bpo-35372: Fix the code page decoder for input > 2 GiB. (GH-10848)
diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index bc98c44..1351eec 100644
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -7211,7 +7211,7 @@
"in the target code page.";
/* each step cannot decode more than 1 character, but a character can be
represented as a surrogate pair */
- wchar_t buffer[2], *startout, *out;
+ wchar_t buffer[2], *out;
int insize;
Py_ssize_t outsize;
PyObject *errorHandler = NULL;
@@ -7248,7 +7248,7 @@
*v = (PyObject*)_PyUnicode_New(size * Py_ARRAY_LENGTH(buffer));
if (*v == NULL)
goto error;
- startout = PyUnicode_AS_UNICODE(*v);
+ out = PyUnicode_AS_UNICODE(*v);
}
else {
/* Extend unicode object */
@@ -7259,11 +7259,10 @@
}
if (unicode_resize(v, n + size * Py_ARRAY_LENGTH(buffer)) < 0)
goto error;
- startout = PyUnicode_AS_UNICODE(*v) + n;
+ out = PyUnicode_AS_UNICODE(*v) + n;
}
/* Decode the byte string character per character */
- out = startout;
while (in < endin)
{
/* Decode a character */
@@ -7318,7 +7317,7 @@
*out = 0;
/* Extend unicode object */
- outsize = out - startout;
+ outsize = out - PyUnicode_AS_UNICODE(*v);
assert(outsize <= PyUnicode_WSTR_LENGTH(*v));
if (unicode_resize(v, outsize) < 0)
goto error;