[3.13] gh-119614: Fix truncation of strings with embedded null characters in Tkinter (GH-120909) (GH-120938)

Now the null character is always represented as \xc0\x80 for
Tcl_NewStringObj().
(cherry picked from commit c38e2f64d0)

Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
This commit is contained in:
Miss Islington (bot) 2024-06-24 11:45:45 +02:00 committed by GitHub
parent 206028dba9
commit 732c00550f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 68 additions and 7 deletions

View File

@ -73,6 +73,18 @@ class TclTest(unittest.TestCase):
tcl.call('set','a','1')
self.assertEqual(tcl.call('set','a'),'1')
def test_call_passing_null(self):
tcl = self.interp
tcl.call('set', 'a', 'a\0b') # ASCII-only
self.assertEqual(tcl.getvar('a'), 'a\x00b')
self.assertEqual(tcl.call('set', 'a'), 'a\x00b')
self.assertEqual(tcl.eval('set a'), 'a\x00b')
tcl.call('set', 'a', '\u20ac\0') # non-ASCII
self.assertEqual(tcl.getvar('a'), '\u20ac\x00')
self.assertEqual(tcl.call('set', 'a'), '\u20ac\x00')
self.assertEqual(tcl.eval('set a'), '\u20ac\x00')
def testCallException(self):
tcl = self.interp
self.assertRaises(TclError,tcl.call,'set','a')
@ -98,6 +110,18 @@ class TclTest(unittest.TestCase):
tcl.setvar('a','1')
self.assertEqual(tcl.eval('set a'),'1')
def test_setvar_passing_null(self):
tcl = self.interp
tcl.setvar('a', 'a\0b') # ASCII-only
self.assertEqual(tcl.getvar('a'), 'a\x00b')
self.assertEqual(tcl.call('set', 'a'), 'a\x00b')
self.assertEqual(tcl.eval('set a'), 'a\x00b')
tcl.setvar('a', '\u20ac\0') # non-ASCII
self.assertEqual(tcl.getvar('a'), '\u20ac\x00')
self.assertEqual(tcl.call('set', 'a'), '\u20ac\x00')
self.assertEqual(tcl.eval('set a'), '\u20ac\x00')
def testSetVarArray(self):
tcl = self.interp
tcl.setvar('a(1)','1')

View File

@ -476,6 +476,15 @@ class MiscTest(AbstractTkTest, unittest.TestCase):
self.assertEqual(vi.micro, 0)
self.assertTrue(str(vi).startswith(f'{vi.major}.{vi.minor}'))
def test_embedded_null(self):
widget = tkinter.Entry(self.root)
widget.insert(0, 'abc\0def') # ASCII-only
widget.selection_range(0, 'end')
self.assertEqual(widget.selection_get(), 'abc\x00def')
widget.insert(0, '\u20ac\0') # non-ASCII
widget.selection_range(0, 'end')
self.assertEqual(widget.selection_get(), '\u20ac\0abc\x00def')
class WmTest(AbstractTkTest, unittest.TestCase):

View File

@ -0,0 +1,2 @@
Fix truncation of strings with embedded null characters in some internal
operations in :mod:`tkinter`.

View File

@ -512,7 +512,7 @@ unicodeFromTclObj(TkappObject *tkapp, Tcl_Obj *value)
else
Py_UNREACHABLE();
}
#endif
#endif /* USE_TCL_UNICODE */
const char *s = Tcl_GetStringFromObj(value, &len);
return unicodeFromTclStringAndSize(s, len);
}
@ -1018,7 +1018,9 @@ AsObj(PyObject *value)
PyErr_SetString(PyExc_OverflowError, "string is too long");
return NULL;
}
if (PyUnicode_IS_ASCII(value)) {
if (PyUnicode_IS_ASCII(value) &&
strlen(PyUnicode_DATA(value)) == (size_t)PyUnicode_GET_LENGTH(value))
{
return Tcl_NewStringObj((const char *)PyUnicode_DATA(value),
(int)size);
}
@ -1033,9 +1035,6 @@ AsObj(PyObject *value)
"surrogatepass", NATIVE_BYTEORDER);
else
Py_UNREACHABLE();
#else
encoded = _PyUnicode_AsUTF8String(value, "surrogateescape");
#endif
if (!encoded) {
return NULL;
}
@ -1045,12 +1044,39 @@ AsObj(PyObject *value)
PyErr_SetString(PyExc_OverflowError, "string is too long");
return NULL;
}
#if USE_TCL_UNICODE
result = Tcl_NewUnicodeObj((const Tcl_UniChar *)PyBytes_AS_STRING(encoded),
(int)(size / sizeof(Tcl_UniChar)));
#else
encoded = _PyUnicode_AsUTF8String(value, "surrogateescape");
if (!encoded) {
return NULL;
}
size = PyBytes_GET_SIZE(encoded);
if (strlen(PyBytes_AS_STRING(encoded)) != (size_t)size) {
/* The string contains embedded null characters.
* Tcl needs a null character to be represented as \xc0\x80 in
* the Modified UTF-8 encoding. Otherwise the string can be
* truncated in some internal operations.
*
* NOTE: stringlib_replace() could be used here, but optimizing
* this obscure case isn't worth it unless stringlib_replace()
* was already exposed in the C API for other reasons. */
Py_SETREF(encoded,
PyObject_CallMethod(encoded, "replace", "y#y#",
"\0", (Py_ssize_t)1,
"\xc0\x80", (Py_ssize_t)2));
if (!encoded) {
return NULL;
}
size = PyBytes_GET_SIZE(encoded);
}
if (size > INT_MAX) {
Py_DECREF(encoded);
PyErr_SetString(PyExc_OverflowError, "string is too long");
return NULL;
}
result = Tcl_NewStringObj(PyBytes_AS_STRING(encoded), (int)size);
#endif
#endif /* USE_TCL_UNICODE */
Py_DECREF(encoded);
return result;
}