Issue #17106: Fix a segmentation fault in io.TextIOWrapper when an underlying
stream or a decoder produces data of an unexpected type (i.e. when
io.TextIOWrapper initialized with text stream or use bytes-to-bytes codec).
diff --git a/Modules/_io/textio.c b/Modules/_io/textio.c
index c9913d4..cd6d443 100644
--- a/Modules/_io/textio.c
+++ b/Modules/_io/textio.c
@@ -236,6 +236,21 @@
Py_TYPE(self)->tp_free((PyObject *)self);
}
+static int
+check_decoded(PyObject *decoded)
+{
+ if (decoded == NULL)
+ return -1;
+ if (!PyUnicode_Check(decoded)) {
+ PyErr_Format(PyExc_TypeError,
+ "decoder should return a string result, not '%.200s'",
+ Py_TYPE(decoded)->tp_name);
+ Py_DECREF(decoded);
+ return -1;
+ }
+ return 0;
+}
+
#define SEEN_CR 1
#define SEEN_LF 2
#define SEEN_CRLF 4
@@ -265,15 +280,9 @@
Py_INCREF(output);
}
- if (output == NULL)
+ if (check_decoded(output) < 0)
return NULL;
- if (!PyUnicode_Check(output)) {
- PyErr_SetString(PyExc_TypeError,
- "decoder should return a string result");
- goto error;
- }
-
output_len = PyUnicode_GET_SIZE(output);
if (self->pendingcr && (final || output_len > 0)) {
Py_UNICODE *out;
@@ -1417,7 +1426,12 @@
Py_DECREF(chunk_size);
if (input_chunk == NULL)
goto fail;
- assert(PyBytes_Check(input_chunk));
+ if (!PyBytes_Check(input_chunk)) {
+ PyErr_Format(PyExc_TypeError,
+ "underlying read1() should have returned a bytes object, "
+ "not '%.200s'", Py_TYPE(input_chunk)->tp_name);
+ goto fail;
+ }
eof = (PyBytes_Size(input_chunk) == 0);
@@ -1430,8 +1444,7 @@
_PyIO_str_decode, input_chunk, eof ? Py_True : Py_False, NULL);
}
- /* TODO sanity check: isinstance(decoded_chars, unicode) */
- if (decoded_chars == NULL)
+ if (check_decoded(decoded_chars) < 0)
goto fail;
textiowrapper_set_decoded_chars(self, decoded_chars);
if (PyUnicode_GET_SIZE(decoded_chars) > 0)
@@ -1444,7 +1457,14 @@
PyObject *next_input = PyNumber_Add(dec_buffer, input_chunk);
if (next_input == NULL)
goto fail;
- assert (PyBytes_Check(next_input));
+ if (!PyBytes_Check(next_input)) {
+ PyErr_Format(PyExc_TypeError,
+ "decoder getstate() should have returned a bytes "
+ "object, not '%.200s'",
+ Py_TYPE(next_input)->tp_name);
+ Py_DECREF(next_input);
+ goto fail;
+ }
Py_DECREF(dec_buffer);
Py_CLEAR(self->snapshot);
self->snapshot = Py_BuildValue("NN", dec_flags, next_input);
@@ -1490,7 +1510,7 @@
decoded = PyObject_CallMethodObjArgs(self->decoder, _PyIO_str_decode,
bytes, Py_True, NULL);
Py_DECREF(bytes);
- if (decoded == NULL)
+ if (check_decoded(decoded) < 0)
goto fail;
result = textiowrapper_get_decoded_chars(self, -1);
@@ -2110,7 +2130,14 @@
if (input_chunk == NULL)
goto fail;
- assert (PyBytes_Check(input_chunk));
+ if (!PyBytes_Check(input_chunk)) {
+ PyErr_Format(PyExc_TypeError,
+ "underlying read() should have returned a bytes "
+ "object, not '%.200s'",
+ Py_TYPE(input_chunk)->tp_name);
+ Py_DECREF(input_chunk);
+ goto fail;
+ }
self->snapshot = Py_BuildValue("iN", cookie.dec_flags, input_chunk);
if (self->snapshot == NULL) {
@@ -2121,7 +2148,7 @@
decoded = PyObject_CallMethod(self->decoder, "decode",
"Oi", input_chunk, (int)cookie.need_eof);
- if (decoded == NULL)
+ if (check_decoded(decoded) < 0)
goto fail;
textiowrapper_set_decoded_chars(self, decoded);
@@ -2245,9 +2272,8 @@
PyObject *decoded = PyObject_CallMethod(
self->decoder, "decode", "s#", input, 1);
- if (decoded == NULL)
+ if (check_decoded(decoded) < 0)
goto fail;
- assert (PyUnicode_Check(decoded));
chars_decoded += PyUnicode_GET_SIZE(decoded);
Py_DECREF(decoded);
@@ -2279,9 +2305,8 @@
/* We didn't get enough decoded data; signal EOF to get more. */
PyObject *decoded = PyObject_CallMethod(
self->decoder, "decode", "si", "", /* final = */ 1);
- if (decoded == NULL)
+ if (check_decoded(decoded) < 0)
goto fail;
- assert (PyUnicode_Check(decoded));
chars_decoded += PyUnicode_GET_SIZE(decoded);
Py_DECREF(decoded);
cookie.need_eof = 1;
@@ -2440,7 +2465,7 @@
Py_DECREF(res);
if (r < 0)
return NULL;
-
+
if (r > 0) {
Py_RETURN_NONE; /* stream already closed */
}