tool+lib: handle non-ASCII filenames [Win]

If file name contains non-ASCII characters, the only reasonable way how
to handle it is to use wide charactes in MSW.

getopt for Win replaced with
<https://github.com/darealshinji/getopt-for-windows> supporting wchar_t.

refer to GH-93

Author: MartinPulec

CMakeLists.txt

@@ -148,6 +148,7 @@ message(STATUS "Configured options: ${COMPILED_OPTIONS}")
 file(GLOB H_FILES libgpujpeg/*.h ${CMAKE_CURRENT_BINARY_DIR}/libgpujpeg/gpujpeg_version.h)
 file(GLOB_RECURSE C_FILES src/*.c src/*.cu)
 list(REMOVE_ITEM C_FILES "${CMAKE_CURRENT_SOURCE_DIR}/src/main.c")
+list(REMOVE_ITEM C_FILES "${CMAKE_CURRENT_SOURCE_DIR}/src/utils/getopt.c")
 add_library(gpujpeg ${H_FILES} ${C_FILES})
 target_include_directories(${PROJECT_NAME}
     PUBLIC ${CMAKE_CUDA_TOOLKIT_INCLUDE_DIRECTORIES}
@@ -165,6 +166,10 @@ endif()
 
 # GPUJPEG commmand-line application
 add_executable(gpujpeg-bin src/main.c)
+if(NOT UNIX)
+    target_sources(gpujpeg-bin PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/src/utils/getopt.c")
+    target_compile_options(gpujpeg-bin PRIVATE $<$<COMPILE_LANGUAGE:C>:/D_UNICODE>)
+endif()
 target_compile_features(gpujpeg-bin PRIVATE ${NEEDED_COMPILER_FEATURES})
 set_property(TARGET gpujpeg-bin PROPERTY C_STANDARD 99)
 target_link_libraries(gpujpeg-bin gpujpeg)

NEWS.md

@@ -1,3 +1,8 @@
+next
+----------
+
+- support for non-ASCII characters in filename (tool+_load/_save)
+
 2025-05-23 - 0.27.4
 ----------
 

libgpujpeg/gpujpeg_common.h

@@ -400,6 +400,8 @@ gpujpeg_image_calculate_size(struct gpujpeg_image_parameters* param);
  *
  * Allocated image must be freed by gpujpeg_image_load_from_file().
  *
+ * Non-ASCII symbols in filename should be UTF-8 encoded.
+ *
  * @param         filaname    Image filename
  * @param[out]    image       Image data buffer allocated as CUDA host buffer
  * @param[in,out] image_size  Image data buffer size (can be specified for verification or 0 for retrieval)
@@ -411,6 +413,8 @@ gpujpeg_image_load_from_file(const char* filename, uint8_t** image, size_t* imag
 /**
  * Save image to a file
  *
+ * Non-ASCII symbols in filename should be UTF-8 encoded.
+ *
  * File format is deduced from the filename extension (if any).
  *
  * Use the extension "XXX" (eg. "image.XXX") to automatically select the

src/gpujpeg_common.c

@@ -108,6 +108,25 @@
             (name).started = 0;                                                                                        \
         } while ( 0 )
 
+#ifdef _WIN32
+static wchar_t*
+mbs_to_wstr_helper(const char* mbstr, wchar_t* wstr_buf, size_t wstr_len)
+{
+    const int size_needed = MultiByteToWideChar(CP_UTF8, 0, mbstr, -1, NULL, 0);
+    if (size_needed <= 0) {
+        ERROR_MSG("MultiByteToWideChar returned: %d (0x%x)!\n", size_needed, size_needed);
+        return NULL;
+    }
+    if (size_needed > (int) wstr_len) {
+        ERROR_MSG("buffer provided to %s too short - needed %d, got %zu!\n", __func__, size_needed, wstr_len);
+        return NULL;
+    }
+    MultiByteToWideChar(CP_UTF8, 0, mbstr, -1, wstr_buf, size_needed);
+    return wstr_buf;
+}
+#define mbs_to_wstr(tstr) mbs_to_wstr_helper(tstr, (wchar_t[1024]){0}, 1024)
+#endif
+
 #define PLANAR   1u
 static const struct {
     enum gpujpeg_pixel_format pixel_format;
@@ -1201,7 +1220,11 @@ gpujpeg_image_load_from_file(const char* filename, uint8_t** image, size_t* imag
     }
 
     FILE* file;
+#ifdef _WIN32
+    file = _wfopen(mbs_to_wstr(filename), L"rb");
+#else
     file = fopen(filename, "rb");
+#endif
     if ( !file ) {
         fprintf(stderr, "[GPUJPEG] [Error] Failed open %s for reading: %s\n", filename, strerror(errno));
         return -1;
@@ -1264,7 +1287,11 @@ gpujpeg_image_save_to_file(const char* filename, const uint8_t* image, size_t im
     }
 
     FILE* file;
+#ifdef _WIN32
+    file = _wfopen(mbs_to_wstr(filename), L"wb");
+#else
     file = fopen(filename, "wb");
+#endif
     if ( !file ) {
         fprintf(stderr, "[GPUJPEG] [Error] Failed open %s for writing: %s\n", filename, strerror(errno));
         return -1;