bpo-37412: os.getcwdb() now uses UTF-8 on Windows (GH-14396)

The os.getcwdb() function now uses the UTF-8 encoding on Windows,
rather than the ANSI code page: see PEP 529 for the rationale. The
function is no longer deprecated on Windows.

os.getcwd() and os.getcwdb() now detect integer overflow on memory
allocations. On Unix, these functions properly report MemoryError on
memory allocation failure.

Doc/library/os.rst

@@ -1730,6 +1730,11 @@ features:
 
    Return a bytestring representing the current working directory.
 
+   .. versionchanged:: 3.8
+      The function now uses the UTF-8 encoding on Windows, rather than the ANSI
+      code page: see :pep:`529` for the rationale. The function is no longer
+      deprecated on Windows.
+
 
 .. function:: lchflags(path, flags)
 

Doc/whatsnew/3.8.rst

@@ -1231,6 +1231,11 @@ Changes in Python behavior
 Changes in the Python API
 -------------------------
 
+* The :func:`os.getcwdb` function now uses the UTF-8 encoding on Windows,
+  rather than the ANSI code page: see :pep:`529` for the rationale. The
+  function is no longer deprecated on Windows.
+  (Contributed by Victor Stinner in :issue:`37412`.)
+
 * :class:`subprocess.Popen` can now use :func:`os.posix_spawn` in some cases
   for better performance. On Windows Subsystem for Linux and QEMU User
   Emulation, Popen constructor using :func:`os.posix_spawn` no longer raise an

Lib/test/test_os.py

@@ -82,6 +82,17 @@ def create_file(filename, content=b'content'):
         fp.write(content)
 
 
+class MiscTests(unittest.TestCase):
+    def test_getcwd(self):
+        cwd = os.getcwd()
+        self.assertIsInstance(cwd, str)
+
+    def test_getcwdb(self):
+        cwd = os.getcwdb()
+        self.assertIsInstance(cwd, bytes)
+        self.assertEqual(os.fsdecode(cwd), os.getcwd())
+
+
 # Tests creating TESTFN
 class FileTests(unittest.TestCase):
     def setUp(self):

Misc/NEWS.d/next/Library/2019-06-26-16-28-59.bpo-37412.lx0VjC.rst

@@ -0,0 +1,3 @@
+The :func:`os.getcwdb` function now uses the UTF-8 encoding on Windows,
+rather than the ANSI code page: see :pep:`529` for the rationale. The function
+is no longer deprecated on Windows.

Modules/posixmodule.c

@@ -506,17 +506,6 @@ void _Py_attribute_data_to_stat(BY_HANDLE_FILE_INFORMATION *,
                                             ULONG, struct _Py_stat_struct *);
 #endif
 
-#ifdef MS_WINDOWS
-static int
-win32_warn_bytes_api()
-{
-    return PyErr_WarnEx(PyExc_DeprecationWarning,
-        "The Windows bytes API has been deprecated, "
-        "use Unicode filenames instead",
-        1);
-}
-#endif
-
 
 #ifndef MS_WINDOWS
 PyObject *
@@ -3334,29 +3323,29 @@ os_lchown_impl(PyObject *module, path_t *path, uid_t uid, gid_t gid)
 static PyObject *
 posix_getcwd(int use_bytes)
 {
-    char *buf, *tmpbuf;
-    char *cwd;
-    const size_t chunk = 1024;
-    size_t buflen = 0;
-    PyObject *obj;
-
 #ifdef MS_WINDOWS
-    if (!use_bytes) {
     wchar_t wbuf[MAXPATHLEN];
     wchar_t *wbuf2 = wbuf;
-        PyObject *resobj;
     DWORD len;
+
     Py_BEGIN_ALLOW_THREADS
     len = GetCurrentDirectoryW(Py_ARRAY_LENGTH(wbuf), wbuf);
     /* If the buffer is large enough, len does not include the
        terminating \0. If the buffer is too small, len includes
        the space needed for the terminator. */
     if (len >= Py_ARRAY_LENGTH(wbuf)) {
+        if (len >= PY_SSIZE_T_MAX / sizeof(wchar_t)) {
             wbuf2 = PyMem_RawMalloc(len * sizeof(wchar_t));
-            if (wbuf2)
+        }
+        else {
+            wbuf2 = NULL;
+        }
+        if (wbuf2) {
             len = GetCurrentDirectoryW(len, wbuf2);
         }
+    }
     Py_END_ALLOW_THREADS
+
     if (!wbuf2) {
         PyErr_NoMemory();
         return NULL;
@@ -3366,51 +3355,67 @@ posix_getcwd(int use_bytes)
             PyMem_RawFree(wbuf2);
         return PyErr_SetFromWindowsErr(0);
     }
-        resobj = PyUnicode_FromWideChar(wbuf2, len);
-        if (wbuf2 != wbuf)
+
+    PyObject *resobj = PyUnicode_FromWideChar(wbuf2, len);
+    if (wbuf2 != wbuf) {
         PyMem_RawFree(wbuf2);
-        return resobj;
     }
 
-    if (win32_warn_bytes_api())
+    if (use_bytes) {
+        if (resobj == NULL) {
             return NULL;
-#endif
+        }
+        Py_SETREF(resobj, PyUnicode_EncodeFSDefault(resobj));
+    }
+
+    return resobj;
+#else
+    const size_t chunk = 1024;
+
+    char *buf = NULL;
+    char *cwd = NULL;
+    size_t buflen = 0;
 
-    buf = cwd = NULL;
     Py_BEGIN_ALLOW_THREADS
     do {
+        char *newbuf;
+        if (buflen <= PY_SSIZE_T_MAX - chunk) {
             buflen += chunk;
-#ifdef MS_WINDOWS
-        if (buflen > INT_MAX) {
-            PyErr_NoMemory();
+            newbuf = PyMem_RawRealloc(buf, buflen);
+        }
+        else {
+            newbuf = NULL;
+        }
+        if (newbuf == NULL) {
+            PyMem_RawFree(buf);
+            buf = NULL;
             break;
         }
-#endif
-        tmpbuf = PyMem_RawRealloc(buf, buflen);
-        if (tmpbuf == NULL)
-            break;
+        buf = newbuf;
 
-        buf = tmpbuf;
-#ifdef MS_WINDOWS
-        cwd = getcwd(buf, (int)buflen);
-#else
         cwd = getcwd(buf, buflen);
-#endif
     } while (cwd == NULL && errno == ERANGE);
     Py_END_ALLOW_THREADS
 
+    if (buf == NULL) {
+        return PyErr_NoMemory();
+    }
     if (cwd == NULL) {
         PyMem_RawFree(buf);
         return posix_error();
     }
 
-    if (use_bytes)
+    PyObject *obj;
+    if (use_bytes) {
         obj = PyBytes_FromStringAndSize(buf, strlen(buf));
-    else
+    }
+    else {
         obj = PyUnicode_DecodeFSDefault(buf);
+    }
     PyMem_RawFree(buf);
 
     return obj;
+#endif   /* !MS_WINDOWS */
 }