Windows Unicode I/O

pull/1257/head
vitalyster 2019-07-21 14:29:32 +03:00 committed by Alexander Medvednikov
parent 6e6f6bc387
commit 3e005074a3
6 changed files with 199 additions and 70 deletions

View File

@ -200,7 +200,11 @@ 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>
#endif
@ -360,7 +364,7 @@ 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(); $cgen.fn_main; return 0; }')
}
else {
println('panic: function `main` is undeclared in the main module')
@ -369,7 +373,7 @@ string _STR_TMP(const char *fmt, ...) {
}
// Generate `main` which calls every single test function
else if v.pref.is_test {
cgen.genln('int main() { init_consts();')
cgen.genln('int main() { \n#ifdef _WIN32\n _setmode(_fileno(stdout), _O_U8TEXT); \n#endif\n init_consts();')
for key, f in v.table.fns {
if f.name.starts_with('test_') {
cgen.genln('$f.name();')
@ -500,7 +504,7 @@ 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)
}
@ -631,6 +635,9 @@ 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(' ')

View File

@ -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 {

View File

@ -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 ''
}
}

View File

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

View File

@ -35,7 +35,8 @@ import const (
INVALID_FILE_ATTRIBUTES
)
struct FILE {
struct C.FILE {
}
struct File {
@ -105,9 +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'
mut fp := &C.FILE{}
$if windows {
fp = C._wfopen(path.to_wide(), mode.to_wide())
} $else {
cpath := path.cstr()
fp := C.fopen(cpath, mode.cstr())
fp = C.fopen(cpath, mode.cstr())
}
if isnil(fp) {
return error('failed to open file "$path"')
}
@ -124,22 +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.cstr(), new.cstr())
}
}
// 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
mode := 'rb'
mut fp := &C.FILE{}
$if windows {
fp = C._wfopen(path.to_wide(), mode.to_wide())
} $else {
cpath := path.cstr()
fp := C.fopen(cpath, 'rb')
fp = C.fopen(cpath, mode.cstr())
}
if isnil(fp) {
// TODO
// return error('failed to open file "$path"')
@ -171,10 +191,19 @@ fn read_ulines(path string) []ustring {
}
pub fn open(path string) ?File {
mut file := File{}
$if windows {
wpath := path.to_wide()
mode := 'rb'
file = File {
cfile: C._wfopen(wpath, mode.to_wide())
}
} $else {
cpath := path.cstr()
file := File {
file = File {
cfile: C.fopen(cpath, 'rb')
}
}
if isnil(file.cfile) {
return error('failed to open file "$path"')
}
@ -183,10 +212,19 @@ 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 {
mut file := File{}
$if windows {
wpath := path.replace('/', '\\').to_wide()
mode := 'wb'
file = File {
cfile: C._wfopen(wpath, mode.to_wide())
}
} $else {
cpath := path.cstr()
file := File {
file = File {
cfile: C.fopen(cpath, 'wb')
}
}
if isnil(file.cfile) {
return error('failed to create file "$path"')
}
@ -194,12 +232,21 @@ pub fn create(path string) ?File {
}
pub fn open_append(path string) ?File {
mut file := File{}
$if windows {
wpath := path.replace('/', '\\').to_wide()
mode := 'ab'
file = File {
cfile: C._wfopen(wpath, mode.to_wide())
}
} $else {
cpath := path.cstr()
file := File {
file = File {
cfile: C.fopen(cpath, 'ab')
}
}
if isnil(file.cfile) {
return error('failed to create file "$path"')
return error('failed to create(append) file "$path"')
}
return file
}
@ -243,7 +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.cstr())
mut ret := int(0)
$if windows {
ret = C._wsystem(cmd.to_wide())
} $else {
ret = C.system(cmd.cstr())
}
if ret == -1 {
os.print_c_errno()
}
@ -251,11 +303,13 @@ pub fn system(cmd string) int {
}
fn popen(path string) *FILE {
cpath := path.cstr()
$if windows {
return C._popen(cpath, 'r')
mode := string('rb')
wpath := path.to_wide()
return C._wpopen(wpath, mode.to_wide())
}
$else {
cpath := path.cstr()
return C.popen(cpath, 'r')
}
}
@ -279,12 +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.cstr())
if isnil(s) {
return ''
}
return string(s)
}
}
pub fn setenv(name string, value string, overwrite bool) int {
$if windows {
@ -315,14 +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.cstr()))
path = path.replace('/', '\\')
attr := int(C.GetFileAttributes(path.to_wide()))
if attr == INVALID_FILE_ATTRIBUTES {
return false
}
@ -345,7 +410,7 @@ pub fn dir_exists(path string) bool {
pub fn mkdir(path string) {
$if windows {
path = path.replace('/', '\\')
C.CreateDirectory(path.cstr(), 0)
C._wmkdir(path.to_wide())
}
$else {
C.mkdir(path.cstr(), 511)// S_IRWXU | S_IRWXG | S_IRWXO
@ -354,7 +419,12 @@ 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.cstr())
}
// C.unlink(path.cstr())
}
@ -365,7 +435,7 @@ pub fn rmdir(path string) {
C.rmdir(path.cstr())
}
$else {
C.RemoveDirectoryA(path.cstr())
C.RemoveDirectoryW(path.to_wide())
}
}
@ -435,20 +505,18 @@ 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
buf := malloc(max)
max := 512 // MAX_PATH * sizeof(wchar_t)
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
// NOTE: Once we have UTF8 encode function to
// convert utf16 to utf8, change to ReadConsoleW
mut nr_chars := 0
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
@ -536,8 +604,8 @@ 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')
@ -545,10 +613,12 @@ pub fn executable() string {
return string(result, count)
}
$if windows {
ret := int(C.GetModuleFileName( 0, result, MAX_PATH ))
return string( result, ret)
mut result := &u16(malloc(512)) // MAX_PATH * sizeof(wchar_t)
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 {
@ -558,6 +628,7 @@ 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)
@ -569,6 +640,7 @@ 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')
@ -576,6 +648,7 @@ 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')
@ -587,9 +660,9 @@ pub fn executable() string {
pub fn is_dir(path string) bool {
$if windows {
val := int(C.GetFileAttributes(path.cstr()))
val := int(C.GetFileAttributes(path.to_wide()))
// Note: this return is broke (wrong). we have dir_exists already how will this differ?
return val &FILE_ATTRIBUTE_DIRECTORY > 0
return (val &FILE_ATTRIBUTE_DIRECTORY) > 0
}
$else {
statbuf := C.stat{}
@ -603,7 +676,7 @@ pub fn is_dir(path string) bool {
pub fn chdir(path string) {
$if windows {
C._chdir(path.cstr())
C._wchdir(path.to_wide())
}
$else {
C.chdir(path.cstr())
@ -611,19 +684,22 @@ 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)
}
$else {
buf := malloc(512)
if C.getcwd(buf, 512) == 0 {
return ''
}
}
return string(buf)
}
}
// win: FILETIME
// https://docs.microsoft.com/en-us/windows/win32/api/minwinbase/ns-minwinbase-filetime
@ -633,7 +709,7 @@ struct filetime {
}
// win: WIN32_FIND_DATA
// https://docs.microsoft.com/en-us/windows/win32/api/minwinbase/ns-minwinbase-_win32_find_dataa
// https://docs.microsoft.com/en-us/windows/win32/api/minwinbase/ns-minwinbase-_win32_find_dataw
struct win32finddata {
mut:
dwFileAttributes u32
@ -670,13 +746,13 @@ pub fn ls(path string) []string {
path_files := '$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.cstr(), &find_file_data)
first_filename := tos(&find_file_data.cFileName, strlen(find_file_data.cFileName))
h_find_files := C.FindFirstFile(path_files.to_wide(), &find_file_data)
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()
}

View File

@ -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.cstr(), mode.cstr())
_fd := C._wfopen(path.to_wide(), mode.cstr())
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 buf := [byte(0); sz] // Not work for GetModuleFileNameW :(
mut sz := int(4096) // Optimized length
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.')