_FIND_DATA findData;
std::string query = FilePath(m_path).fullPath() + "/*";
- HANDLE findHandle = FindFirstFileA(query.c_str(), &findData);
+ HANDLE findHandle = FindFirstFile(UTF8Convert_UTF8toW(query).c_str(), &findData);
if (findHandle == INVALID_HANDLE_VALUE)
{
@@ -116,7 +116,7 @@ std::vector<std::string> LocalFileHandle::listFiles() const
do
{
// Get name
- std::string name = findData.cFileName;
+ std::string name = UTF8Convert_WtoUTF8(findData.cFileName);
// Ignore . and ..
if (name != ".." && name != ".")
@@ -213,7 +213,7 @@ bool LocalFileHandle::createDirectory()
if (exists()) return false;
// Create directory
- if (!CreateDirectoryA(m_path.c_str(), nullptr))
+ if (!CreateDirectory(UTF8Convert_UTF8toW(m_path).c_str(), nullptr))
{
return false;
}
@@ -229,7 +229,7 @@ bool LocalFileHandle::removeDirectory()
if (!isDirectory()) return false;
// Remove directory
- if (!RemoveDirectoryA(m_path.c_str()))
+ if (!RemoveDirectory(UTF8Convert_UTF8toW(m_path).c_str()))
{
return false;
}
@@ -255,7 +255,7 @@ bool LocalFileHandle::copy(AbstractFileHandleBackend & dest)
}
// Copy file
- if (!CopyFileA(src.c_str(), dst.c_str(), FALSE))
+ if (!CopyFile(UTF8Convert_UTF8toW(src).c_str(), UTF8Convert_UTF8toW(dst).c_str(), FALSE))
{
// Error!
return false;
@@ -282,7 +282,7 @@ bool LocalFileHandle::move(AbstractFileHandleBackend & dest)
}
// Move file
- if (!MoveFileA(src.c_str(), dst.c_str()))
+ if (!MoveFile(UTF8Convert_UTF8toW(src).c_str(), UTF8Convert_UTF8toW(dst).c_str()))
{
// Error!
return false;
@@ -312,7 +312,7 @@ bool LocalFileHandle::createLink(AbstractFileHandleBackend & dest)
}
// Copy file
- if (!CreateHardLinkA(dst.c_str(), src.c_str(), 0))
+ if (!CreateHardLink(UTF8Convert_UTF8toW(dst).c_str(), UTF8Convert_UTF8toW(src).c_str(), 0))
{
// Error!
return false;
@@ -338,7 +338,7 @@ bool LocalFileHandle::createSymbolicLink(AbstractFileHandleBackend & dest)
}
// Copy file
- if (!CreateSymbolicLinkA(dst.c_str(), src.c_str(), 0))
+ if (!CreateSymbolicLink(UTF8Convert_UTF8toW(dst).c_str(), UTF8Convert_UTF8toW(src).c_str(), 0))
{
// Error!
return false;
@@ -357,7 +357,7 @@ bool LocalFileHandle::rename(const std::string & filename)
std::string path = FilePath(FilePath(m_path).directoryPath()).resolve(filename).fullPath();
// Rename
- if (!MoveFileA(m_path.c_str(), path.c_str()))
+ if (!MoveFile(UTF8Convert_UTF8toW(m_path).c_str(), UTF8Convert_UTF8toW(path).c_str()))
{
// Error!
return false;
@@ -377,7 +377,7 @@ bool LocalFileHandle::remove()
if (!isFile()) return false;
// Delete file
- if (!DeleteFileA(m_path.c_str()))
+ if (!DeleteFile(UTF8Convert_UTF8toW(m_path).c_str()))
{
return false;
}
@@ -406,7 +406,7 @@ void LocalFileHandle::readFileInfo() const
m_fileInfo = (void *)new WIN32_FILE_ATTRIBUTE_DATA;
// Get file info
- if (!GetFileAttributesExA(m_path.c_str(), GetFileExInfoStandard, (WIN32_FILE_ATTRIBUTE_DATA*)m_fileInfo))
+ if (!GetFileAttributesEx(UTF8Convert_UTF8toW(m_path).c_str(), GetFileExInfoStandard, (WIN32_FILE_ATTRIBUTE_DATA*)m_fileInfo))
{
// Error!
delete (WIN32_FILE_ATTRIBUTE_DATA *)m_fileInfo;
diff --git a/source/cppfs/source/windows/LocalFileIterator.cpp b/source/cppfs/source/windows/LocalFileIterator.cpp
index 5e01f0e..c01a0a2 100644
--- a/source/cppfs/source/windows/LocalFileIterator.cpp
+++ b/source/cppfs/source/windows/LocalFileIterator.cpp
@@ -5,6 +5,7 @@
#include <cppfs/FilePath.h>
#include <cppfs/windows/LocalFileSystem.h>
+#include <cppfs/windows/UTF8Handler.h>
namespace cppfs
@@ -78,7 +79,7 @@ std::string LocalFileIterator::name() const
}
// Return filename of current item
- return std::string(static_cast<WIN32_FIND_DATA *>(m_findData)->cFileName);
+ return UTF8Convert_WtoUTF8(std::wstring(static_cast<WIN32_FIND_DATA *>(m_findData)->cFileName));
}
void LocalFileIterator::next()
@@ -97,7 +98,7 @@ void LocalFileIterator::readNextEntry()
{
// Open directory
std::string query = FilePath(m_path).fullPath() + "/*";
- m_findHandle = FindFirstFileA(query.c_str(), static_cast<WIN32_FIND_DATA *>(m_findData));
+ m_findHandle = FindFirstFile(UTF8Handler::utf8_to_wstring(query).c_str(), static_cast<WIN32_FIND_DATA *>(m_findData));
// Abort if directory could not be opened
if (m_findHandle == INVALID_HANDLE_VALUE)
@@ -122,7 +123,7 @@ void LocalFileIterator::readNextEntry()
m_index++;
// Get filename
- filename = std::string(static_cast<WIN32_FIND_DATA *>(m_findData)->cFileName);
+ filename = UTF8Handler::wstring_to_utf8(std::wstring(static_cast<WIN32_FIND_DATA *>(m_findData)->cFileName));
} while (filename == ".." || filename == ".");
}
diff --git a/source/cppfs/source/windows/LocalFileWatcher.cpp b/source/cppfs/source/windows/LocalFileWatcher.cpp
index b9d6664..c0d69c7 100644
--- a/source/cppfs/source/windows/LocalFileWatcher.cpp
+++ b/source/cppfs/source/windows/LocalFileWatcher.cpp
@@ -5,7 +5,7 @@
#include <cppfs/FilePath.h>
#include <cppfs/windows/LocalFileSystem.h>
-
+#include <cppfs/windows/UTF8Handler.h>
namespace
{
@@ -64,8 +64,8 @@ AbstractFileSystem * LocalFileWatcher::fs() const
void LocalFileWatcher::add(FileHandle & dir, unsigned int events, RecursiveMode recursive)
{
// Open directory
- ::HANDLE dirHandle = ::CreateFileA(
- dir.path().c_str(), // Pointer to the directory name
+ ::HANDLE dirHandle = ::CreateFile(
+ UTF8Convert_UTF8toW(dir.path()).c_str(), // Pointer to the directory name
FILE_LIST_DIRECTORY, // Access (read/write) mode
FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, // File share mode
NULL, // Security descriptor
@@ -189,50 +189,36 @@ void LocalFileWatcher::watch(int timeout)
// Get file event notification
FILE_NOTIFY_INFORMATION * fileEvent = reinterpret_cast<FILE_NOTIFY_INFORMATION*>(entry);
- // Convert filename to 8-bit character string
- char fileName[4096];
- int numBytes = ::WideCharToMultiByte(CP_ACP,
- 0,
- fileEvent->FileName,
- fileEvent->FileNameLength / sizeof(WCHAR),
- fileName,
- sizeof(fileName),
- NULL,
- NULL);
-
- // Check if conversion was successful
- if (numBytes != 0) {
- // Get filename in unified format
- std::string fname = FilePath(std::string(fileName, fileName + numBytes)).path();
-
- // Determine event type
- FileEvent eventType = (FileEvent)0;
- switch (fileEvent->Action) {
- case FILE_ACTION_ADDED:
- eventType = FileCreated;
- break;
-
- case FILE_ACTION_REMOVED:
- eventType = FileRemoved;
- break;
-
- case FILE_ACTION_MODIFIED:
- case FILE_ACTION_RENAMED_NEW_NAME:
- eventType = FileModified;
- break;
+ // Get filename in unified format
+ std::string fname = FilePath(UTF8Handler::wstring_to_utf8(fileEvent->FileName)).path();
+
+ // Determine event type
+ FileEvent eventType = (FileEvent)0;
+ switch (fileEvent->Action) {
+ case FILE_ACTION_ADDED:
+ eventType = FileCreated;
+ break;
+
+ case FILE_ACTION_REMOVED:
+ eventType = FileRemoved;
+ break;
+
+ case FILE_ACTION_MODIFIED:
+ case FILE_ACTION_RENAMED_NEW_NAME:
+ eventType = FileModified;
+ break;
- default:
- break;
- }
-
- // Check if event is watched for
- if (watcher.events & eventType) {
- // Get file handle
- FileHandle fh = watcher.dir.open(fname);
-
- // Invoke callback function
- onFileEvent(fh, eventType);
- }
+ default:
+ break;
+ }
+
+ // Check if event is watched for
+ if (watcher.events & eventType) {
+ // Get file handle
+ FileHandle fh = watcher.dir.open(fname);
+
+ // Invoke callback function
+ onFileEvent(fh, eventType);
}
// Get next event
diff --git a/source/cppfs/source/windows/UTF8Handler.cpp b/source/cppfs/source/windows/UTF8Handler.cpp
new file mode 100644
index 0000000..be11d58
--- /dev/null
+++ b/source/cppfs/source/windows/UTF8Handler.cpp
@@ -0,0 +1,50 @@
+
+#include <cppfs/windows/UTF8Handler.h>
+#include <windows.h>
+#include <stdexcept>
+
+
+namespace cppfs
+{
+ namespace UTF8Handler
+ {
+ std::wstring utf8_to_wstring(const std::string& utf8_str) {
+ if (utf8_str.empty()) {
+ return std::wstring();
+ }
+
+ const int wstr_len = MultiByteToWideChar(CP_UTF8, 0, utf8_str.c_str(), -1, nullptr, 0);
+ if (wstr_len == 0) {
+ throw std::runtime_error("Error converting UTF-8 string to UTF-16.");
+ }
+
+ std::wstring wstr(wstr_len, L'\0');
+
+ if (MultiByteToWideChar(CP_UTF8, 0, utf8_str.c_str(), -1, &wstr[0], wstr_len) == 0) {
+ throw std::runtime_error("Error converting UTF-8 string to UTF-16.");
+ }
+ wstr.resize(static_cast<size_t>(wstr_len) - 1); // -1 because MultiByteToWideChar includes the null terminator in its count
+
+ return wstr;
+ }
+
+ std::string wstring_to_utf8(const std::wstring& wstr) {
+ if (wstr.empty()) {
+ return std::string();
+ }
+
+ const int utf8_len = WideCharToMultiByte(CP_UTF8, 0, wstr.c_str(), -1, nullptr, 0, nullptr, nullptr);
+ if (utf8_len == 0) {
+ throw std::runtime_error("Error converting UTF-16 string to UTF-8.");
+ }
+
+ std::string utf8_str(utf8_len, '\0');
+ if (WideCharToMultiByte(CP_UTF8, 0, wstr.c_str(), -1, &utf8_str[0], utf8_len, nullptr, nullptr) == 0) {
+ throw std::runtime_error("Error converting UTF-16 string to UTF-8.");
+ }
+ utf8_str.resize(static_cast<size_t>(utf8_len) - 1); // -1 because WideCharToMultiByte includes the null terminator in its count
+
+ return utf8_str;
+ }
+ }
+} // namespace cppfs