filterstring() and filterunicode() in Python/bltinmodule.c

blindly assumed that tp_as_sequence->sq_item always returns
a str or unicode object. This might fail with str or unicode
subclasses.

This patch checks whether the object returned from __getitem__
is a str/unicode object and raises a TypeError if not (and
the filter function returned true).

Furthermore the result for __getitem__ can be more than one
character long, so checks for enough memory have to be done.

Lib/test/test_builtin.py

@@ -367,6 +367,16 @@ class BuiltinTest(unittest.TestCase):
                 raise ValueError
         self.assertRaises(ValueError, filter, lambda x: x >="3", badstr("1234"))
 
+        class badstr2(str):
+            def __getitem__(self, index):
+                return 42
+        self.assertRaises(TypeError, filter, lambda x: x >=42, badstr2("1234"))
+
+        class weirdstr(str):
+            def __getitem__(self, index):
+                return weirdstr(2*str.__getitem__(self, index))
+        self.assertEqual(filter(lambda x: x>="33", weirdstr("1234")), "3344")
+
         if have_unicode:
             # test bltinmodule.c::filterunicode()
             self.assertEqual(filter(None, unicode("12")), unicode("12"))
@@ -374,6 +384,17 @@ class BuiltinTest(unittest.TestCase):
             self.assertRaises(TypeError, filter, 42, unicode("12"))
             self.assertRaises(ValueError, filter, lambda x: x >="3", badstr(unicode("1234")))
 
+            class badunicode(unicode):
+                def __getitem__(self, index):
+                    return 42
+            self.assertRaises(TypeError, filter, lambda x: x >=42, badunicode("1234"))
+
+            class weirdunicode(unicode):
+                def __getitem__(self, index):
+                    return weirdunicode(2*unicode.__getitem__(self, index))
+            self.assertEqual(
+                filter(lambda x: x>=unicode("33"), weirdunicode("1234")), unicode("3344"))
+
     def test_float(self):
         self.assertEqual(float(3.14), 3.14)
         self.assertEqual(float(314), 314.0)

Python/bltinmodule.c

@@ -1892,6 +1892,7 @@ filterstring(PyObject *func, PyObject *strobj)
 	PyObject *result;
 	register int i, j;
 	int len = PyString_Size(strobj);
+	int outlen = len;
 
 	if (func == Py_None) {
 		/* No character is ever false -- share input string */
@@ -1921,13 +1922,43 @@ filterstring(PyObject *func, PyObject *strobj)
 		}
 		ok = PyObject_IsTrue(good);
 		Py_DECREF(good);
-		if (ok)
-			PyString_AS_STRING((PyStringObject *)result)[j++] =
-				PyString_AS_STRING((PyStringObject *)item)[0];
+		if (ok) {
+			int reslen;
+			if (!PyString_Check(item)) {
+				PyErr_SetString(PyExc_TypeError, "can't filter str to str:"
+					" __getitem__ returned different type");
+				Py_DECREF(item);
+				goto Fail_1;
+			}
+			reslen = PyString_GET_SIZE(item);
+			if (reslen == 1) {
+				PyString_AS_STRING(result)[j++] =
+					PyString_AS_STRING(item)[0];
+			} else {
+				/* do we need more space? */
+				int need = j + reslen + len-i-1;
+				if (need > outlen) {
+					/* overallocate, to avoid reallocations */
+					if (need<2*outlen)
+						need = 2*outlen;
+					if (_PyString_Resize(&result, need)) {
+						Py_DECREF(item);
+						return NULL;
+					}
+					outlen = need;
+				}
+				memcpy(
+					PyString_AS_STRING(result) + j,
+					PyString_AS_STRING(item),
+					reslen
+				);
+				j += reslen;
+			}
+		}
 		Py_DECREF(item);
 	}
 
-	if (j < len)
+	if (j < outlen)
 		_PyString_Resize(&result, j);
 
 	return result;
@@ -1946,6 +1977,7 @@ filterunicode(PyObject *func, PyObject *strobj)
 	PyObject *result;
 	register int i, j;
 	int len = PyUnicode_GetSize(strobj);
+	int outlen = len;
 
 	if (func == Py_None) {
 		/* No character is ever false -- share input string */
@@ -1975,13 +2007,43 @@ filterunicode(PyObject *func, PyObject *strobj)
 		}
 		ok = PyObject_IsTrue(good);
 		Py_DECREF(good);
-		if (ok)
-			PyUnicode_AS_UNICODE((PyStringObject *)result)[j++] =
-				PyUnicode_AS_UNICODE((PyStringObject *)item)[0];
+		if (ok) {
+			int reslen;
+			if (!PyUnicode_Check(item)) {
+				PyErr_SetString(PyExc_TypeError, "can't filter unicode to unicode:"
+					" __getitem__ returned different type");
+				Py_DECREF(item);
+				goto Fail_1;
+			}
+			reslen = PyUnicode_GET_SIZE(item);
+			if (reslen == 1) {
+				PyUnicode_AS_UNICODE(result)[j++] =
+					PyUnicode_AS_UNICODE(item)[0];
+			} else {
+				/* do we need more space? */
+				int need = j + reslen + len-i-1;
+				if (need > outlen) {
+					/* overallocate, to avoid reallocations */
+					if (need<2*outlen)
+						need = 2*outlen;
+					if (PyUnicode_Resize(&result, need)) {
+						Py_DECREF(item);
+						return NULL;
+					}
+					outlen = need;
+				}
+				memcpy(
+					PyUnicode_AS_UNICODE(result) + j,
+					PyUnicode_AS_UNICODE(item),
+					reslen*sizeof(Py_UNICODE)
+				);
+				j += reslen;
+			}
+		}
 		Py_DECREF(item);
 	}
 
-	if (j < len)
+	if (j < outlen)
 		PyUnicode_Resize(&result, j);
 
 	return result;