cannam@154: #if defined(_WIN32)
cannam@154: # include <stdio.h>
cannam@154: # include <stdlib.h>
cannam@154: # include <wchar.h>
cannam@154: /*We need the following two to set stdin/stdout to binary.*/
cannam@154: # include <io.h>
cannam@154: # include <fcntl.h>
cannam@154: # define WIN32_LEAN_AND_MEAN
cannam@154: # define WIN32_EXTRA_LEAN
cannam@154: # include <windows.h>
cannam@154: # include "win32utf8.h"
cannam@154: 
cannam@154: static char *utf16_to_utf8(const wchar_t *_src){
cannam@154:   char   *dst;
cannam@154:   size_t  len;
cannam@154:   size_t  si;
cannam@154:   size_t  di;
cannam@154:   len=wcslen(_src);
cannam@154:   dst=(char *)malloc(sizeof(*dst)*(3*len+1));
cannam@154:   if(dst==NULL)return dst;
cannam@154:   for(di=si=0;si<len;si++){
cannam@154:     unsigned c0;
cannam@154:     c0=_src[si];
cannam@154:     if(c0<0x80){
cannam@154:       /*Can be represented by a 1-byte sequence.*/
cannam@154:       dst[di++]=(char)c0;
cannam@154:       continue;
cannam@154:     }
cannam@154:     else if(c0<0x800){
cannam@154:       /*Can be represented by a 2-byte sequence.*/
cannam@154:       dst[di++]=(char)(0xC0|c0>>6);
cannam@154:       dst[di++]=(char)(0x80|c0&0x3F);
cannam@154:       continue;
cannam@154:     }
cannam@154:     else if(c0>=0xD800&&c0<0xDC00){
cannam@154:       unsigned c1;
cannam@154:       /*This is safe, because c0 was not 0 and _src is NUL-terminated.*/
cannam@154:       c1=_src[si+1];
cannam@154:       if(c1>=0xDC00&&c1<0xE000){
cannam@154:         unsigned w;
cannam@154:         /*Surrogate pair.*/
cannam@154:         w=((c0&0x3FF)<<10|c1&0x3FF)+0x10000;
cannam@154:         /*Can be represented by a 4-byte sequence.*/
cannam@154:         dst[di++]=(char)(0xF0|w>>18);
cannam@154:         dst[di++]=(char)(0x80|w>>12&0x3F);
cannam@154:         dst[di++]=(char)(0x80|w>>6&0x3F);
cannam@154:         dst[di++]=(char)(0x80|w&0x3F);
cannam@154:         si++;
cannam@154:         continue;
cannam@154:       }
cannam@154:     }
cannam@154:     /*Anything else is either a valid 3-byte sequence, an invalid surrogate
cannam@154:        pair, or 'not a character'.
cannam@154:       In the latter two cases, we just encode the value as a 3-byte
cannam@154:        sequence anyway (producing technically invalid UTF-8).
cannam@154:       Later error handling will detect the problem, with a better
cannam@154:        chance of giving a useful error message.*/
cannam@154:     dst[di++]=(char)(0xE0|c0>>12);
cannam@154:     dst[di++]=(char)(0x80|c0>>6&0x3F);
cannam@154:     dst[di++]=(char)(0x80|c0&0x3F);
cannam@154:   }
cannam@154:   dst[di++]='\0';
cannam@154:   return dst;
cannam@154: }
cannam@154: 
cannam@154: typedef LPWSTR *(APIENTRY *command_line_to_argv_w_func)(LPCWSTR cmd_line,
cannam@154:  int *num_args);
cannam@154: 
cannam@154: /*Make a best-effort attempt to support UTF-8 on Windows.*/
cannam@154: void win32_utf8_setup(int *_argc,const char ***_argv){
cannam@154:   HMODULE hlib;
cannam@154:   /*We need to set stdin/stdout to binary mode.
cannam@154:     This is unrelated to UTF-8 support, but it's platform specific and we need
cannam@154:      to do it in the same places.*/
cannam@154:   _setmode(_fileno(stdin),_O_BINARY);
cannam@154:   _setmode(_fileno(stdout),_O_BINARY);
cannam@154:   hlib=LoadLibraryA("shell32.dll");
cannam@154:   if(hlib!=NULL){
cannam@154:     command_line_to_argv_w_func command_line_to_argv_w;
cannam@154:     /*This function is only available on Windows 2000 or later.*/
cannam@154:     command_line_to_argv_w=(command_line_to_argv_w_func)GetProcAddress(hlib,
cannam@154:      "CommandLineToArgvW");
cannam@154:     if(command_line_to_argv_w!=NULL){
cannam@154:       wchar_t **argvw;
cannam@154:       int       argc;
cannam@154:       argvw=(*command_line_to_argv_w)(GetCommandLineW(),&argc);
cannam@154:       if(argvw!=NULL){
cannam@154:         int ai;
cannam@154:         /*Really, I don't see why argc would ever differ from *_argc, but let's
cannam@154:            be paranoid.*/
cannam@154:         if(argc>*_argc)argc=*_argc;
cannam@154:         for(ai=0;ai<argc;ai++){
cannam@154:           char *argv;
cannam@154:           argv=utf16_to_utf8(argvw[ai]);
cannam@154:           if(argv!=NULL)(*_argv)[ai]=argv;
cannam@154:         }
cannam@154:         *_argc=argc;
cannam@154:         LocalFree(argvw);
cannam@154:       }
cannam@154:     }
cannam@154:     FreeLibrary(hlib);
cannam@154:   }
cannam@154: # if defined(CP_UTF8)
cannam@154:   /*This does not work correctly in all environments (it breaks output in
cannam@154:      mingw32 for me), and requires a Unicode font (e.g., when using the default
cannam@154:      Raster font, even characters that are available in the font's codepage
cannam@154:      won't display properly).*/
cannam@154:   /*SetConsoleOutputCP(CP_UTF8);*/
cannam@154: # endif
cannam@154: }
cannam@154: #endif