Use Py_ssize_t to count the length.
diff --git a/Parser/tokenizer.c b/Parser/tokenizer.c
index 646a7c1..dee8e84 100644
--- a/Parser/tokenizer.c
+++ b/Parser/tokenizer.c
@@ -370,7 +370,7 @@
 	PyObject* utf8 = NULL;
 	PyObject* buf = tok->decoding_buffer;
 	char *str;
-	int utf8len;
+	Py_ssize_t utf8len;
 
 	/* Ask for one less byte so we can terminate it */
 	assert(size > 0);