bpo-32922: dbm.open() now encodes filename with the filesystem encoding. (GH-5832)

(cherry picked from commit 6f600ff1734ca2fdcdd37a809adf8130f0d8cc4e)

Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>

4 files changed, 45 insertions, 18 deletions

diff --git a/Modules/_dbmmodule.c b/Modules/_dbmmodule.c
index 7e1344177b..8afd92cf3c 100644
--- a/Modules/_dbmmodule.c
+++ b/Modules/_dbmmodule.c
@@ -412,7 +412,7 @@ static PyTypeObject Dbmtype = {
 
 _dbm.open as dbmopen
 
-    filename: str
+    filename: unicode
         The filename to open.
 
     flags: str="r"
@@ -429,9 +429,9 @@ Return a database object.
 [clinic start generated code]*/
 
 static PyObject *
-dbmopen_impl(PyObject *module, const char *filename, const char *flags,
+dbmopen_impl(PyObject *module, PyObject *filename, const char *flags,
              int mode)
-/*[clinic end generated code: output=5fade8cf16e0755f input=226334bade5764e6]*/
+/*[clinic end generated code: output=9527750f5df90764 input=376a9d903a50df59]*/
 {
     int iflags;
 
@@ -450,7 +450,20 @@ dbmopen_impl(PyObject *module, const char *filename, const char *flags,
                         "arg 2 to open should be 'r', 'w', 'c', or 'n'");
         return NULL;
     }
-    return newdbmobject(filename, iflags, mode);
+
+    PyObject *filenamebytes = PyUnicode_EncodeFSDefault(filename);
+    if (filenamebytes == NULL) {
+        return NULL;
+    }
+    const char *name = PyBytes_AS_STRING(filenamebytes);
+    if (strlen(name) != (size_t)PyBytes_GET_SIZE(filenamebytes)) {
+        Py_DECREF(filenamebytes);
+        PyErr_SetString(PyExc_ValueError, "embedded null character");
+        return NULL;
+    }
+    PyObject *self = newdbmobject(name, iflags, mode);
+    Py_DECREF(filenamebytes);
+    return self;
 }
 
 static PyMethodDef dbmmodule_methods[] = {
diff --git a/Modules/_gdbmmodule.c b/Modules/_gdbmmodule.c
index 12d973b5ce..9996d8c26f 100644
--- a/Modules/_gdbmmodule.c
+++ b/Modules/_gdbmmodule.c
@@ -527,7 +527,7 @@ static PyTypeObject Dbmtype = {
 
 /*[clinic input]
 _gdbm.open as dbmopen
-    filename as name: str
+    filename: unicode
     flags: str="r"
     mode: int(py_default="0o666") = 0o666
     /
@@ -557,8 +557,9 @@ when the database has to be created.  It defaults to octal 0o666.
 [clinic start generated code]*/
 
 static PyObject *
-dbmopen_impl(PyObject *module, const char *name, const char *flags, int mode)
-/*[clinic end generated code: output=31aa1bafdf5da688 input=55563cd60e51984a]*/
+dbmopen_impl(PyObject *module, PyObject *filename, const char *flags,
+             int mode)
+/*[clinic end generated code: output=9527750f5df90764 input=3be0b0875974b928]*/
 {
     int iflags;
 
@@ -606,7 +607,19 @@ dbmopen_impl(PyObject *module, const char *name, const char *flags, int mode)
         }
     }
 
-    return newdbmobject(name, iflags, mode);
+    PyObject *filenamebytes = PyUnicode_EncodeFSDefault(filename);
+    if (filenamebytes == NULL) {
+        return NULL;
+    }
+    const char *name = PyBytes_AS_STRING(filenamebytes);
+    if (strlen(name) != (size_t)PyBytes_GET_SIZE(filenamebytes)) {
+        Py_DECREF(filenamebytes);
+        PyErr_SetString(PyExc_ValueError, "embedded null character");
+        return NULL;
+    }
+    PyObject *self = newdbmobject(name, iflags, mode);
+    Py_DECREF(filenamebytes);
+    return self;
 }
 
 static const char dbmmodule_open_flags[] = "rwcn"
diff --git a/Modules/clinic/_dbmmodule.c.h b/Modules/clinic/_dbmmodule.c.h
index 63d5b1a41f..0f831c9eec 100644
--- a/Modules/clinic/_dbmmodule.c.h
+++ b/Modules/clinic/_dbmmodule.c.h
@@ -121,18 +121,18 @@ PyDoc_STRVAR(dbmopen__doc__,
     {"open", (PyCFunction)dbmopen, METH_FASTCALL, dbmopen__doc__},
 
 static PyObject *
-dbmopen_impl(PyObject *module, const char *filename, const char *flags,
+dbmopen_impl(PyObject *module, PyObject *filename, const char *flags,
              int mode);
 
 static PyObject *
 dbmopen(PyObject *module, PyObject *const *args, Py_ssize_t nargs)
 {
     PyObject *return_value = NULL;
-    const char *filename;
+    PyObject *filename;
     const char *flags = "r";
     int mode = 438;
 
-    if (!_PyArg_ParseStack(args, nargs, "s|si:open",
+    if (!_PyArg_ParseStack(args, nargs, "U|si:open",
         &filename, &flags, &mode)) {
         goto exit;
     }
@@ -141,4 +141,4 @@ dbmopen(PyObject *module, PyObject *const *args, Py_ssize_t nargs)
 exit:
     return return_value;
 }
-/*[clinic end generated code: output=8ce71abac849155f input=a9049054013a1b77]*/
+/*[clinic end generated code: output=5c858b4080a011a4 input=a9049054013a1b77]*/
diff --git a/Modules/clinic/_gdbmmodule.c.h b/Modules/clinic/_gdbmmodule.c.h
index 2222967aaa..7bdc4321df 100644
--- a/Modules/clinic/_gdbmmodule.c.h
+++ b/Modules/clinic/_gdbmmodule.c.h
@@ -234,23 +234,24 @@ PyDoc_STRVAR(dbmopen__doc__,
     {"open", (PyCFunction)dbmopen, METH_FASTCALL, dbmopen__doc__},
 
 static PyObject *
-dbmopen_impl(PyObject *module, const char *name, const char *flags, int mode);
+dbmopen_impl(PyObject *module, PyObject *filename, const char *flags,
+             int mode);
 
 static PyObject *
 dbmopen(PyObject *module, PyObject *const *args, Py_ssize_t nargs)
 {
     PyObject *return_value = NULL;
-    const char *name;
+    PyObject *filename;
     const char *flags = "r";
     int mode = 438;
 
-    if (!_PyArg_ParseStack(args, nargs, "s|si:open",
-        &name, &flags, &mode)) {
+    if (!_PyArg_ParseStack(args, nargs, "U|si:open",
+        &filename, &flags, &mode)) {
         goto exit;
     }
-    return_value = dbmopen_impl(module, name, flags, mode);
+    return_value = dbmopen_impl(module, filename, flags, mode);
 
 exit:
     return return_value;
 }
-/*[clinic end generated code: output=dc0aca8c00055d02 input=a9049054013a1b77]*/
+/*[clinic end generated code: output=dec05ff9c5aeaeae input=a9049054013a1b77]*/