[python] Fix grammar non-terminal memory bug

2 files changed, 121 insertions, 90 deletions

diff --git a/python/src/_cdec.cpp b/python/src/_cdec.cpp
index 20b86169..88c03520 100644
--- a/python/src/_cdec.cpp
+++ b/python/src/_cdec.cpp
@@ -1,4 +1,4 @@
-/* Generated by Cython 0.17.beta1 on Fri Jul 27 22:15:28 2012 */
+/* Generated by Cython 0.17.beta1 on Sat Jul 28 20:10:21 2012 */
 
 #define PY_SSIZE_T_CLEAN
 #include "Python.h"
@@ -482,7 +482,7 @@ struct __pyx_obj_5_cdec_Scorer {
  * 
  * cdef class NTRef:             # <<<<<<<<<<<<<<
  *     cdef public unsigned ref
- *     def __init__(self, ref):
+ *     def __init__(self, unsigned ref):
  */
 struct __pyx_obj_5_cdec_NTRef {
   PyObject_HEAD
@@ -718,12 +718,12 @@ struct __pyx_obj_5_cdec___pyx_scope_struct_6_genexpr {
  *     return ' '.join(w.encode('utf8') if isinstance(w, unicode) else str(w) for w in phrase)
  * 
  * cdef class NT:             # <<<<<<<<<<<<<<
- *     cdef public char* cat
+ *     cdef public bytes cat
  *     cdef public unsigned ref
  */
 struct __pyx_obj_5_cdec_NT {
   PyObject_HEAD
-  char *cat;
+  PyObject *cat;
   unsigned int ref;
 };
 
@@ -1784,13 +1784,14 @@ static PyObject *__pyx_pf_5_cdec_12SparseVector_37__div__(PyObject *__pyx_v_x, P
 #endif
 static PyObject *__pyx_pf_5_cdec_7_phrase_genexpr(PyObject *__pyx_self); /* proto */
 static PyObject *__pyx_pf_5_cdec__phrase(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_phrase); /* proto */
-static int __pyx_pf_5_cdec_2NT___init__(struct __pyx_obj_5_cdec_NT *__pyx_v_self, PyObject *__pyx_v_cat, PyObject *__pyx_v_ref); /* proto */
+static int __pyx_pf_5_cdec_2NT___init__(struct __pyx_obj_5_cdec_NT *__pyx_v_self, char *__pyx_v_cat, unsigned int __pyx_v_ref); /* proto */
 static PyObject *__pyx_pf_5_cdec_2NT_2__str__(struct __pyx_obj_5_cdec_NT *__pyx_v_self); /* proto */
 static PyObject *__pyx_pf_5_cdec_2NT_3cat___get__(struct __pyx_obj_5_cdec_NT *__pyx_v_self); /* proto */
 static int __pyx_pf_5_cdec_2NT_3cat_2__set__(struct __pyx_obj_5_cdec_NT *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
+static int __pyx_pf_5_cdec_2NT_3cat_4__del__(struct __pyx_obj_5_cdec_NT *__pyx_v_self); /* proto */
 static PyObject *__pyx_pf_5_cdec_2NT_3ref___get__(struct __pyx_obj_5_cdec_NT *__pyx_v_self); /* proto */
 static int __pyx_pf_5_cdec_2NT_3ref_2__set__(struct __pyx_obj_5_cdec_NT *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
-static int __pyx_pf_5_cdec_5NTRef___init__(struct __pyx_obj_5_cdec_NTRef *__pyx_v_self, PyObject *__pyx_v_ref); /* proto */
+static int __pyx_pf_5_cdec_5NTRef___init__(struct __pyx_obj_5_cdec_NTRef *__pyx_v_self, unsigned int __pyx_v_ref); /* proto */
 static PyObject *__pyx_pf_5_cdec_5NTRef_2__str__(struct __pyx_obj_5_cdec_NTRef *__pyx_v_self); /* proto */
 static PyObject *__pyx_pf_5_cdec_5NTRef_3ref___get__(struct __pyx_obj_5_cdec_NTRef *__pyx_v_self); /* proto */
 static int __pyx_pf_5_cdec_5NTRef_3ref_2__set__(struct __pyx_obj_5_cdec_NTRef *__pyx_v_self, PyObject *__pyx_v_value); /* proto */
@@ -4832,15 +4833,14 @@ static PyObject *__pyx_pf_5_cdec__phrase(CYTHON_UNUSED PyObject *__pyx_self, PyO
 /* Python wrapper */
 static int __pyx_pw_5_cdec_2NT_1__init__(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
 static int __pyx_pw_5_cdec_2NT_1__init__(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
-  PyObject *__pyx_v_cat = 0;
-  PyObject *__pyx_v_ref = 0;
+  char *__pyx_v_cat;
+  unsigned int __pyx_v_ref;
   int __pyx_r;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("__init__ (wrapper)", 0);
   {
     static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__cat,&__pyx_n_s__ref,0};
     PyObject* values[2] = {0,0};
-    values[1] = ((PyObject *)__pyx_int_0);
     if (unlikely(__pyx_kwds)) {
       Py_ssize_t kw_args;
       const Py_ssize_t pos_args = PyTuple_GET_SIZE(__pyx_args);
@@ -4872,8 +4872,12 @@ static int __pyx_pw_5_cdec_2NT_1__init__(PyObject *__pyx_v_self, PyObject *__pyx
         default: goto __pyx_L5_argtuple_error;
       }
     }
-    __pyx_v_cat = values[0];
-    __pyx_v_ref = values[1];
+    __pyx_v_cat = PyBytes_AsString(values[0]); if (unlikely((!__pyx_v_cat) && PyErr_Occurred())) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 11; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    if (values[1]) {
+      __pyx_v_ref = __Pyx_PyInt_AsUnsignedInt(values[1]); if (unlikely((__pyx_v_ref == (unsigned int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 11; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    } else {
+      __pyx_v_ref = ((unsigned int)0);
+    }
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
@@ -4889,18 +4893,17 @@ static int __pyx_pw_5_cdec_2NT_1__init__(PyObject *__pyx_v_self, PyObject *__pyx
 }
 
 /* "/Users/vchahun/Sandbox/cdec/python/src/grammar.pxi":11
- *     cdef public char* cat
+ *     cdef public bytes cat
  *     cdef public unsigned ref
- *     def __init__(self, cat, ref=0):             # <<<<<<<<<<<<<<
+ *     def __init__(self, char* cat, unsigned ref=0):             # <<<<<<<<<<<<<<
  *         self.cat = cat
  *         self.ref = ref
  */
 
-static int __pyx_pf_5_cdec_2NT___init__(struct __pyx_obj_5_cdec_NT *__pyx_v_self, PyObject *__pyx_v_cat, PyObject *__pyx_v_ref) {
+static int __pyx_pf_5_cdec_2NT___init__(struct __pyx_obj_5_cdec_NT *__pyx_v_self, char *__pyx_v_cat, unsigned int __pyx_v_ref) {
   int __pyx_r;
   __Pyx_RefNannyDeclarations
-  char *__pyx_t_1;
-  unsigned int __pyx_t_2;
+  PyObject *__pyx_t_1 = NULL;
   int __pyx_lineno = 0;
   const char *__pyx_filename = NULL;
   int __pyx_clineno = 0;
@@ -4908,27 +4911,32 @@ static int __pyx_pf_5_cdec_2NT___init__(struct __pyx_obj_5_cdec_NT *__pyx_v_self
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/grammar.pxi":12
  *     cdef public unsigned ref
- *     def __init__(self, cat, ref=0):
+ *     def __init__(self, char* cat, unsigned ref=0):
  *         self.cat = cat             # <<<<<<<<<<<<<<
  *         self.ref = ref
  * 
  */
-  __pyx_t_1 = PyBytes_AsString(__pyx_v_cat); if (unlikely((!__pyx_t_1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 12; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyBytes_FromString(__pyx_v_cat); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 12; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_t_1));
+  __Pyx_GIVEREF(((PyObject *)__pyx_t_1));
+  __Pyx_GOTREF(__pyx_v_self->cat);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->cat));
   __pyx_v_self->cat = __pyx_t_1;
+  __pyx_t_1 = 0;
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/grammar.pxi":13
- *     def __init__(self, cat, ref=0):
+ *     def __init__(self, char* cat, unsigned ref=0):
  *         self.cat = cat
  *         self.ref = ref             # <<<<<<<<<<<<<<
  * 
  *     def __str__(self):
  */
-  __pyx_t_2 = __Pyx_PyInt_AsUnsignedInt(__pyx_v_ref); if (unlikely((__pyx_t_2 == (unsigned int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 13; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_v_self->ref = __pyx_t_2;
+  __pyx_v_self->ref = __pyx_v_ref;
 
   __pyx_r = 0;
   goto __pyx_L0;
   __pyx_L1_error:;
+  __Pyx_XDECREF(__pyx_t_1);
   __Pyx_AddTraceback("_cdec.NT.__init__", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = -1;
   __pyx_L0:;
@@ -4961,7 +4969,6 @@ static PyObject *__pyx_pf_5_cdec_2NT_2__str__(struct __pyx_obj_5_cdec_NT *__pyx_
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
-  PyObject *__pyx_t_4 = NULL;
   int __pyx_lineno = 0;
   const char *__pyx_filename = NULL;
   int __pyx_clineno = 0;
@@ -4985,23 +4992,21 @@ static PyObject *__pyx_pf_5_cdec_2NT_2__str__(struct __pyx_obj_5_cdec_NT *__pyx_
  * 
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = PyBytes_FromString(__pyx_v_self->cat); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 17; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_GOTREF(((PyObject *)__pyx_t_2));
-    __pyx_t_3 = PyLong_FromUnsignedLong(__pyx_v_self->ref); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 17; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyLong_FromUnsignedLong(__pyx_v_self->ref); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 17; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_2);
+    __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 17; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 17; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_GOTREF(__pyx_t_4);
-    PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_t_2));
-    __Pyx_GIVEREF(((PyObject *)__pyx_t_2));
-    PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_3);
-    __Pyx_GIVEREF(__pyx_t_3);
+    __Pyx_INCREF(((PyObject *)__pyx_v_self->cat));
+    PyTuple_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_v_self->cat));
+    __Pyx_GIVEREF(((PyObject *)__pyx_v_self->cat));
+    PyTuple_SET_ITEM(__pyx_t_3, 1, __pyx_t_2);
+    __Pyx_GIVEREF(__pyx_t_2);
+    __pyx_t_2 = 0;
+    __pyx_t_2 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_8), ((PyObject *)__pyx_t_3)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 17; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(((PyObject *)__pyx_t_2));
+    __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
+    __pyx_r = ((PyObject *)__pyx_t_2);
     __pyx_t_2 = 0;
-    __pyx_t_3 = 0;
-    __pyx_t_3 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_8), ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 17; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_GOTREF(((PyObject *)__pyx_t_3));
-    __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-    __pyx_r = ((PyObject *)__pyx_t_3);
-    __pyx_t_3 = 0;
     goto __pyx_L0;
     goto __pyx_L3;
   }
@@ -5015,13 +5020,10 @@ static PyObject *__pyx_pf_5_cdec_2NT_2__str__(struct __pyx_obj_5_cdec_NT *__pyx_
  * cdef class NTRef:
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_3 = PyBytes_FromString(__pyx_v_self->cat); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 18; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_t_3));
-  __pyx_t_4 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_9), ((PyObject *)__pyx_t_3)); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 18; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_t_4));
-  __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_r = ((PyObject *)__pyx_t_4);
-  __pyx_t_4 = 0;
+  __pyx_t_2 = PyNumber_Remainder(((PyObject *)__pyx_kp_s_9), ((PyObject *)__pyx_v_self->cat)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 18; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_t_2));
+  __pyx_r = ((PyObject *)__pyx_t_2);
+  __pyx_t_2 = 0;
   goto __pyx_L0;
 
   __pyx_r = Py_None; __Pyx_INCREF(Py_None);
@@ -5029,7 +5031,6 @@ static PyObject *__pyx_pf_5_cdec_2NT_2__str__(struct __pyx_obj_5_cdec_NT *__pyx_
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_XDECREF(__pyx_t_4);
   __Pyx_AddTraceback("_cdec.NT.__str__", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
@@ -5052,32 +5053,21 @@ static PyObject *__pyx_pw_5_cdec_2NT_3cat_1__get__(PyObject *__pyx_v_self) {
 /* "/Users/vchahun/Sandbox/cdec/python/src/grammar.pxi":9
  * 
  * cdef class NT:
- *     cdef public char* cat             # <<<<<<<<<<<<<<
+ *     cdef public bytes cat             # <<<<<<<<<<<<<<
  *     cdef public unsigned ref
- *     def __init__(self, cat, ref=0):
+ *     def __init__(self, char* cat, unsigned ref=0):
  */
 
 static PyObject *__pyx_pf_5_cdec_2NT_3cat___get__(struct __pyx_obj_5_cdec_NT *__pyx_v_self) {
   PyObject *__pyx_r = NULL;
   __Pyx_RefNannyDeclarations
-  PyObject *__pyx_t_1 = NULL;
-  int __pyx_lineno = 0;
-  const char *__pyx_filename = NULL;
-  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__get__", 0);
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = PyBytes_FromString(__pyx_v_self->cat); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 9; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_t_1));
-  __pyx_r = ((PyObject *)__pyx_t_1);
-  __pyx_t_1 = 0;
+  __Pyx_INCREF(((PyObject *)__pyx_v_self->cat));
+  __pyx_r = ((PyObject *)__pyx_v_self->cat);
   goto __pyx_L0;
 
   __pyx_r = Py_None; __Pyx_INCREF(Py_None);
-  goto __pyx_L0;
-  __pyx_L1_error:;
-  __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_AddTraceback("_cdec.NT.cat.__get__", __pyx_clineno, __pyx_lineno, __pyx_filename);
-  __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
@@ -5098,13 +5088,16 @@ static int __pyx_pw_5_cdec_2NT_3cat_3__set__(PyObject *__pyx_v_self, PyObject *_
 static int __pyx_pf_5_cdec_2NT_3cat_2__set__(struct __pyx_obj_5_cdec_NT *__pyx_v_self, PyObject *__pyx_v_value) {
   int __pyx_r;
   __Pyx_RefNannyDeclarations
-  char *__pyx_t_1;
   int __pyx_lineno = 0;
   const char *__pyx_filename = NULL;
   int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__set__", 0);
-  __pyx_t_1 = PyBytes_AsString(__pyx_v_value); if (unlikely((!__pyx_t_1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 9; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_v_self->cat = __pyx_t_1;
+  if (!(likely(PyBytes_CheckExact(__pyx_v_value))||((__pyx_v_value) == Py_None)||(PyErr_Format(PyExc_TypeError, "Expected bytes, got %.200s", Py_TYPE(__pyx_v_value)->tp_name), 0))) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 9; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_INCREF(__pyx_v_value);
+  __Pyx_GIVEREF(__pyx_v_value);
+  __Pyx_GOTREF(__pyx_v_self->cat);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->cat));
+  __pyx_v_self->cat = ((PyObject*)__pyx_v_value);
 
   __pyx_r = 0;
   goto __pyx_L0;
@@ -5117,6 +5110,32 @@ static int __pyx_pf_5_cdec_2NT_3cat_2__set__(struct __pyx_obj_5_cdec_NT *__pyx_v
 }
 
 /* Python wrapper */
+static int __pyx_pw_5_cdec_2NT_3cat_5__del__(PyObject *__pyx_v_self); /*proto*/
+static int __pyx_pw_5_cdec_2NT_3cat_5__del__(PyObject *__pyx_v_self) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__del__ (wrapper)", 0);
+  __pyx_r = __pyx_pf_5_cdec_2NT_3cat_4__del__(((struct __pyx_obj_5_cdec_NT *)__pyx_v_self));
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+static int __pyx_pf_5_cdec_2NT_3cat_4__del__(struct __pyx_obj_5_cdec_NT *__pyx_v_self) {
+  int __pyx_r;
+  __Pyx_RefNannyDeclarations
+  __Pyx_RefNannySetupContext("__del__", 0);
+  __Pyx_INCREF(Py_None);
+  __Pyx_GIVEREF(Py_None);
+  __Pyx_GOTREF(__pyx_v_self->cat);
+  __Pyx_DECREF(((PyObject *)__pyx_v_self->cat));
+  __pyx_v_self->cat = ((PyObject*)Py_None);
+
+  __pyx_r = 0;
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* Python wrapper */
 static PyObject *__pyx_pw_5_cdec_2NT_3ref_1__get__(PyObject *__pyx_v_self); /*proto*/
 static PyObject *__pyx_pw_5_cdec_2NT_3ref_1__get__(PyObject *__pyx_v_self) {
   PyObject *__pyx_r = 0;
@@ -5129,9 +5148,9 @@ static PyObject *__pyx_pw_5_cdec_2NT_3ref_1__get__(PyObject *__pyx_v_self) {
 
 /* "/Users/vchahun/Sandbox/cdec/python/src/grammar.pxi":10
  * cdef class NT:
- *     cdef public char* cat
+ *     cdef public bytes cat
  *     cdef public unsigned ref             # <<<<<<<<<<<<<<
- *     def __init__(self, cat, ref=0):
+ *     def __init__(self, char* cat, unsigned ref=0):
  *         self.cat = cat
  */
 
@@ -5197,7 +5216,7 @@ static int __pyx_pf_5_cdec_2NT_3ref_2__set__(struct __pyx_obj_5_cdec_NT *__pyx_v
 /* Python wrapper */
 static int __pyx_pw_5_cdec_5NTRef_1__init__(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
 static int __pyx_pw_5_cdec_5NTRef_1__init__(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
-  PyObject *__pyx_v_ref = 0;
+  unsigned int __pyx_v_ref;
   int __pyx_r;
   __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("__init__ (wrapper)", 0);
@@ -5226,7 +5245,7 @@ static int __pyx_pw_5_cdec_5NTRef_1__init__(PyObject *__pyx_v_self, PyObject *__
     } else {
       values[0] = PyTuple_GET_ITEM(__pyx_args, 0);
     }
-    __pyx_v_ref = values[0];
+    __pyx_v_ref = __Pyx_PyInt_AsUnsignedInt(values[0]); if (unlikely((__pyx_v_ref == (unsigned int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 22; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
@@ -5244,36 +5263,26 @@ static int __pyx_pw_5_cdec_5NTRef_1__init__(PyObject *__pyx_v_self, PyObject *__
 /* "/Users/vchahun/Sandbox/cdec/python/src/grammar.pxi":22
  * cdef class NTRef:
  *     cdef public unsigned ref
- *     def __init__(self, ref):             # <<<<<<<<<<<<<<
+ *     def __init__(self, unsigned ref):             # <<<<<<<<<<<<<<
  *         self.ref = ref
  * 
  */
 
-static int __pyx_pf_5_cdec_5NTRef___init__(struct __pyx_obj_5_cdec_NTRef *__pyx_v_self, PyObject *__pyx_v_ref) {
+static int __pyx_pf_5_cdec_5NTRef___init__(struct __pyx_obj_5_cdec_NTRef *__pyx_v_self, unsigned int __pyx_v_ref) {
   int __pyx_r;
   __Pyx_RefNannyDeclarations
-  unsigned int __pyx_t_1;
-  int __pyx_lineno = 0;
-  const char *__pyx_filename = NULL;
-  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__init__", 0);
 
   /* "/Users/vchahun/Sandbox/cdec/python/src/grammar.pxi":23
  *     cdef public unsigned ref
- *     def __init__(self, ref):
+ *     def __init__(self, unsigned ref):
  *         self.ref = ref             # <<<<<<<<<<<<<<
  * 
  *     def __str__(self):
  */
-  __pyx_t_1 = __Pyx_PyInt_AsUnsignedInt(__pyx_v_ref); if (unlikely((__pyx_t_1 == (unsigned int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[2]; __pyx_lineno = 23; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_v_self->ref = __pyx_t_1;
+  __pyx_v_self->ref = __pyx_v_ref;
 
   __pyx_r = 0;
-  goto __pyx_L0;
-  __pyx_L1_error:;
-  __Pyx_AddTraceback("_cdec.NTRef.__init__", __pyx_clineno, __pyx_lineno, __pyx_filename);
-  __pyx_r = -1;
-  __pyx_L0:;
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
@@ -5352,7 +5361,7 @@ static PyObject *__pyx_pw_5_cdec_5NTRef_3ref_1__get__(PyObject *__pyx_v_self) {
  * 
  * cdef class NTRef:
  *     cdef public unsigned ref             # <<<<<<<<<<<<<<
- *     def __init__(self, ref):
+ *     def __init__(self, unsigned ref):
  *         self.ref = ref
  */
 
@@ -18748,15 +18757,38 @@ static PyTypeObject __pyx_type_5_cdec_SparseVector = {
 };
 
 static PyObject *__pyx_tp_new_5_cdec_NT(PyTypeObject *t, CYTHON_UNUSED PyObject *a, CYTHON_UNUSED PyObject *k) {
+  struct __pyx_obj_5_cdec_NT *p;
   PyObject *o = (*t->tp_alloc)(t, 0);
   if (!o) return 0;
+  p = ((struct __pyx_obj_5_cdec_NT *)o);
+  p->cat = ((PyObject*)Py_None); Py_INCREF(Py_None);
   return o;
 }
 
 static void __pyx_tp_dealloc_5_cdec_NT(PyObject *o) {
+  struct __pyx_obj_5_cdec_NT *p = (struct __pyx_obj_5_cdec_NT *)o;
+  Py_XDECREF(((PyObject *)p->cat));
   (*Py_TYPE(o)->tp_free)(o);
 }
 
+static int __pyx_tp_traverse_5_cdec_NT(PyObject *o, visitproc v, void *a) {
+  int e;
+  struct __pyx_obj_5_cdec_NT *p = (struct __pyx_obj_5_cdec_NT *)o;
+  if (p->cat) {
+    e = (*v)(p->cat, a); if (e) return e;
+  }
+  return 0;
+}
+
+static int __pyx_tp_clear_5_cdec_NT(PyObject *o) {
+  struct __pyx_obj_5_cdec_NT *p = (struct __pyx_obj_5_cdec_NT *)o;
+  PyObject* tmp;
+  tmp = ((PyObject*)p->cat);
+  p->cat = ((PyObject*)Py_None); Py_INCREF(Py_None);
+  Py_XDECREF(tmp);
+  return 0;
+}
+
 static PyObject *__pyx_getprop_5_cdec_2NT_cat(PyObject *o, CYTHON_UNUSED void *x) {
   return __pyx_pw_5_cdec_2NT_3cat_1__get__(o);
 }
@@ -18766,8 +18798,7 @@ static int __pyx_setprop_5_cdec_2NT_cat(PyObject *o, PyObject *v, CYTHON_UNUSED
     return __pyx_pw_5_cdec_2NT_3cat_3__set__(o, v);
   }
   else {
-    PyErr_SetString(PyExc_NotImplementedError, "__del__");
-    return -1;
+    return __pyx_pw_5_cdec_2NT_3cat_5__del__(o);
   }
 }
 
@@ -18917,10 +18948,10 @@ static PyTypeObject __pyx_type_5_cdec_NT = {
   0, /*tp_getattro*/
   0, /*tp_setattro*/
   &__pyx_tp_as_buffer_NT, /*tp_as_buffer*/
-  Py_TPFLAGS_DEFAULT|Py_TPFLAGS_CHECKTYPES|Py_TPFLAGS_HAVE_NEWBUFFER|Py_TPFLAGS_BASETYPE, /*tp_flags*/
+  Py_TPFLAGS_DEFAULT|Py_TPFLAGS_CHECKTYPES|Py_TPFLAGS_HAVE_NEWBUFFER|Py_TPFLAGS_BASETYPE|Py_TPFLAGS_HAVE_GC, /*tp_flags*/
   0, /*tp_doc*/
-  0, /*tp_traverse*/
-  0, /*tp_clear*/
+  __pyx_tp_traverse_5_cdec_NT, /*tp_traverse*/
+  __pyx_tp_clear_5_cdec_NT, /*tp_clear*/
   0, /*tp_richcompare*/
   0, /*tp_weaklistoffset*/
   0, /*tp_iter*/
diff --git a/python/src/grammar.pxi b/python/src/grammar.pxi
index 5ec21422..59266238 100644
--- a/python/src/grammar.pxi
+++ b/python/src/grammar.pxi
@@ -6,9 +6,9 @@ def _phrase(phrase):
     return ' '.join(w.encode('utf8') if isinstance(w, unicode) else str(w) for w in phrase)
 
 cdef class NT:
-    cdef public char* cat
+    cdef public bytes cat
     cdef public unsigned ref
-    def __init__(self, cat, ref=0):
+    def __init__(self, char* cat, unsigned ref=0):
         self.cat = cat
         self.ref = ref
 
@@ -19,7 +19,7 @@ cdef class NT:
 
 cdef class NTRef:
     cdef public unsigned ref
-    def __init__(self, ref):
+    def __init__(self, unsigned ref):
         self.ref = ref
 
     def __str__(self):