Rename old function to _path_splitanchor

Co-authored-by: Steve Dower <steve.dower@microsoft.com>
python · zooba · Apr 25, 2024 · Apr 19, 2024 · Apr 19, 2024 · Apr 19, 2024
commit ca0761a50d90c0eb524554556b69bf1424b0763f
diff --git a/Lib/importlib/_bootstrap_external.py b/Lib/importlib/_bootstrap_external.py
@@ -104,29 +104,28 @@ def _path_join(*path_parts):
             return ""
         if len(path_parts) == 1:
             return path_parts[0]
-        anchor = ""
+        root = ""
         path = []
-        for drive, root, tail in map(_os._path_splitroot, path_parts):
-            new_anchor = drive + root
-            if new_anchor.startswith(path_sep_tuple) or new_anchor.endswith(path_sep_tuple):
-                anchor = new_anchor.rstrip(path_separators) or anchor
+        for new_root, tail in map(_os._path_splitanchor, path_parts):
+            if new_root.startswith(path_sep_tuple) or new_root.endswith(path_sep_tuple):
+                root = new_root.rstrip(path_separators) or root
                 path = [path_sep + tail]
-            elif new_anchor.endswith(':'):
-                if anchor.casefold() != new_anchor.casefold():
+            elif new_root.endswith(':'):
+                if root.casefold() != new_root.casefold():
                     # Drive relative paths have to be resolved by the OS, so we reset the
                     # tail but do not add a path_sep prefix.
-                    anchor = new_anchor
+                    root = new_root
                     path = [tail]
                 else:
                     path.append(tail)
             else:
-                anchor = new_anchor or anchor
+                root = new_root or root
                 path.append(tail)
         path = [p.rstrip(path_separators) for p in path if p]
         if len(path) == 1 and not path[0]:
-            # Avoid losing the anchor's trailing separator when joining with nothing
-            return anchor + path_sep
-        return anchor + path_sep.join(path)
+            # Avoid losing the root's trailing separator when joining with nothing
+            return root + path_sep
+        return root + path_sep.join(path)
 
 else:
     def _path_join(*path_parts):
@@ -179,9 +178,8 @@ def _path_isabs(path):
         """Replacement for os.path.isabs."""
         if not path:
             return False
-        drive, root, _ = _os._path_splitroot(path)
-        anchor = (drive + root).replace('/', '\\')
-        return len(anchor) > 1 and (anchor.startswith('\\\\') or anchor.endswith('\\'))
+        root = _os._path_splitanchor(path)[0].replace('/', '\\')
+        return len(root) > 1 and (root.startswith('\\\\') or root.endswith('\\'))
 
 else:
     def _path_isabs(path):

diff --git a/Modules/clinic/posixmodule.c.h b/Modules/clinic/posixmodule.c.h
diff --git a/Modules/posixmodule.c b/Modules/posixmodule.c
@@ -5040,6 +5040,50 @@ os__getvolumepathname_impl(PyObject *module, path_t *path)
 }
 
 
+/*[clinic input]
+os._path_splitanchor
+    path: path_t
+Removes everything after the root on Win32.
+[clinic start generated code]*/
+
+static PyObject *
+os__path_splitanchor_impl(PyObject *module, path_t *path)
+/*[clinic end generated code: output=37b687463b40c424 input=00dd0c06233b8cff]*/
+{
+    wchar_t *buffer;
+    wchar_t *end;
+    PyObject *result = NULL;
+    HRESULT ret;
+
+    buffer = (wchar_t*)PyMem_Malloc(sizeof(wchar_t) * (wcslen(path->wide) + 1));
+    if (!buffer) {
+        return NULL;
+    }
+    wcscpy(buffer, path->wide);
+    for (wchar_t *p = wcschr(buffer, L'/'); p; p = wcschr(p, L'/')) {
+        *p = L'\\';
+    }
+
+    Py_BEGIN_ALLOW_THREADS
+    ret = PathCchSkipRoot(buffer, &end);
+    Py_END_ALLOW_THREADS
+    if (FAILED(ret)) {
+        result = Py_BuildValue("sO", "", path->object);
+    } else if (end != buffer) {
+        size_t rootLen = (size_t)(end - buffer);
+        result = Py_BuildValue("NN",
+            PyUnicode_FromWideChar(path->wide, rootLen),
+            PyUnicode_FromWideChar(path->wide + rootLen, -1)
+        );
+    } else {
+        result = Py_BuildValue("Os", path->object, "");
+    }
+    PyMem_Free(buffer);
+
+    return result;
+}
+
+
 /*[clinic input]
 os._path_isdir
 
@@ -5434,13 +5478,21 @@ os__path_splitroot_impl(PyObject *module, PyObject *path)
 /*[clinic end generated code: output=6904e00a6a970b9b input=4ef301247820b583]*/
 {
     Py_ssize_t len, drvsize, rootsize;
+    PyObject *drv, *root, *tail, *result = NULL;
     wchar_t *buffer = PyUnicode_AsWideCharString(path, &len);
+    if (!buffer) {
+        goto exit;
+    }
     _Py_skiproot(buffer, len, &drvsize, &rootsize);
-    PyObject *drv = PyUnicode_FromWideChar(buffer, drvsize);
-    PyObject *root = PyUnicode_FromWideChar(&buffer[drvsize], rootsize);
-    PyObject *tail = PyUnicode_FromWideChar(&buffer[drvsize + rootsize], len - drvsize - rootsize);
+    if (!(drv = PyUnicode_FromWideChar(buffer, drvsize)) ||
+        !(root = PyUnicode_FromWideChar(&buffer[drvsize], rootsize)) ||
+        !(tail = PyUnicode_FromWideChar(&buffer[drvsize + rootsize], len - drvsize - rootsize)))
+    {
+        goto exit;
+    }
+    result = Py_BuildValue("(OOO)", drv, root, tail);
+exit:
     PyMem_Free(buffer);
-    PyObject *result = Py_BuildValue("(OOO)", drv, root, tail);
     Py_DECREF(drv);
     Py_DECREF(root);
     Py_DECREF(tail);
@@ -16779,6 +16831,7 @@ static PyMethodDef posix_methods[] = {
     OS__GETFINALPATHNAME_METHODDEF
     OS__FINDFIRSTFILE_METHODDEF
     OS__GETVOLUMEPATHNAME_METHODDEF
+    OS__PATH_SPLITANCHOR_METHODDEF
     OS__PATH_SPLITROOT_METHODDEF
     OS__PATH_NORMPATH_METHODDEF
     OS_GETLOADAVG_METHODDEF

diff --git a/Python/fileutils.c b/Python/fileutils.c
@@ -2296,8 +2296,10 @@ PathCchCombineEx(wchar_t *buffer, size_t bufsize, const wchar_t *dirname,
 #endif /* defined(MS_WINDOWS_GAMES) && !defined(MS_WINDOWS_DESKTOP) */
 
 void
-_Py_skiproot(wchar_t *path, Py_ssize_t size, Py_ssize_t *drvsize, Py_ssize_t *rootsize)
+_Py_skiproot(const wchar_t *path, Py_ssize_t size, Py_ssize_t *drvsize, Py_ssize_t *rootsize)
 {
+    assert(drvsize);
+    assert(rootsize);
 #ifndef MS_WINDOWS
 #define IS_SEP(x) (*(x) == SEP)
     *drvsize = 0;