Windows Unicode I/O

.travis.yml

@@ -35,7 +35,8 @@ script:
     if [[ "${TRAVIS_OS_NAME}" == "windows" ]]; then
       echo "Running make.bat..."
       ./make.bat
-      ./v.exe -o v2.exe compiler
+      ./v.exe -debug -o v2.exe compiler
+      gcc -Ithirdparty/microsoft_craziness -DUNICODE -D_UNICODE .v2.exe.c thirdparty/microsoft_craziness/microsoft_craziness.c -o v2.exe -lole32 -ladvapi32 -loleaut32
       ./v2.exe -o v3.exe compiler
     fi
   - |

compiler/main.v

@@ -193,6 +193,9 @@ fn (v mut V) compile() {
 #ifdef _WIN32
 #define WIN32_LEAN_AND_MEAN
 #include <windows.h>
+#include <io.h> // _waccess
+#include <fcntl.h> // _O_U8TEXT
+#include <direct.h> // _wgetcwd
 //#include <WinSock2.h>
 #ifdef _MSC_VER
 // On MSVC these are the same (as long as /volatile:ms is passed)
@@ -385,7 +388,7 @@ string _STR_TMP(const char *fmt, ...) {
 			// It can be skipped in single file programs
 			if v.pref.is_script {
 				//println('Generating main()...')
-				cgen.genln('int main() { init_consts(); $cgen.fn_main; return 0; }')
+				cgen.genln('int main() { \n#ifdef _WIN32\n _setmode(_fileno(stdout), _O_U8TEXT); \n#endif\n init_consts(); $cgen.fn_main; return 0; }')
 			}
 			else {
 				println('panic: function `main` is undeclared in the main module')
@@ -394,7 +397,7 @@ string _STR_TMP(const char *fmt, ...) {
 		}
 		// Generate `main` which calls every single test function
 		else if v.pref.is_test {
-			cgen.genln('int main() { init_consts();')
+			cgen.genln('int main() { \n#ifdef _WIN32\n _setmode(_fileno(stdout), _O_U8TEXT); \n#endif\n init_consts();')
 			for key, f in v.table.fns { 
 				if f.name.starts_with('test_') {
 					cgen.genln('$f.name();')
@@ -622,7 +625,7 @@ fn (c &V) cc_windows_cross() {
                obj_name = obj_name.replace('.exe', '')
                obj_name = obj_name.replace('.o.o', '.o')
                mut include := '-I $winroot/include '
-               cmd := 'clang -o $obj_name -w $include -m32 -c -target x86_64-win32 $ModPath/$c.out_name_c'
+               cmd := 'clang -o $obj_name -w $include -DUNICODE -D_UNICODE -m32 -c -target x86_64-win32 $ModPath/$c.out_name_c'
                if c.pref.show_c_cmd {
                        println(cmd)
                }
@@ -757,6 +760,9 @@ mut args := ''
 			a << ' -ldl ' 
 		} 
 	}
+	if v.os == .windows {
+		a << '-DUNICODE -D_UNICODE'
+	}
 	// Find clang executable
 	//fast_clang := '/usr/local/Cellar/llvm/8.0.0/bin/clang'
 	args := a.join(' ')

vlib/builtin/builtin.v

@@ -37,8 +37,12 @@ pub fn println(s string) {
 	if isnil(s.str) {
 		panic('println(NIL)')
 	}
+	$if windows {
+		C._putws(s.to_wide())
+	} $else {
 		C.printf('%.*s\n', s.len, s.str)
 	}
+}
 
 pub fn eprintln(s string) {
 	if isnil(s.str) {
@@ -54,8 +58,12 @@ pub fn eprintln(s string) {
 }
 
 pub fn print(s string) {
+	$if windows {
+		C.wprintf(s.to_wide())
+	} $else {
 		C.printf('%.*s', s.len, s.str)
 	}
+}
 
 __global total_m i64 = 0
 pub fn malloc(n int) byteptr {

vlib/builtin/utf8.v

@@ -91,3 +91,43 @@ pub fn (_rune string) utf32_code() int {
 	return res
 }
 
+const (
+		CP_UTF8 = 65001
+	)
+
+pub fn (_str string) to_wide() &u16 {
+    $if windows {
+   	    num_chars := int(C.MultiByteToWideChar(CP_UTF8, 0, _str.str, _str.len, 0, 0))
+	    mut wstr := &u16(malloc((num_chars + 1) * 2)) // sizeof(wchar_t)
+	    if wstr > 0 {
+		    C.MultiByteToWideChar(CP_UTF8, 0, _str.str, _str.len, wstr, num_chars)
+		    C.memset(&byte(wstr) + num_chars * 2, 0, 2)
+	    }
+	    return wstr
+    } $else {
+        return 0
+    }
+}
+
+pub fn string_from_wide(_wstr &u16) string {
+    $if windows {
+	    wstr_len := int(C.wcslen(_wstr))
+	    return string_from_wide2(_wstr, wstr_len)
+    } $else {
+        return ''
+    }
+}
+
+pub fn string_from_wide2(_wstr &u16, len int) string {
+    $if windows {
+    	num_chars := int(C.WideCharToMultiByte(CP_UTF8, 0, _wstr, len, 0, 0, 0, 0))
+    	mut str_to := &byte(malloc(num_chars + 1))
+    	if str_to > 0 {
+    		C.WideCharToMultiByte(CP_UTF8, 0, _wstr, len, str_to, num_chars, 0, 0)
+		    C.memset(&byte(str_to) + num_chars, 0, 1)
+	    }
+	    return tos2(str_to)
+    } $else {
+        return ''
+    }
+}

vlib/math/math.v

@@ -4,6 +4,7 @@
 
 module math
 
+#include <math.h>
 
 // NOTE
 // When adding a new function, please make sure it's in the right place.

vlib/os/os.v

@@ -35,7 +35,8 @@ import const (
 	INVALID_FILE_ATTRIBUTES
 )
 
-struct FILE {
+struct C.FILE {
+	
 }
 
 struct File {
@@ -105,8 +106,14 @@ fn parse_windows_cmd_line(cmd byteptr) []string {
 
 // read_file reads the file in `path` and returns the contents.
 pub fn read_file(path string) ?string { 
-	mut mode := 'rb' 
+	mode := 'rb'
-	fp := C.fopen(path.str, mode.str) 
+	mut fp := &C.FILE{}
+	$if windows {
+		fp = C._wfopen(path.to_wide(), mode.to_wide())
+	} $else {
+		cpath := path.str
+		fp = C.fopen(cpath, mode.str) 
+	}
 	if isnil(fp) {
 		return error('failed to open file "$path"')
 	}
@@ -123,21 +130,36 @@ pub fn read_file(path string) ?string {
 
 // file_size returns the size of the file located in `path`.
 pub fn file_size(path string) int {
-	s := C.stat{}
+	mut s := C.stat{}
+	$if windows {
+		C._wstat(path.to_wide(), &s)
+	} $else {
 		C.stat(path.str, &s)
+	}
 	return s.st_size
 }
 
 pub fn mv(old, new string) {
+	$if windows {
+		C._wrename(old.to_wide(), new.to_wide())
+	} $else {
 		C.rename(old.str, new.str)
 	}
+}
 
 // read_lines reads the file in `path` into an array of lines.
 // TODO return `?[]string` TODO implement `?[]` support
 pub fn read_lines(path string) []string {
 	mut res := []string
 	mut buf := [1000]byte
-	fp := C.fopen(path.str, 'rb')
+	mode := 'rb'
+	mut fp := &C.FILE{}
+	$if windows {
+		fp = C._wfopen(path.to_wide(), mode.to_wide())
+	} $else {
+		cpath := path.str
+		fp = C.fopen(cpath, mode.str) 
+	}
 	if isnil(fp) {
 		// TODO
 		// return error('failed to open file "$path"')
@@ -169,8 +191,18 @@ fn read_ulines(path string) []ustring {
 }
 
 pub fn open(path string) ?File {
-	file := File {
+	mut file := File{}
-		cfile: C.fopen(path.str, 'rb') 
+	$if windows {
+		wpath := path.to_wide()
+		mode := 'rb'
+		file = File {			
+			cfile: C._wfopen(wpath, mode.to_wide())
+		}
+	} $else {
+		cpath := path.str 
+		file = File {
+			cfile: C.fopen(cpath, 'rb') 
+		}
 	}
 	if isnil(file.cfile) {
 		return error('failed to open file "$path"')
@@ -180,8 +212,18 @@ pub fn open(path string) ?File {
 
 // create creates a file at a specified location and returns a writable `File` object.
 pub fn create(path string) ?File {
-	file := File {
+	mut file := File{}
-		cfile: C.fopen(path.str, 'wb') 
+	$if windows {
+		wpath := path.replace('/', '\\').to_wide()
+		mode := 'wb'
+		file = File {			
+			cfile: C._wfopen(wpath, mode.to_wide())
+		}
+	} $else {
+		cpath := path.str 
+		file = File {
+			cfile: C.fopen(cpath, 'wb') 
+		}
 	}
 	if isnil(file.cfile) {
 		return error('failed to create file "$path"')
@@ -190,11 +232,21 @@ pub fn create(path string) ?File {
 }
 
 pub fn open_append(path string) ?File {
-	file := File {
+	mut file := File{}
-		cfile: C.fopen(path.str, 'ab') 
+	$if windows {
+		wpath := path.replace('/', '\\').to_wide()
+		mode := 'ab'
+		file = File {			
+			cfile: C._wfopen(wpath, mode.to_wide())
+		}
+	} $else {
+		cpath := path.str 
+		file = File {
+			cfile: C.fopen(cpath, 'ab') 
+		}
 	}
 	if isnil(file.cfile) {
-		return error('failed to create file "$path"')
+		return error('failed to create(append) file "$path"')
 	}
 	return file 
 }
@@ -238,7 +290,12 @@ pub fn (f File) close() {
 
 // system starts the specified command, waits for it to complete, and returns its code.
 pub fn system(cmd string) int {
-	ret := C.system(cmd.str) 
+	mut ret := int(0)
+	$if windows {
+		ret = C._wsystem(cmd.to_wide())
+	} $else {
+		ret = C.system(cmd.str) 
+	}
 	if ret == -1 {
 		os.print_c_errno()
 	}
@@ -246,11 +303,13 @@ pub fn system(cmd string) int {
 }
 
 fn popen(path string) *FILE {
-	cpath := path.str
 	$if windows {
-		return C._popen(cpath, 'r')
+		mode := 'rb'
+		wpath := path.to_wide()
+		return C._wpopen(wpath, mode.to_wide())
 	}
 	$else {
+		cpath := path.str
 		return C.popen(cpath, 'r')
 	}
 }
@@ -274,12 +333,20 @@ pub fn exec(cmd string) string {
 
 // `getenv` returns the value of the environment variable named by the key.
 pub fn getenv(key string) string {	
+	$if windows {
+		s := C._wgetenv(key.to_wide())
+		if isnil(s) {
+			return ''
+		}
+		return string_from_wide(s)
+	} $else {
 		s := C.getenv(key.str)
 		if isnil(s) {
 			return ''
 		}
 		return string(s)
 	}
+}
 
 pub fn setenv(name string, value string, overwrite bool) int {
 	$if windows {
@@ -310,14 +377,17 @@ pub fn unsetenv(name string) int {
 // `file_exists` returns true if `path` exists.
 pub fn file_exists(path string) bool {
 	$if windows {
-		return C._access( path.str, 0 ) != -1
+		path = path.replace('/', '\\')
-	}
+		return C._waccess( path.to_wide(), 0 ) != -1
+	} $else {
 		return C.access( path.str, 0 ) != -1
 	}
+}
 
 pub fn dir_exists(path string) bool {
 	$if windows {
-		attr := int(C.GetFileAttributes(path.str)) 
+		path = path.replace('/', '\\')
+		attr := int(C.GetFileAttributes(path.to_wide()))
 		if attr == INVALID_FILE_ATTRIBUTES {
 			return false
 		}
@@ -345,7 +415,7 @@ pub fn mkdir(path string) {
 		if path.last_index('\\') != -1 {
 			mkdir(path.all_before_last('\\'))
 		}
-		C.CreateDirectory(path.str, 0)
+		C.CreateDirectory(path.to_wide(), 0)
 	}
 	$else {
 		C.mkdir(path.str, 511)// S_IRWXU | S_IRWXG | S_IRWXO
@@ -354,8 +424,13 @@ pub fn mkdir(path string) {
 
 // rm removes file in `path`.
 pub fn rm(path string) {
+	$if windows {
+		C._wremove(path.to_wide())
+	}
+	$else {
 		C.remove(path.str)
-	// C.unlink(path.str)
+	}
+	// C.unlink(path.cstr())
 }
 
 
@@ -365,7 +440,7 @@ pub fn rmdir(path string) {
 		C.rmdir(path.str)		
 	}
 	$else {
-		C.RemoveDirectoryA(path.str)
+		C.RemoveDirectory(path.to_wide())
 	}
 }
 
@@ -435,20 +510,18 @@ pub fn get_line() string {
 // get_raw_line returns a one-line string from stdin along with '\n' if there is any
 pub fn get_raw_line() string {
 	$if windows {
-		max := 256
+		max := 512 // MAX_PATH * sizeof(wchar_t)
-		buf := malloc(max)
+		buf := &u16(malloc(max))
 		h_input := C.GetStdHandle(STD_INPUT_HANDLE)
 		if h_input == INVALID_HANDLE_VALUE {
 			panic('get_raw_line() error getting input handle.')
 		}
-		nr_chars := 0
+		mut nr_chars := 0
-		// NOTE: Once we have UTF8 encode function to
-		// convert utf16 to utf8, change to ReadConsoleW
 		C.ReadConsole(h_input, buf, max, &nr_chars, 0)
 		if nr_chars == 0 {
 			return ''
 		}
-		return tos(buf, nr_chars)
+		return string_from_wide2(buf, nr_chars)
 	}
 	$else {
 		//u64 is used because C.getline needs a size_t as second argument
@@ -539,8 +612,8 @@ fn on_segfault(f voidptr) {
 }
 
 pub fn executable() string {
-	mut result := malloc(MAX_PATH) 
 	$if linux {
+		mut result := malloc(MAX_PATH)
 		count := int(C.readlink('/proc/self/exe', result, MAX_PATH ))
 		if count < 0 {
 			panic('error reading /proc/self/exe to get exe path')
@@ -548,10 +621,12 @@ pub fn executable() string {
 		return string(result, count)
 	}
 	$if windows {
-		ret := int(C.GetModuleFileName( 0, result, MAX_PATH ))
+		mut result := &u16(malloc(512)) // MAX_PATH * sizeof(wchar_t)
-		return string( result, ret)
+		len := int(C.GetModuleFileName( 0, result, MAX_PATH ))
+		return string_from_wide2(result, len)
 	}
 	$if mac {
+		mut result := malloc(MAX_PATH)
 		pid := C.getpid() 
 		ret := C.proc_pidpath (pid, result, MAX_PATH) 
 		if ret <= 0  {
@@ -561,6 +636,7 @@ pub fn executable() string {
 		return string(result) 
 	}
 	$if freebsd {
+		mut result := malloc(MAX_PATH)
 		mut mib := [1 /* CTL_KERN */, 14 /* KERN_PROC */, 12 /* KERN_PROC_PATHNAME */, -1]!! 
 		size := MAX_PATH 
 		C.sysctl(mib, 4, result, &size, 0, 0) 
@@ -572,6 +648,7 @@ pub fn executable() string {
 		return os.args[0] 
 	} 
 	$if netbsd {
+		mut result := malloc(MAX_PATH)
 		count := int(C.readlink('/proc/curproc/exe', result, MAX_PATH ))
 		if count < 0 {
 			panic('error reading /proc/curproc/exe to get exe path')
@@ -579,6 +656,7 @@ pub fn executable() string {
 		return string(result, count)
 	} 
 	$if dragonfly {
+		mut result := malloc(MAX_PATH)
 		count := int(C.readlink('/proc/curproc/file', result, MAX_PATH ))
 		if count < 0 {
 			panic('error reading /proc/curproc/file to get exe path')
@@ -607,7 +685,7 @@ pub fn is_dir(path string) bool {
 
 pub fn chdir(path string) {
 	$if windows {
-		C._chdir(path.str)
+		C._wchdir(path.to_wide())
 	}
 	$else { 
 		C.chdir(path.str)
@@ -615,19 +693,22 @@ pub fn chdir(path string) {
 }
 
 pub fn getwd() string {	
-	buf := malloc(512)
 	$if windows {
-		if C._getcwd(buf, 512) == 0 {
+		max := 512 // MAX_PATH * sizeof(wchar_t)
+		buf := &u16(malloc(max))
+		if C._wgetcwd(buf, max) == 0 {
 			return ''
 		}
+		return string_from_wide(buf)
 	}
 	$else {
+		buf := malloc(512) 
 		if C.getcwd(buf, 512) == 0 {
 			return ''
 		}
-	} 
 		return string(buf)
 	}
+}
 
 // win: FILETIME
 // https://docs.microsoft.com/en-us/windows/win32/api/minwinbase/ns-minwinbase-filetime
@@ -637,7 +718,7 @@ struct filetime {
 }
 
 // win: WIN32_FIND_DATA
-// https://docs.microsoft.com/en-us/windows/win32/api/minwinbase/ns-minwinbase-_win32_find_dataa
+// https://docs.microsoft.com/en-us/windows/win32/api/minwinbase/ns-minwinbase-_win32_find_dataw
 struct win32finddata {
 mut:
     dwFileAttributes u32
@@ -674,13 +755,13 @@ pub fn ls(path string) []string {
 		path_files := '$path\\*' 
 		// NOTE:TODO: once we have a way to convert utf16 wide character to utf8
 		// we should use FindFirstFileW and FindNextFileW
-		h_find_files := C.FindFirstFile(path_files.str, &find_file_data)
+		h_find_files := C.FindFirstFile(path_files.to_wide(), &find_file_data)
-		first_filename := tos(&find_file_data.cFileName, strlen(find_file_data.cFileName))
+		first_filename := string_from_wide(&u16(find_file_data.cFileName))
 		if first_filename != '.' && first_filename != '..' {
 			dir_files << first_filename
 		}
 		for C.FindNextFile(h_find_files, &find_file_data) {
-			filename := tos(&find_file_data.cFileName, strlen(find_file_data.cFileName))
+			filename := string_from_wide(&u16(find_file_data.cFileName))
 			if filename != '.' && filename != '..' {
 				dir_files << filename.clone()
 			}

vlib/os/os_win.v

@@ -15,7 +15,7 @@ type HANDLE voidptr
 // get_file_handle retrieves the operating-system file handle that is associated with the specified file descriptor.
 pub fn get_file_handle(path string) HANDLE {
     mode := 'rb'
-    _fd := C.fopen(path.str, mode.str)
+    _fd := C._wfopen(path.to_wide(), mode.to_wide())
     if _fd == 0 {
 	    return HANDLE(INVALID_HANDLE_VALUE)
     }
@@ -27,17 +27,14 @@ pub fn get_file_handle(path string) HANDLE {
 // get_module_filename retrieves the fully qualified path for the file that contains the specified module. 
 // The module must have been loaded by the current process.
 pub fn get_module_filename(handle HANDLE) ?string {
-    mut sz := int(1024) // Optimized length 
+    mut sz := int(4096) // Optimized length 
-    mut buf := [byte(0); sz] // Not work for GetModuleFileNameW :(
+    mut buf := &u16(malloc(4096))
     for {
         status := C.GetModuleFileName(handle, &buf, sz)
         switch status {
         case SUCCESS:
-            _filename := tos(buf.data, sz)
+            _filename := string_from_wide2(buf, sz)
             return _filename
-        case ERROR_INSUFFICIENT_BUFFER:
-            sz += 1024 // increment buffer cluster by 1024
-            buf = [byte(0); sz] // clear buffer
         default:
             // Must handled with GetLastError and converted by FormatMessage
             return error('Cannot get file name from handle.')