fixed non-ascii path handling on windows

2 files changed, 45 insertions, 26 deletions

diff --git a/filesystem-win.c b/filesystem-win.c
index a16dc6b..dc17eaf 100644
--- a/filesystem-win.c
+++ b/filesystem-win.c
@@ -13,7 +13,11 @@ static FsType windows_file_attributes_to_type(DWORD attrs) {
 }
 
 FsType fs_path_type(char const *path) {
-	return windows_file_attributes_to_type(GetFileAttributesA(path));
+	WCHAR wide_path[4100];
+	if (MultiByteToWideChar(CP_UTF8, 0, path, -1, wide_path, sizeof wide_path) == 0) {
+		return FS_NON_EXISTENT;
+	}
+	return windows_file_attributes_to_type(GetFileAttributesW(wide_path));
 }
 
 FsPermission fs_path_permission(char const *path) {
@@ -28,39 +32,46 @@ bool fs_file_exists(char const *path) {
 }
 
 FsDirectoryEntry **fs_list_directory(char const *dirname) {
-	char file_pattern[256] = {0};
+	char file_pattern[1000] = {0};
 	FsDirectoryEntry **files = NULL;
-	WIN32_FIND_DATA find_data;
+	WIN32_FIND_DATAW find_data;
 	HANDLE fhandle;
 	assert(*dirname);
 	sprintf_s(file_pattern, sizeof file_pattern, "%s%s*", dirname,
 		dirname[strlen(dirname) - 1] == PATH_SEPARATOR ? "" : PATH_SEPARATOR_STR);
-	fhandle = FindFirstFileA(file_pattern, &find_data);
+	wchar_t wide_pattern[1024] = {0};
+	MultiByteToWideChar(CP_UTF8, 0, file_pattern, -1, wide_pattern, sizeof wide_pattern - 1);
+	
+	fhandle = FindFirstFileW(wide_pattern, &find_data);
 	if (fhandle != INVALID_HANDLE_VALUE) {
 		// first, figure out number of files
 		int nfiles = 1, idx = 0;
-		while (FindNextFile(fhandle, &find_data))  {
+		while (FindNextFileW(fhandle, &find_data))  {
 			++nfiles;
 		}
 		FindClose(fhandle);
 		// now, fill out files array
 		files = calloc(nfiles + 1, sizeof *files);
 		if (files) {
-			fhandle = FindFirstFileA(file_pattern, &find_data);
+			fhandle = FindFirstFileW(wide_pattern, &find_data);
 			if (fhandle != INVALID_HANDLE_VALUE) {
 				do {
 					if (idx < nfiles) {
-						const char *filename = find_data.cFileName;
-						size_t len = strlen(filename);
-						FsDirectoryEntry *entry = calloc(1, sizeof *entry + len + 1);
+						LPWSTR wide_filename = find_data.cFileName;
+						size_t wide_len = wcslen(wide_filename);
+						size_t cap = 4 * wide_len + 4;
+						FsDirectoryEntry *entry = calloc(1, sizeof *entry + cap);
+						
+						
 						if (entry) {
+							if (WideCharToMultiByte(CP_UTF8, 0, wide_filename, (int)wide_len, entry->name, (int)cap - 1, NULL, NULL) == 0)
+								break;
 							DWORD attrs = find_data.dwFileAttributes;
 							entry->type = windows_file_attributes_to_type(attrs);
-							memcpy(entry->name, filename, len);
 							files[idx++] = entry;
 						} else break; // stop now
 					}
-				} while (FindNextFile(fhandle, &find_data));
+				} while (FindNextFileW(fhandle, &find_data));
 				FindClose(fhandle);
 			}
 		}
@@ -69,7 +80,11 @@ FsDirectoryEntry **fs_list_directory(char const *dirname) {
 }
 
 int fs_mkdir(char const *path) {
-	if (CreateDirectoryA(path, NULL)) {
+	WCHAR wide_path[4100];
+	if (MultiByteToWideChar(CP_UTF8, 0, path, -1, wide_path, sizeof wide_path) == 0)
+		return -1;
+	
+	if (CreateDirectoryW(wide_path, NULL)) {
 		// directory created successfully
 		return 1;
 	} else {
@@ -82,12 +97,10 @@ int fs_mkdir(char const *path) {
 
 int fs_get_cwd(char *buf, size_t buflen) {
 	assert(buf && buflen);
-	DWORD pathlen = GetCurrentDirectory((DWORD)buflen, buf);
-	if (pathlen == 0) {
-		return -1;
-	} else if (pathlen < buflen) { // it's confusing, but this is < and not <=
-		return 1;
-	} else {
+	wchar_t wide_path[4100];
+	DWORD wide_pathlen = GetCurrentDirectoryW(sizeof wide_path - 1, wide_path);
+	if (wide_pathlen == 0) return -1;
+	if (WideCharToMultiByte(CP_UTF8, 0, wide_path, (int)wide_pathlen, buf, (int)buflen, NULL, NULL) == 0)
 		return 0;
-	}
+	return 1;
 }
diff --git a/main.c b/main.c
index 2857869..791af49 100644
--- a/main.c
+++ b/main.c
@@ -245,17 +245,17 @@ INT WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance,
 	(void)hInstance; (void)hPrevInstance; (void)lpCmdLine; (void)nCmdShow;
 	int argc = 0;
     LPWSTR* wide_argv = CommandLineToArgvW(GetCommandLineW(), &argc);
-    char** argv = malloc(argc * sizeof *argv);
+    char** argv = calloc(argc + 1, sizeof *argv);
 	if (!argv) {
 		die("Out of memory.");
 	}
     for (int i = 0; i < argc; i++) {
         LPWSTR wide_arg = wide_argv[i];
         int len = (int)wcslen(wide_arg);
-        argv[i] = malloc(len + 1);
-		if (!argv[i]) die("Out of memory.");
-        for (int j = 0; j <= len; j++)
-            argv[i][j] = (char)wide_arg[j];
+        int bufsz = len * 4 + 8;
+        argv[i] = calloc((size_t)bufsz, 1);
+	if (!argv[i]) die("Out of memory.");
+	WideCharToMultiByte(CP_UTF8, 0, wide_arg, len, argv[i], bufsz - 1, NULL, NULL);
     }
     LocalFree(wide_argv);
 #else
@@ -281,7 +281,11 @@ int main(int argc, char **argv) {
 	#error "Unrecognized operating system."
 #endif
 	
-	setlocale(LC_ALL, ""); // allow unicode
+	#if _WIN32
+	setlocale(LC_ALL, ".65001");
+	#else
+	setlocale(LC_ALL, "C.UTF-8");
+	#endif
 	
 	// read command-line arguments
 	char const *starting_filename = NULL;
@@ -335,8 +339,10 @@ int main(int argc, char **argv) {
 		}
 		
 		// on Windows, the global data directory is just the directory where the executable is.
+		WCHAR executable_wide_path[TED_PATH_MAX] = {0};
 		char executable_path[TED_PATH_MAX] = {0};
-		if (GetModuleFileNameA(NULL, executable_path, sizeof executable_path) > 0) {
+		if (GetModuleFileNameW(NULL, executable_wide_path, sizeof executable_wide_path - 1) > 0) {
+			WideCharToMultiByte(CP_UTF8, 0, executable_wide_path, -1, executable_path, sizeof executable_path, NULL, NULL);
 			char *last_backslash = strrchr(executable_path, '\\');
 			if (last_backslash) {
 				*last_backslash = '\0';