[PATCH v3 0/6] MR6937: msvcrt: Prepare directory related functions to handle utf-8 encoded file names.

[Piotr Caban (＠piotr)]

The idea is to use common helper to convert between CP_ACP and WCHAR strings first. Later all functions will be switched to support UTF-8 at the same time.

--
  v3: msvcrt: Call _wfullpath in _fullpath function.
      msvcrt: Call _wgetdcwd in _getdcwd function.
      msvcrt: Call _wgetcwd in _getcwd function.

https://gitlab.winehq.org/wine/wine/-/merge_requests/6937

[PATCH v3 1/6] msvcrt: Call _wmkdir in _mkdir function.

[Piotr Caban]

From: Piotr Caban

---
 dlls/msvcrt/dir.c          | 11 +++++++----
 dlls/msvcrt/msvcrt.h       | 18 ++++++++++++++++++
 dlls/ucrtbase/tests/file.c | 35 +++++++++++++++++++++++++++++++++++
 3 files changed, 60 insertions(+), 4 deletions(-)

diff --git a/dlls/msvcrt/dir.c b/dlls/msvcrt/dir.c
index 546e8d49a7b..648c8d80884 100644
--- a/dlls/msvcrt/dir.c
+++ b/dlls/msvcrt/dir.c
@@ -1004,10 +1004,13 @@ unsigned int CDECL _getdiskfree(unsigned int disk, struct _diskfree_t * d)
  */
 int CDECL _mkdir(const char * newdir)
 {
-  if (CreateDirectoryA(newdir,NULL))
-    return 0;
-  msvcrt_set_errno(GetLastError());
-  return -1;
+    wchar_t *newdirW = NULL;
+    int ret;
+
+    if (newdir && !(newdirW = wstrdupa_utf8(newdir))) return -1;
+    ret = _wmkdir(newdirW);
+    free(newdirW);
+    return ret;
 }

/*********************************************************************
diff --git a/dlls/msvcrt/msvcrt.h b/dlls/msvcrt/msvcrt.h
index eaaf8b0034d..71b0f9039f1 100644
--- a/dlls/msvcrt/msvcrt.h
+++ b/dlls/msvcrt/msvcrt.h
@@ -62,6 +62,7 @@ typedef struct _iobuf

#include "windef.h"
 #include "winbase.h"
+#include "winnls.h"
 #undef strncpy
 #undef wcsncpy

@@ -411,4 +412,21 @@ extern char* __cdecl __unDName(char *,const char*,int,malloc_func_t,free_func_t,

#define INHERIT_THREAD_PRIORITY 0xF000

+static inline int convert_acp_utf8_to_wcs(const char *str, wchar_t *wstr, int len)
+{
+    return MultiByteToWideChar(CP_ACP, MB_PRECOMPOSED, str, -1, wstr, len);
+}
+
+static inline wchar_t* wstrdupa_utf8(const char *str)
+{
+    int len = convert_acp_utf8_to_wcs(str, NULL, 0);
+    wchar_t *wstr;
+
+    if (!len) return NULL;
+    wstr = malloc(len * sizeof(wchar_t));
+    if (!wstr) return NULL;
+    convert_acp_utf8_to_wcs(str, wstr, len);
+    return wstr;
+}
+
 #endif /* __WINE_MSVCRT_H */
diff --git a/dlls/ucrtbase/tests/file.c b/dlls/ucrtbase/tests/file.c
index 99372a391e3..ef62db88062 100644
--- a/dlls/ucrtbase/tests/file.c
+++ b/dlls/ucrtbase/tests/file.c
@@ -18,6 +18,8 @@
  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
  */

+#include <errno.h>
+#include <direct.h>
 #include <stdarg.h>
 #include <locale.h>
 #include <share.h>
@@ -216,10 +218,43 @@ static void test_fopen(void)
     setlocale(LC_ALL, "C");
 }

+static void test_utf8(void)
+{
+    const char dir[] = "dir\xc4\x99\xc5\x9b\xc4\x87";
+    const WCHAR dirW[] = L"dir\x0119\x015b\x0107";
+
+    int ret;
+
+    if (!setlocale(LC_ALL, ".utf8"))
+    {
+        win_skip("utf-8 tests\n");
+        return;
+    }
+
+    ret = _mkdir(dir);
+    if (ret == -1 && errno == ENOENT)
+    {
+        skip("can't create test environment\n");
+        return;
+    }
+    ok(!ret, "_mkdir returned %d, error %d\n", ret, errno);
+
+    ret = _wrmdir(dirW);
+    todo_wine_if(GetACP() != CP_UTF8) ok(!ret, "_wrmdir returned %d, errno %d\n", ret, errno);
+    if (ret)
+    {
+        ret = _rmdir(dir);
+        ok(!ret, "_rmdir returned %d, errno %d\n", ret, errno);
+    }
+
+    setlocale(LC_ALL, "C");
+}
+
 START_TEST(file)
 {
     test_std_stream_buffering();
     test_iobuf_layout();
     test_std_stream_open();
     test_fopen();
+    test_utf8();
 }

[PATCH v3 2/6] msvcrt: Call _wrmdir in _rmdir function.

[Piotr Caban]

From: Piotr Caban

---
 dlls/msvcrt/dir.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/dlls/msvcrt/dir.c b/dlls/msvcrt/dir.c
index 648c8d80884..4fd0d200500 100644
--- a/dlls/msvcrt/dir.c
+++ b/dlls/msvcrt/dir.c
@@ -1043,10 +1043,13 @@ int CDECL _wmkdir(const wchar_t* newdir)
  */
 int CDECL _rmdir(const char * dir)
 {
-  if (RemoveDirectoryA(dir))
-    return 0;
-  msvcrt_set_errno(GetLastError());
-  return -1;
+    wchar_t *dirW = NULL;
+    int ret;
+
+    if (dir && !(dirW = wstrdupa_utf8(dir))) return -1;
+    ret = _wrmdir(dirW);
+    free(dirW);
+    return ret;
 }

/*********************************************************************

[PATCH v3 3/6] msvcrt: Call _wchdir in _chdir function.

[Piotr Caban]

From: Piotr Caban

---
 dlls/msvcrt/dir.c          | 13 +++++++------
 dlls/ucrtbase/tests/file.c |  5 +++++
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/dlls/msvcrt/dir.c b/dlls/msvcrt/dir.c
index 4fd0d200500..d574a4ae22e 100644
--- a/dlls/msvcrt/dir.c
+++ b/dlls/msvcrt/dir.c
@@ -274,12 +274,13 @@ static void msvcrt_wfttofd64i32( const WIN32_FIND_DATAW *fd, struct _wfinddata64
  */
 int CDECL _chdir(const char * newdir)
 {
-  if (!SetCurrentDirectoryA(newdir))
-  {
-    msvcrt_set_errno(newdir?GetLastError():0);
-    return -1;
-  }
-  return 0;
+    wchar_t *newdirW = NULL;
+    int ret;
+
+    if (newdir && !(newdirW = wstrdupa_utf8(newdir))) return -1;
+    ret = _wchdir(newdirW);
+    free(newdirW);
+    return ret;
 }

/*********************************************************************
diff --git a/dlls/ucrtbase/tests/file.c b/dlls/ucrtbase/tests/file.c
index ef62db88062..bb01a8a7b8f 100644
--- a/dlls/ucrtbase/tests/file.c
+++ b/dlls/ucrtbase/tests/file.c
@@ -239,6 +239,11 @@ static void test_utf8(void)
     }
     ok(!ret, "_mkdir returned %d, error %d\n", ret, errno);

+    ret = _chdir(dir);
+    ok(!ret, "_chdir returned %d, error %d\n", ret, errno);
+    ret = _chdir("..");
+    ok(!ret, "_chdir returned %d, error %d\n", ret, errno);
+
     ret = _wrmdir(dirW);
     todo_wine_if(GetACP() != CP_UTF8) ok(!ret, "_wrmdir returned %d, errno %d\n", ret, errno);
     if (ret)

[PATCH v3 4/6] msvcrt: Call _wgetcwd in _getcwd function.

[Piotr Caban]

From: Piotr Caban

---
 dlls/msvcrt/dir.c          | 29 +++++++++++++----------------
 dlls/msvcrt/msvcrt.h       | 17 +++++++++++++++++
 dlls/ucrtbase/tests/file.c | 21 +++++++++++++++++++++
 3 files changed, 51 insertions(+), 16 deletions(-)

diff --git a/dlls/msvcrt/dir.c b/dlls/msvcrt/dir.c
index d574a4ae22e..abb3bc21388 100644
--- a/dlls/msvcrt/dir.c
+++ b/dlls/msvcrt/dir.c
@@ -787,24 +787,21 @@ int CDECL _wfindnext64i32(intptr_t hand, struct _wfinddata64i32_t * ft)
  */
 char* CDECL _getcwd(char * buf, int size)
 {
-  char dir[MAX_PATH];
-  int dir_len = GetCurrentDirectoryA(MAX_PATH,dir);
+    wchar_t dirW[MAX_PATH];
+    int len;

-  if (dir_len < 1)
-    return NULL; /* FIXME: Real return value untested */
+    if (!_wgetcwd(dirW, ARRAY_SIZE(dirW))) return NULL;

-  if (!buf)
-  {
-      if (size <= dir_len) size = dir_len + 1;
-      if (!(buf = malloc( size ))) return NULL;
-  }
-  else if (dir_len >= size)
-  {
-    *_errno() = ERANGE;
-    return NULL; /* buf too small */
-  }
-  strcpy(buf,dir);
-  return buf;
+    if (!buf) return astrdupw_utf8(dirW);
+    len = convert_wcs_to_acp_utf8(dirW, NULL, 0);
+    if (!len) return NULL;
+    if (len > size)
+    {
+        *_errno() = ERANGE;
+        return NULL;
+    }
+    convert_wcs_to_acp_utf8(dirW, buf, size);
+    return buf;
 }

/*********************************************************************
diff --git a/dlls/msvcrt/msvcrt.h b/dlls/msvcrt/msvcrt.h
index 71b0f9039f1..39349345de0 100644
--- a/dlls/msvcrt/msvcrt.h
+++ b/dlls/msvcrt/msvcrt.h
@@ -417,6 +417,11 @@ static inline int convert_acp_utf8_to_wcs(const char *str, wchar_t *wstr, int le
     return MultiByteToWideChar(CP_ACP, MB_PRECOMPOSED, str, -1, wstr, len);
 }

+static inline int convert_wcs_to_acp_utf8(const wchar_t *wstr, char *str, int len)
+{
+    return WideCharToMultiByte(CP_ACP, 0, wstr, -1, str, len, NULL, NULL);
+}
+
 static inline wchar_t* wstrdupa_utf8(const char *str)
 {
     int len = convert_acp_utf8_to_wcs(str, NULL, 0);
@@ -429,4 +434,16 @@ static inline wchar_t* wstrdupa_utf8(const char *str)
     return wstr;
 }

+static inline char* astrdupw_utf8(const wchar_t *wstr)
+{
+    int len = convert_wcs_to_acp_utf8(wstr, NULL, 0);
+    char *str;
+
+    if (!len) return NULL;
+    str = malloc(len * sizeof(char));
+    if (!str) return NULL;
+    convert_wcs_to_acp_utf8(wstr, str, len);
+    return str;
+}
+
 #endif /* __WINE_MSVCRT_H */
diff --git a/dlls/ucrtbase/tests/file.c b/dlls/ucrtbase/tests/file.c
index bb01a8a7b8f..e89d91be36d 100644
--- a/dlls/ucrtbase/tests/file.c
+++ b/dlls/ucrtbase/tests/file.c
@@ -218,11 +218,24 @@ static void test_fopen(void)
     setlocale(LC_ALL, "C");
 }

+static BOOL is_lossless_convertion(const char *str)
+{
+    wchar_t bufW[32];
+    char buf[32];
+
+    if (!MultiByteToWideChar(CP_ACP, 0, str, -1, bufW, ARRAY_SIZE(bufW)))
+        return FALSE;
+    if (!WideCharToMultiByte(CP_ACP, 0, bufW, -1, buf, ARRAY_SIZE(buf), NULL, NULL))
+        return FALSE;
+    return !strcmp(str, buf);
+}
+
 static void test_utf8(void)
 {
     const char dir[] = "dir\xc4\x99\xc5\x9b\xc4\x87";
     const WCHAR dirW[] = L"dir\x0119\x015b\x0107";

+    char buf[256], *p;
     int ret;

if (!setlocale(LC_ALL, ".utf8"))
@@ -241,6 +254,14 @@ static void test_utf8(void)

ret = _chdir(dir);
     ok(!ret, "_chdir returned %d, error %d\n", ret, errno);
+
+    p = _getcwd(buf, sizeof(buf));
+    ok(p == buf, "_getcwd returned %p, errno %d\n", p, errno);
+    p = strrchr(p, '\');
+    ok(!!p, "strrchr returned NULL, buf = %s\n", debugstr_a(buf));
+    todo_wine_if(!is_lossless_convertion(dir))
+        ok(!strcmp(p + 1, dir), "unexpected working directory: %s\n", debugstr_a(buf));
+
     ret = _chdir("..");
     ok(!ret, "_chdir returned %d, error %d\n", ret, errno);

[PATCH v3 5/6] msvcrt: Call _wgetdcwd in _getdcwd function.

[Piotr Caban]

From: Piotr Caban

---
 dlls/msvcrt/dir.c          | 40 +++++++++++---------------------------
 dlls/ucrtbase/tests/file.c |  7 +++++++
 2 files changed, 18 insertions(+), 29 deletions(-)

diff --git a/dlls/msvcrt/dir.c b/dlls/msvcrt/dir.c
index abb3bc21388..b878a5ab1a9 100644
--- a/dlls/msvcrt/dir.c
+++ b/dlls/msvcrt/dir.c
@@ -869,39 +869,21 @@ int CDECL _getdrive(void)
  */
 char* CDECL _getdcwd(int drive, char * buf, int size)
 {
-  static char* dummy;
-
-  TRACE(":drive %d(%c), size %d\n",drive, drive + 'A' - 1, size);
-
-  if (!drive || drive == _getdrive())
-    return _getcwd(buf,size); /* current */
-  else
-  {
-    char dir[MAX_PATH];
-    char drivespec[] = "A:";
-    int dir_len;
+    wchar_t dirW[MAX_PATH];
+    int len;

-    drivespec[0] += drive - 1;
-    if (GetDriveTypeA(drivespec) < DRIVE_REMOVABLE)
-    {
-      *_errno() = EACCES;
-      return NULL;
-    }
+    if (!_wgetdcwd(drive, dirW, ARRAY_SIZE(dirW))) return NULL;

-    dir_len = GetFullPathNameA(drivespec,MAX_PATH,dir,&dummy);
-    if (dir_len >= size || dir_len < 1)
+    if (!buf) return astrdupw_utf8(dirW);
+    len = convert_wcs_to_acp_utf8(dirW, NULL, 0);
+    if (!len) return NULL;
+    if (len > size)
     {
-      *_errno() = ERANGE;
-      return NULL; /* buf too small */
+        *_errno() = ERANGE;
+        return NULL;
     }
-
-    TRACE(":returning '%s'\n", dir);
-    if (!buf)
-      return _strdup(dir); /* allocate */
-
-    strcpy(buf,dir);
-  }
-  return buf;
+    convert_wcs_to_acp_utf8(dirW, buf, size);
+    return buf;
 }

/*********************************************************************
diff --git a/dlls/ucrtbase/tests/file.c b/dlls/ucrtbase/tests/file.c
index e89d91be36d..53a7ab7e4a6 100644
--- a/dlls/ucrtbase/tests/file.c
+++ b/dlls/ucrtbase/tests/file.c
@@ -262,6 +262,13 @@ static void test_utf8(void)
     todo_wine_if(!is_lossless_convertion(dir))
         ok(!strcmp(p + 1, dir), "unexpected working directory: %s\n", debugstr_a(buf));

+    p = _getdcwd(_getdrive(), buf, sizeof(buf));
+    ok(p == buf, "_getdcwd returned %p, errno %d\n", p, errno);
+    p = strrchr(p, '\');
+    ok(!!p, "strrchr returned NULL, buf = %s\n", debugstr_a(buf));
+    todo_wine_if(!is_lossless_convertion(dir))
+        ok(!strcmp(p + 1, dir), "unexpected working directory: %s\n", debugstr_a(buf));
+
     ret = _chdir("..");
     ok(!ret, "_chdir returned %d, error %d\n", ret, errno);

[PATCH v3 6/6] msvcrt: Call _wfullpath in _fullpath function.

[Piotr Caban]

From: Piotr Caban

---
 dlls/msvcrt/dir.c          | 49 +++++++++++++-------------------------
 dlls/ucrtbase/tests/file.c |  7 ++++++
 2 files changed, 23 insertions(+), 33 deletions(-)

diff --git a/dlls/msvcrt/dir.c b/dlls/msvcrt/dir.c
index b878a5ab1a9..ff26a0ce2b8 100644
--- a/dlls/msvcrt/dir.c
+++ b/dlls/msvcrt/dir.c
@@ -1296,43 +1296,26 @@ wchar_t * CDECL _wfullpath(wchar_t * absPath, const wchar_t* relPath, size_t siz
  *          Otherwise populates absPath with the path and returns it.
  * Failure: NULL. errno indicates the error.
  */
-char * CDECL _fullpath(char * absPath, const char* relPath, size_t size)
+char * CDECL _fullpath(char *abs_path, const char *rel_path, size_t size)
 {
-  DWORD rc;
-  char* lastpart;
-  char* buffer;
-  BOOL alloced = FALSE;
-
-  if (!relPath || !*relPath)
-    return _getcwd(absPath, size);
-
-  if (absPath == NULL)
-  {
-      buffer = malloc(MAX_PATH);
-      size = MAX_PATH;
-      alloced = TRUE;
-  }
-  else
-      buffer = absPath;
-
-  if (size < 4)
-  {
-    *_errno() = ERANGE;
-    return NULL;
-  }
-
-  TRACE(":resolving relative path '%s'\n",relPath);
+    wchar_t abs_pathW[MAX_PATH], *rel_pathW = NULL, *retW;
+    size_t len;

-  rc = GetFullPathNameA(relPath,size,buffer,&lastpart);
+    if (rel_path && !(rel_pathW = wstrdupa_utf8(rel_path))) return NULL;
+    retW = _wfullpath(abs_pathW, rel_pathW, ARRAY_SIZE(abs_pathW));
+    free(rel_pathW);
+    if (!retW) return NULL;

-  if (rc > 0 && rc <= size)
-    return buffer;
-  else
-  {
-      if (alloced)
-          free(buffer);
+    if (!abs_path) return astrdupw_utf8(abs_pathW);
+    len = convert_wcs_to_acp_utf8(abs_pathW, NULL, 0);
+    if (!len) return NULL;
+    if (len > size)
+    {
+        *_errno() = ERANGE;
         return NULL;
-  }
+    }
+    convert_wcs_to_acp_utf8(abs_pathW, abs_path, size);
+    return abs_path;
 }

/*********************************************************************
diff --git a/dlls/ucrtbase/tests/file.c b/dlls/ucrtbase/tests/file.c
index 53a7ab7e4a6..da0103265be 100644
--- a/dlls/ucrtbase/tests/file.c
+++ b/dlls/ucrtbase/tests/file.c
@@ -269,6 +269,13 @@ static void test_utf8(void)
     todo_wine_if(!is_lossless_convertion(dir))
         ok(!strcmp(p + 1, dir), "unexpected working directory: %s\n", debugstr_a(buf));

+    p = _fullpath(buf, NULL, sizeof(buf));
+    ok(p == buf, "_fulpath returned %p, errno %d\n", p, errno);
+    p = strrchr(p, '\');
+    ok(!!p, "strrchr returned NULL, buf = %s\n", debugstr_a(buf));
+    todo_wine_if(!is_lossless_convertion(dir))
+        ok(!strcmp(p + 1, dir), "unexpected working directory: %s\n", debugstr_a(buf));
+
     ret = _chdir("..");
     ok(!ret, "_chdir returned %d, error %d\n", ret, errno);

[Aida Jonikienė (＠DodoGTA)]

Is this a preparation for Windows native UTF-8 support?: https://learn.microsoft.com/en-us/windows/apps/design/globalizing/use-utf8-c...

[Piotr Caban (＠piotr)]

On Sun Dec  1 16:48:42 2024 +0000, Aida Jonikienė wrote:
> Is this a preparation for Windows native UTF-8 support?: https://learn.microsoft.com/en-us/windows/apps/design/globalizing/use-utf8-c...

No, it's ucrtbase feature. It supports utf-8 strings in functions that used to work only with ANSI strings.